Ader 1985

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8, 379-394
Printed in the United States of America
CNS-immune system interactions:

Conditioning phenomena
Robert Ader
Division of Behavioral and Psychosocial Medicine, Department of
Psychiatry, University of Rochester School of Medicine and Dentistry,
Rochester, N.Y. 14642
Nicholas Cohen
Division of Immunology, Department of Microbiology, University of
Rochester School of Medicine and Dentistry, Rochester, N.Y. 14642
Abstract: Converging data from different disciplines indicate that central nervous system processes are capable of influencing
immune responses. This paper concentrates on recent studies documenting behaviorally conditioned suppression and enhancement
of immunity. Exposing rats or mice to a conditioned stimulus previously paired with an immunomodulating agent results in
alterations in humoral and cell-mediated immune responses to antigenic stimuli, and unreinforced reexposures to the conditioned
stimuli result in extinction of the conditioned response. Although the magnitude of such conditioning effects has not been large, the
phenomenon has been independently verified under a variety of experimental conditions. The biological impact of conditioned
alterations in immune function is illustrated by studies in which conditioning operations were applied in the pharmacotherapy of
autoimmune disease in New Zealand mice. In conditioned animals, substituting conditioned stimuli for active drugs delays the onset
of autoimmune disease relative to nonconditioned animals using a dose of immunosuppressive drug that, by itself, is ineffective in
modifying the progression of disease. The hypothesis that such conditioning effects are mediated by elevations in adrenocortical
steroid levels receives no support from available data. Despite its capacity for self-regulation, it appears that the immune system is
integrated with other psychophysiological processes and subject to modulation by the brain.
Keywords: adrenocortical steroids; autoimmune disease; cell-mediated immunity; conditioning; cyclophosphamide; humoral
immunity; immunity; immunosuppression; psychosomatics; stress; taste aversion learning
The purpose of this paper is to direct attention to an function-and behavior-psychoneuroimmunology may

emerging field of interdisciplinary research concerned be defined more generally and simply as the study of the
with integrated central nervous system and immune relationships between the brain and the immune system.
system function. Until recently, all of the effort devoted It has been known for some time that lesions and
to understanding the regulation of immune responses has electrical stimulation of the hypothalamus alter humoral
been based on the assumption that such regulation is and cell-mediated immune responses (e.g., Cross, Mark-
achieved by interacting components of the immune sys- esberry, Brooks & Roszman 1980; Dann, Wachtel &
tem itself. In vitro studies of interactions among sub- Rubin 1979; Jankovic & Isakovic 1973; Korneva & Khai
populations of lymphocytes have supported this assump- 1963; Roszman, Cross, Brooks, & Markesberry 1982;
tion and led to the concept that the immune system is Stein, Schleifer & Keller 1981). Notonlydohypothalamic
autonomous. This concept of autonomy, however, ig- interventions influence immunological reactivity and,
nores the fact that in vivo immunoregulatory processes perhaps, reproduce responses originally elicited by anti-
occur within a neuroendocrine environment that is congenic stimulation (Benetato & Baciu, cited by Spector
stantly changing in response to the demands of the 1980), but elicitation of an immune response influences
external world. Although no one can question that the hypothalamic activity. Besedovsky, Sorkin, Felix, and
immune system is indeed capable of considerable self- Haas (1977), for example, have recorded an increase in
regulation, one must also consider converging evidence the firing rate of neurons in the ventromedial hypothala-
from immunology and from the behavioral and brain mus that corresponded to the time of peak antibody titer
sciences of the significant, if not critical, role for the in response to different antigenic stimuli.
nervous system in the regulation or modulation of immu- The central role of the hypothalamus in regulating
nity (Ader 1981). The term "psychoneuroimmunology" neuroendocrine function and autonomic nervous system
(Ader 1980; 1981d) refers to the study of the neuroen- activity suggests that hormones and neurotransmitter
docrine mediation of the effects of behavior in modifying substances play a role in the modulation of immunological
immune function. Since it is now appreciated that im- reactivity. This possibility is supported by the recent
mune responses also influence neural and endocrine documentation of sympathetic and parasympathetic in-
© 1985 Cambridge University Press OUO-525XI85IO3O379-16I$O6.OO 379

Ader & Cohen: CNS-iminune system interactions
nervation of lymphoid tissue (e.g., Bulloch & Moore 1983; Persinger, Carrey, Lafreniere & Mazzuchin 1978).
1980; 1981; Felten, Overhage, Felten & Schmedtje 1981; Inescapable electric shock stimulation, for example, has
Giron, Crutcher & Davis 1980; Williams, Peterson, been reported to suppress iminunological reactivity
Shea, Schmedtje, Bauer & Felten 1981). The notion that (Laudenslager et al. 1983); escapable shock does not have
endocrine and neurochemical signals provide relevant this effect. The inescapable footshock used in this experi-
information to the immune system is further reinforced ment produces an analgesia to subsequent footshock that
by the finding that lymphocytes and macrophages bear is presumed to be mediated by the endogenous release of
receptors for a variety of hormones and neurotransinitters opioid peptides. The suppression of immunological reac-
(Abraham & Bugg 1976; Arrenbrecht 1974; Cake & Lit- tivity may therefore have been mediated by a central or
wak 1975; Gillette & Gillette 1979; Hadden, Hadden & peripheral release of opioids that did not occur in animals
Middleton 1970; Hilderman & Strom 1978; Hollenberg& exposed to escapable shock. Shavit, Lewis, Terman,
Cuatrecasas 1974; Roszkowski, Plaut & Lichtenstein Gale, and Liebeskind (1984) subjected rats to an intermit-
1977; Singh, Milson, Smith & Owen 1979; Strom, Syt- tent or continuous regimen of inescapable shock, both of
kowsky, Carpenter & Merrill 1974). Not only do varia- which result in a subsequent analgesic response. Only the
tions in hormonal state and level of neurotransmitter intermittent shock, however, was associated with the
activity influence immune responses (e.g., Berczi, Nagy, inferred release of opioids, and only the intermittent
Kovacs & Horvath 1981; Coinsa, Leonhardt & Wekerle shock suppressed the cytotoxic activity of natural killer
1982; Hall & Goldstein 1981; Nagy & Berczi 1978), but cells.' How opioid peptides might be exerting effects on
antigenic stimulation (or, perhaps, the immune response different aspects of immune function remains to be
to antigenic stimulation) results in neuroendocrine determined.
changes (e.g., Besedovsky, del Rey & Sorkin 1983; Be- Parametric analyses of the effects of stress and the
sedovsky, del Rey, Sorkin, Da Prada, Burri & Honegger capacity to cope with such environmental circumstances
1983; Besedovsky & Sorkin 1981; Shek & Sabiston on immunocompetence need to be undertaken before we
1983). proceed to examine the relationship between "stress-
Behavioral evidence suggesting that there is a link induced" alterations in immunological reactivity and the
between the brain and the immune system actually ante- onset and course of experimentally induced infectious,
dates the mounting data indicating that there are multiple autoimmune, and neoplastic disease processes. It is not
pathways by which the brain and the immune system the purpose of this review to provide an extensive, critical
might be monitoring and communicating with each evaluation of the effects of stress on immune function.
other. For example, there is a large literature on the The available literature may be summarized as indicating
relationship between personality factors, ("stressful") life that such effects depend upon: (a) the quality and quantity
events, and experimentally induced or spontaneously (intensity, frequency, and duration) of stressful stimula-
occurring disease processes in man (e.g., Plaut & Fried- tion and the availability of means for coping with the
man 1981; Solomon 1981a; 1981b). Also, there are re- environmental demands; (b) the quality and quantity of
views covering the effects of psychosocial variables and immunogenic stimulation; (c) the temporal relationship
"stress" on measures of immune function perse in animal between stressful stimulation and immunogenic stimula-
subjects (e.g., Ader & Cohen 1984; Kelley 1980; Monjan tion; (d) the parameters of immunological reactivity and
1981; Solomon & Amkraut 1981). Interest in the effects of the time(s) at which measurements are made; (e) the
stress on immune function, however, has grown rapidly, social (e.g., housing) and environmental (e.g., tem-
and although more sophisticated research has been pub- perature, time of day) conditions on which stressful and
lished over the past few years, a clarification of the effects immunogenic stimulation are superimposed; (f) a variety
of stress on immune function or the mediation of such of host factors such as species, strain, age, gender, and
effects has not been achieved. nutritional state; and (g) the interaction among these
Most often, stress has been reported to suppress im- several variables. However complex the phenomena may
mune responses, and this suppression has been at- be, stress-related changes in immunity provide one body
tributed to an increase in adrenocortical steroids. In fact, of data that bear on interactions between the immune and
the same stressor can enhance or suppress different central nervous systems.
immune responses, and only some of these effects appear One of the newer and more dramatic lines of behavioral
to be mediated by adrenal changes. Different stressors research providing evidence of a link between the brain
have different effects on the same parameter of iin- and the immune system comes from studies of condi-
munological reactivity; an elevation in adrenocortical tioned alterations in immunological reactivity. We be-
steroids has been associated with both suppression and lieve that conditioning, as a potential immunoregulatory
enhancement of immune function. This variability is mechanism, should be considered apart from the immu-
evident from a review of the available literature as well as nomodulating effects of stress. Attempts to dismiss condi-
from the data from single studies (e.g., Blecha, Kelley & tioning phenomena as nonspecific stress (i.e., adrenocor-
Satterlee 1982; Nieburgs, Weiss, Navarrete, Strax, Teir- tically) mediated effects (Dwyer 1983) have not been
stein, Grillione & Siedlecki 1979) or a series of experi- persuasive (Bovbjerg, Cohen & Ader 1983) and are dis-
ments from a single laboratory (e.g., Keller, Weiss, cussed below. The present paper, then, will focus on
Schleifer, Miller & Stein 1981; 1983). studies of conditioning. We will review this emerging
Recent data suggest that the organism's capacity to field of interdisciplinary research rather than attempt to
exert some control over "stressful" environmental cir- develop a theoretical position based on what is still a
cumstances (i.e., coping behavior) can influence immune limited amount of data on CNS-immune system inter-
function (Laudenslager, Ryan, Drugan, Hyson & Maier actions.
380 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

Ader & Cohen: CNS-immune system interactions
1. Historical perspectives trials in which a CS was paired with injection of a

staphylococcus filtrate. After a 10-day rest period, the CS
References to the possibility that conditioning processes was repeatedly presented without the UCS, and on the
could be involved in the etiology of certain pathophysio- next day these two experimental animals and a noncondi-
logical conditions are scattered throughout the literature tioned control guinea pig were injected with a lethal close
in psychosomatic medicine. Apart from the recent liter- of a culture of Vibrio cholera. The control animal died
ature on the clinical implications of classically and instru- whereas the two experimental animals survived the chol-
mentally conditioned alterations in autonomic and cer- era injection. An additional control was introduced in
tain visceral responses (e.g., Miller 1978), one of the most mother experiment in which one of the conditioned
frequently recited clinical vignettes concerns the induc- inimals was not reexposed to the CS before being in-
tion of allergic reactions in particularly sensitive indi- ected with the cholera culture. Again, only the experi-
viduals who are exposed to symbolic, nonallergenic en- mental animal that was reexposed to the CS survived the
vironmental stimuli that have been associated in the past infection.
with an allergen (e.g., Hill 1930; Mackenzie 1886). There Because of the novelty and the importance of the
are, however, laboratory studies in humans (Dekker, phenomena described by Metal'nikov and Chorine,
Pelser & Groen, 1957) and animals (Ottenberg, Stein, several experiments were conducted to confirm their
Lewis & Hamilton 1958) providing evidence that, in observations (Nicolau & Antinescu-Dimitriu 1929a;
some proportion of subjects, conditioned stimuli are Podkopaeff & Saatchian 1929; Vygodchikov & Barykini
capable of eliciting asthmatic symptoms. 1927). The most extensive study was conducted by Os-
The first systematic studies of conditioning in the travskaya (1930). Conditioned guinea pigs were subjected
modification of host defense mechanisms - and immune to kinesthetic stimuli in the form of heat, scratching, or
function in particular - were initiated by Russian investi- electrical stimulation (the CS) for a period of 3-5 min.
gators. These studies were adapted from Pavlov's (1928) The CS was followed by an injection of foreign material
work on the conditioning of physiological responses. In once each day for three weeks. Nonconditioned guinea
the classical (Pavlovian) paradigm, a stimulus (e.g., food) pigs received either the CS without the UCS or the UCS
that unconditionally elicits a particular response (sali- without the CS. After a 10-15-day interval, the animals
vation) is repeatedly paired with a neutral stimulus that were again exposed to the CS. Peritoneal exudates were
does not elicit that same response. Eventually, the neu- examined at different times before and after exposure to
tral stimulus, the conditional stimulus (CS), will elicit the UCS and before and after presentation of the CS on
salivation in the absence of food, the unconditional stim- the test day. In response to the CS, there was an increase
ulus (UCS). In the same way, a tone (CS) that has been in the percentage of PMNs in the peritoneal exudate in
repeatedly paired with an electric shock (UCS) to an 67% of the conditioned animals. In contrast, only 23% of
animal's leg will eventually elicit a leg flexion in the the control animals showed an increase in polynucleated
absence of the UCS. In addition to the production of cells in response to the kinesthetic stimulation.
antibody, the injection of foreign material into the per- In the aforementioned studies, the unconditional re-
itoneum of guinea pigs unconditionally elicits a non- sponse was a nonspecific inflammatory reaction. Subse-
specific defense reaction characterized by an increase in quently, Metal'nikov and Chorine (1928) attempted to
polymorphonuclear leukocytes (PMNs).2 Using Pavlo- condition changes in the specific antibody response to
vian procedures, Metal'nikov and Chorine (1926) at- Vibrio cholera. One group of three rabbits received 12-
tempted to condition the increase in PMNs by pairing 15 trials on which a heat or scratch stimulus was followed
injections of foreign material, the UCS, with immuno- by an intraperitoneal injection of the antigen. Three
logically neutral conditional stimuli. Guinea pigs re- weeks later (when antibody titers were still relatively
ceived intraperitoneal injections of small doses of either high), two animals (Nos. 92 and 93) were reexposed to the
Bacillus anthracis, a staphylococcus filtrate, or a tapioca CS alone and showed an additional elevation in antibody
emulsion that was associated with the scratching or heat- titer. Two months later, the CS was presented again -
ing of a single area of the skin. These CS-UCS pairings this time, to rabbits 93 and 96, with No. 92 serving as the
occurred once daily for 18-25 days. After a 12-15-day control. Although the experimentally induced response
rest period to allow the cellularity of the peritoneal was not large, these data suggested that immune as well
exudate to return to baseline levels, the CSs were applied as nonspecific defense reactions were subject to central
several times in the absence of antigenic stimulation. nervous system processes. Studies by Nicolau and Anti-
As part of the unconditioned response, there was a nescu-Dimitriu (1929b), Ostravskaya (1930), and Polet-
rapid influx of PMNs that subsided after 1 or 2 days. In tini (1929) yielded additional data indicating that condi-
one guinea pig, for example, PMNs constituted 90% of tioning processes were capable of altering antibody
the peritoneal exudate cells within 5 hr of the injection of levels. Other studies by Friedberger and Gurwitz (1931)
foreign material. The same animal experienced 21 CS- and by Kopeloffand his colleagues (1933; 1935), however,
UCS conditioning trials and, after a 13-day rest period, failed to uncover conditioning effects.
was exposed to the CS alone. PMNs increased from a These very early studies of conditioned alterations of
resting level of 0.6% to 62% by 5 hr after exposure to the immunological reactivity stimulated considerable in-
CS. Similar observations were made in two other ani- terest and attention, particularly in the Soviet Union.3
mals. The conditioned response was weaker and more Although the implication and underlying premise of such
transient than the unconditioned response, but it was research were consistent with the predominant view that
clearly eident. all physiological processes were regulated by the central
Two guinea pigs were subjected to 12 conditioning nervous system, some of the data were viewed with
THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 381

appropriate skepticism since there were no known mech- tualization provided the framework for the early studies
anisms to account for such findings. There was little of conditioning, and, unfortunately, rejection of the un-
controversy over the conditioning of nonspecific de- derlying concept resulted in rejection of the experimental
fenses, but conditioned changes in antibody levels were data.
considered unlikely since, even at that time, antibody Even today, we can only speculate about the mecha-
production was known to be an immunologically specific nisms underlying conditioned alterations in immunologi-
response to each new antigenic stimulus. There followed, cal reactivity. We do not yet have a complete picture of
then, a 20-year hiatus in this area of research. the cellular interactions that result in the synthesis and
Interest in the area resurfaced in the 1950s with the release of antibodies, and we have only a rudimentary
critiques and studies reported by Bereznykh (1955), Dol- knowledge of the neuroendocrine factors that appear
in and Krylov (1952), Doroshkevich (1954), Vygodchikov capable of influencing such responses. As reviewed by
(1955), Zdrodovskii (1956), and Zeitlenok and Bychkova Luk'ianenko, the data available, even then, provided
(1954), and in 1957-58 there was another series of more sufficent reason to suppose that conditioning processes
or less vitriolic discussions of the role of the central operating through the CNS could alter the threshold of
nervous system in immunity (Gordienko 1957; Pletsityi sensitivity of the organism to antigenic stimuli and modu-
1957; Vygodchikov 1957). Several studies purporting to late the level of immunological reactivity. As is the case
document a variety of conditioning effects were subse- with so many other psychobiological phenomena, partic-
quently reported (Dolin, Krylov, Luk'ianenko & Flerov ularly those involving pathophysiological processes, the
I960; Il'enko & Kovaleva 1960; Luk'ianenko 1959; Sakan- available data suggest to us that, although antigenic
jan & Kostanjan 1957; Savchuk 1958), but they did not stimulation may be a necessary factor for the initiation of
resolve all the methodological or theoretical issues that antibody production, it is not sufficient in itself for defin-
had been raised. ing the complex chain of events that assures activation of
In one of the last papers on this subject to come from the immunoregulatory defense mechanisms that serve
the organism under in vivo conditions. It is in this context
the Soviet Union, Luk'ianenko (1961) attempted to sum-
that, for all their flaws, the early attempts to use condi-
marize the available literature. There was considerable tioning procedures to modify immune responses are of
evidence, for example, that nonspecific defense reactions considerable interest. A more detailed review of this
such as leukocyte infiltration could be conditioned. Other literature is available elsewhere (Ader 1981c).
studies had documented conditioned changes in phago-
cytosis, complement, and lysozyme activity. It was gen-
erally assumed that these nonspecific cellular and humor-
al defense reactions were mediated by hypothalamic- 2. Conditioned suppression of humoral immunity
pituitary-adrenal mechanisms and that the demonstra-
tion of conditioning effects thus implied an involvement Our own entree into the study of conditioned changes in
of the central nervous system. immunological reactivity resulted from the serendipitous
In summary, by 1960 several Russian studies had observation of mortality among animals that were being
indicated that it was possible to condition alterations in tested in a taste aversion learning situation. This is a
specific immune responses (i.e., antibody production). passive avoidance conditioning paradigm in which con-
However, the failure to confirm these observations in sumption of a distinctively flavored drinking solution is
some laboratories and the lack of any notion as to how paired, most frequently, with an injection of a phar-
conditioned immunomodulation might occur left the is- macological agent. Typically, the pharmacological agent
sue open. Also, the variety of experimental paradigms is one that is known to produce noxious effects and the
made it difficult to discern the nature of any functional appetitive response to the flavored solution is thereby
relationships between parameters of the conditioning punished. A single pairing of the distinctive solution and
process and immunological changes. For example, the the drug can result in subsequent avoidance of the solu-
nature of the antigen as well as the dose, route of inoculation. In our study, cyclophosphamide, a potent immu-
tion, frequency of application, and the temporal rela- nosuppressive drug, was paired with the consumption of
tionship among conditioning, antigenic stimulation, and different volumes of a saccharin drinking solution (Ader
reexposure to conditioned stimuli are all relevant param- 1974). The single pairing of saccharin and cyclophospha-
eters that could influence the observation of conditioning mide (CY) was followed by extinction trials at 3-day
effects. Similarly, the qualitative and quantitative charac- intervals. The magnitude of the initial aversion to sac-
teristics of the CS, the CS-UCS interval, and the number charin and resistance to extinction were, as hypothesized,
of conditioning trials varied among experiments or, in a directly related to the volume of the saccharin solution
single experiment, had been found to influence the con- consumed on the conditioning trial. During the course of
ditioned response. Some of these same issues, inciden- repeated, unreinforced exposures to saccharin, however,
tally, also apply to current studies of conditioned (and some of the animals died, and mortality rate also tended
"stress-induced") alterations in immune function. to vary directly with the volume of saccharin consumed
As much as anything else, the demise of conditioning during the single conditioning trial. Based on these ob-
studies from the Soviet Union can be attributed to the servations, it was hypothesized that the pairing of a
intransigence of the proponents as well as the critics of neutral (conditioned) stimulus with a drug that uncondi-
such work. For many, the observation of conditioning tionally suppresses immunity would result in the condi-
effects was interpreted as evidence for the neural trans- tioning of an immunosuppressive response. Mortality,
mission of antigenic stimulation and led to the hypothesis then, might have resulted from repeated exposure to
that the central nervous system was capable of stimulat- such a conditioned stimulus, which, by compromising the
ing antibody production de novo. This restrictive concep- integrity of the immune system, might have rendered

these animals susceptible to any latent pathogens in the ICh ADER S COHEN (75) ROGERS el al ( 76) WAYNER el al I 78)
laboratory environment in proportion to the magnitude of
the conditioned response. These speculations were trans-
lated into the following controlled experiment specifically
designed to determine whether it was possible to condi-
tion an immunosuppressive response (Ader & Cohen 4'
1975).
Individually caged rats were gradually adapted to
drinking their total daily allotment of water during a
single 15-min period that occurred at the same time each
day.4 On the day of conditioning, conditioned animals
2-
cs0 cs, cs, us
COND
L
cs0 cs, cs, us
COND
cs0 cs, cs, us
COND
were given a 0.1% solution of sodium saccharin in tap Figure 1. Hemagglutinating antibody titers (mean ± SE)
water instead of plain water during the 15-min drinking measured 6 days after the injection of SRBC. NC, noncondi-
period. Drinking was followed by an intraperitoneal (ip) tioned rats; CSo, conditioned animals that were not reexposed
injection of 50 mg/kg CY. Nonconditioned animals in this to the CS after immunization; CS! and CS 2 , conditioned animals
initial experiment were provided with plain water as reexposed to the CS on one or two occasions, respectively; US,'
usual and injected with 50 mg/kg CY. In subsequent conditioned animals injected with CY at the time of immuniza-
studies, nonconditioned animals received saccharin and tion with SRBC; P, placebo-treated animals. Reprinted from
CY in a noncontingent (unpaired) fashion. Placebo ani- Ader (1981c) by permission of Pergamon Press.
mals in this and subsequent studies drank plain water or
saccharin and were injected with vehicle.
Three days after conditioning, all animals were in- with the titers in Group NC and in Group CSo, condi-
jected ip (intraperitoneally) with antigen, sheep red tioned animals reexposed to the CS on one or two occa-
blood cells (SRBC). Thirty min after immunization, each sions after inoculation with SRBC showed attenuated
animal in one subgroup of conditioned animals (Group antibody responses. These initial results, then, sup-
CS) was provided with a single drinking bottle containing ported the hypothesis that pairing saccharin consumption
the saccharin drinking solution for 15 min and was then with the injection of an immunosuppressive drug would
injected with saline. To control for the effects of condi- enable saccharin to elicit an immunosuppressive re-
tioning per se, a second subgroup of conditioned animals sponse.5
(Group CSo) was provided with plain water and injected Using essentially, the same experimental paradigm,
with saline. To define the unconditioned effects of CY, a Rogers, Reich, Strom, and Carpenter (1976) and Wayner,
third subgroup of conditioned animals (Group US) re- Flannery, and Singer (1978) obtained essentially the
ceived plain water followed by an injection of CY. After same results (Fig. 1). Neither study uncovered an effect of
immunization, nonconditioned rats (Group NC) were a single reexposure to the CS, but two reexposures to
given saccharin-flavored water and injected with saline; saccharin did significantly attenuate the antibody re-
placebo-treated animals (Group P) were given plain water sponse to SRBC in conditioned animals. In these experi-
and no injection. In this first experiment, independent ments, two reexposures to the CS also decreased the
subgroups of conditioned animals were reexposed to magnitude of the conditioned aversion to saccharin, a
saccharin on the day of immunization (Day 0) and/or 3 change that was not observed in the Ader and Cohen
days after being injected with SRBC. For each subgroup (1975) study. In the study by Wayner et al. (1978), the
of conditioned animals there was a subgroup of noncondi- antibody response in conditioned animals that were reex-
tioned animals that received saccharin at the correspond- posed to the CS on three occasions did not differ from that
ing time(s). Six days after immunization, blood samples in Groups NC or CSo, but these control values were not
were obtained, coded, and titrated for hemagglutinating obtained on Day 12 when the experimental animals
antibody activity. reexposed to the CS on three occasions were sampled.
Behaviorally, conditioned rats reexposed to saccharin Nevertheless, the possibility that there may be a dissocia-
reduced their intake of this solution on the day or days tion between the conditioned behavioral and immu-
when it was presented in place of plain water relative to nosuppressive responses is supported by recent data on
the intake of control animals provided with either sac- the rate of extinction of a conditioned taste aversion and a
charin or plain water. Figure 1 contains the immunologi- cell-mediated immune response (Bovbjerg, Ader & Co-
cal data from this initial experiment. The relationship hen 1984; discussed below).
among the several groups was exactly as predicted. In addition to studying immunity to SRBC, a humoral
Placebo-treated animals, having experienced no immu- (antibody) response that involves interactions between
nosuppressive treatment, had the highest antibody titers thymus-dependent helper T lymphocytes and antibody-
and animals injected with CY at the time of immunization producing bone marrow-derived B lymphocytes, Wayner
had the lowest antibody titers. Antibody titers of serum et al. (1978) examined conditioning using Brucella abor-
from nonconditioned animals (Group NC) and from con- tus, a so-called T-cell independent antigen. Their experi-
ditioned animals that were not reexposed to the saccharin ment did not uncover conditioning effects. Cohen, Ader,
solution (Group CSo) did not differ. That both of these Green, and Bovbjerg (1979), however, did observe condi-
groups had lower titers than Group P probably reflects tioned immunosuppression in two experiments with mice
residual immunosuppressive effects of CY administered immunized with the hapten 2,4,6-trinitrophenyl coupled
three days earlier. Groups NC and CSo, then, are the to another relatively thymus-independent carrier,
appropriate control groups against which to evaluate the lipopolysaccharide. Although these data suggest that the
effects of conditioning and CS reexposure. Compared effects of conditioning could be confined to some direct

effect on B lymphocytes, this conclusion would be pre- was for this reason, perhaps, that the net effect of this
mature considering the available data on the conditioning procedural modification did not exaggerate the condi-
of cell-mediated (T-cell) responses described below. At tioned immunosuppressive response in comparison to
this juncture, it seems prudent to allow for the possibility that seen in previous studies.
that there are multiple targets of conditioned immu- In still another attempt to magnify the effects of condi-
nopharmacological effects. tioning (Ader, Cohen & Bovbjerg 1982), procedural
If one varies the dose of CY (the magnitude of the changes did yield a more pervasive effect than we had
unconditioned drug effect) or the iinmunosuppressive previously reported (Ader & Cohen 1981). As noted
drug itself, the magnitude or kinetics of the conditioned above, the immunosuppressive effects of CY are dose-
response are affected. Increasing the dose of a single related (Ader, Cohen & Grota 1979; Hill 1975; Shand
injection of CY from 50 to 75 mg/kg suppresses immuno- 1979; Turk & Parker 1979). If, during the period of
logical reactivity for a longer period of time, with the recovery from the effects of CY, conditioned animals
result that antibody titers measured six days after immu- were reexposed to a CS paired with the original CY
nization with SRBC are still relatively low and there are treatment, this might alter the baseline level of immu-
no treatment effects at that time. Eight days after anti- nocompetence on which subsequent immunization was
genic stimulation, however, hemagglutinating antibody superimposed. Moreover, CS reexposure before immu-
titers have risen and, again, one can observe an attenu- nization would also obviate the coincidental presentation
ated response in conditioned animals that were reex- of a conditioned stimulus for suppression of immunity
posed to the CS (Ader & Cohen 1981). Methotrexate is with an unconditioned stimulus (antigen) for activation of
also capable of inducing a conditioned taste aversion and a an immune response. The procedures of this experiment
conditioned immunosuppressive response, but the mag- were basically the same as those described previously.
nitude of the effect in the one study that we conducted Rats in this study, however, were given 75 mg/kg CY
was not as great as the effect observed with CY (experi- immediately after drinking a saccharin-flavored solution.
mental rats had the lowest antibody titers but the dif- Independent subgroups of conditioned animals were re-
ferences reached statistical significance only in com- exposed to the CS 5 and 10 days after conditioning, and
parison with a group composed of combined control independent samples from all groups were immunized
with SRBC 10, 15, or 25 days after conditioning. Inde-
animals) (Ader & Cohen 1981).
pendent samples from each subgroup were sampled for
Studies on cell-mediated immune responses (de- hemagglutinating antibody titers 4, 6, or 8 days after
scribed below) support the conclusion that conditioned immunization. Also, taste aversion conditioning in this
alterations in immunological reactivity are not confined to experiment was assessed using a two-bottle preference
the idiosyncratic effects of CY as an immunomodulating testing procedure. As can be seen in Figure 2, the pairing
agent. Similarly, the conditioning does not appear to be of saccharin consumption with CY induced an aversion to
attributable to some esoteric effect of saccharin. We have saccharin-flavored water. It is also evident that the use of
obtained conditioning effects using a sucrose solution a two-bottle procedure obviated any difference in the
paired with CY (Ader & Cohen 1981), and Wayner total fluid consumption of experimental and control
(personal communication) has observed conditioned im- animals.
munosuppression using a mild citric acid solution as the
CS. In an unpublished study, we have also observed The interval between conditioning and antigenic stim-
conditioned immunomodulation using a commercially
available chocolate milk as the CS. This study was actually
conducted in an attempt to mask the taste of CY and 25 TOTAL
H2O
introduce the immunosuppressive drug orally. We had
previously argued that injection procedures become a SAC
part of the CS in pharmacological studies (Dolin et al. 20-
1960; Hutton, Woods & Makous 1970; Pavlov 1928;
Siegel 1975). Thus, at the time of immunization, even
nonconditioned animals were reexposed to stimuli that UJ 15-
had previously been paired with immunosuppressive 5
drug treatment. If animals would voluntarily consume a
volume of a solution containing CY sufficient to uncondi- Q 10-
tionally suppress immunological reactivity, it would elim- 3
inate the need for the ip injection that has accompanied
administration of both CY and antigen. Some proportion 5-
of rats do consume a volume of chocolate milk plus CY
equivalent to a dose in excess of 40 mg/kg. This oral dose
is minimally effective in suppressing antibody produc-
tion, and these conditioned animals reexposed to plain P CS P CS
chocolate milk did show a conditioned immunosup- 5 10
pressive response. However, as evidenced by taste aver- DAYS POST-CONDITIONING
sion behavior, the animals can discriminate between the
taste of plain chocolate milk and chocolate milk contain- Figure 2. Mean intake of plain water and a saccharin solution
under a two-bottle preference testing procedure in conditioned
ing CY (i.e., reexposure to unadulterated chocolate milk
(CS) and nonconditioned, placebo-treated (P) rats. Reprinted
was apparently "not the same taste" previously experi- from Ader et al. (1982) by permission of the American Psycho-
enced and there was only a mild aversion to the "CS"). It logical Association. Copyright 1982 by the APA.

ulation did not influence antibody titers. Conditioned docrine responses conditioned by the same operations
animals reexposed to the CS before immunization had are long lasting. It seems more reasonable, however, to
lower antibody titers than placebo-treated animals and consider seriously the possibility that one or more of the
conditioned animals that were not reexposed to the CS at effects of CY on the overall or relative number of B-cells
each of the sample times (Fig. 3). These data extend our and/or helper or suppressor T lymphocyte subsets (Mc-
previous observations in several ways: (a) in contrast to Connell, Munro & Waldman 1981) could be influenced
previous experiments in which conditioning effects were by conditioning processes.
discerned at only a single sample time (Ader & Cohen Conditioned suppression of humoral immunity has also
1981), the differences observed in this study were more been verified by Gorczynski, Macrae, and Kennedy
pervasive; (b) the longest interval between conditioning (1984). In an initial series of experiments on mice, the
and immunization in previous studies of humoral immu- results were equivocal: conditioned immunosuppression
nity was 14 days (Cohen et al. 1979), whereas this study was observed in five of eight studies, one study showed no
included animals immunized 25 days after conditioning; effects, and an enhancement of immunological reactivity
(c) the imposition of a preference testing procedure ruled was observed in two experiments. These studies were
out differences in the animals' state of dehydration as an carried out in group-housed mice subjected to three
explanation of the reduced immunological reactivity in pairings of saccharin consumption and CY (100 mg/kg) at •
conditioned animals reexposed to the CS; (d) preference 14-day intervals. The plaque-forming cell response6 to
testing also eliminated the component of conflict inherent SRBC was measured 6 days after immunization. Since the
in a single-bottle exposure to the CS (Rabin, Hunt & Lee mice were group-caged, it is not possible to confirm that a
1983); (e) the observation that CS reexposure before taste aversion had in fact been induced in every member
immunization could still support conditioned immu- of the group, that is, that every member of the group had
nosuppressive responses indicated that an antigen-driven actually been exposed to the saccharin-CY contingency.
activation of the immune system is not a requirement for The variability within and between experiments could be
conditioning to occur. This latter point may be particu- reduced, however, by attending to circadian rhythmicity,
larly important in elucidating the mechanisms underlying that is, the time at which conditioning and reexposure to
conditioned alterations in immunologic reactivity. One the CS took place. When the time of CS reexposure was
might still argue that the neuroendocrine concomitants of held constant (1:00 P.M.), mice conditioned at 7:00 A.M
conditioned taste aversion behavior or the neuroen- invariably displayed a conditioned immunosuppressive
6:30 a.m. 6:30 p.m 6 30 a.i

100-
6-
C3
O 4-
11
1
2-
P CS 0 CS P CS 0 CS P C S 0 CS
4 6
DAYS POST-ANTIGEN
Figure 3. Hemagglutinating antibody titers (mean ± SE)
obtained 4, 6, and 8 days after immunization with SRBC from
independent groups of placebo-treated rats (P), and conditioned Figure 4. Plaque-forming cell response to SRBC in mice
animals that were (CS) and were not (CSo) reexposed to the CS conditioned at different times of the day. NC, nonconditioned
during the interval between conditioning and immunization. mice; CSo, conditioned mice that were not reexposed to the CS;
Reprinted from Ader et al. (1982) by permission of the American CS, conditioned mice reexposed to the CS. Adapted from
Psychological Association. Copyright 1982 by the APA. Gorczynski et al. (1984) by permission of the authors.

response; mice conditioned at 1:00 P.M. showed condi- immunological reactivity. Following irradiation, an ex-
tioned immunosuppression in two of three experiments; perimental group was reexposed to the experimental
and mice conditioned at 7:00 P.M. showed conditioned chamber and experienced the auditory stimulation only
enhancement in one experiment arid failed to differ from (5 times during each of 6 sessions). The plaque-forming
controls in two other experiments (Fig. 4). cell response was measured 4 days after immunization
That the psychophysiological background upon which with SRBC (introduced after the fourth session of reex-
conditioning is superimposed exerts some effect on the posure to the CS). In the absence of irradiation, "stress"
subsequent manifestation of conditioned changes in im- did not influence the number of splenic antibody-forming
munological reactivity is also supported by Gorczynski et cells. Among the irradiated groups, however, reexposure
al.'s (1984) observation that, following "acute stress" (45 to the environmental stimuli previously associated with
min of rotation on a single occasion), conditioned immu- electric shock suppressed immunological reactivity rela-
nosuppression was observed in two experiments. Follow- tive to animals that experienced buzzer-shock trials but
ing "chronic stress" (20 min of rotation on each of 5 were not reexposed to the CS. The magnitude of suppres-
consecutive days), mice showed no conditioned changes sion was not great. Conditioned animals that were reex-
in their plaque-forming cell response in one experiment posed to the CS differed only marginally from animals
and a conditioned enhancement in the second. Additional that were subjected to radiation plus handling (placement
studies indicated that 3-month-old mice would display in the experimental chamber) or irradiation plus place-
conditioned immunosuppressive responses whereas such ment in the experimental chamber and auditory stimula-
changes could not be observed in 24-month-old animals. tion, but, in terms of the treatment of these latter groups,
Without some independent measure of conditioning (ob- the conditioning of the "experimental" and control
viated by the group housing), one cannot be sure that groups was probably only a matter of degree. We in-
these results reflect the modifiability of immunological terpret these observations as suggesting that the impact
reactivity, differences in sensitivity to a constant dosage of "relatively mild stress" or conditioned stress responses
of CY, or differences in conditionability of a taste aversion may be capable of exerting their subtle effects only in an
in relatively old as compared to young mice. immunocompromised host. Such an hypothesis would
Klosterhalfen and Klosterhalfen (1983) confirmed the not be inconsistent with the data bearing on the timing of
stress in relation to immunogenical (pathogenic) stimula-
phenomenon of conditioned immunosuppression by
tion or to data derived from experimental studies of the
using a conditioning paradigm to modify adjuvant-in- effects of behavioral factors in susceptibility to disease
duced arthritis in rats (Pearson & Wood 1959). Condi- (Ader 1980; 1981d).
tioned animals consumed a saccharin-vanilla drinking
solution and were injected with 100 mg/kg CY in one
experiment and 80 mg/kg CY in a second. Noncondi-
tioned animals were exposed to the saccharin-vanilla 3. Conditioning and cell-mediated immunity
solution and CY in a noncontingent manner. Arthritis was
induced by injecting Freund's complete adjuvant (CFA)7 Conditioned alterations in immunological reactivity are
in a hind paw. Within 24 hr the injected paw showed not confined to humoral immunity (antibody mediated),
dramatic swelling, and in approximately 12 days the but extend to cell-mediated reactions, which are medi-
uninjected hind paw also showed signs of inflammation. ated by T lymphocytes. Taking advantage of the observa-
There were no differences among the groups in the tion that multiple, low-dose injections of CY could sup-
degree of swelling of the injected paw. Reexposure to the press a graft-versus-host (GvH) response8 (Whitehouse,
CS at the time of injection of adjuvant and 2 and 4 days Levy & Beck 1973), Bovbjerg, Ader, and Cohen (1982)
later, however, obviated the swelling in the uninjected modified the experimental protocol for conditioning an
paw that was observed in more than half the control immunopharmacological suppression of antibody titer
animals in each of the two experiments. The authors and applied it to the conditioned suppression of a local
entertained the possibility that the difference between GvH response.
the injected and noninjected paws could relate to the (Lewis X Brown Norwegian)Fj female rats were condi-
dose of CY. It should also be considered that the (su- tioned by pairing saccharin consumption with an injec-
prathreshold) concentration of adjuvant may have tion of 50 mg/kg CY 48 days before these animals re-
masked the effects of conditioning in the injected paw and ceived a subdermal injection of splenic leukocytes
that this dose, which induced swelling in the contralateral obtained from female Lewis donors. On that same day,
paw in only 50-60% of control animals, was essentially the experimental subgroup of previously conditioned rats
equivalent to a subthreshold dose of adjuvant and thus was reexposed to the CS (saccharin plus an ip injection of
allowed the observation of conditioning effects. In a very saline); on the next day, they were again reexposed to
recent study using a different strain of rats (Klosterhalfen saccharin and given an injection of 10 mg/kg CY; and, on
& Klosterhalfen 1984), the concentration of CFA was the next day, they were again reexposed to the CS only.
reduced and a conditioned immunosuppressive response The effect of reexposure to the CS in conditioned animals
was observed in the injected paw of conditioned rats that is shown in Figure 5. As expected, rats treated with 10
were reexposed to the CS. mg/kg on the day of grafting and on the following two days
Conditioning effects have also been reported by Sato, (Group US) showed a marked reduction in the weight of
Flood, and Makinodan (1984) in still another conditioning popliteal lymph nodes harvested 5 days after introduction
paradigm. Mice were subjected to 25 pairings of a buzzer of the cellular graft. Group NC (which previously re-
and 3 sec of inescapable electric shock during each of 5 ceived saccharin and CY in a nonpaired manner) and
conditioning sessions. The animals were then subjected Group CSo (previously conditioned but not reexposed to
to a low dose (200R) of ionizing radiation to suppress saccharin) confirmed our preliminary observation that a

the mechanisms, the experimental demonstration of the

30- acquisition and extinction of an immunosuppressive re-
sponse implies that the environmental contingencies that
INJ.
modify the behavioral expression of the association be-
25" tween saccharin and CY are also capable of modifying the
immunological expression of the association between
saccharin and CY. That is, the change in immunological
I 20- reactivity reflects a sensitivity to the same environmental
interventions that modify behavioral responses. Even so,
: WEIGH!
these results also suggest that the conditioned immu-

nosuppression can be dissociated from the conditioned
15- taste aversion. Such findings are quite consonant with the
dissociation between behavioral and endocrine responses
in the taste aversion conditioning paradigm observed by
8 10- Smotherman, Margolis, and Levine (1980) and are remi-
niscent of Gantt's (1953) description of dissociation in the
acquisition and/or extinction of other conditioned motor
and physiological responses (schizokinesis).
5- Using an in vitro measure of alloantigen-induced9 T-
cell proliferation and a biological rather than an immu-
mopharmacological immunosuppressant, Kusnecov, Siv-
yer, King, Husband, Cripps, and Clancy (1983) have also
documented conditioned suppression of cell-mediated
CONDITIONED
immunity. In this experiment, rats were conditioned by
pairing saccharin consumption with an ip injection of
Figure 5. Popliteal lymph node weights in (Brown Norwegian) rabbit antirat lymphocyte serum (ALS), which selectively
F, rats determined 5 days after inoculation with splenic leuko-
cytes from Lewis F, donor animals. Mean (± SE) for injected
and contralateral footpads are given for placebo-treated (P) rats;
for nonconditioned (NCr) animals exposed to a single low dose of
CY (10 mg/kg) injected one day after the cellular graft; and for
conditioned animals injected with a single low dose of CY and 22
provided with plain water (CSo), conditioned animals given a
single low-dose injection of CY and reexposed to the CS on Days .;,;cs0 ;;.•;.
•'•:i :•'••''. •'
0, 1, and 2 after the cellular graft (CSr), and conditioned animals 20 ! * • * • • • '
(CONTROL)
given three low-dose injections of CY on Days 0, 1, and 2.
Reprinted from Ader and Cohen (1981) by permission from
Academic Press. 18
x
E_ /
single low-dose injection of CY - even in animals pre-
/
viously treated with 50 mg/kg CY and even in conjunction H 16 •
X /
with saccharin consumption (in Group NC) - yielded only o /
a modest reduction in the GvH response in relation to the UJ j
f
effects seen in the experimental group. Conditioned 14
animals that received only a single, low-dose injection of UJ
CY and reexposure to the CS showed a significant sup- Q
pression of the GvH response relative to each of these o
z 12
control groups.
Since CS reexposure was able to suppress the GvH
response 7 weeks after conditioning, it became feasible to
introduce unreinforced CS presentations between condi- 3
[ A —ft—
tioning and immunogenic stimulation in order to examine 2 >
extinction of the conditioned immunosuppressive re-
sponse (Bovbjerg et al. 1984). Behaviorally, rats that 1 •
received 4, 9, or 18 extinction trials did not differ among

themselves but showed a greater preference for saccharin
than the experimental group that received no extinction
CS-0 CS-4 CS-9 CS-I8
trials. As shown in Figure 6, animals that experienced no GROUP (NO. EXTINCTION TRIALS)
extinction trials displayed a conditioned suppression of
the GvH response, confirming the results described Figure 6. Popliteal lymph node weights in conditioned ani-
above. Four extinction trials did not attenuate the condi- mals that were not reexposed to the CS following conditioning
tioned response, but animals that received 9 or 18 extinc- (CSo) and in conditioned animals reexposed to the CS following
0, 4, 9, or 18 unreinforced CS presentations. Results represent
tion trials did show extinction; these animals did not differ mean (± SE) weights of draining (•) and unstimulated (O) nodes.
from conditioned animals that were not reexposed to the Reprinted from Bovbjerg et al. (1984) by permission from the
CS during initiation of the cellular response. Whatever Journal of Immunology.

Ader & Cohen: CNS-iminune system interactions
cytes from the CBA recipients were analyzed to assess the

degree of antigen-specific response by determining the
• H2O - ALS number of precursors of cytotoxic T lymphocytes (CTLp)
A SAC • NRS that could react against alloantigens of the C57BL/6J
O SAC - ALS donor of the grafted tissue. This procedure was repeated
three times at 40-day intervals, which was the time
required for CTLp frequency to return to the baseline
level given by experimentally naive control mice. On the
fourth trial, animals were sham grafted rather than skin
allografted. In each of two experiments, 50-60% of the
animals so treated showed an increase in CTLp numbers
in response to sham grafting. These "responders" were
then divided into groups that received two additional
conditioning trials and those that experienced extinction
trials (the grafting procedure alone). When subsequently
tested with the CS alone, all the mice that received
additional conditioning trials displayed a conditioned
elevation of CTLp, whereas there was no increase in
CTLp numbers in any of the mice that experienced
extinction trials.
Conditioned enhancement of a delayed-type hyper-
sensitivity (DTH) response11 has also been observed
(Bovbjerg 1983). The effects of CY on DTH reactions
depend upon several factors, prominent among which is
1 2 3 4 5 the timing of drug treatment in relation to sensitization
DAYS IN CULTURE
(initial immunization) and subsequent antigenic chal-
lenge. CY administered before sensitization enhances
Figure 7. Primary mixed lymphocyte culture response (mean DTH (by a selective deleterious effect on suppressor
± SE) of mesenteric lymph node cells prepared 21 days after T-cells), while CY administered after sensitization but
conditioning in conditioned animals treated with saccharin and before challenge suppresses the response (Schwartz
antilymphocyte serum and subsequently reexposed to saccharin Askenase & Gershon 1978). In a series of experiments
(SAC-ALS), and in control groups reexposed to saccharin follow-
ing the pairing of saccharin and normal rabbit serum (SAC-NRS) designed to define parameters appropriate for a study of
or nonconditioned animals (H2O-ALS). For each animal, the conditioning effects, Bovbjerg (1983) further established
mean counts per minute (CPM) of control cultures were sub- that, while CY administered before challenge would
tracted from the CPM of experimental cultures to provide a suppress the DTH response, these mice showed en-
measure of in vitro alloantigen-induced cell proliferation. Re- hanced DTH responses to several subsequent chal-
printed from Kusnecov et al. (1983) by permission from the lenges. He then proceeded to take advantage of this
Journal of Immunology. pattern of responses to examine conditioned suppression
and enhancement of DTH. Two strategies were adopted:
one was based on the original conditioning paradigm
destroys lymphocytes without apparently inducing the described by Ader and Cohen (1975), and the other was
noxious gastrointestinal effects of drugs such as CY (Bach based on the successful conditioning of a depressed GvH
1975). Control animals were injected with normal rabbit response in which the basic paradigm was modified by
serum (NRS) devoid of antilymphocyte activity. Fourteen introducing multiple reexposures to a CS previously
days after conditioning, animals were reexposed to sac- paired with CY in conjunction with a single reexposure to
charin. Seven days later, mesenteric lymph nodes were a low-dose injection of CY that was not, in itself, sufficient
removed and the dissociated cells were assayed for pro- to alter the immune response (Bovbjerg, Ader & Cohen
liferative reactivity in a mixed lymphocyte culture.10 In 1982). Although a complicated pattern of results was
contrast to animals treated with NRS and controls that observed, conditioned suppression and enhancement of
were injected with ALS after consuming plain water, rats the DTH response were found. When the original, un-
that had ALS paired with saccharin showed an aversion to modified paradigm was used, conditioned animals reex-
saccharin when tested 14 days after conditioning. Lym- posed to the CS showed suppression of the DTH re-
phocytes from conditioned animals reexposed to sac- sponse when challenged for the first time but did not
charin also displayed a decreased immunological reac- evidence enhanced reactivity to rechallenge. Groups that
tivity (Fig. 7). experienced a low-dose injection of CY before challenge
Conditioned enhancement of a cell-mediated response showed a marked suppression of the DTH, and no condi-
has been described by Gorczynski et al. (1982). This is an tioned immunosuppressive response could be detected
important study, not only because it extends the phe- when animals were challenged for the first time. Upon
nomenon of conditioned alterations in immunological rechallenge, however, conditioned animals reexposed to
reactivity by demonstrating enhanced rather than sup- the CS showed a significantly enhanced DTH response in
pressed responses, but because antigen itself served as comparison with conditioned animals that were not reex-
the unconditional stimulus. CBA mice were shaved, posed to the CS and a nonconditioned group that had
anesthetized, grafted with skin from C57BL/6J mice, and previously experienced noncontingent exposure to sac-
then left bandaged for 9 days. Peripheral blood leuko- charin and CY.

Ader & Cohen: CNS-imnuine system interactions
Preliminary data suggest that conditioned alterations component in the precipitation of allergic reactions (e.g.,
in DTH reactions may also be observed in human sub- Dekker et al. 1957).
jects. Based on the studies in animals and the report by These several studies using different antibody- and T-
Black, Humphrey, and Niven (1963) that hypnotized cell-mediated immune responses, different uncondi-
subjects instructed not to respond to tuberculin skin tioned stimuli, and different readout systems provide
testing showed attenuated DTH reactions, Smith and compelling evidence for the acquisition and extinction of
McDaniels (1983) explored the possibility that the DTH conditioned suppression and enhancement of immuno-
response in humans could be reduced using conditioning logical reactivity and reinforce the proposition that asso-
techniques. Seven volunteer subjects completed an ex- ciative processes are involved in the modulation of im-
perimental regimen in which they were subjected to mune responses.
tuberculin skin testing six times at monthly intervals. On
each of the first five trials, a nurse "blinded" to the
experimental protocol performed the skin testing under 4. Conditioning and the development of
precisely the same conditions each month. One arm was autoimmune disease
consistently treated with material (tuberculin) drawn
from a green vial; the other arm was consistently treated Although the effects of conditioning have been consistent
with material (saline) drawn from a red vial. On the sixth and reproducible, the magnitude of the changes in immu-
(experimental) trial, the contents of the vials were re- nological reactivity has been relatively small. What, then,
versed: tuberculin was applied to the arm that had been are the biological effects or implications of a small condi-
treated with saline and vice versa. Treatment with saline tioned change in the level of immunocompetence? To
did not elicit a skin reaction as might have been expected evaluate the biological impact of conditioned alterations
if there was a simple conditioning effect that could induce in immunocompetence, we chose an experimental situa-
a DTH reaction. However, when tuberculin was applied tion in which a diminution in reactivity would be in the
to the arm that had been previously treated with saline, survival interests of the organism. Systemic lupus ery-
there was a significant diminution in erythema and indu- thematosus (SLE) is an autoimmune disease12 for which
ration compared with the stable level of response re- the female New Zealand (NZBxNZW)F, mouse has be-
corded on the preceding control trials. Consistent with come a standard experimental model (Steinberg, Huston,
the results described by Soviet investigators in similar Taurog, Cowdery & Raveche 1981; Talal 1976; The-
experimental paradigms (Ader 1981c), Smith and ofilopoulos & Dixon 1981). These hybrid mice develop a
McDaniels were able to document a (conditioned) immu- lethal glomerulonephritis that can be delayed by treat-
nodepressed response. They were unable, however, to ment with cyclophosphamide(Hahn, Knotts, Ng& Ham-
detect any enhancement of immunological reactivity. It ilton 1975; Lehman, Wilson & Dixon 1976; Russell &
is possible, though, that the superimposition of a CS on Hicks 1968; Steinberg, Gelfand, Hardin & Lowenthal
the effects of subthreshold immunogenic stimulation, 1975). Based on our conditioning data described above, it
analogous to the subthreshold immunopharmacologic was hypothesized that the substitution of conditioned
stimulation used by Bovbjerg, Ader, and Cohen (1982), stimuli for immunosuppressive therapy in conditioned
might have yielded a potentiation of immunological reac- animals would delay the development of autoimmune
tivity. disease (Ader & Cohen 1982).
Histamine released from mast cells, besides being an Beginning at 4 months of age, individually caged
effector molecule of the immune system, may also mod- hybrid mice were provided with a saccharin drinking
ulate the response of other lymphoid cells. This has solution by pipette once each week for 8 weeks. Mice
been regarded as one of the self-regulatory mechanisms treated under a traditional chemotherapeutic regimen
in inflammation and immunity (Bourne, Lichtenstein, (Group C100) were injected with 30 mg/kg CY after each
Melmon, Henney, Weinstein & Shearer 1974). Russell, exposure to saccharin. As expected, the development of
Dark, Cummins, Ellman, Callaway, and Peeke (1984) proteinuria and mortality was delayed in these animals.
have recently provided evidence for a conditioned re- For Group C50, CY followed saccharin on only half of the
lease of histamine. Guinea pigs were sensitized to bovine weekly trials. Group NC50 received the same number of
serum albumin (BSA) and, 4 weeks later, began a series of saccharin and CY presentations, but in an unpaired (non-
weekly conditioning trials comprised of the random pre- contingent) manner. Group NC50 did not differ from a
sentation of an odor (CS + ) that was invariably paired with no-drug control group. Group C50, conditioned mice also
BSA on five of the trials and a different odor (CS~) that treated with half the amount of CY administered to Group
was always paired with saline on five other trials. Begin- C100, however, developed an unremitting proteinuria
ning two weeks later, the animals were reexposed only to significantly more slowly than untreated controls and
these odors on test trials administered at 2-week inter- significantly more slowly than nonconditioned animals
vals. Reexposure to the odor previously paired with treated with the same amount of drug (Fig. 8). The
antigen resulted in the release of histamine. The response differences are particularly striking when one considers
to the CS + was significantly greater than the response to the progression of disease using as a point of reference the
the CS ~, which in turn did not differ from the pretraining rate of development of proteinuria in the initial 50% of the
baseline. Histamine was also released in response to the population that developed SLE.
second test trial with the CS + odor, but the conditioned The mortality data (Fig. 9) also revealed overall group
response was markedly attenuated, perhaps as a result of differences. Again, since all these mice could be expected
the previously unreinforced presentation of this odor. to develop SLE and die, the longer one follows the
Such results are consistent with the clinical observations progression of disease the less likely it is that one will
and hypotheses regarding a psychosocial or conditioning discern treatment effects. An analysis of these data in

100-i co
100-i
i C 100
so -
5
2
H. 60-
fc 60-1
Uj
o (3
UJ
CD • C 100% Q NO TRT
40-
A C 50% 20- ACS
o' O CS + US
I
& NC 50%
20- O CONTROL
10 20 30 40 50 60
WEEKS (POST-CONDITIONING)
10 20 30 40 50 60
Figure 10. Mortality rate in (NZBxNZW)F, female mice treat-
WEEKS (POST-CONDITIONING) ed with saccharin and CY weekly and then continued on a
regimen of saccharin and CY (CS + US), continued on placebo
Figure 8. Rate of development of an unremitting proteinuria medication alone (CS), or deprived of both saccharin and CY (No
in (NZBxNZW)F, female mice under different chemotherapeu- treatment). Reprinted from Ader (1985) by permission from
tic regimens. Croup C100 received saccharin followed by an Guilford Press.
injection of CY weekly; Group C50 received saccharin weekly
and CY on 50% of the saccharin trials; Group NC50 received the
same saccharin and CY as Group C50 except that saccharin and
CY were not paired in time; control mice received saccharin
period of chemotherapy, the C100, C33, and NC33
weekly but were not injected with CY. Reprinted from Ader and groups were divided into subgroups that (a) continued to
Cohen (1982) by permission from the American Association for receive saccharin and CY on the same schedule as before,
the Advancement of Science. (b) received saccharin but no CY, or (c) received neither
saccharin nor CY. As hypothesized, unreinforced presen-
tations of the CS influenced the development of autoim-
terms of the rate at which a 50% mortality was reached mune disease in conditioned but not in unconditioned
yielded the same results as the proteinuria data. Noncon- mice. In animals "trained" under a schedule of continu-
ditioned animals did not differ from untreated animals, ous reinforcement, mice that continued to receive CS
whereas conditioned mice treated with the same amount presentations (saccharin plus ip injections of saline) after
of drug survived significantly longer than untreated con- the termination of CY therapy died at a slower rate than
trols and nonconditioned mice. In terms of mortality, animals deprived of both saccharin and CY. Moreover,
Group C50 did not differ statistically from Group C100, the mortality of mice that continued to receive weekly
animals that received twice as much CY. exposure to the CS without CY did not differ from that in
Chemotherapy in this study consisted of a series of mice that continued to receive both saccharin and CY
weekly conditioning trials on which presentation of a CS (Fig. 10).
was pharmacologically reinforced by CY either 100% or Among the animals that were originally treated with
50% of the time. Preliminary data on extinction were CY on only i of the weeks during which saccharin was
obtained in a second experiment that was, during the presented, conditioned mice that were deprived of CY
initial period of therapy, identical to the first, except that but continued to receive saccharin died more slowly than
the experimental animals (Group C33) received CY after mice deprived of both saccharin and CY and, at least for
only 4 of the weekly saccharin trials. Following the initial the initial half of these subgroups, the mortality rate of
mice that received only the CS did not differ from that of
mice continuing on drug treatment. In contrast, con-
100-i tinued exposure to saccharin in unconditioned mice was
without effect; these animals did not differ from animals
that received neither saccharin nor CY (Fig. 11). It was
not possible to assess resistance to extinction as a function
of reinforcement schedule in this experiment because of
the difference in the cumulative amount of drug received
by mice under the continuous and partial treatment
regimens. Such a study, however, is currently in prog-
ress.
Even though these data must be considered prelimi-
nary, the (conditioned) modifications in the onset of
autoimmune disease in (NZBxNZW)F! mice are con-
30 40 SO 60 sistent with the immunological effects of conditioning
WEEKS (POST-CONDITIONING) described above and the hypothesis that the conditioning
of immunosuppression would delay the onset of SLE
Figure 9. Cumulative mortality rate in (NZBxNZW)F1 mice
maintained under different chemotherapeutic regimens. Group
under a regimen of chemotherapy that was not in itself
designations are the same as in Figure 8. Reprinted from Ader sufficient to influence the development of autoimmune
and Cohen (1982) by permission from the American Association disease. Confirmation of these results is provided by a
for the Advancement of Science. Copyright 1982 by the AAAS. recent study by Gorczynski, Kennedy, and Ciampi

100 n 9"
C33
8-
60-
V)
* 20-
O
Uj
9 lOOn
NC33
D NO TRT
4 CS
20- O CS US PLACEBO NC CS 0 CS
4.-1,
CONDITIONED
10 20 30 40 50 60
Figure 12. Hemagglutinating antibody titers (mean ± SE)
WEEKS (POST-CONDITIONING) obtained 6 days after immunization with SRBC in rats condi-
tioned with LiCl as the UCS. P, placebo-treated animals; NC,
Figure 11. Mortality rate in conditioned (C33) and noncondi-
nonconditioned animals; CSo, conditioned animals that were
tioned (NC33) (NZBxNZW)F, mice that continued to receive
saccharin and CY (CS + US), continued to receive only sac- not reexposed to the CS; CS, conditioned animals that were
charin (CS), or received neither saccharin nor CY (No treat- reexposed to the CS; US, conditioned animals injected with
ment). Reprinted from Ader (1985) by permission from Guilford LiCl at the time of immunization. Reprinted from Ader and
Press. Cohen (1975) by permission from Elsevier North-Holland, Inc.
(1985). In this instance, repeated reexposures to a CS tioned stimuli to yield a conditioned adrenocortical re-
previously associated with CY accelerated tumor growth sponse (Ader 1976). However, in our protocol, LiCl does
and mortality in response to challenge with a syngeneic not suppress the antibody responses to SRBC, and rats
plasmacytoma. These results, then, constitute an elab- conditioned with LiCl instead of CY do not show an
oration of the biological impact of conditioned immuno- attenuated immune response (Fig. 12). To determine
pharmacological responses. Our results also indicate that whether an elevated corticosterone level superimposed
there may be considerable heuristic value in viewing upon the residual immunosuppressive effects of CY could
some pharmacotherapeutic regimens as conditioning par- account for the conditioned suppression of the antibody
adigms. For example, the introduction of partial sched- response, two experiments were conducted in which
ules of pharmacotherapeutic reinforcement may enable subgroups of animals conditioned with CY were injected
the physician to maintain some physiological state within with LiCl or with corticosterone instead of being reex-
normal limits with a lower cumulative amount of active posed to the CS. While antibody titers were lower in
drug than is currently being used. The definition of the conditioned animals reexposed to the CS than in controls,
"placebo effect" as a learned response and the implica- groups that received LiCl or corticosterone did not differ
tions of a conditioning model of pharmacotherapy are from nonconditioned animals or conditioned animals that
elaborated elsewhere (Ader, 1985). were not reexposed to the CS. These experiments, then,
provided no support for the hypothesis that the condi-
tioned immunosuppressive response was mediated by an
5. Adrenocortical mediation of conditioned experimentally induced elevation in glucocorticoids.
immunopharmacological effects Antibody production in response to antigenic stimula-
tion is the net result of a complex chain of events that, at
It has been argued that conditioned alterations in immu- any of several levels, might be influenced by neural and
nological reactivity actually represent adrenocortically endocrine changes that accompany humoral responses
mediated "stress" effects rather than conditioning pro- (e.g., Besedovsky & Sorkin 1981), neuroendocrine
cesses (Dwyer 1983; see also Bovbjerg et al. 1983). In changes that accompany conditioned responses (e.g.,
contrast, we have argued (Ader & Cohen 1975) that these Harris & Brady 1974), or neuroendocrine responses that
are conditioning effects that might be mediated by are themselves conditioned responses (e.g., Ader 1976).
changes in corticosterone levels. The experiments de- The neuroendocrine changes that occur in response to
scribed below were designed to evaluate this possibility antigenic stimulation, however, are common to all the
(Ader & Cohen 1975; Ader et al. 1979) and failed to animals in the conditioning experiments, and we know of
support the hypothesis. no data on the long-term neuroendocrine changes result-
Both LiCl and CY are effective stimuli for inducing a ing from a single, brief exposure to a CS. The adrenocor-
taste aversion, both unconditionally elicit elevations in tical response to brief environmental stimulation, howev-
corticosterone level, and both can be used as uncondi- er, is relatively short-lived (Ader & Friedman 1968). It

would be difficult, therefore, to attribute an attenuated immune network governing immunoregulatory pro-
antibody response to the persistent effect of a conditioned cesses included the observation that certain hormonal or
elevation in plasma corticosterone level when reexposure neurotransinitter manipulations may exert their effects in
to the CS occurred several days before antigenic stimula- vivo only within a host that has a functioning adrenal
tion (Ader et al. 1982). One could still hypothesize that gland (e.g., Besedovsky, del Rey, Sorkin, Da Prada &
conditioned neuroendocrine changes exert their effects Keller 1979) or that certain antibody responses may be
on one and/or another subpopulation of B- and/or T-cells detected in vitro only when steroids are added to the
or macrophages and that such effects are sufficiently long- culture medium (Ambrose 1970). If endogenous cor-
lasting to influence a response to subsequent iinmu- ticosteroids play some "permissive" role in iimnu-
nogenic stimulation. noregulation, interpretations of the immunologica) ef-
It should be noted, though, that Gorczynski et al. fects of conditioning (or other neuroendocrine manipula-
(1984) were unable to observe a conditioned iminunosup- tions) in adrenalectomized animals must be advanced
pressive response in adrenalectomized mice. As men- with caution. It is tempting to speculate, however, that,
tioned previously, these experiments were conducted on in the absence of maintenance steroid therapy, adre-
group-housed animals, so there are no data to verify that nalectomy, in addition to abolishing the adrenal meduF
any association had been made between saccharin and CY lary influences on immunological reactivity, would have
- or how many animals in the group were actually resulted in chronically elevated ACTH levels and that
conditioned.l3 There is, too, the potential confounding of ACTH and ACTH peptide fragments could be operating
altered sensory thresholds that can accompany adrenal to influence immunocompetence. That such substances
insufficiency (Henkin 1970) and that might modify the may act as vehicles of communication between the brain
discriminability of the CS and/or the UCS (and thus the and the immune system is implied by the observations
strength of any conditioned behavioral or immunophar- that lymphocytes are capable of producing immunoreac-
macological response). Assuming, however, that the ef- tive ACTH-like material (Blalock 1984; Smith, Meyer &
fects of adrenalectomy can be taken at face value, these Blalock 1982).
data are interesting in relation to the results obtained by The situation with respect to the role of adrenocortical
varying the time of day at which conditioning occurs. steroids in mediating the effects of conditioning on cell-
Mice were maintained on a 12-hr light-dark schedule mediated responses is less clear because only indirect
with light beginning at 6:30 A.M. Animals conditione it evidence of relevance is available. The effects of glucocor-
7:00 A.M. and, to only a slightly lesser degree, at 1:00 P.M. ticoids on cell-mediated responses have been considered
showed conditioned immunosuppressive responses, minimal (Ahlqvist 1981). In the case of a GvH response,
whereas animals conditioned at 7:00 P.M. did not for example, the reacting donor cells are resistant to
(Gorczynski et al. 1984). On this light-dark regimen, corticosterone effects, but treating recipient animals with
mice conditioned at 7:00 A.M. were manipulated during hydrocortisone during the response does reduce spleno-
the trough in the 24-hour corticosterone rhythm, and megaly (Cohen & Claman, 1971). Plasma corticosterone
animals conditioned at 7:00 P.M. would have been near levels were not measured in the Bovbjerg et al. (1982)
the peak in the daily corticosterone rhythm. That is, study, but animals were reexposed to the CS in a prefer-
steroid levels would have been low in animals condi- ence testing procedure that does not elevate steroid
tioned at 7:00 A.M., high in animals conditioned at 7:00 levels (Smotherman, Hennesey & Levine 1976). In our
P.M., and between these extremes in animals conditioned study of extinction of the GvH response (Bovbjerg et al.
at 1:00 P.M. To the extent that adrenocortical steroids 1984), two placebo groups were used: one was manipu-
were involved in the conditioning of immunosuppres- lated in the same manner as the experimental group that
sion, this relationship between steroid levels and the received the maximum number of extinction trials (i.e.,
conditioned immunosuppressive response would appear manipulations that would have elevated steroid levels),
to be contrary to the effects of adrenalectomy. One could whereas the other remained totally unmanipulated. The
still speculate that, since adrenocortical reactivity is fact that the GvH response of these two groups did not
greater at the trough than at the peak in the daily cycle differ is not in keeping with the hypothesis that an
(Ader & Friedman 1968), some component of the pitui- elevation in steroid levels is a critical factor for the
tary-adrenal axis was involved in the positive effects seen expression of immunological reactivity in these experi-
in mice conditioned early in the light-dark cycle. By the ments.
same token, control animals (e.g., animals that experi- "Stress" in the form of physical restraint, suppresses
ence noncontingent CS and UCS presentations, han- DTH whether it is administered before sensitization or
dling, and ip injection of antigen) would also be more before challenge with SRBC (Blecha, Barry & Kelley
reactive at the trough than at the peak in the daily 1982). Since restraint elevates corticosterone level and
adrenocortical cycle, and there is no reason to believe adrenalectomy and metyrapone block the stress-induced
that there would be a differential elevation in cor- suppression of immunological reactivity under precisely
ticosterone level between conditioned and noncondi- these conditions (Blecha, Barry & Kelley 1982), the
tioned animals, that is, that a conditioned elevation of conditioned enhancement of a DTH response observed
steroid level would exceed the unconditioned adrenocor- by Bovbjerg (1983) would appear to be inconsistent with
tical response to the manipulations imposed upon control the effects that would be expected to occur if a stress-
animals in order to assess immune function. induced or conditioned elevation of adrenocortical
The data obtained by Gorczynski et al. from control steroids were the mediating physiological event. Unless
animals did not suggest any alterations in immunocompe- one hypothesizes that a difference in the magnitude or
tence in the adrenalectomized mice. Nonetheless, a part duration of the restraint-induced and any conditioned
of the evidence for an integrated neural-endocrine- elevation in steroid level would result in a selective effect

of conditioning on steroid-sensitive T-suppressor cells, a The vast majority of the research on hormones and
conditioned elevation of corticosterone level might actu- immunity has concentrated on the thymus (Comsa et al.
ally have acted to reduce the magnitude of the condi- 1982). Because of its central role in the ontogeny of
tioned response. immunocompetence, the neuroendocrine functions of
It is also reasonable to infer that the manipulations the thymus have not, until recently, received the atten-
imposed by Gorczynski et al. (1982) in grafting allogeneic tion they apparently deserve (Hall & Goldstein 1983).
skin tissue would have elicited an adrenocortical re- Conversely, hormones other than those of the thymus
sponse. These glucocorticoid elevations, however, were have not received sufficient attention as modulators of
evidently insufficient to obviate a conditioned enhance- immune function. Considering the fact that endogenous
ment of immunological reactivity as reflected in eleva- levels of glucocorticoids have been associated with both
tions of CTLp. As in the studies cited above, control suppression and enhancement of immunological reac-
animals in the study by Kusnecov et al. (1983) also tivity, any generalization regarding a mediational role of
received the same manipulations as the experimental adrenocortical steroids in the conditioning of immune
animals that displayed conditioned suppression of a cell- responses would be premature and counterproductive
mediated response to rabbit antirat lymphocyte serum, with respect to the possibilities that have yet to be fully
and, according to unpublished data from their laboratory, evaluated - or even identified.
conditioned immunosuppression using ALS was not asso-
ciated with an increase in plasma corticosterone level.
Based on these studies and the larger literature on the 6. Concluding remarks
effects of stress on immune function, it is apparent that
there is a dissociation between alterations in immunologi- If one adopts the position that immunology is confined to
cal and adrenocortical activity. The available data provide the study of the cellular and molecular mechanisms by
no support for the hypothesis that conditioned alterations which the organism is capable of recognizing what is a
in humoral immunity are mediated by nonspecific or part of itself and what is not a part of itself and, in the latter
conditioned elevations in adrenocortical steroids. Al- instance, defending itself against what is foreign, none of
though the evidence is less direct, it also appears that the material reviewed in this paper can be construed as
conditioned alterations in cell-mediated immune re- "immunology." Indeed, much has been learned about
sponses are not simply due to changes in steroid level. immunological defense mechanisms by in vitro analysis of
Since different subsets of regulatory T-cells (helpers and immune reactions, and the application of these principles
suppressors) appear to be involved, for example, in the has virtually eliminated- certain diseases. This may be
DTH response to SRBC and in contact hypersensitivity only part of the story, however. While the achievements
reactions (Huber, Devinsky, Gershon & Cantor 1976; of the classic biomedical approach are legion, it has been
McKenzie & Potter 1979), stress, conditioning, or other argued that the strategies handed down by Koch,
behavioral manipulations may exert different effects on Pasteur, and their followers do not apply to all disease
different subpopulations of T lymphocytes rather than (or processes (e.g., Dubos 1959). Dixon (1978) has pointed
in addition to) affecting endocrine (e.g., adrenocortical out that "the dazzling achievements of specific etiology
steroid) responses that are common responses to a wide have now been followed by a situation where all our major
range of environmental circumstances. health problems - most obviously cardiovascular disease,
We do not mean to exclude direct or indirect participa- cancers, and much mental illness - represent areas where
tion of neuroendocrine processes in conditioned altera- the theory has failed." In another context, Smith and
tions of immune processes. Indeed, growing evidence for Steinberg (1983) have stressed the futility of seeking
the immunomodulating effects of a variety of hormones single causes for the multifactorial etiology of the disor-
and neurotransmitter substances provides myriad candi- dered immunoregulatory processes that culminate in
dates for the mediation of behaviorally induced altera- autoimmune diseases.
tions in immune function. Steroid hormones, including There is an alternative to the classic biomedical model
glucocorticoids, androgens, estrogens, and pro- to which the majority of immunologists or immunological
gesterone, are generally immunosuppressive, whereas research seems to have subscribed, and that is to view and
growth hormone, thyroxine, and insulin have been study immunological phenomena as part of an integrated
shown to potentiate immune responses (Ahlqvist 1981; physiological system. Such a position does not deny the
Comsa et al. 1982). These reviews indicate that the effects self-regulating capacities of the immune system; it com-
of hormones and neurotransmitters depend upon a
plements them by adding levels of organization that
number of factors, including the dose of hormone or the
neurotransmitter agonist or antagonist, the timing of acknowledge that the adaptive functions of the immune
administration in relation to immunization, the dose of system of ultimate significance are those that occur in vivo
antigen, and the parameter of immunological reactivity as part of an integrated network of adaptive processes,
under consideration. Pharmacological doses of glucocor- each of which is capable of influencing and being influ-
ticoids, for example, may be immunosuppressive, but the enced by others.
in vivo effects of endogenous adrenocortical steroids are Certainly, it is easier to analyze the mechanisms under-
less clear; and while many of the effects of stress are lying immunity by reducing complex processes to simpler
immunosuppressive, different stressors have different or more discrete elements. Such analyses are essential for
effects and such effects are not always attributable to an understanding of how the components of the immune
elevated steroid levels (Blecha, Kelley & Satterlee 1982; system work. There is a danger, however, of losing sight
Keller et al. 1983; Monjan 1981; Monjan & Collector of the original questions - in forgetting that such reac-
1976). tions are a part of the whole and that the test-tube
medium in which immune reactions can be made to take

place is not the same as the medium in which they take reciprocal monitoring and modulation of neuroendocrine
place in the living organism. It is inevitable, then, that and immune responses. Bovbjerg, Cohen, and Ader
immunity will come to be studied in relation to other (1982) have discussed the implications of conditioned
physiological processes in much the same way that hor- alterations in immunological reactivity in terms of "feed-
monal responses have come to be studied as part of an forward" regulatory mechanisms that characterize other
organized neuroendocrine system. anticipatory physiological and pharmacological responses
The study of behavior is also an integral part of biology, (e.g., Eikelboom & Stewart 1982). Admittedly, the CNS
and the reciprocal relationship between behavior and mechanisms underlying feedforward regulatory pro-
physiological function apparently extends to immune cesses (of which conditioning is a primary example) are
function. As noted above, behavioral factors contribute to poorly understood. The role of conditioning and the
the individual's susceptibility or response to infectious adaptive significance of feedforward regulation in deter-
disease. Conversely, infectious disease can alter behavior mining some of the behavioral and physiological effects of
(e.g., McFarland & Hotchin 1983). Are these effects (in pharmacological stimulation, for example, have been
either direction) mediated by immune changes? Are the described (e.g., Siegel, in press). The contribution of
behavioral influences on disease susceptibility, for exam- conditioning to the readiness of the organism to respond
ple, mediated by the immunological changes that accom- or to the rapidity of immune responses - or, even, to the
pany affective responses such as depression? In hind- existence of compensatory (paradoxical) responses that
sight, immune function is conspicuously absent from modulate immunity - remains to be studied.
psychobiological studies of the emotions [see Panksepp: Because alterations in the state of the central nervous
"Toward a General Psychobiological Theory of Emo- system, including those that accompany learned re-
tions" BBS 5(3) 1982]. The relationship between behav- sponses, can influence immune function, and alterations
ioral disorders and immune function (e.g., Solomon of immune status can influence behavior, one might also
1981b) represents another point of departure for studying ask whether a deviation from homeostasis at the immuno-
brain, behavior, and immune system relationships. It has logical level is sufficient to motivate behavior, that is,
been suggested (e.g., Pierpaoli, Kopp, Muller & Keller could it serve as a stimulus to support learning in order to
1977) that there is a reciprocal relationship between the correct a homeostatic imbalance? Miller, DiCara, and
ontogeny of neuroendocrine and immune function. The Wolf (1968) have hypothesized that in those cases in
impact of early life experiences on subsequent psycho- which homeostasis is mediated by central nervous system
physiological function might profitably be extended to mechanisms, sufficiently large deviations from homeo-
the development and level of immunocompetence seen stasis might function as a drive and the restoration of
in adult organisms (Ader 1983). The relationship between homeostasis might function as a reward. Granted that
the major histocompatibility complex and behavior sufficiently large deviations from homeostasis may occur
(Boyse, Beauchamp & Yamazaki 1983) suggests interest- only under abnormal circumstances; nevertheless, a dis-
ing directions for studies in behavioral genetics. Although ruption of immunoregulatory circuits such as those that
the literature in most of thesefieldsis not yet very large, a characterize autoimmune disorders (Smith & Steinberg
review of this material is well beyond the scope of this 1983) might, in view of the data reported above, con-
paper. stitute an appropriate model in which to examine the role
The present discussion has been limited to the growing of learning in the maintenance of homeostasis and the
number of studies that have documented conditioned impact of behavioral processes of adaptation on levels of
alterations in immunity. The fact that learning processes immunocompetence.
can modify immune responses provides dramatic evi- Behaviorally conditioned alterations in immunological
dence that the central nervous system may play a role in reactivity represent only one of several converging lines
the modulation of immunity. Since the immune system is of evidence implicating the central nervous system in the
a part of an integrated physiological system, these studies modulation of immunity. Speaking only of this area, a
can be viewed as extensions of the voluminous literature delineation of the effects of conditioning - the circum-
on conditioned physiological responses. We know less stances under which conditioning may attenuate or am-
about the neuroanatomic and neurochemical connections plify nonspecific defense reactions and specific immune
between the brain and the immune system than we do responses and the multiple pathways by which condition-
about the innervation of other physiological systems, but ing may modulate immunity - has not yet been achieved.
that is a quantitative distinction that additional research Considerably more data on the unconditioned immu-
will remedy. Indeed, the conditioned modulation of nomodulating effects of hormones and neurotransmit-
immune responses should stimulate the search for path- ters, for example, would be required in order to examine
ways capable of serving as channels of communication just the one hypothesis that subsets of lymphocytes are
between the brain and the immune system. It should also differentially sensitive to the neuroendocrine changes
prompt more serious consideration of experimental para- that accompany conditioning or are themselves condi-
digms for assessing hypotheses about the similarities and tioned responses to environmental, pharmacological, or
dissimilarities in the mechanisms underlying immunity immunogenic stimuli. Although we cannot yet specify the
and brain function, particularly learning (e.g., Cohn pathways involved in conditioned alterations of immuno-
1968). logical reactivity, we can point to a literature that is
As mentioned at the outset, some of these channels of providing a growing list of potential candidates. Despite
communication have already been identified. Be- the current inability to specify the underlying mecha-
sedovsky, del Rey, and Sorkin (1983) have described what nisms, we are suggesting that new research strategies and
the brain and the immune system know about each other new advances are likely to emerge from viewing immune
in terms of feedback networks that could provide for processes as an integrated part of psychophysiological

Commentary I Ader & Cohen: CNS-immune system interactions
function. Adapting the strategies of psychobiological re- within a species. Histocompatibility antigens that elicit graft
search to an understanding of integrated immune func- rejection are examples of alloantigens.
tion will increase the armamentarium of the immu- 10. In the mixed leukocyte culture reaction, lymphocytes
nologist, and a greater understanding of the complexity of from two histoincompatible animals are cocultured for several
immune processes will extend the reach of psycho- days. The ensuing proliferation of T-cells is quantified by scin-
biological research. tillation spectrometry of the cultures that were pulsed with
tritiated thymidine for several hours prior to the termination of
the culture period. Proliferation reflects the recognition of
ACKNOWLEDGMENTS foreign histocompatibility alloantigens by T-cells that do not
Preparation of this manuscript was supported by a Research themselves display the same antigens. If lymphocytes from one
Scientist Award to R.A. (K05-MH06318) from the National of the animals are prevented from proliferating, then the thy-
Institute of Mental Health and by research grant NS-15071 midine incorporation reflects proliferation of cells from the
from the National Institute of Neurological and Communicative animal that provided the responder cells (i.e., the animal that
Disorders and Stroke, USPHS. was treated with ALS).
We are indebted to Shmuel Livnat and Lee Grota for their 11. DTH reactions are inflammatory responses that are ini-
critical comments on earlier versions of this manuscript and on tially mediated by T lymphocytes. They are measured by a local
our response to the commentators. skin reaction that occurs 24-48 hours after the cutaneous chal-
lenge with an antigen to which the animal had previously been
NOTES immunized (sensitized). In the experiments of Bovbjerg(1983),
1. Natural killer (NK) cells are large granuloleukocytes that mice were immunized with SRBC and challenged some time
can kill target cells, particularly tumor cells, without previous later in the footpad with SRBC. DTH was measured by record-
sensitization (and by poorly understood mechanisms). The lin- ing the thickness of the footpad.
eage(s) of these NK cells is not clear. 12. Autoimmune diseases such as SLE are the consequence
2. By definition, and antigen is a molecule that stimulates of cellular and/or antibody-mediated immune reactions di-
immune responses (e.g., antibodies) by activating only those rected against self-antigens.
lymphocytes that bear surface receptors for that antigen. Like 13. In subsequent experiments, Gorczynski (personal com-
surface receptors, the antibodies that are elicited will react only munication) has established that adrenalectomized mice are
with the antigen that induced their production. Such reactions capable of acquiring a taste aversion based on the experimental
are referred to as immunologically specific. In contrast, anti- paradigm used in his laboratory.
genic as well as nonantigenic materials can also elicit defense
responses characterized by the production of nonantibody hu-
moral factors. These factors (e.g., chemotactic, mitogenic, lytic,
macrophage-activating) interact with a variety of leukocytes in a
nonantigen-specific way and thereby effect elimination of any
foreign material that happens to be in the vicinity. Such reac-
tions are referred to as "nonspecific."
3. These data were reviewed by Clark Hull in 1934, but they
evidently had no impact on studies of conditioning in the United
Open Peer Commentary
States.
4. Food intake was not measured but, after adaptation to the Commentaries submitted by the qualified professional readership of
drinking schedule, all animals gained weight during the course this journal will be considered for publication in a later issue as
of the experiment. Continuing Commentary on this article. Integrative overviews and
5. We do not yet know which of the direct and/or indirect syntheses are especially encouraged.
effects of CY that are associated with a suppression of immune
function were responsible for establishing the connection be-
tween CS and UCS. As will be seen in the following literature
review, however, it is not necessary to use a drug as the UCS to
effect a conditioned change in immunological reactivity. Brain and the immune system: Multiple sites
6. Rather than measure serum antibody titers, one can enu- of interaction
merate individual antibody-forming lymphocytes in a plaque
assay. Immune lymphocytes from SRBC-immunized animals Hymie Anisman and Robert M. Zacharko
are incubated with the antigen and complement in a semisolid Department of Psychology, Carieton University, Ottawa, Ont., K1S 586,
supporting medium (e.g., agar). A clear zone of hemolysis (i.e., Canada
a plaque) occurs around each antibody-releasing cell and these
plaques can be counted. Peak PFC responses occur before peak It is curious that although environmental events are often
serum antibody titers can be detected. thought to contribute to disease processes through their effect
7. Freund's complete adjuvant contains killed Mycobac- on the central nervous system (CNS) (e.g., stressors increasing
terium tuberculosis incorporated in a water-oil emulsion. vulnerability to illness), hard data are often greeted with skep-
8. In a GvH reaction, grafted T lymphocytes recognize histo- ticism. It seems that the data may sometimes be less compelling
compatibility alloantigens on cells of the host (but not vice than our intuitions! Nevertheless, in recent years the traditional
versa). In the local GvH reaction in which donor cells are view that immunological and CNS processes are independent of
injected into the plantar surface of the hind foot, this recognition one another has been repeatedly challenged. As indicated by
of nonself results in proliferation and recruitment of donor and Ader & Cohen (A & C), manipulations of CNS activity have been
host cells in the regional draining lymph node (popliteal node). shown to influence immune functioning, and conversely, anti-
This is quantified by comparing the weights of the lymph node genic stimulation may affect central neuronal activity (e.g.,
that drains the site of injection and the contralateral node. Besedovsky, Sorkin, Felix & Haas 1977; Besedovsky, del Rey,
9. Alloantigens are antigens obtainedfromone individual (or Sorkin, Da Prada, Burri & Honegger 1983; Stein, Schleifer &
inbred line) that will incite a specific immune reaction when Keller 1981). The work conducted by A & C has not only been
they are introduced into another individual (or inbred line) of fundamental in strengthening the view that CNS activity and
that same species. Alloantigens reflect genetic polymorphism immunological processes may be reciprocally modulating sys-

Commentary I Atler & Cohen: CNS-immune system interactions
turns, but it has also revealed that environmental stimuli are The seven veils of Immune conditioning
subject to conditioning and may thus influence immune status.
These findings are of obvious heuristic value, and they have R. E. Ballieux
important practical applications as well. For instance, the dem- Department of Clinical Immunology, University Hospital, 3511 GV Utrecht,
onstration that conditioned stimuli may have an impact on The Netherlands
autoimmune disorders in mice raises the possibility that similar
processes may be operative in humans. Thus, when fairly toxic C. J. Heijnen
agents are used to induce an immunosuppressant effect, it may Department of Immunology, University Children's Hospital, 3512 LK
be most efficacious to use the conditioned stimuli to engender or Utrecht, The Netherlands
maintain the immunosuppression, without the necessity of For a long time research in the Held of immunology has been
continuously using the immunosuppressant compound itself. directed toward understanding the intricate processes of activa-
Given that environmental events may influence immune tion and regulation of the immune system. The information
functioning and hence disease processes, it is necessary at this obtained has given insight into the structure and function of cells
juncture to determine those conditions that influence vul- of the immune system, the genes governing the synthesis of
nerability to immunological changes. In their target article, A & specific antibodies, the major histocompatibility gene complex
C indicate that although the conditioned immunosuppression and its products regulating cell interactions, and the diseases
has been independently verified in several laboratories, the that develop as a result of a dysfunction of the immune system.
magnitude of the effect is quite small and, I assume, variable. A The science of immunology has until recently been mainly self-
similar outcome was reported in the analysis of stressor effects oriented. As Ader & Cohen (A & C) state in the introduction of
on immune functioning. For instance, in their exciting report, their well-written and stimulating target article, most iminu-
Gamzu, Vincent, Tare, Benjamin, Farrar & Sullivan (1984) nologists studying immune regulation emphasize the "internal
indicated that various stressors and social housing conditions character" of the various processes involved. Indeed, T-helper
influenced plaque-forming cells in the spleen and provoked a and T-suppressor lymphocytes, which play an important role in
decline of the mitogen response. However, these investigators regulating immune responses, are part of the immune system
emphasized that such effects were not always evident, and they itself. Nevertheless, we hesitate to support A & C's view that
were unable to isolate the variables that accounted for the immunologists in general adhere to the concept that the im-
between-experiment differences. These findings may be in- mune system is autonomous.
terpreted to suggest that the phenomenon is not a particularly There is increasing awareness of the fact that in the living
reliable one and probably has little bearing on pathology. creature the immune system is integrated with other physiologi-
Alternatively, the position might be entertained that the orga- cal systems like the brain and the endocrine system. One of the
nism is a fairly adaptable one, and even in the face of appreciable main reasons, however, that studies in the interdisciplinary area
insults immune functioning will not necessarily be compro- of behavior sciences, neuroendocrinology, and immunology are
mised. However, given an appropriate environmental, experi- few, is the complexity of the systems involved. To paraphrase
ential, or organismic backdrop, immunological disturbances Sir P. B. Medawar, this involves working very close to the
may be provoked by stressors, and hence pathology may arise. frontier between bewilderment and understanding. Thus the
Unfortunately, in the case of both conditioned immune re- studies of the Rochester group are important because they are
sponses and stress effects on immunological functioning, insuffi- based on firm knowledge of behavior as well as immunology.
cient data are currently available concerning the conditions that
Furthermore, the experimental protocols are carefully designed
permit expression of particular variations in response to en-
vironmental challenges. so as to reduce the problem to be analyzed. But even under
these conditions the results obtained are difficult to explain as
Assuming that the immune system is, in fact, modulated by regards the biological processes involved in conditioning of the
CNS activity, it is clear that advances in psychoneuroim- immune response.
munology necessitate the concurrent assessment of the effects of One important point should be mentioned. The immune
environmental events on immune processes, central neuro- system consists of specialized cells that occur in organized,
transmitters, and hormones. Data are available indicating that lymphoid tissues and which, in addition, form a pool of cells
hypothalamic manipulations will influence immune system ac- (re)circulating in the lymph, blood, and interstitial fluid. These
tivity. Moreover, it has been demonstrated that psychological cells can be considered as part of a network that is in dynamic
and physical injury influence neurotransmitter activity in the equilibrium, regulated by a number of activating and sup-
hypothalamus as well as several other brain regions (Anisman pressive forces.
1984). Indeed, it appears that cues associated with aversive Some of the regulatory signals are antigen-specific, whereas
stimulation not only influence immune functioning as reported others are not. It seems as if conditioning, as an immunoregula-
by A & C, but may affect the activity of central norepinephrine tory mechanism, is operating at the nonspecific level. This
(Anisman & Sklar 1979; Cassens, Roffman, Kuruc, Orsulak & assumption is supported by the observation that exposure to a
Schildkraut 1980), dopamine (Herman, Guillonneau, Dantzer,
conditioned stimulus before immunization can still lead to
Scatton, Semerdjian-Rouquier & Le Moal 1982), acetylcholine
(Hingtgen, Smith, Shea, Aprison & Gaff 1976), andendorphins immunosuppression. As A & C state, the implication of this
(Chance, White, Krynock & Rosecrans 1978). It is necessary at observation is that perturbation of the immune network by
this point to determine whether those environmental variables antigen is not a requirement for conditioning to occur. This view
that maximally influence neurochemical and hormonal respon- no longer seems tenable when antigen is used as the uncondi-
sivity likewise have the greatest influence on the immune tioned stimulus. In the paradigm described by Gorczynski,
response. In addition, given that the effects of environmental Macrae, and Kennedy (1982), mice were skin allografted. In this
stressors or cues associated with stressors influence neu- situation the foreign alloantigen represents the unconditioned
rochemical activity in a number of brain regions, it is necessary stimulus, whereas the grafting procedure serves as the condi-
to evaluate the contribution of different neuronal pathways on tioned stimulus. Gorczynski and his coworkers were able to
immune functioning. In effect, at this juncture less time needs show a "conditioned" increase in frequency of antigen-specific
to be devoted to the demonstration that immunological respon- cytotoxic T lymphocyte precursors (CTL ) in the peripheral
sivity is modifiable by environmental stimuli, and more atten- blood circulation after skin grafting mice that had been exposed
tion must be devoted to the central mechanisms subserving to these conditioning trials. Although the conditioned response
these immunological changes. clearly was alloantigen-specific, the increase in frequency of
circulating CTL )|t,t. could still be based on a nonspecific mecha-

Commentary I Ader & Cohen: CNS-iinmune system interactions
nism. It is conceivable that the perception of the conditioned blood cell counts, sterility, anaemia, cystitis, marrow depres-
stimulus results in a signal, mediated by the nervous system, to sion, and other unpleasant symptoms, including an intense
lymphoid tissues that may lead to release into the circulation of experience of sickness. (Incidentally, it would be most interest-
(memory) CTL|)rt.t. from the expanded allospecific clone. If true, ing to see if some of these other effects are conditioned along
the presumed antigen specificity in the conditioned response is with the immune depression.) Such broad toxicity provides
merely a reflection of pool size of reactive lymphocytes. opportunity for indirect effects on immune function, for exam-
If one accepts the view that conditioning alters immune ple, by alteration of blood flow through lymphoid organs.
reactivity in a nonantigen-specific fashion, how then is the effect Whether or not one calls this "stress" seems of semantic impor-
mediated? AfitC present evidence that conditioned changes in tance only - the underlying question concerns the mechanism
immune response are not simply due to changes in adrenocor- of the immunosuppression and how specific or direct its trigger-
tical activity. As they point out, however, from the literature on ing is. Anticipation of illness is likely to induce a stress response,
the effects of stress on the immune system, one has to consider in the Selyean sense, with its wide range of nervous and
seriously that neuroendocrine factors are involved. Recent work hormonal fluctuations, many of which might directly or indi-
has revealed that lymphoid tissues are subject to nervous inner- rectly influence the immune system, for example, thyroxine,
vation. Furthermore, it has been documented that cells of the parathyroid hormone, insulin, catecholamines, or direct sym-
immune system bear receptors for a great number of hormones, pathetic nervous effects on lymphoid organs. It is difficult to rule
neurotransmitters, and neuropeptides. Indeed, substances like out this stress hypothesis by reconstructive experiments: that is,
endorphins can dramatically alter lymphoid cell activity in vitro, injections of corticosterone may not mimic the natural produc-
and are most probably also involved in stress-induced iminu- tion of this hormone after the conditioned stimulus (CS). And as
nomodulation in vivo. It seems logical that part of the research A & C point out early in the target article, prior research has
effort in the area of behaviorally induced changes in immune shown that "stress" may have a variety of effects on immune
reactivity will be focused on the mechanisms involved. These response, only some of which are mediated by adrenal changes.
"veil-lifting" studies may contribute not only to the science of There is no guarantee that lithium chloride, as an alternative
biomedicine but also to clinical medicine. The results obtained way of inducing taste aversion, has stress-inducing effects analo-
of Ader & Cohen with their murine experimental model of gous to those of CY. Gorczynski et al.'s (1984) demonstration of
autoimmune disease are significant in this respect. lack of conditioning in adrenalectomized animals requires spe-
cial pleading to dismiss. The strongest argument against stress is
that the CS can be given several days before antigen - but even
here, as A & C themselves argue, one can readily imagine long-
Conditioned immunosuppression: An lasting imrnunodepressive effects of quite transient stress reac-
tions; the difficulty would be the same, in any case, for more
important but probably nonspecific specific mechanisms.
phenomenon Ader & Cohen should not be dismayed by the cogency of
these arguments: If conditioned immunosuppression can only
Alastair J. Cunningham be elicited when the UCS is a specific immune suppressive
Ontario Cancer Institute, Toronto, Ont. M4X 1K9, Canada agent, then their phenomenon is of rather academic impor-
Over the last 20 years or so in the West (and longer in the USSR) tance. How many such agents do we encounter outside of the
it has been established, by research in conditioning and biofeed- cancer clinic? But if a range of stressors, which indirectly
back, that many functions previously thought to be "involun- suppress immunity, can become coupled to a symbolic CS, we
tary," notably smooth muscle and visceral responses, may be have a potentially important pathogenic pathway that may help
deliberately influencable by the mind. This has exciting implica- to explain how stress promotes many diseases normally con-
tions for health. Ader, Cohen, Bovbjerg, and their colleagues trolled by the immune system, notably infectious disease and
have done an important job in demonstrating rigorously that this perhaps also cancer. A practical direction for future research is
mental influence extends to a bodily system formerly thought by to attempt to broaden both the kinds of response conditioned
many to be autonomous: the immune system. Many years of (e.g., to parameters of haemopoiesis) and the kinds of UCS,
effort by a relatively small band of workers had previously shown specific and nonspecific, that may be mimicked by various
the often small and variable effects of neuroendocrine mecha- conditioned psychological stimuli.
nisms on immune reactions, but the Ader work (which might be
better called "psychoimmunology" rather than "psychoneuro-")
has had a dramatic impact on the consciousness of those wishing
to explain how mind could affect disease, an impact that has
already filtered down into the popular press. More evidence for the role of learning in
How does this conditioned immunomodulation of immune homeostasis
response occur? Is the mind/brain communicating rather specif-
ically with elements of the immune system, for example, via a Barry Dworkin
hormone released by the nervous system that suppresses some Department of Behavioral Science, Pennsylvania State University College
lymphocyte function? Or is the immune system simply a casu- of Medicine, Hershey, Penn. 17033
alty of a global stress response? Ader & Cohen (A & C) seem to There is good evidence that the brain, through the autonomic
lean toward the former view and are at pains to argue that their nervous system, continuously influences steady-state visceral
effects are not mediated by circulating corticosteroids. For function (Dworkin, in press), and the data and interpretation
reasons discussed below, I suggest that the latter explanation is presented by Ader & Cohen (A & C) indicate a role for the
more probable, but that far from trivializing the phenomenon of nervous system in both the regulation of immunocompetence
conditioned immune suppression, it broadens its importance. and the pathophysiology of allergic reactions. Although large
In the most common experimental paradigm, cyclophospha- effects remain to be described, the learning phenomena that
mide (CY) has been used as the unconditioned stimulus (UCS). they have demonstrated are reliable, reproducible, and appear
This alkylating agent has a wide variety of effects on the body in to verify the existence of previously unappreciated central
mice and humans, including destruction of dividing cells in nervous system (CNS) regulatory pathways.
many sites (e.g., gut endothelium, hair follicle, gonadal and The pharmacological conditioning experiments of Siegel
haemopoietic cells) producing depression of peripheral white (1972; 1975; 1976; 1978) similarly seem to indicate that learning

modification of visceral function can be demonstrated. Howev- Taste aversion proneness: A selective
er, putting aside the admittedly dramatic evidence that extero- breeding strategy for studies of Immune
ceptive conditioned stimuli can influence mortality rate (Siegel, system condltlonability
Hinson, Krank & McCully 1982; target article), are these phe-
nomena important in the normal regulation of visceral function
and the maintenance of homeostasis? Learning is clearly impor- Ralph L. Elkins
tant to homeostasis through the regulation of skeletal behavior. Psychology Research, VA Medical Center, and Department of Psychiatry
A & C specifically refer to the learned homeostasis experiments and Health Behavior, Medical College of Georgia, Augusta, Ga. 30910
of Miller, DiCara, and Wolf (1968), which showed that rats Ader & Cohen (A & C) are to be commended for their organiza-
would perform an instrumental or type II skeletal response that tion and interpretation of a multidisciplinary data base that
corrected an artificially induced blood electrolyte imbalance. clearly implicates conditioning as a significant contributor to at
The thermoregulation experiments of Weiss and Laties (1960) or least some aspects of immune system functioning. Immune
Rozin and Mayer (1961) and the oxygen tension regulation system conditionability is of considerable basic theoretical in-
experiments of van Sommers (1962) are among other examples terest and may prove to have important applications in ame-
of homeostatically motivated type II learning. liorative or preventive medicine. The following comments will
Whereas A & C's and Siegel's experiments all follow the address a methodological strategy that may improve the power
protocol of type I or respondent conditioning, the conditioned of experiments that use taste aversion (TA) learning to study
responses (CR) that Siegel observes are not usual for type I conditionable aspects of immune system reactions.
conditioning in that ordinarily, the CR is expected more or less The independent studies of Ader and Cohen (1975); Rogers,
to mimic the unconditioned response (UCR), whereas Siegel Reich, Strom, and Carpenter (1976); and Wayner, Flannery,
usually observes a CR that is diametrically opposite to the UCR and Singer (1978) and other related reports reviewed by A & C
- in other words, compensatory. Because of this outcome, depict a TA conditioning methodology that has reliably demon-
Siegel's experiments could in fact be examples of type II rather strated conditioned immunosuppressive responses. Despite
than type I learning (Dworkin 1984). their statistical reliability, however, these conditioned immu-
If an anticipatory compensatory CR ameliorates the physio- nosuppressive responses have all been of a relatively small
logical disturbance (UCR) elicited by unconditioned stimulus absolute magnitude as explicated by Ader and Cohen (1981 and
(UCS), it would be reinforced by preservation or restoration of target article). The small magnitude of these conditioned
homeostatic balance: the response-reinforcement contingency changes may accurately reflect the typical limits of conditioned
being entirely contained within the organism. Is there a similar immunosuppressive responses. As emphasized by Ader and
interpretation of A & C's experiments? The immune response Cohen (1981), however, it is likely that optimal methodologies
itself can be seen as inherently compensatory; the antigen for studying conditioned alterations of immune system re-
producing a physiological disturbance and the response elim- sponses are yet to be developed. Attention to individual dif-
inating or stabilizing the antigen. Unlike the pharmacological ferences in conditionability within strains of laboratory subjects
example, however, this interpretation is more an analogical in one approach that may increase the magnitude of observable
metaphor than a behaviorally rigorous analysis of the processes conditionable alterations of immune system functions.
involved. Furthermore, it is a direct and necessary implication Conditioned TAs are exceptionally efficient learned adjust-
of the type II learning analysis that some mechanism exists for ments that are readily induced in many species, including
sensing fluctuations in homeostatic balance. For the phar- rodents and humans (e.g., Riley&Tuck, in press). Long-lasting
macological responses of Siegel there are documented in- TAs have been observed in some rats following a single pairing
teroceptors that would suffice; if A & C's reference to Miller, of a novel saccharinflavorwith a relatively low dose (12.5 to 25
DiCara, and Wolf (1968) is intended to suggest a type II process mg/kg) of cyclophosphamide (Elkins 1973). However, the
with the improvement of immunocompetence as the reward, Sprague-Dawley-derived rats that were the subjects of that
then it would follow that the nervous system must be capable of experiment displayed considerable individual variability in
sensing as well as influencing specific immunocompetence. This
aversion acquisition, and similar differences in aversion reten-
fact should be demonstrated before a type II analysis is applied
to their experiments. tion have also been reported (Elkins 1984). Comparable indi-
vidual differences in TA acquisition and retention of humans
Is the type I-type II distinction of genuine importance? I were observed during a consummatory aversion approach to
think it is. The brain has a recognized and important role in the alcoholism treatment (Elkins 1980).
integration of skeletal motor behavior through instrumental Individual differences in TA acquisition and retention are
learning (McFarland 1971). To be effective, an integrative consistent with the concept of TA proneness, a hypothesized
physiological mechanism must include some form of outcome- intrasubject modulator of TA conditionability (Elkins fit Hobbs
dependent response selection, in which future response proba- 1982). Interest in TA proneness has led to an ongoing selective
bility is partially determined by the homeostatic consequences breeding development of strains of TA-prone (TAP) and TA-
of previous response occurrences. Type I learning does not have resistant (TAR) rats. The immediate objective of the selective
this property. breeding is the development of TAP and TAR phenotypes for
The conventional systems conception of autonomic regula- use in studies of the biological bases of individual differences in
tion, in which the physiological state is stabilized by negative TA conditionability. The selection methodology may also be
feedback from peripheral interoceptors, is on the brink of a producing TAP rats that will be useful in studies of conditioned
Kuhnian paradigmatic crisis because the commonly observed immune system responses. As in the demonstrations of condi-
adaptation characteristics of all known receptors are inconsis- tioned immunosuppression of Ader and Cohen (1975), Rogers et
tent with the requirements of the regulation model. For the al. (1976), and Wayner et al. (1978), TA conditioning in the
cardiovascular system the best solution offered involves total selective breeding project features cyclophosphamide-induced
exclusion of CNS mechanisms from steady-state regulation aversions to a saccharin solution. Obtained strain differences in
(Guyton 1977). Visceral learning could be a missing concept in TA proneness, however, are not restricted to the
understanding central participation in homeostatic mecha- cyclophosphamide-induced illness or to the saccharin flavored
nisms. The body of work described and reviewed in Ader & solution that are the unconditioned stimulus (US) and condi-
Cohen's target article adds to the evidence that at least the tioned stimulus (CS) bases of strain selection. Strain differences
necessary central and peripheral efferent pathways exist. [See in TA proneness have generalized to different modes of US
also Toates: "Homeostasis and Drinking" BBS 2(1) 1979.] induction, including rotational stimulation (Elkins & Harrison

1983b) and x-ray exposure (Peacock & Fuller 1982). Strain Immune behavior
differences have also been observed with new CS flavors that
include vanilla- and beef-flavored solutions (Elkins & Harrison Bernard T. Engel
1983a; 1984). Gerontology Research Center (Baltimore), National Institute on Aging,
Details of strain development appear in Elkins (in press). A National Institutes of Health, PHS, U.S. Department of Health and Human
summary of conditioning and selection methodologies is as Services, Bethesda, and Francis Scott Key Medical Center, Baltimore, Md.
21224
follows: Sprague-Dawley-derived male and female rats were
given intraperitoneal cyclophosphamide injections following The discovery (rediscovery would be more accurate) that the
the ingestion of a novel saccharin solution. Aversion acquisition central nervous system (CNS) can influence immune responses
was assessed with standard two-bottle preference tests that is one of the many exciting developments in neuroscience over
provided ad lib home-cage access to two separate bottles con- the last 15 years. Ader & Cohen (A & C) have made important
taining the saccharin solution and plain tap water. Daily prefer- experimental contributions to this field, and the target article
ence percentage scores were computed to express the degree to provides a valuable review of the current state of knowledge and
which saccharin-solution ingestion contributed to each subject's the historical precedents that led to this state. A & C should be
total daily fluid consumption. The best male and female TA applauded for the comprehensiveness, clarity, and, above all,
learners were mated through nonsibling pairings, and the most excellent organization of this paper.
inefficient learners were similarly paired to produce the S-l There are two conceptual issues that I want to raise; the first is
generation. This process has been repeated through 12 genera- relatively minor, the second significant. In the section on
tions. Significant strain differences were observed within the historical perspectives A & C cite a number of studies from the
S-2 generation and thereafter. Mean TAP-TAR preference Soviet Union. They note that while there was general accep-
score differences of 40 or more have been observed in several tance of the notion that nonspecific immune responses could be
recently conditioned generations. These differences are asso- conditioned, there was strong resistance to the notion that
ciative as indicated by the traditional use of unconditioned specific responses could be conditioned. There is a message in
control subjects that received unpaired exposures to the CS and this for all of us. I would hope that by now we have learned that
US (Elkins & Boudewyns 1984). There is no indication that the distinction between specific and nonspecific effects is a
maximum divergence has been obtained. Continued diver- measure of our ignorance about mechanism rather than about
gence approximating the previously observed rate would pro- specificity of function. In a causal world nonspecificity is a
duce mean conditioned TAP-TAR preference score differences meaningless notion. Perhaps it would be useful to recall the
in excess of 60 by the S-20 generation. state of knowledge about intermediary metabolism before 1950.
Available data are consistent with the possibility that selective Very few enzymes had been characterized; however, this did
breeding is exerting an effect that is highly specific to TA not prevent biochemists from inventing enzymes to account for
conditionability. No consistent strain-related differences have the substrate interactions they observed. Their faith in specifici-
been observed with two different schedules of food-reinforced ty/causality was justified: There are no nonspecific metabolic
bar-press responding (Hobbs & Elkins 1983), during the ac- products in living systems.
quisition of a shock-motivated environmental avoidance re- My second point has to do with the implicit dualism that
sponse (Elkins & Boudewyns 1984), or during tests of open-field pervades the target article. Throughout the paper A & C
and enclosed-conpartment reactivity (Elkins, Harrison, Mur- carefully distinguish physiologic functions from behavior - for
phy & Hobbs 1984). example, see their discussion of the definition of psycho-
The value of the TAP and TAR strains in relation to studies of neuroimmunology in the first paragraph or their notion of a
immune system conditionability is an empirical issue to be reciprocal relationship between behavior and physiological
resolved by future research. As noted by A & C, the precise function in the concluding remarks. The dualism is especially
nature of the relationship between the behavioral manifesta- noteworthy since it is clear from their own research and from
tions of cyclophosphamide-induced TAs and the accompanying their conclusions that what they really want to assert is that
immunosuppressive responses is unknown. Important findings immune responses are behavior. Like somatomotor responses,
about immune system conditionability may or may not emerge immune responses can be elicited by a number of stimuli; in
as a result of the TA propensity and attenuated individual both cases some of the stimulus effects are neurally mediated;
variability of the TAP rats. Irrespective of any specific findings and in both cases the responses can be trained and can be
that may be forthcoming from these strains, the value of other emitted as well as elicited. No doubt there are important
selection approaches to the development of useful phenotypes differences among these response systems just as there are
for studies of immune system conditionability should be ex- important differences among component responses within each
plored. As recognized by Monjan (1981), "interstrain genetic system. Conceptually, however, both response systems must be
heterogeneity enters into both the neuroendocrine and immune behavior; if not, what is the message of the finding that immune
systems. Inconsistencies between studies may be rectified by responses can be conditioned? As far as I can judge, a major
understanding this potential source of variance" (p. 216). In importance of the work of A & C and others lies in the discovery
addition, as noted by Ader and Cohen (1981): that there is no such thing as a reciprocal relationship between
Conceptually, the capacity of conditioning to suppress or facilitate behavior and physiological function. There are only interactions
immune responses raises innumberable issues with respect to the among behaviors in their manifold forms.
normal operation and modifiability of the immune system and could As a gerontologist I would be remiss if I did not call attention
lead to very basic studies of the mediation of individual differences in to the relationship between aging and the subject of A & C's
the body's natural armamentarium for adaptation and survival along paper. First, I want to note that age as an independent variable
lines that are currently poorly understood or almost totally unrecog- could be used to address some of the unsolved issues noted by A
nized, (p. 316) & C (see the recent review by Nagel [1983]). One should be able
to test the hypothesis that there is a relationship between CS
ACKNOWLEDGMENTS (conditioned stimulus) strength and host immunocompetence
Preparation of this manuscript was supported by the Medical Research on the magnitude of the CR (conditioned response) by testing
Service of the Veterans Administration and by the National Institute on animals of different ages. In this same vein, I wonder whether
Alcohol Abuse and Alcoholism research grant #1 R01 AA06465-01. the "relatively small" magnitude of change in immunological
1 am indebted to Mary F. Mobley and to Lelon J. Peacock for their activity that concerns A & C might appear not so small in older
reviews of a preliminary version of this manuscript. cohorts where host resistance is reduced. I wonder if age
THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 • 399

Commentary I Ader & Cohen: CNS-iminune system interactions
differences in conditionability of immune responses exist, and if learning process. 1'he most fascinating possibility is that the IS
these differences can predict some of the age-related changes in might affect the learning process directly through neurotrans-
morbidity or mortality risk? mitter changes. We know that lymphocytes have receptor sites
for neurotransmitters and peptides. The ability to release as well
as to receive them would provide an important possible route for
a direct effect on the IS on the nervous system, especially the
learning process, whether through Pavlovian conditioning, op-
Disease is a stepchild In erant conditioning, or other modes of learning.
psychoneuroimmunology Why stop there? Why might the IS not affect other brain
functions such as memory (amygdala and hippocampus arc as
Bernard H. Fox closely connected to hypothalamus as other cortical loci) and
Department of Biobehavioral Sciences, Boston University School of perception? The limiting factors are mechanisms of transmission
Medicine, Boston, Mass. 02118
of lymphocyte information and the transmission time factor.
The contention of Ader & Cohen (A & C) that the immune Intermediate effects are possible, for example, emotional
system (IS) and neuroendocrine system (NES) interact is, at this changes, which in turn can influence learning, since affect is
stage, so buttressed by supporting data that one cannot seriously known to be a potent variable in the learning process.
doubt it. Conditioning is one mode of such interaction - an Last, the demonstrations by Ader and colleagues have been
interesting one, to be sure. It is joined by many others, and called less than convincing because of the weakness of the effects
support for it is capped by the series of papers and reviews demonstrated. I view this as an ill-considered cavil. It if happens
submitted to the First International Workshop on Neuroim- that the route of the conditioning effect is through an auxiliary
munomodulation, held at the National Institutes of Health in mechanism with less potency than the direct mechanism by
December 1984. As A & C are aware - but the point needs which cyclophosphamide exerts its effects, we would have an
emphasis - the extant data, although convincingly indicating important demonstration that such effects were separate. Just
that there is an interaction between the NES and IS, are such a phenomenon was found when stress, stimulating adrenal
sparsely distributed tiles of a complex mosaic. We need to know corticosterone, suppressed the immune system; but stress in
the mechanisms by which each affects the other, although some adrenalectomized animals also did so, to a lesser degree. If the
hints about the NES-IS pathway(s) are emerging. On the other route is the same mechanism by which cyclophosphamide
hand, the IS-NES mechanisms and routes of action are, except operates, the question of why the effect is less powerful cries for
for a few papers, unknown. In the NES-IS case, however, many an answer. In either case the finding is not one to be decried (I
of the data are based on partial models whose outcomes are believe the evidence of the conditioning effect is incontroverti-
immune process changes. This is also the case with the series on ble), but rather one that challenges the researcher to new
conditioning of Ader and his colleagues except for their work on insights. If the demonstration of the basic effect is called into
New Zealand mice with a syndrome resembling lupus. question, it is up to the questioner to show how such results
There is a third element that should constitute an important could have been obtained in the absence of an IS conditioning
outcome when the whole picture is viewed: bodily state associ- process.
ated with IS response. The state of greatest concern, of course,
is disease, but changes in the well organism are also of interest.
Some conditioning studies have been done on disease processes
where presumptive manipulation of the IS through NES altera-
tion is the initiating event, even though accompanying IS CNS-immune system interaction: A
measurements have usually not been taken (e.g., Klosterhalfen psychosomatic model
& Klosterhalfen 1983; Ader & Cohen 1982). There are also a
large number of studies other than conditioning ones where Stanford B. Friedman
differences in NES status have been associated with variations Division of Behavioral Pediatrics, Department of Pediatrics, University of
in disease outcome (for specific reviews, see Cohen 1979; Fox Maryland School of Medicine, Baltimore, Md. 21201
1983; Jacobs and Ostfeld 1977). In these cases, the IS was Traditionally, the field of psychosomatic medicine has identified
conjectured to be involved. Examples are the relation of person- specific diseases as being etiologically linked to specific psycho-
ality to cancer in humans, of stress to cancer in animals, and, in logical phenomena. For instance, Alexander (1950) described
some cases, of stress to cancer in humans (Fox 1978; Newberry seven diseases (ulcerative colitis, regional enteritis, peptic ul-
1981; Sklar&Anisman 1981). The bulk of the work in these areas cer, hyperthyroidism, bronchial asthma, essential hyperten-
has ignored the IS. A few exceptions exist (e.g., Greer, Morris sion, rheumatoid arthritis) as psychosomatic in etiology, with
& Pettingale 1979; Riley 1981). A number of studies have been
each disease entity associated with a specific psychic conflict. In
done in which personality (usually state rather than trait mea-
sures) and stress have been related to the IS. Again, in this work a similar fashion, Dunbar (1943) associated "personality pro-
the simultaneous investigation of disease was, by and large, not files" as having etiologic importance in certain disease states.
prominent (Jemmott and Locke 1984). The work on condition- Clinical observations such as these led to the theory of
ing and the IS is similar in that little work has been attempted on specificity, whereby identifiable psychological factors were
disease outcome. In sum, there is a large empty space and it etiologically associated with specific diseases. This concept
should be filled. fostered an "either-or" dichotomy - either a disease was caused
by psychological factors or it was not (i.e., it was organic or
After recounting their conditioning studies, A & C make some physical in etiology). In reality, even early investigators in
interesting concluding remarks. Among them is the suggestion psychosomatic medicine acknowledged that nonpsychological
that learning may contribute to immunoregulation and behav- factors had an etiologic role in so-called psychosomatic illnesses;
ioral adaptation, presumably on a broader base than Pavlovian however, the contribution of genetic and other biological at-
conditioning experiments have already demonstrated. Perhaps tributes of the host were often minimized or totally ignored.
at least as interesting a speculation is the possibility that changes Only the careful appraisal of clinical experience and subsequent
in the IS affect not only the response of the NES (e.g., Be- research over the years has demonstrated the oversimplistic
sedovsky, del Rey, Sorkin, De Prada, Burri & Honegger [1983], nature of this cause and effect conceptualization of the etiology
who report alteration of hypothalamic discharge with IS of disease. Yet, even in today's practice of medicine, it is,
changes), but also its ability to learn. The demonstration of such unfortunately, common to hear of an illness in a particular
an effect is likely to take the form of IS effects on the endocrine patient described as having either an organic or a psychological
system; that system is in turn known to be able to affect the basis.

Cominentary I Ader& Cohen: CNS-immune system interactions
Engcl (1954) has been a leader in advocating the "multi- integration of findings from experimental and clinical studies in
factorial" etiology of disease, and more recently (1977), the this area of investigation might do much toward discarding the
biopsychosocial model of disease. In our own writings (Plaut & outdated notion that disease is either psychosomatic or physical,
Friedman 1981), we have suggested that all diseases may be thus providing further confirmation that the very term "psycho-
influenced by psychosocial factors, and that vulnerability to somatic disease" may have outlived its usefulness as a clinical
such factors may be viewed along three continua. First is the concept.
nature of the disease under consideration, with some illnesses or
disease states particularly susceptible to psychological factors
(e.g., ulcerative colitis, hypertension) and others less so (per-
haps CNS [central nervous system] tumors). The second con- "Relatively mild stress" depresses cellular
tinuum comprises the unique biological and psychosocial char-
acteristics of a given patient, both genetic and acquired, with a immunity in healthy adults
range of vulnerability to any given disease. Third, we have Ronald Glaser
emphasized that changes in vulnerability to disease may be
Department of Medical Microbiology and Immunology and Comprehensive
related to temporal or developmental factors.
Cancer Center, Ohio State University College of Medicine, Columbus, Ohio
With these considerations in mind, infectious disease pro- 43210
vides an interesting model for the understanding of psychoso-
matic concepts (Friedman & Glasgow 1966). The usual medical Janice K. Kiecolt-Glaser
thinking is that infectious agents (e.g., bacteria, viruses) cause Department of Psychiatry, Ohio State University College of Medicine,
infectious disease, without due consideration of numerous fac- Columbus, Ohio 43210
tors that modify the host-parasite relationship. That exposure to Ader & Cohen (A & C) have provided an excellent discussion of
an infectious agent is a necessary condition for developing an conditioned immunopharmacological responses. Their work in
infectious disease is not in doubt, yet mere exposure to, or even this area is clearly outstanding. They suggest, however, that
the harboring of, a pathogen does not necessarily result in illness relatively mild stress or conditioned stress may be capable of
or disease. Thus,. only a small proportion of children with exerting their subtle effects only in an already immunocompro-
positive throat cultures for beta-hemolytic streptococcus will miscd host. Though such speculation is consistent with data
develop symptoms of infection (e.g., fever, pharyngitis). The obtained in some studies with rodents as discussed by A & C, it
question that arises is whether psychosocial factors influence is in conflict with the human experimental data discussed below.
which children becoi le clinically ill from the streptococcal Work from our laboratory has demonstrated the immunosup-
organism, and there is evidence that this is indeed the case pressive effects of commonplace acute "stressful" events in
(Myer & Haggerty 1962). otherwise healthy adults. For example, we found replicable and
There are now abundant clinical and experimental (animal) significant decrements in natural killer (NK) cell activity in
studies demonstrating the limitations of the "germ theory" in medical student blood samples obtained on the day of examina-
understanding the development of infectious disease (for re- tions in comparison to baseline samples obtained one month
views of this work see Friedman & Glasgow 1966; Plaut & previously (Kiecolt-Glaser, Garner, Speicher, Penn, Holliday
Friedman 1981). In one early study, for example, we showed & Glaser 1984; Kiecolt-Glaser, Glaser, Strain, Stout, Tarr, Holli-
that the clinical course of Coxsackie B virus infection in the day, & Speicher, in press). Self-report data documented the
mouse reflected, at the very least, the dosage of inoculated virus significantly greater distress associated with examinations.
and the age of the animal, as well as the psychosocial (experi- Other immune functions were also responsive to this relatively
mental) manipulation; a demonstration of the interaction of mild stressful event, including the transformation of B lympho-
biological and psychosocial factors (Friedman, Ader & Glasgow cytes by Epstein-Barr virus (Kiecolt-Glaser, Speicher, Holliday
1965). In this and in numerous other experiments using a variety & Glaser 1984), the percentage of helper T lymphocytes
of murine viruses, however, we were unable to delineate the (Kiecolt-Glaser et al., in press), and the production of inter-
physiological mechanisms that accounted for the modified re- ferons by lymphocytes stimulated with concanavalin A (Con A).
sponse to the infectious agents. Additional demonstrations of the immunodepressive effects
Ader & Cohen (A & C) in their eloquent review of their of acute stress in otherwise healthy adults are seen in data from
experiments and the work of others, clearly demonstrate that Jemmott, Borysenko, Borysenko, McClelland, Chapman,
the conditioning of immune responses may represent a major Meyer, and Benson (1983), who showed that alterations in
mechanism in altering host resistance to a wide range of micro- salivary immunoglobin A in dental students were related to high
organisms. Furthermore, they have suggested logical ap- and low stress periods. Acute stress-related alterations in hu-
proaches to further defining the interactions between the cen- moral immunity have also been shown in nonhuman primates
tral nervous and immunological systems, including the possible (Coe, Wiener, Rosenberg & Levine, in press).
role of adrenocorticoids and other hormones. A & C characterize as "subtle" the immunological conse-
From the point of view of a physician, a practical clinical quences of relatively mild stress or conditioned stress. Although
question remains, namely, will A & C's experimental findings the size of conditioned effects is generally small (while having
ultimately prove important in the daily practice of medicine? survival value, as described by A & C), the effects of relatively
This is not meant to imply that these studies of A & C are not mild stress are frequently of considerable magnitude. For exam-
important in their own right and to the further understanding of ple, the production of interferon by lymphocytes stimulated by
the physiology and pathophysiology of immunity. Nevertheless, Con A declined from a baseline value of 2003.03 U/ml 6 weeks
this commentator would argue for concurrent studies address- before final examinations to a mean of 80.00 U/ml on the day of
ing the question of clinical importance, to the practitioner, of final examination in 40 second-year medical student blood
psychosocial factors in infectious diseases. This would include samples (data submitted for publication). Similarly, exam-relat-
investigations in human subjects related to the etiology of ed decrements in natural killer (NK) cell activity have ranged
infectious illnesses, the clinical course of such illnesses, and from 16% to 43%, depending on the subject sample, the NK
their response to medical therapy and intervention, as well as target cell used, and the effectontarget-cell ratio across three
studies on the underlying physiological mechanisms. Though samples of medical students; the modal decrease is around 33%
numerous clinical studies have focused on these issues, data do of baseline values.
not exist that are influential to the clinician. If psychosocial In further work, we have assessed the possible enhancement
factors are clearly involved in the outcome of individual patients of immune function by simple positive interventions (Kiecolt-
exposed to infectious agents, the area of CNS-immune system Glaser, Glaser, Williger, Stout, Messick, Sheppard, Ricker,
interactions would take on even more importance. Last, the Romisher, Briner, Bonnell & Donnerberg 1985). Forty-five

geriatric residents of independent living facilities were ran- other researchers obtained conditioned changes in blood
domly assigned to one of three protocols: (1) progressive relaxa- glucose levels using insulin and glucose as unconditioned stim-
tion training, (2) social contact, or (3) no intervention. Subjects uli (e.g., Alvarez-Buylla & de Alvarez-Buylla 1975; Matysiak &
in the relaxation and social contact protocols were seen indi- Green 1984; Siegel 1975; Woods 1976). In still other work,
vidually three times a week for a month. Blood samples and self- tolerance to morphine has been shown to be a classically condi-
report data were collected at baseline before the intervention tioned response (Siegel 1976), and there is accumulating evi-
began, at the end of the one-month intervention, and at a one- dence that the endogenous opiate system may also be triggered
month follow-up. There was a significant increase in NK cell by conditioned stimuli previously paired with aversive uncondi-
activity in the relaxation subjects at the end of the month-long tioned stimuli. The value of this research, both for the basic
intervention without significant changes in the other two understanding of conditioning effects and the interrelation of
groups; NK cell activity returned to near baseline levels in the brain and behavior and for its possible biomedical relevance,
relaxation group at the follow-up, when subjects reported little should be obvious.
continued relaxation practice. There was also a significant de- However, the mechanisms mediating these conditioned ef-
crease in antibody titers to herpes simplex virus in the relaxation fects remain to be determined. Because the unconditioned
group at the end of the intervention, a change which was response has not been directly measured and because confusion
maintained at follow-up. Lower antibody titers to latent herpes exists as to the response elicited in some of these preparations,
viruses probably reflect better cellular immune system control conflicting theoretical analyses of the conditioned response and
of virus latency (Glaser, Kiecolt-Glaser, Speicher & Holliday, in considerable speculation as to mechanism abound. In A & C's
press; Claser & Gottlieb-Stematsky 1982). exciting research, the saccharin solution and injection pro-
Taken together, these data suggest that the immune system in cedure constitute the CS, which, when paired with cyclophos-
otherwise healthy individuals is responsive to relatively com- phamide (the US) apparently conies to evoke a suppression of
monplace stressful events (and perhaps positive events) across immunological reactivity when the CS is later presented in the
populations. These data are not consistent with the hypothesis absence of the US. While several possible mechanisms respon-
that the effects of relatively mild stress may be evidenced only in sible for this effect might be unlikely, such as an elevation in
an immunocompromised host, in contrast to the rodent data adrenocortical steroid levels, we still do not have clear evidence
discussed by A & C. Although the functioning of the immune for conditioned immunosuppression itself. The result may be
system is similar across mammalian species, there are also a due to other, still unknown, factors that influence immunologi-
number of important differences, including the greater sen- cal reactivity.
sitivity of the rodent immune system to the immunosuppressive Similarly, conditioned changes in the endogenous analgesic
effects of the adrenal glucocorticosteroids (Claman 1972). More- system are suggested but not directly measured in preparations
over, it is not clear how to compare the physical stressors used in pairing a Pavlovian CS with painful stimulation as the US. In the
animal studies with the cognitive stressors that are of primary investigation of conditioned glycemic responses, the problem is
interest in human psychoneuroimmunology. manifest. If a distinctive stimulus (CS) is repeatedly paired with
Immune function is depressed by commonplace aversive insulin administration (the US), then presentation of the CS
events in otherwise healthy adults. Recent data suggest that alone may come to elicit a change in blood-glucose levels (the
more profound distress may be associated with dysfunction at conditioned response, CR). These obtained changes have some-
the molecular level, in the speed and quality of DNA repair times been conditioned increases, while in other studies condi-
(Kiecolt-Glaser, Stephens, Lipetz, Speicher & Glaser, in press). tioned decreases are found. Indeed, Flaherty, Uzwiak, Levine,
The health-related consequences of such alterations are not yet Smith, Hall, and Schuler (1980) reported both hypo- and hyper-
known. glycemic conditioned responses in the presence of different
CSs. It is not at all clear what variables are responsible for such
differences in results. However, as Eikelboom and Stewart
(1982) have pointed out in their discussion of conditioned drug
Pavlovian conditioned responses: Some effects, "it is the failure to define and identify correctly the
unconditioned stimulus and unconditioned response and to
elusive results and an indeterminate determine their relation to the observed drug effect that is the
explanation source of the apparent contradictions and confusions."
Leonard Green
We have made a similar claim with regard to conditioned
glycemia in particular, in which insulin and glucose were used as
Department of Psychology, Washington University, St. Louis, Mo. 63130
USs (Matysiak & Green 1984). For example, in conditioned
For those who may have thought that little more was to be glycemia using glucose as the US, the unconditioned response
learned from the study of Pavlovian conditioning, the implica- (UR) is generally considered the increase in blood glucose
tions of which were already known or of esoteric interest to (hyperglycemia) following glucose administration. The CR ob-
animal-learning theorists, the Ader & Cohen (A & C) target tained on test trials is a reduction in blood-glucose levels
article should give pause. In their research on the Pavlovian (hypoglycemia), a change in the direction opposite to that of the
conditioning of immunological reactivity, a saccharin solution presumed UR. Consequently, this CR is viewed as a compen-
(the conditioned stimulus, CS) is paired several times with the satory response (e.g.,Siegel 1975). However, we noted that,
administration of an immunosuppressive drug, cyclophospha- contrary to what has generally been stated, changes in blood
mide (the unconditioned stimulus, US). When later adminis- glucose levels cannot be treated as the UR:
tered an antigen and tested in the presence of the CS, such
conditioned animals show an attenuation in the antibody re- To claim, or example, that the administration of glucose produces
sponse as compared to nonconditioned animals. Such condition- an unconditioned response of increased levels of glucose in the blood
ing of immunosuppressive reactivity is also shown to modify the would be like claiming that the presentation of food in Pavlov"s
development of an autoimmune disease. Specifically, the rate of experiments led to a UR of increased amounts of food in the dog's
development of proteinuria and mortality are significantly re- mouth. A UR must be the response of the central nervous system to
tarded in female New Zealand hybrid mice (Ader & Cohen 1982) the US, and not the direct effect of the substance injected. (Matysiak
who received pairings of the saccharin solution with and Green 1984, p. 7)
cyclophosphamide as compared to control animals who received We suggested that the administration of glucose stimu-
equivalent amounts of the saccharin and the drug, but which lates endogenous insulin secretion. If so, the obtained
were explicitly unpaired. Also using a Pavlovian procedure, hypoglycemic CR might be attributed to a conditioned

Commentary I Ader& Cohen: CNS-immune system interactions
insulin release, and therefore this CR is not compensato- there is an optimal range of functioning for the body's immune
ry and not in a direction opposite to that of the UR. A system. If the immune system is suppressed, the organism
similar reinterpretation can be offered for the "compen- becomes more susceptible to disease. Conversely, a tonically
satory" CR presumed by the conditioned-morphine- activated immune system may also have other pernicious ef-
fects. How is regulation of the immune system achieved? There
tolerance research. Until such time as the UR is well
are undoubtedly many mechanisms. A & C have provided
specified, we cannot be sure about the exact nature of the evidence that one of these mechanisms involves Pavlovian
conditioning effect, or whether the CR is a compensatory conditioning. However, the results described by A & C raise
or a standard Pavlovian response. some questions. Paramount is the question of the form of the
A & C have made most intriguing and valuable begin- CR. In most of the studies described, rats were injected with
nings in the field of psychoneuroimmunology. Further cyclophosphamide (CY) on a single occasion following ingestion
specification of the mechanisms involved in the condi- of a saccharin-flavored solution. At some later time, a test of the
tioning of immunosuppressive reactivity as well as in conditioned reaction to the saccharin-flavored solution was
conditioned drug tolerance and glycemia will provide a given. As A & C note, CY is a potent immunosuppressive drug.
firmer grasp of the phenomena and a more convincing When the conditioned response to saccharin was tested, the
usual result was also immunosuppression. If one considers the
explanation. It should, at the very least, be clear, as A & C
adaptive nature of Pavlovian conditioning, conditioned immu-
state, that "the study of behavior is also an integral part of nosuppression is paradoxical. If the immune system is home-
biology" as biology is a part of the study of behavior. One ostatically regulated, then during the initial training, when CY
cannot be reduced to the other. As the Pavlovian condi- is injected and produces immunosuppression, one would expect
tioning of the immune function, of drug tolerance, and of the activation of compensatory mechanisms to restore the level
glycemic responses shows, the relation between phys- of immune competence (cf. Obal 1966; Solomon 1980). As a
iology and behavior is a dialectical one: Each depends on result of the saccharin-CY pairing, saccharin would be expected
the other and each modifies the other. to elicit the compensatory immune response, not the immu-
nosuppressive response. This amounts to saying that the orga-
nism will respond to a stimulus signalling immunosuppression
(i.e., the saccharin CS) with an anticipatory immunoregulatory
response. This analysis is based on the identification of the
immunosuppression produced by CY as the unconditional stim-
Conditioned immunosuppression and the ulus (US) that reflexively elicits the unconditional response (UR)
adaptive function of Pavlovian conditioning of compensatory immunoregulation (cf. Eikelbomm & Stewart
1982). In fact, A & C cite a study by Gorczynski, Macrae and
Riley E. Hinson Kennedy (1984) that demonstrated conditioned enhancement of
Department of Psychology, University of Western Ontario, London, Ont., immunological reactivity following saccharin-CY pairings (there
NBA 5C2, Canada was also evidence of conditioned immunosuppression in this
Pavlovian conditioning has been described by many authors as study).
providing a vital adaptive capacity to organisms (e.g., Pavlov I have suggested that conditioned immunosuppression fol-
1927; Obal 1966; Hollis, 1982). For example, in discussing the lowing saccharin—CY pairings is paradoxical since it does not
occurrence of a salivary conditional response (CR) in dogs appear that it would be adaptive. Thus, if conditioned immu-
resulting from the pairing of dilute acid with a conditional nosuppression occurs to a CS paired with CY, it is to be expected
stimulus (CS), Pavlov (1927) said: "The great advantage to the that the level of immunosuppression observed over the course
organism of a capacity to react to the [conditioned stimulus] is of repeated CS-CY pairings will become augmented - the
evident, for it is in virtue of this action that . . . rejectable unconditional immunosuppression of CY would be added to by
substances, often nocuous to the mucous membrane, find a the ever-increasing conditioned immunosuppression elicited by
layer of protective saliva already in the mouth which rapidly the CS. The phenomenon would be analogous to the "sensitiza-
dilutes and washes them out" (pp. 13-14). tion" to behavioral effects of a variety of psychomotor stimulants
A major tenet of modern biology is that most physiological that has been shown to arise from conditioning (e.g., Hinson &
systems have an optimal, or homeostatic, level of functioning, Poulos 1981; Post, Lockfeld, Squillace & Contel 1980). There
and that deviations from this homeostatic level elicit reflex are no data presented in A & C's target article that are directly
reactions to bring the system back into balance. [See Toates: relevant to this prediction. However, the serendipitous finding
"Homeostasis and Drinking" BBS 2(1) 1979.] An obvious exam- (see the section on conditioned suppression of humoral activity
ple is thermoregulation. If by some means (e.g., injection of a in the target article) that leads to the series of reported studies
drug) the body temperature of an organism is depressed, this would seem to indicate that such augmentation occurs. Ob-
hypothermia elicits processes that will tend to restore body viously, conditioning that increases the organism's susceptibili-
temperature to its homeostatic level. Recent research has dem- ty to infection and mortality is, on the surface, maladaptive.
onstrated that this homeostasis-restoring reflex can he condi- There may be instances where augmented immunosuppression
tioned. In one experiment (Crowell, Hinson & Siegel 1981), rats would be beneficial to the organism (as, for example, in the
were injected with ethyl alcohol in a dose that depressed body autoimmune diseases discussed by A & C). But why conditioned
temperature. Each injection of alcohol was paired, in a Pavlo- immunosuppression should result from conditioning involving
vian conditioning procedure, with a CS that initially elicited CY is unclear at this time. As A & C indicate, there is still much
little thermic response. After a number of such pairings of the to be learned about the pathways and mechanisms involved in
CS with alcohol-induced hypothermia, the CS was followed by
the functioning of the immune system. It is possible that with a
an injection of saline (a placebo test trial). The finding was that
fuller understanding of the immune system, the conditioning
the CS elicited a /it/perthermie response. There are many other
effects reported by A & C could be incorporated into an adaptive
demonstrations of conditional responses that appear to serve a
framework involving Pavlovian conditioning (cf, Eikelbomm &
similar homeostasis-restoring function (for a review, see Siegel
Stewart 1982). The research reported by A & C is thus doubly
1983).
important, in that it not only demonstrated an interesting link
I would like to discuss the research summarized by Ader & between the brain and the immune system, but may also
Cohen (A & C) from the perspective of Pavlovian conditioning provide very useful clues to how the immune system is
serving a homeostatic regulatory function. It is obvious that organized.

Commentary I Ader & Cohen: CNS-imimme system interactions
Conditioning of immunosuppression in the er, there does not seem to be a single experiment on condi-
treatment of transplant tissue rejection tioned immunomodulation (CIM) in which treatments of experi-
mental and control animals differ only with respect to condition-
H. D. Kimmel ing. Although in most studies on CIM experimental and control
Department of Psychology, University ol South Florida, Tampa, Fla. 33620 animals can be distinguished by one factor on operational
grounds, this may not be sufficient to make an inference to
Ader & Cohen's (A & C's) review of CNS-iimnunc system
learning. We will give a few examples to illustrate that in the
interactions summarizes and evaluates recent research on the
experiments on CIM, manipulations in control groups are not
conditioning of immune reactions. Their scholarly and compre-
limited to the elimination of conditioning but affect other factors
hensive treatment deliberately remains within appropriately
as well (see also Rescorla 1967; Seliginan 1969).
modest limits of speculation, concentrating mainly on empirical
The typical design used by Ader and his colleagues is outlined
findings. They conclude their assessment with the assertion that
in Bovbjerg, Ader, and Cohen (1982). The critical comparisons
current research increasingly documents the possibility of plas-
are said to be between the experimental and two control groups:
ticity in the neural modulation of both enhancement and sup-
group CS (conditioned and reexposed to CS [conditional stim-
pression of immunity to specific antigens. The only speculations
ulus]) and CS 0 (conditioned but not reexposed to CS) or NC
they allow themselves to make are theoretical, concerning
(unconditioned, i.e., receives US [unconditional stimulus] but
possible neurophysiological mechanisms underlying the ob-
is exposed to 'CS' when group CS is exposed to CS). Groups CS,
served effects.
CS0, and NC obviously differ with respect to several aspects (see
A more or less noninvolved commentator, however, may
also Klosterhalfen and Klosterhalfen 1983a).
enjoy the luxury of being somewhat less restrained in speculat-
1) Group CS receives more total CS exposure than groups CS 0
ing regarding possible applications of suppression. With the
or NC. As long as we do not know that the CS has no uncondi-
highly publicized baboon-to-human heart transplant in the
"Baby Fae" case so recently in the news (e.g., Newsweek, Nov. tioned effects whatsoever on the immune system, the total
26, 1984), A & C's target article suggests the exciting possibility number of CS exposures should be kept constant across groups.
that an otherwise innocuous conditioned stimulus might be 2) To demonstrate Pavlovian conditioning (i.e., that an ob-
presented in close temporal contiguity to an immunosuppres- served effect was due to the pairing of CS and US), the control
sant drug (in Baby Fae's case, the drug was cyclosporin), in group should receive the same number of CS and US in an
advance of the transplant surgery, so that conditioned antirejec- unpaired (or noncontingent) manner. This kind of control has
tion, or conditioning-assisted antirejection, methods might be actually been used in the following experiments on CIM, but
attempted. not without introducing some new factor. In Ader and Cohen's
The reason why conditioned immunosuppression in animal- (1982) experiment on an autoimmune disease, the groups unfor-
to-human transplantation might be a genuine behavioral-med- tunately differ in the number of handling and intraperitoneal (ip)
ical technological advance is that the use of cyclosporin and injections. Gorczynski, Macrae, and Kennedy (1982) used dif-
related antirejection drugs is known to place the transplant ferent USs for their paired versus unpaired groups. In Kloster-
recipient's kidneys under severe stress. Indeed, in the Baby Fae halfen and Klosterhalfen's (1983b, 1984) experiments on adju-
case (and in some other recently publicized organ-transplant vant arthritis (which is commonly regarded as the result of a
cases), renal system failure contributed to the patient's eventual DTH (delayed-type hypersensitivity) reaction or, alternately,
demise. As A & C have pointed out, conditioned modulation of an autoimmune disease), groups drank different amounts of a
the immunoregulation system may be accomplished without saccharin-vanilla solution (SV). What is lacking is the demon-
accompanying (mediating) changes in adrenocortical steroid stration of CIM in an experiment in which the amount of
levels. It is thus quite possible that some substantial portion of saccharin (Sac), SV, or any other substance used as CS (e.g.,
the antirejection effect of a drug like cyclosporin might be odor) is controlled and in which conditioned and unconditioned
achieved without the imposition of unusual stress on the kid- groups do not differ procedurally in any other respect.
neys. At the very least, A & C's article suggests that research on 3) Group CS consumes less fluid than groups CS 0 or NC. In
possible attenuation of damage to the renal (or other) system by several experiments a two-bottle-choice situation was estab-
means of the use of conditioned immunosuppression (or condi- lished to control for the amount of fluid consumption (Ader,
tioned antirejection) would be highly appropriate. Cohen & Bovbjerg 1982; Bovbjerg, Ader & Cohen 1982, 1984;
Animal-to-human (and human-to-human) organ transplan- Cohen, Ader, Green & Bovbjerg 1979). However, in these
tation has become almost medically ubiquitous, encompassing experiments groups differed with respect to the amount of
much more than the most-highly publicized heart transplant handling and ip injections.
cases. In many of these transplant cases, the recipient is not in a 4) Group CS shows an elevation in plasma corticosterone to
medically critical condition prior to the surgery. Thus, the the CS (Ader 1976) and presumably also other - as yet unex-
conditioning procedure could be carried out without life-threat- plored -humoral alterations. Several attempts have been made
ening insult to the recipient. The effectiveness of the condition- to test the assumption that CIM is mediated by conditioned
ing procedure could be fully evaluated prior to the surgery as arousal due to 'CS' presentation. Sac previously paired with
well (i.e., did the conditioned immunosuppression "take"?). LiCl (instead of cyclophosphamide, CY) did not suppress an in
Thus, a program of combined conditioned and unconditioned vitro humoral response to SRBC (sheep red blood cells) (Ader &
immunosuppressions could be selected in advance for use fol- Cohen 1975); however, Sac suppressed an in vivo SRBC-DTH
lowing the transplant surgery. reaction (Kelley, Dantzer, Mormede, Salmon & Aynaud 1984),
although LiCl itself (presented somewhat earlier than the 'CS')
The possibilities of this type of application are almost endless.
did not suppress either of the immune responses. In their
Ader & Cohen deserve our thanks for their heuristic article.
extinction study, Bovbjerg et al. (1984) conclude that condi-
tioned immunosuppression could be dissociated from condi-
tioned taste aversion. However, stress effects (from ip injec-
On demonstrating that conditioned
tions) may act differentially in experimental (immunocompro-
immunomodulation is conditioned mised) and control rats. This hypothesis is supported by the
results of Sato, Flood, and Makinodan (1984), who showed that
Wolfgang Klosterhalfen and Sibylle Klosterhalfen
stress treatments affected the immunological response to SRBC
Institute of Medical Psychology, University of Dusseldorf, 4000 Dusseldorf, only in immunocompromised animals.
Federal Republic of Germany
A design that is not a perfect solution but a refinement in
Ader & Cohen (A & C) cite an impressive bulk of data suggesting dealing with the problem of conditioned arousal is one in which
that learning processes can modify immune responses. Howev- all animals receive two USs (e.g., CY and LiCl) paired with two

CSs, respectively (cf. Rcscorla & Holland 1976); then one half This functional analysis predicts that both immunosuppression
should be tested for conditioned CY effects and the other half for and immunoactivation improve the animal's ability to interact
conditioned LiCl effects separately, rather than testing effects of with the conditioning agent. Although antigen-induced immu-
the CSs sequentially in one group (Russel, Dark, Cummins, noactivation is consistent with Hollis's approach, unless some
Ellman, Callaway & Peeke 1984). general adaptive function can be demonstrated for CY-induced
We believe that the phenomenon of CIM is so important that immunosuppression, this observation is problematic for func-
every effort should be made to control for nonassociative contri- tional interpretations of conditioning.
butions that might artefactually affect immune responses and Another area of concern in drug conditioning research rele-
erroneously lead one prematurely to infer a learning process. vant to the studies reviewed is the nature of the CS. The stimuli
Trust is good, control is better. manipulated in drug conditioning studies have included audito-
ry, visual, tactile, thermal, olfactory, and/or gustatory compo-
ACKNOWLEDGMENT nents. Although investigators may focus on any one of these
We thank J. Bruce Overmier for his critical comments on an earlier stimulus dimensions, the procedures used often inadvertently
version of the manuscript. manipulate additional stimulus components that may not be
identified explicitly. This issue may be important because there
is considerable evidence in the conditioning literature that the
development of a CR depends, in part, on the modality of the
Conditioning the immune system: New CS. For example, certain stimulus components are selectively
evidence for the modification of associated with specific UCRs and elicit specific CRs. Thus,
aversions produced by association with drug-induced illness are
physiological responses by drug-associated often specific to gustatory or olfactory components of a CS (e.g.,
cues Garcia & Koelling 1967). In contrast, investigators of drug
conditioning of physiological responses have emphasized the
Marvin D. Krank association between auditory or visual components of the CS
Department of Psychology, Mount Allison University, Sackville, N.B. EOA and the drug UCR (e.g., Siegel, Hinson & Krank 1978). A & C's
SCO, Canada
review of conditioned suppression of the immune system by a
Ader & Cohen (A & C) provide a convincing review of data taste CS associated with CY suggests a departure from the
indicating that Pavlovian conditioning modifies activity in the emphasis on environmental CSs in many drug conditioning
immune system. This commentary will discuss the implications studies.
of the reviewed data within the larger context of drug condition- The concentration on taste as the relevant CS in studies of the
ing. Conditioned modifications of the immune system add to the immune system appears to be appropriate for much of the
existing evidence that many physiological responses are sen- reviewed data, but it should not distract one from considering
sitive to the acquisition of associations between environmental the possibility that other cues may also enter into associations
events, conditioned stimuli (CSs), and the systemic effects of with conditioning agents such as CY. Lack of attention to
drug administration, unconditioned stimuli (UCSs) (for reviews potential environmental CSs may have important consequences
see Siegel 1983; Eikelboom & Stewart 1982). Because the for the outcome of a conditioning procedure (see, for example,
conditioning procedures used and the type of response modifi- the discussion of Gorczynski et al. [1984] below). Perhaps most
cations acquired are similar to but distinctive from those typ- important is the possibility that conditioning effects in the
ically studied in drug conditioning, the present observations immune system may appear small because they are masked by
may have implications for several theoretical issues that have the presence of environmental cues. Environmental cues could
arisen in the drug conditioning literature. In addition, existing mask conditioning effects either by influencing comparison
studies in the drug conditioning framework provide a valuable control conditions or by exerting an associative competition with
heuristic for predicting the range of conditioning effects that the nominal taste CS (e.g., Braveman 1979).
may be revealed within the immune system. The drug conditioning literature may be helpful in suggesting
A major theoretical debate addressed by drug conditioning possible explanations of what might otherwise appear to be
research is the nature of the conditioned response (CR) and its enigmatic findings in the conditioning of the immune system.
relationship to the effective UCS and unconditioned response For example, our work with conditioned tolerance to opiate
(UCR). Although drug CRs sometimes mimic the uncondi- effects has demonstrated the importance of conditioned inhibi-
tioned effects of the drug, frequently they are opposite or tion (Siegel, Hinson & Krank 1981). When an environmental
compensatory to the drug's effects (Siegel 1983). Eikelboom and stimulus reliably signals the absence of morphine for a period of
Stewart (1982) have suggested that the locus of action of a drug several hours, tolerance is not displayed in the presence of that
determines when drug mimicking and drug compensatory CRs stimulus. Moreover, tolerance development is delayed in the
should occur. A drug that exerts its action on the afferent side of presence of that stimulus when morphine is later presented
the CNS (central nervous system) will condition CRs that mimic repeatedly in conjunction with it. These findings indicate that
the drug UCR. A drug that exerts its action on the efferent side the environmental cue had acquired the ability to inhibit ac-
of the CNS will condition compensatory CRs that are opposite to tively the display of morphine tolerance.
the drug UCR. Conditioned modifications of the immune sys- Thefindingsreviewed by A & C suggest that such inhibitory
tem mimic the UCR produced by an immunomodulating agent. phenomena will be revealed in studies of conditioned immune
(The exception in the Gorczynski, Macrae, and Kennedy (1984) reactions. In fact, the presence of extinction and partial rein-
study has an alternative interpretation, discussed below.) The forcement effects in conditioned suppression of immune re-
Eikelboom and Stewart (1982) model clearly predicts that both sponses suggests that inhibition is active in the immune system.
cyclophosphamide (CY)-induced immunosuppression and anti- More suggestive evidence comes from the study of Gorczynski
gen-induced immunoactivation should be mediated by afferent et al. (1984). Although A & C have chosen to consider this study
sites of action. Although it is unclear where CY exerts its with respect to the levels of corticosteroids present in the
immunosuppressive action, antigens would be expected to af- various segments of the diurnal cycle, a simple consideration of
fect peripheral effector mechanisms, and hence conditioned the environmental stimuli present at training and testing pro-
immunoactivation would be inconsistent with the Eikelboom vides a clearer interpretation of the outcome. For two condi-
and Stewart model. In an alternative view, Hollis (1982) has tions, 7:00 A.M. and 1:00 P.M., training took place in the light,
proposed a general theory of CRs based on their functional day portion of the cycle. Testing was similarly conducted in the
properties. Specifically, she has argued that CRs prepare the day portion of the cycle at 1:00 P.M. In contrast, in the third
animal to optimize its interaction with the effects of the UCS. condition training was conducted at the start of the dark, night

portion of the cycle. For animals in this condition not only were shops not considered fundable as recently as five years ago have
many of the temporal and environmental cues not available at been found worthy of support and their proceedings publishablc
testing, but also the cues at testing had been consistently paired (Levy 1982; Melnechuk 1983; Cooper 1984; Guillemin, Cohn &
with the absence ofCY. This operation is similar to the one that Melnechuk 1985). Most of the workshops held in the United
had effectively produced an inhibitory CS in our previous States and abroad have comprised only advocates. Accordingly,
morphine tolerance study (Siegel et al. 1981). The present one in which leading workers in the field were confronted with
analysis suggests that the apparent conditioned enhancement of an audience of neutral (i.e., skeptical) immunologists and other
the immune system in this condition may actually reflect the relevant biologists was recently organized by Roger Guillemin,
inhibitory properties of the prevalent testing conditions. Ana- a founder of neuroendocrinology; the late Richard K. Gershon,
lyzing this study in terms of inhibition makes better sense of the discoverer of suppressor T lymphocytes, and me. (After
data than proposing that the direction of the immune CR Gershon's death, Melvin Cohn, a founder of the Salk Institute,
changes as a function of circadian rhythmicity in corticosteroid replaced him as cochairman.)
levels. Immunologists, who have had to be genetically minded, have
Finally, conditioned modifications of the immune system, in general discounted psychoneuroimmunology; this negative
like drug conditioning in general, appear to have practical view has recently been expressed in an editorial (Maddox 1984).
implications. Of particular interest is the novel suggestion that In the postpresentation and general discussions at our meeting
partial reinforcement could be an effective procedure for reduc- (Guillemin et al. 1985), the skeptics found it hard to imagine
ing the amount of toxic drug necessary to maintain treatment of a how conditioned immunosuppression could work; they crit-
disorder such as autoimmune disease. It should be noted, icized the unsophisticated sampling of stress-evoked changes in
however, that partial reinforcement interferes with the acquisi- blood hormone levels and the use of changes in lymphocyte
tion of drug CRs (Krank, Hinson & Siegel 1984). Consequently, responses to mitogens as an assay of psychological effects; and
partial reinforcement procedures might be better applied as they pointed out that "no investigator had yet delineated a
maintenance procedures after CR acquisition had been com- complete causal chain showing that a particular neural manip-
pleted. Nevertheless, Ader & Cohen's demonstration that im- ulation, psychological or physical, caused a particular change in
munosuppressive CRs can enhance survival of autoimmune some intermediary that then brought about a particular change
disease complements our work showing that conditioned drug in an immune response" (Melnechuk 1984b). However, Wigzell
CRs are crucial to the survival of the "overdose" effects of heroin (1985) stated that "reactions and systems regulated via the CNS
(Siegel, Hinson, Krank & McCully 1982). Both these findings may selectively affect some more-or-less well-defined functional
underscore that although conditioning effects are sometimes area of the immune responsefield."Cohn, though adamant that
small and difficult to measure, their contribution to successful this area could not include those aspects of immune regulation
adaptation should not be underestimated. that are antigen-dependent, on the grounds that neurons lack
receptors for distinguishing antigen conformation specificities,
conceded that the area of potential neural influence might
include those aspects of immunity that are independent of
antigen recognition, such as the development, migration pat-
Progress toward a general theory of health terns, and response thresholds of lymphocytes.
This last is a crucial point. If messengers of neural origination
Theodore Melnechuk can modulate lymphocyte response thresholds, the "mind" can
Department of Neurosciences, University of California, San Diego, La Jolla, affect immune tolerance and hyperreactivity. Optimal regula-
Calif. 92093
tion of immune responsiveness to antigens is literally vital, as
Ader & Cohen (A & C) have effectively reviewed their own and shown by AIDS, an immune deficiency disease. Immune re-
other workers' evidence that it is possible to suppress and sponses that are too weak to nonself entities from without permit
enhance immune responses by Pavlovian conditioning. They infections; responses too weak to nonself entities from within
also cite other kinds of evidence that points toward the reality of permit cancers; responses too strong to nonself entities from
psychoneural modulation of immunity. Besides the kinds they without permit allergies and potentially fatal anaphylactic
mention, there is still other recent evidence that cortical lesions shock; and responses too strong to nonself entities from within -
affect immune responses and that hemispheric differences are i.e., self entities misperceived as nonself- permit autoimmune
involved (Biziere, Guillemin, Degenne, Bardos, Renoux & disorders.
Renoux 1985; Behan & Geschwind 1985), and that the thymus Assuming that the methodological objections of the skeptics
gland, besides being autonomically innervated from brain stem will shortly be met and the psychoneural modulation of immu-
and spinal cord nuclei, is linked to a sensory nerve circuit that nity (and vice versa) will be established and characterized,
ascends the vagus nerve toward the brain (Bulloch, Roth & several conceptualizations are possible. One is that "perhaps
Cullen 1984). the nervous and immune systems, with their hormonal links in
Despite all the suggestive evidence, however, the scope, both directions, could be considered part of a general adaptive
power, and mechanisms of the modulation have by no means supersystem that must deal with an infinite variety of internal
been elucidated, and it is in some danger of becoming a received and external novelties from antigens to ideas" (Melnechuk
idea independent of its factual basis. Investigators trying to 1983). A second conceptualization would take its departure from
study it have been handicapped by the paucity of funding for this the apparent ability of messengers dispatched neurally to modu-
(as for any) interdisciplinary research. Nevertheless, a large late the responsiveness of immune cells to the local growth
body of knowledge has accumulated and it has an accelerating factors, specific to the immune system, that are released by its
growth rate. A recent computerized search of the literature cells as part of a response to antigen binding. Taken together
made in order to compile a bibliography annotated with ab- with the increasing number of tissue-specific growth and differ-
stracts (Locke & Hornig-Rohan 1983) found that about as many entiation factors found in other tissue systems besides the
items (1,450) were published between 1976 and 1982 as had immune system (James & Bradshaw 1984), it is possible to
been published before 1976. Advances, quarterly journal of the conceive of the psychoneural modulation of growth, and thus of
new nonprofit Institute for the Advancement of Health in New healing, in every organ and tissue. Such a mechanism could
York, which monitors this and other research that studies the underlie "miraculous" remissions (Dowling 1984) and cures
phenomena and mechanisms of psychological effects on health, effected by various noncanonical therapies in this and other
cultures.
has found almost 150 more psychoneuroimmunological ab-
stracts to publish in its first year (1983-84). Such modulation would be caused not only by stress but by
As this growing body of work has attracted attention, work- "eustress" - by positive emotions, which, as Norman Cousins

Commentary I Ader& Cohen: CNS-immune system interactions
(1979) has been arguing, have been ignored as experimental permit animals to avoid foods that make them sick and thus
variables by establishment science and even by psychosomatic avoid sickness. Hence the main conditioning is that the taste
medicine, both of which have concentrated on negative emo- becomes aversive. Presumably the taste does not strongly elicit
tions in the 50 years, from Dunbar (1935) to Teinoshok, Van CRs in preparation for an inevitable US because its role is to
Dyke, and Zegans (1983). A small number of investigators have prevent the US from occurring. In contrast, if a drug state, such
begun to study how positive emotions might affect health by as pentobarbital, is paired with toxicosis, it does not become
means of the the modulation not only of immune-system reg- aversive (Revusky, Taukulis, Parker & Coombes 1979). In fact,
ulatory factor effects but also of the activity of growth factors in even a drug state that directly precedes toxicosis is not very
other tissues (Melnechuk 1984a). I believe that such a psycho- useful as a cue for cessation of eating or drinking (Revusky,
biology of growth and healing, which includes the psychoneural Coombes & Pohl 1982; Rusiniak, Garcia & Hankins, 1976).
modulation of immunity, will come to be itself included in an Instead, a drug taste that precedes toxicosis tends to elicit an
eventual multidisciplinary, multicultural, and multifactorial antisickness CR that prepares the animal for an inevitable
general theory of health. sickness (Lett 1983). I infer from this, without direct evidence,
that a state ought to be a far less effective signal than a drug state
for a conditioned immune response.
Questions about conditioned

immunosuppression and biological
adaptation The meaning of learning
Sam Revusky Anthony L. Riley
Department of Psychology, Memorial University of Newfoundland, St. Psychopharmacology Laboratory, Department of Psychology, American
John's.Newfoundland A1B 3X9, Canada University, Washington, D.C. 20016
The apparent conditioning of immunosuppressive reactions to Ader and Cohen's (A & C's) review offers a clear and thorough
saccharin as a result of the pairing of saccharin drinking with summary of the current research findings on conditioning and
cyclophosphamide (CY) seems intuitively unreasonable to me the immune system. What remains to be determined is exactly
on the basis of various things I believe about conditioning and what such findings mean. Although many interpretations may
particularly taste aversion learning. As a result, I have two be available, several possibilities immediately come to mind.
questions for Ader & Cohen (A & C) that are unfair by most For example, one clear conclusion is that the immune system is
standards and may well reflect my ignorance of immunology; subject to conditioning, i.e., the immune system is condition-
however, I believe that if the authors can supply even spec- able. This is certainly evident both from the fact that under a
ulative answers, they will strengthen their case for conditioned wide range of conditions animals receiving pairings of a sac-
immunosuppression. (Note that I am not dealing with condi- charin taste with an injection of an immunosuppressant drug
tioned enhancement of immune reactions, which is not coun- display conditioned immunosuppression, and from the fact that
terintuitive to me.) these responses are subject to a number of manipulations known
1) Since the immunosuppression produced by CY is not due to affect conditioning, for example, partial reinforcement, ex-
to a UR (unconditioned response) produced through the ani- tinction, and repeated trials (see Mackintosh 1974). A second
mal's nervous system, why is the CR (conditioned response) conclusion is that such conditioning contributes to immune
immunosuppression rather than excitation of the immune sys- functioning, i.e., the actual display of immune responses in a
tem? In agreement with recent analyses of Pavlovian condition- normal, functioning organism is in part mediated by condition-
ing of internal responses (e.g., Eikelboom & Stewart 1982; ing. Although possible, the latter conclusion must remain a
Siegel 1975), I would consider the immunosuppression pro- speculation at this point in time.
duced by CY as an external assault upon the rat (a US: uncondi- Questioning the potential role of conditioning in normal
tioned stimulus). The UR would be a defensive response to this immune function may seem at odds with the clear acceptance of
US and the CR would be an anticipatory defensive reaction. So I the fact that such responses are conditionable. The two would
would expect conditioned excitation of the immune system appear to go hand in hand. The caution in accepting that
where A & C report conditioned immunosuppression. My conditioning may mediate or contribute to the normal display of
expected result would, as far as I can see, be adaptive, since the immune responses is based on a number of issues. For example,
conditioning would counter the usually harmful immunosup- if conditioning is contributing to immune functioning, some
pression of the CY. Of course, both conditioning and immune reasonably strict relationships would have to be maintained
reactions are often maladaptive in specific situations. But there between the stimuli surrounding drug exposure and the effects
is always a burden of proof to explain maladaptations in terms of of the drug itself. As noted by Mackintosh (1983), conditioning is
a larger adaptive process. dependent upon the contingent (and possibly contiguous) rela-
More generally, there seems to be little in the way of a larger tionship between the CS (conditioned stimulus) and US (uncon-
scientific context for conditioned immunosuppression. Taste ditioned stimulus). (The importance of contingency is actually
aversion learning is an obvious adaptation to allow association demonstrated by A & C, who note that when the US followed
between the flavor of food and the delayed physiological conse- the CS only 50% of the time conditioned immunosuppression
quences of ingestion. Its fit into the larger context of the was substantially reduced, almost to the level of nonconditioned
regulation of food intake gives it face-validity. I am not aware of control subjects.) How can either contingency or contiguity be
such a larger context for the conditioning of immunosuppression assumed to be present and reliable in the setting in which
to a saccharin taste. These shortcomings are particularly serious humans would be administered an immunosuppressant drug,
because, in comparison to conditioned taste aversion, the effect especially when it is not clear in the human situation exactly
seems very weak. what stimuli would be paired with the effects of the drug?
2) What is the relationship of CS modality to conditioned A second issue that might call into question the likelihood
immunosuppression? I was surprised that a saccharin taste was (and thereby the potential contribution) of conditioning in the
an effective CS for immunosuppression (CY was the US) and human setting concerns the parameters under which condi-
would expect a drug state produced by injection to be a better tioned immunosuppression has been reported in the laboratory.
CS. My surprise was based on the following somewhat spec- A & C report a range of studies demonstrating conditioned
ulative ideas about conditioned taste aversions and selective immunosuppression, some dealing with humoral immune re-
association (Revusky 1984). The biological role of taste is to sponses, others dealing with cell-mediated responses. Indepen-

dent of the specific response system examined, conditioned are neither more nor less likely to overinterpret their reduc-
immunosuppression was reliably produced only in a specific tionist and in vitro successes than are physiologists who study
paradigm, the conditioned taste aversion procedure (sec Riley nerve-muscle preparations or cndocrinologists who do
& Tuck, in press). The taste aversion paradigm refers to the Scatchard plots to measure hormone-receptor binding.
situation in which the animal is injected with some toxin shortly 2. I believe A & C's review doe not distinguish clearly or
after exposure to a drinking solution (or food). This procedure is sharply enough between the study of the inductive phase and
somewhat unique in that it supports very robust conditioning, the elicited or effector phase of immune responses. Indeed, the
often under circumstances in which more traditional paradigms experimental separation may be blurred and difficult.
do not, for example, when long delays intervene between the Psycho (neuro) modulation of effector responses is hardly
CS and US or when only a single conditioning trial is given (see controversial. Indeed, the placebo effect has confounded the
Garcia & Ervin 1968). As described by A & C, when a more evaluation of treatment of allergy and, viewed differently, can
traditional paradigm is used in the analysis of conditioned be an effective and important form of therapy. [See also Rosen-
immunosuppression, conditioning is only marginal, even when thai & Rubin: "Interpersonal Expectancy Effects" BBS 1(3)
multiple trials are given. It is not clear what this apparent 1978.]
specificity means in relation to the limitation of a role for My favorite concerns the guinea pigs that were being condi-
conditioning in normal immune functioning, but until the effect tioned in the classical Pavlovian paradigm to express anaphylac-
is clearly demonstrated in other conditioning paradigms it is tic symptoms resembling those seen in bronchial asthma, with a
hard to imagine a scenario in which the laboratory condition bell as the conditioned stimulus. In time, the investigative team
could be approximated in a natural setting with humans. recognized that the entrance of a white-coated biped into the
Questioning the possible role of conditioning in the natural animal quarters was the guinea pigs' choice of conditioned
display of a phenomenon, despite a clear demonstration of its stimulus! The animals had learned to wheeze before rather than
conditionability, is not an issue limited to conditioned immu- on cue!
nosuppression. A similar point could be made for a number of These comments seem to me to pertain to the conditioned
conditioned behaviors. For example, although stimuli that relia- modulation of disease states, including lupus-like conditions or
bly precede either nonprecipitated or naloxone-precipitated graft rejection.
withdrawal will elicit withdrawal themselves, even in animals 3. There is a strange dilemma in this business. In some sense,
that have had no morphine in their system for weeks (see Wilder to accept a finding is to diminish its significance. The unex-
1973a, b), there is little evidence that conditioning underlies the plained, mysterious- quality of certain observations endows
withdrawal response in animals (see Zellner, Dacanay & Riley them with an importance that understanding threatens to strip
1984). Also, none of this suggests that conditioning cannot away.
contribute to the functioning of the immune system, only that The conditioned diminution in antibody response does not
the question remains open. It is certainly possible that in normal subvert any of the dogma of fundamental immunology, which
immune activity, a myriad of factors are acting and interacting, asserts that antigen as initiator plays a specific and selective role
one of which may be conditioning. in inducing synthesis of preprogrammed progenitor cells, each
Of the two possible interpretations that one could draw from of which expresses a specificity arrived at by essentially random
Ader & Cohen's review, the one that is clearest is that immune combination of sets of minigenes. Conditioned enhancement or
responses can be conditioned. The one that may have the most depression of immune response occurring as learning, by neural
importance is the conclusion involving the extension and ap- or hormonal means, or pharmacologically induced processes fits
plication of the basic data. It is a question that can be (see Siegel easily into current immune dogma. Such is the situation with
1984) and should be discussed. respect to the cylcophosphamide studies.
4. Ader & Cohen perform an important service in defining
and refining experimental protocols that facilitate the study of
neural and humoral modulation of immune responses. The
detailed analysis of how such effects come about will, in my
The condition of immunology view, require reductionist approaches. Assessment of the
clinical importance of such changes as can be achieved, which
Leon T. Rosenberg are not of great magnitude, will require experiments carried out
Department of Medical Microbiology, Stanford University School of at the most complex of levels, that is, organisms subject to
Medicine. Stanford, Calif. 94305 infectious or other challenge.
Let me first state my viewpoint. I am an immunologist with an
abiding interest in the host-parasite interaction. I have been
teaching immunology to undergraduates, graduate students,
and medical students for the past 25 years. During that quarter
century the subject has been transformed. Immunology used to Behavioral conditioning of
be thought of as subsidiary to or a branch of microbiology. Now, immunomodulation
the discipline commands departmental status in many institu-
tions. Key issues in molecular biology, in the regulation of Thomas Roszman
differentiation, and in the study of intercellular communication Department of Medical Microbiology and Immunology, University of
are fruitfully addressed with research in immunology. The Kentucky Medical Center, Lexington, Ky. 40536-0084
subject constitutes the cutting edge (to use a cliche) of bio- The immune system has been characterized as a self-regulatory
medical research. system that constantly monitors its own potential through the
Such success promotes an ecumenical attitude. Psychoim- action of various suppressor-helper systems (Gershon 1974),
munology, neuroimmunology, immunopharmacology, and im- idiotypic networks (Jerne 1974), and genetic controls (McDevitt
munotoxicology are all most welcome. 1980). Stated more directly, the immune system is thought to be
Now as to the Ader & Cohen (A & C) target article: autonomous and requires no input from the central nervous
1. A point is made about the "concept of autonomy" of system. Recently, data have accrued that indicate that factors
immune responses. I say "thank God" for the in vitro systems under neural control, that is, hormones (reviewed by Maestroni
with which one can study so effectively so much of the immune and Pierpaoli 1981), neurotransmitters (reviewed by Hall and
response. The use to which in vitro systems are put in no way Goldstein 1981), and sympathetic nervefibers(Williams, Peter-
suggests that "autonomy" is the reductionist's creed. I think the son, Shea, Schmedtje, Bauer & Felten 1981), may serve as
concept of autonomy is a straw man. Immunologists as a breed intermediates in a neuroimmunomodulatory network. Of a

more demonstrational nature are the observations of Ader and conditions of the conditioned immunosuppression phe-
colleagues that illustrate that immunological responses are mod- nomenon. This can be accomplished with a series of experi-
ifiable when certain behavior conditioning techniques are ap- ments that begin with a straightforward parametric examination
plied (Ader & Cohen 1981). These experiments also lend strong of conditioned-stimulus/unconditioned stimulus intensity rela-
support to the concept of a relationship between the brain and tionships and proceeding logically toward an examination of the
the immune response. phenomenon with more complex Pavlovian paradigms such as
Ader & Cohen (A & C) used the taste aversion paradigm to compound conditioned stimulus conditioning and contextual
establish that the immune response can be behaviorally modi- conditioning. In addition, A & C should be encouraged to
fied. Using this paradigm they have over the years, examined a develop other behavioral conditioning paradigms, particularly
number of behavioral and immunological parameters and dem- those that result in enhanced immunomodulation.
onstrated the validity of this model. Of all the areas of neuroim-
munomodulation, however, perhaps the most difficult for im-
munologists to come to grips with is behavioral conditioning of
immunomodulation. The reason for this is in part that behav- Psychoneuroimmunology,
ioral conditioning experiments have yielded mainly descriptive psychopharmacology, and synthetic
data rather than data that can be used to explicate the mecha-
nism responsible for the phenomenon. However, this has not physiology
deterred attempts to explain these data as all attributable to
"stress." Thus, the immunosuppression observed in the taste Shepard Siegel and Michael T. Scoles
aversion paradigm discussed by A & C could result from in- Department of Psychology, Me Master University, Hamilton, Ont. L8S 4K1,
Canada
creased levels of plasma corticosterone. A & C have conducted a
number of experiments, including adrenalectomizing their ani- A distinguishing feature of the living, intact organism is its
mals prior to conditioning, and they have concluded from their ability not only to respond reflexively to stimulation, but also to
data that glucocorticosteroids are not responsible for the ob- anticipate a stimulus. For example, a variety of interrelated
served conditioning effects. Their conclusion that even if an behavioral, hormonal, and other physiological responses (e.g.,
acute rise in corticosterone does occur it is not sufficient to postural adjustments, sympathetic activation, release of endog-
account for the immunosuppression observed is probably cor- enous opiates) are displayed following a signal for an aversive
rect. However, temporal measurement of plasma cortico- stimulus. These anticipatory reactions are normally beneficial in
sterone levels in conditioned animals would provide valuable that they enhance the organism's ability to deal with the im-
information toward ultimately resolving the issue. pending unpleasant event. These reactions have also been
Related to this stress issue are the results indicating that when implicated, however, in a variety of maladaptive behavioral and
LiCl is used instead of cyclophosphamide in the taste aversion physiological symptoms. Similarly, many hormonal and gas-
paradigm, no suppression of the immune response is observed. trointestinal alterations are exhibited in response to a signal for a
Thus, both agents can be used as the unconditioned stimulus meal [see LeMagnen: "The Metabolic Basis of Dual Periodicity'
and taste aversion is observed with an increase in plasma BBS 4(4) 1981]. Such food preparatory responses aid normal
corticosterone, but, significantly, LiCl does not suppress the digestive functioning, but they may also be involved in patho-
sheep red blood cell (SRBC) antibody response. This is a strong logical symptoms such as obesity and ulceration.
argument against a role for corticosterone in the taste aversion As indicated by Ader & Cohen's (A & C's) review, responses
paradigm. In a recent report using the same procedures de- are also displayed in anticipation of immunological stimuli. The
scribed by Ader, however, the opposite results were found, that majority of the studies A & C discuss report conditioned immu-
is, that LiCl could in fact suppress the SRBC antibody response nosuppressive effects that, in most cases, would be expected to
(Kelley, Dantzer, Mormede, Salmon & Ayaud 1984). The major decrease the animal's chances of survival. It should be noted
difference between these two studies is that Ader's used rats however, that the unconditioned stimulus (US) in these studies
while the other used mice. This is an important experiment and has typically been a synthetic drug to which the animal would
bears repeating. not normally be exposed. It appears that in cases where antigens
While the conditioned immunosuppression of the SRBC have been used as USs, a situation more similar to naturally
antibody response is consistent, reproducible, and statistically occurring events, a potentially adaptive enhancement of im-
significant, it is relatively small in magnitude. It may be argued mune response has been reported (Gorzynski, Macrae & Ken-
that this phenomenon should not be "all or none" but rather a nedy 1982). An integration of these data on the conditioning of
modulation of the normal self-regulating events responsible for suppressive and enhancing effects on immune function with
the initiation or maintenance of an immune response. Greater current theories on the conditioning of other bidirectional
differences between the conditioned animals and appropriate systems (e.g., Eikelboom & Stewart 1982; Siegel 1983; Solomon
control animals might be evidenced, however, if more quan- & Corbit 1974) would be particularly beneficial. Clearly, in
titative immunological techniques were employed. For exam- these and other systems that have in the past been studied
ple, rather than using the agglutination test to measure serum primarily as autonomous mechanisms, information concerning
antibody titers, quantification of splenic plaque-forming cells the role of the organism's past experience is essential to an
would provide a more accurate reflection of the ongoing anti- understanding of normal and pathological functioning and to the
body response. In parallel with these studies it would also be of development of therapeutic tactics.
interest to determine the effects of conditioning on other cel- The existence of immunological conditioning may be ex-
lular parameters such as percentage and absolute number of ploited to alter the course of immune disease. Thus, in certain
thymus-dependent lymphocytes (T cells) and bone-marrow de- regimens of drug administration, it may be possible and
rived lymphocytes (B cells). Functional analysis of these T and B therapeutically useful occasionally to administer drug-predic-
cells would also provide valuable information. Such experi- tive cues, rather than the drug. As discussed by A & C, the use of
ments are a necessary first step in understanding the mecha- such partial-reinforcement schedules of drug administration
nism^) responsible for behavioral conditioning of immunomod- (i.e., interspersing placebo administrations among drug admin-
ulation. istrations), which uses lower cumulative amounts of the drug
The work of A & C demonstrates that the behavioral condi- than the usual continuous-reinforcement schedule of drug ad-
tioning of immune responses is an important and viable phe- ministration, may well be a procedure for maximizing therapeu-
nomenon, but it must be extended beyond the mere demonstra- tic efficacy while minimizing undesirable drug side effects. A
tion stage. It is now necessary to extend this research to other similar partial-reinforcement strategy, based on parallel reason-
paradigms in an attempt to establish the behavioral boundary ing, has been useful in minimizing the deleterious effects of

Commentary I Ader& Cohen: CNS-imrnune system interactions
repeated administration of high morphine doses (Krank, flinson that there is no evidence to support glucocorticoid involvement
& Siegel 1984). In this study, the use of the partial-reinforce- in the conditioning phenomenon.
ment schedule decreased the weight loss and attenuated the The pituitary adrenal system is a highly sensitive index of
development of analgesic tolerance, compared with a continu- environmental change (Hennessy & Lcvine 1979) and is acti-
ous-reinforcement schedule. In this morphine study, as in Ader vated by a broad range of stimuli (Hart, Coover, Shnerson &
and Cohen's (1982) cyclophosphamide study, extinction was not Smotherman 1980; Coover, Satterfield, Smotherman, Steinkc
evaluated. Although A & C indicate that such a study of extinc- & Dorsa 1983). Goldman, Coover, and Levine (1973) have
tion of immunological conditioning following acquisition with demonstrated, however, that this system is capable of bidirec-
various reinforcement schedules is in progress, it is not clear tional shifts. In other words, both elevations and suppressions in
what pattern of results would be expected. Inasmuch as there circulating glucocorticoids occur, and both types of changes can
are conflicting reports concerning the effects of partial reinforce- be conditioned to environmental cues. In trying to rule out
ment on resistance to extinction in traditional Pavlovian condi- adrenocortical hormones as mediating the conditioned immu-
tioning preparations (Mackintosh 1974, pp. 73-75), there is nopharmacotherapeutic effects, A & C have not considered the
little basis in the existing learning literature to predict the role that suppressions in circulating glucocorticoids might play
outcome of an immunological-pharmacological conditioning in the conditioning process.
study of the effect of reinforcement schedule on resistance to Levine and Coover (1976) achieved such suppressions in
extinction. glucocorticoids by placing rats on a schedule in which their daily
Results of recent research suggest that the strength of drug intake of water was restricted to a 15-min period (which took
conditioning can be affected by a variety of nonpharmacological place at the same time each day). After several weeks on this
manipulations of predrug cues, in addition to extinction. Non- regimen: (1) there was a reentrainment of the circadian peri-
contingent presentations of an environmental CS (conditioned odicity of corticosterone in plasma such that peak levels were
stimulus) and a pharmacological US, as well as explicitly un- evidenced during the 60 min before water was presented; (2)
paired presentations of these stimuli, have powerful effects on there was a rapid suppression (within 5-10 min of drinking
the response to the drug (Fanselow & German 1982; Siegel, onset) in circulating glucocorticoids after water was presented;
Hinson, Krank & McCully, 1982). The important therapeutic and (3) these rapid suppressions in corticosterone could be
implications of these findings, as outlined by Fanselow and conditioned. Environmental cues associated with the presenta-
German (1982), may well be relevant to psychoneuroim- tion of water (e.g., the water bottle on the cage, or the presence
munology, as well as to psychopharmacology. [See also Bolles & of the experimenter in the colony room when animals would
Fanselow: "A Perceptual-Defensive-Recuperative Model of normally be watered) became conditioned stimuli and were
Fear and Pain" BBS 3(2) 1980] capable, on subsequent occasions, in the absence of water, of
A & C's findings concerning the importance of the interaction triggering suppressions in pituitary adrenocortical activity.
between conditioned and unconditioned responses in immuno- These findings are pertinent to a discussion of the conditioned
logical phenomena are part of a growing body of data demon- changes in the immune system that A & C summarize. The
strating the importance of conditioned responses in the normal paradigm used by Ader and others for conditioning immune
and pathological functioning of a variety of physiological systems system changes (e.g., taste aversion), involves a regimen in
in many species, including humans. This work on the interac- which water consumption is restricted to the same 15-min
tion of learning and physiology, conducted (until recently) period each day. The maintenance of rats under these condi-
mostly by Eastern European and Soviet scientists, is the foundations for several weeks, apart from any other procedures, should
tion of a "synthetic physiology," "a science of the course of vital be expected to alter the dynamics of the pituitary adrenal
processes in an integral organism during its various natural system. Circadian periodicity would change and reorganize
relations with the surrounding medium" (Bykov 1960, p. 25; itself such that the peak level of corticosterone in plasma would
emphasis added). It would appear that because immunological occur during the hour prior to the scheduled watering. In
events are often accompanied by reliable environmental cues, addition, rats would experience suppressions in circulating
the immunological responses of an "integral organism" can be levels of glucocorticoids when consuming their daily allotment
best understood as a combination of unconditioned and condi- of water. Finally, any disturbance in the colony room prior to
tioned immunological responses. watering or any cue that came to predict watering would trigger
these conditioned glucocorticoid suppressions.
Pituitary-adrenal system involvement in Levine and Coover (1976) showed that these environmentally
induced suppressions in corticosterone titres were transient.
conditioned immune changes: Perhaps Rats that consumed their daily allotment of water showed
suppressions are playing a role prolonged suppressions in plasma titers of corticosterone. Rats
that were not watered but were exposed to the cues associated
William P. Smotherman with watering (e.g., an empty water bottle) first showed the
Laboratory tor Psychobiological Research, Department of Psychology, conditioned suppression in corticosterone and then experi-
Oregon State University, Corvallis, Ore. 97331 enced adrenocortical activation that resulted in a release of
Ader's work has demonstrated that the immune system is corticosterone, which increased circulating levels of cor-
integrated with the body's other defense systems and, more ticosterone above the already elevated presession levels. In a
importantly, that this system is subject to the brain's modulating taste aversion paradigm, rats in control and experimental groups
influences. Specifically, they have provided compelling evi- would show very different patterns of drinking and different pat-
dence that immunological reactivity can be conditioned. A & C's terns of adrenocortical reactivity - perhaps prolonged suppres-
findings are important and they will have far-reaching implica- sions or perhaps transient suppressions followed by prolonged
tions for psychosomatic medicine and psychosomatic research. activations - depending on whether or not they consumed
The possibility of using conditioning techniques in other phar- the fluids available to them and the timing of such consumption.
macotherapeutic domains is most exciting and warrants clinical At this time it would seem a good strategy to investigate
investigation. further the pituitary adrenal system involvement in the condi-
My comments will be restricted to a discussion of pituitary tioning of immune system changes by making use of testing
adrenal system involvement in the conditioning of immune procedures that do not alter the dynamics of the pituitary
system changes (i.e., immunosuppression). In discussing the adrenal system. It is possible that the magnitude of the condi-
adrenocortical mediation of immunopharmacological effects, A tioned changes in immune system reactivity would be more
& C focus on situations where hormones secreted by the pitui- pronounced in an organism with a pituitary adrenocortical
tary adrenocortical system would be elevated. They conclude system that was responding in a more "normal" fashion.

The emerging field of 5. Hormones and other substances regulated or elaborated
psychoneuroimmunology by the CNS should influence immune mechanisms.
6. Immunologically competent cells should have receptor
George Freeman Solomon sites for neurohormones neurotransmitters, neuropeptides, and
Department of Psychiatry, University of California, Los Angeles, Calif. for substances regulated by them.
90024 and VA Medical Center, Sepulveda, Calif. 91343
7. Experimental behavioral manipulation (e.g., stress, con-
It is to Robert Ader's seminal, methodologically elegant re- ditioning, differential early experience) should have immu-
search and his capacity for visionary integration - a rare com- nologic consequences.
bination - that the now rapidly developing field of psychoneuro- 8. Experimental manipulation of appropriate portions of the
immunology largely owes its definition and growing acceptance. CNS should have immunologic consequences.
(The term neuroimmunomodulation, preferred by some, sub- 9. Activation of the immune system (e.g., immunization)
sumes psychic function under neurology.) It is through work should be accompanied by CNS phenomena.
such as his, in essential interdisciplinary collaboration (in Ader's 10. Feedback mechanisms in immune regulation should act,
case, with the bold immunologist Nicholas Cohen), that the at least in part, via CNS mediation.
biopsychosocial nature of all disease, physical and mental, as 11. Factors elaborated by the immune system should affect
envisioned by Ader's distinguished University of Rochester the CNS and substances regulated by it.
mentor and colleague, George Engel, is being clarified through 12. Biochemical and functional similarities might be ex-
gradual understanding of underlying neuroendocrine, immune, pected between substances modulating CNS function and reac-
and other mechanisms. It was most appropriate, thus, that Ader tivity (neuropeptides) and those having comparable actions
added "neuro" to my earlier (1964) term "psychoiminunology," within the immune system (lymphokines).
a then-speculative theoretical integration of "emotions, immu- 13. Thymic hormones regulating immune function should be
nity, and disease " (Solomon and Moos 1964), deriving largely influenced by the CNS.
from clinical research on the emotional and personality factors in 14. Behavioral interventions (e.g., psychotherapy, relaxa-
the onset and cause of the autoimmune disease rheumatoid tion techniques, imagery, biofeedback, hypnosis) should be
arthritis (Solomon 1981a) and from early hints of immune defi- able to enhance or optimize immune function.
ciency states underlying autoimmunity. Ader & Cohen's (A & Whither psychoneuroimmunology? Undoubtedly, much
C's) convincing body of work emphasizes that early pioneering more is to be learned about the similarity and overlap of
efforts, such as those of Metal'nikovov and Chorine (1926) in the communication-modulation-regulation mechanisms within
Soviet Union, should not be dismissed only because meth- central nervous and immune systems. That so many neurotrans-
odology and technology cannot yet "prove" the point. mitters, neurohormones and peptides act at the cellular levels
A & C's convincing demonstration of conditioned immu- within both systems is truly remarkable. A & C wisely suggest
nosuppression made it hard to controvert that the central the importance of early experience on later immune function -
nervous system (CNS) plays a role in immune regulation, direct as we know well to be the case in psychic function. (Levine and I
evidence for which first was reported by Korneva and Khai demonstrated that daily handling of preweanling rat pups en-
(1963) and is now becoming overwhelming. It is most exciting hanced their humoral immune response as adults [Solomon,
that, as discussed by A & C, conditioning phenomena have been Levine & Kraft 1968].) A & C laudably point to the importance
extended from humoral immunity to cellular immunity (Bovb- and relationship of histocompatibility loci for both behavior and
jerg, Ader & Cohen 1982), to enhancement of immunity immunity.
(Gorczynski, Macrae & Kennedy 1982), including natural killer I suggest two other areas as ripe for contemporary psychologi-
(NK) cell activity (Solvason, Ghanta, Hiramoto & Spector 1984), cally oriented research in the effects of psychoneuroim-
and to conditioning of alteration of immunologically mediated munology. One is the effect of behavioral-psychotherapeutic
disease processes (Ader & Cohen 1982; Klosterhalfen & intervention on both immune function and immunologically
Klosterhalfen 1983). In some concluding remarks at a Leningrad resisted or mediated disease processes in humans, a meth-
1982 Soviet Academy of Sciences-sponsored conference on odologically difficult area that must be approached with great
neuroimmunoregulation, a field currently in an almost vertical care but one that is obviously of great potential clinical signifi-
slope of exponential growth and credibility, I stated (as the only cance. The other is correlation of a variety of immunological
psychiatrist present), "We have heard a good deal here about functions with emotional, personality, and coping-style vari-
multiple levels of integration,' including the CNS (even the ables in "healthy" (physical and mental disease-free) individuals
frontal cortex), the autonomic nervous system, the endocrine in order to assess the possibility of "psychoimmunological vul-
system, and the immune system. We have heard nothing about nerability" to disease - with important implications for preven-
thoughts or feelings. Among other things, the CNS thinks and tive medicine. (Moos and I found such correlates with the
feels." presence of "rheumatoid factor," an IgM anti-IgG autoan-
In an in-press paper (Solomon, in press), I list 14 hypotheses tibody, in healthy relatives of patients with rheumatoid arthritis
that should be able to be verified if the CNS and immune system [Solomon & Moos 1965].) The emerging field of psycho-
are linked - a far cry from the single "stress should be immu- neuroimmunology offers hope not only of contributing to treat-
nosuppressive" I set out to verify over 15 years ago (Solomon ment and prophylaxis of disease but of increasing our under-
1969). There are data to support all 14; there no doubt will be standing of cellular and molecular mechanisms in the immune
more data but also more hypotheses. The hypotheses are: system and in the CNS.
1. Personality (as well as genetic) characteristics should influ-
ence the susceptibility of an individual's immune system to
alteration of function induced by exogenous factors, including
experiential ones. Is conditioned immunosuppression an
2. Emotional upset and distress should alter the incidence, adequate research strategy?
severity, or course of those diseases to which there is immu-
nologic resistance (infectious and neoplastic) and those associ- H. Dick Veldhuis and David De Wied
ated with aberrant immunologic function (allergic, autoim- Rudolf Magnus Institute for Pharmacology, Medical Faculty, University of
mune, and AIDS [acquired immune deficiency syndrome]). Utrecht, 3521 GD Utrecht, The Netherlands
3. Severe emotional disturbance and mental dysfunction The central nervous system and the immune system have a
should be accompanied by immunologic abnormalities. certain structural similarity. The assumption that these two
4. Diseases of immunologic aberration should, at times, be systems could cooperate as part of an integrated physiological
accompanied by psychological and/or neurological symptoms. entity within the whole network of adaptive processes poses the

Commentary I Ader & Cohen: CNS-iinmune system interactions
fascinating challenge of studying the regulation of such a cooper- As A & C make clear, well-controlled and well-executed
ative system. Aside from the enormous complexity in each of the experiments demonstrate that immune responses can be modi-
two systems, reflected in the existence of numerous specialty fied by the presentation of otherwise neutral stimuli that have
subdivisions in their two fields of study, the search for interac- been previously paired with (i.e., associated with) treatments
tions requires thorough knowledge of both systems. In the that alter immune function. By and large, the effects of immune
target article Ader & Cohen (A & C) try to unravel one of the responses produced by presenting such classically conditioned
interactions - the influence of learning processes on iinmu- stimuli have been small, as these investigators point out, but in
nocompetence - by using conditioned immunosuppression as a and of itself this does not detract from the validity of the
model. In this commentary we wish to make some remarks on phenomenon they describe. The size of effects bears only on the
the use of these conditioned alterations in iminunological issue of how much difference classical conditioning might make
reactivity. in overall immune function, and this question will remain
A & C consider conditioning a potential immunoregulating unanswered until the phenomenon has been studied quite
mechanism independent of the iminunomodulating effects of extensively.
"stress." It is, of course, inevitable to call into question the term At this early stage in the research, perhaps the major issue is
stress, but, apart from this, conditioning can still be considered how classically conditioned effects on the immune system are
a rather dramatic "threat" to the organism, especially in the mediated, for it is possible that the mediation of effects could
form in which A & C use it (i.e., by pairing an appetitive make the phenomenon less remarkable than it would otherwise
response with the injection of a potent immunosuppressive seem.The importance of this issue accounts for A & C's concern
drug). Although we agree with A & C that, in the present, still with the possibility that the effects that have been found are due
immature stage of research, one has to use research strategies to the conditioning of nonspecific stress responses. In this
that induce large deviations from hoineostasis — deviations that event, changes in immune function would come about simply
only occur in abnormal circumstances - we would strongly because the response that became classically conditioned to the
advocate the use of less dramatic, more "natural" influences on previously neutral stimulus was an emotional response, and this
the organism's adaptive capacity to unravel the relation between in turn caused physiological reactions, such as steroid secretion,
brain and the immune system. which altered immune function. Were this the case, aspects of
The second remark we wish to make is related to the first one immune function would be subject to classical conditioning, but
and concerns the actual use of pharmacological agents, such as this would be the result of the conditioning of emotional re-
cyclophosphamide and methotrexate, antirabbit lymphocyte sponses (such as fear) that have long been known to be subject to
serum, or radiation. Such treatments rather severely impair the classical conditioning.
immunological apparatus for quite a long time, making the The section entitled "Adrenocortical mediation of condi-
recipient a compromised host, even at the time of behavioral tioned immunopharmaeological effects" attempts to deal with
testing. In the context of an "integrated network of adaptive the most conspicuous possibility of this type, namely, that
processes" subserving the maintenance of homeostasis in the conditioned aversive responses gave rise to elevated steroid
brain and immune system, one should try to minimize the secretion which suppressed immune function. The data pre-
impact on the capacity of an organism to cope with the environ- sented, and the interpretations made, marshall a series of
mental circumstances. arguments that secretion of adrenal steroids is not the sole
Since conditioned immunosuppression, as described by A & mediator or even the principal mediator of the classical condi-
C, implies the (in our opinion, avoidable) shortcomings noted tioning phenomenon under discussion in this review. In reading
above, we wish to encourage the use of other behavioral para- this discussion, however, one wonders why such indirect argu-
digms in which the mechanism of brain-iminune system interac- ments are made, and why, given the sophisticated and complex
tion could be studied. For instance, the one-trial learning, research that has been carried out to examine this phenomenon,
passive avoidance test situation (Ader, Weijnen & Moleman emphasis is opt placed on carrying out the relevant conditioning
1972) contains all characteristics of such a behavioral learning experiments in adrenalectomized animals. When we (Keller,
paradigm. The performance of an animal in this test system Weiss, Schleifer, Miller & Stein 1981) were faced with similar
could be coupled with both in vivo and in vitro parameters of questions regarding the mediation of T cell suppression by
immune functioning. In fact, such an approach, which lacks the stressors, Neal Miller and I immediately realized, and conse-
drawbacks of any immunosuppressive measures, has been used quently suggested to our colleagues, that we had no alternative
successfully in a combined effort of the research group of but to repeat our manipulations in adrenalectomized animals to
Ballieux and ourselves (Heijnen, Croiset, Bevers, Veldhuis, determine whether similar effects would still occur. Moreover,
DeWied & Ballieux, in preparation) to study the interaction the methodology exists to maintain low or moderate circulating
between brain functioning and immune competence. levels of steroids (by pellet implants or pumps) to rule out the
possibility that some steroid is needed simply to permit the
effect to occur (hence, a "permissive" effect). It seems to this
commentator that much of the controversy would be eliminated
Conditioned immune responses: How are by carrying out the relevant experiments in adrenalectomized
they mediated and how are they related to animals. A & C do cite one paper that studied adrenalectomized
other classically conditioned responses? animals and reported failure to obtain conditioned effects
(Gorczynski, Macrae & Kennedy 1984). These studies did not,
Jay M. Weiss however, determine whether some circulating level of steroid
Department of Psychiatry, Duke University Medical Center, Durham, N.C. might simply be required in order to permit conditioned effects
27710 to occur. This remains to be done.
Ader & Cohen (A & C) present us with an exemplary and It should also be mentioned that, if effects were indeed
informative review of the phenomenon of conditioning in rela- mediated by the classical conditioning of fear, emotionality, or
tion to the immune system. It should be indicated at the outset "stress responses," it is not necessarily true that these effects
that A & C's review refers to classical Pavlovian conditioning. would be eliminated by adrenalectomy. We have found (Keller,
The term conditioning has also been used in physiology to refer Weiss, Schleifer, Miller & Stein 1983) that effects of stressful
to habituational or adaptational processes that occur when conditions on T cell mitogenesis were still apparent in adre-
organisms are repeatedly exposed to stimuli. What A & C nalectomized animals. Adrenalectomy will eliminate the major
comprehensively review in their target article is the phe- peripheral hormones released in stressful situations (cor-
nomenon by which a previously neutral stimulus comes to elicit ticosteroids and epinephrine), but less pronounced hormonal
or modify immune responses. responses can still occur (e.g., elevations in plasma nor-

Response I Ader & Cohen: CNS-immune system interactions
epinephrine). Thus, in addition to studies using adrenalectomy, fields have progressed in parallel, and until recently, few
it will be necessary to discover and utilize iinmunosuppressive serious attempts had been made to integrate CNS and
techniques that show no evidence of being aversive. Whether immune system function. Our studies of conditioned
such manipulations can be found remains a question, but with- modulation of immune responses represent one such
out their discovery it will be difficult to determine definitively attempt. The impetus for this target article was our desire
whether or not conditioned iinmunosuppression actually de-
rives from conditioned emotional responses. Regarding this (and the desire of the editor) to draw attention to the
issue, loss of conditioned taste aversion will not be a sufficient development of this new interdisciplinary field of
indicator that conditioned "stress responses" are absent. In research.
humans, progressive relaxation techniques have been shown to In responding to the commentators' remarks, we
eliminate avoidance behavior for phobic objects, but at the same would like to say, first, that we hope in the near future to
time autonomic components of the emotional responses can be able to agree with Rosenberg's contention that the
remain intact (e.g., Hodgson & Rachman 1974). Thus, the loss of "autonomy" of the immune system is a "straw man. "
taste aversion could represent a similar phenomenon. Eventually, we hope it will not be necessary to mention
Finally, an interesting aspect, perhaps even a puzzling one, the issue at all, because immunology will be studied and
concerning the phenomenon of the classically conditioned im- taught, or at least recognized, as an integrated part of
munosuppressive response is why the conditioned response
(CR) occurs in the same direction as the unconditioned response physiology. In our experience, however, the comments
(UCR). Whereas this might have been expected a number of made by Cunningham, Melnechuk, and Roszman more
years ago, the concept of opponent processes has received a accurately describe the current reactions of many immu-
good deal of attention in the last few years (e.g., Siegel, Hinson nologists to the development of psychoneuroimmunolog-
& Krank 1978; Solomon 1980). This concept has answered a ical research. We are confident that studies of CNS-
number of questions concerning classical conditioning. The immune system interactions will become accepted -
concept states that in classical conditioning, presenting the even welcomed - as sufficient data accumulate and influ-
unconditioned stimulus (UCS) not only elicits the UGR but also ence the research that immunologists do.
elicits compensatory physiological responses that tend to main- Rosenberg is quite right in pointing out the specific
tain homeostasis in the face of the UCR. In many cases, what and selective role of antigens in initiating immune re-
actually get associated with the CS are the compensatory re-
sponses that serve to prepare the animal for the impact of the sponses, a point emphasized by Ballieux & Heijnen and
unconditioned stimulus, so that the conditioned response is by Melnechuk. We don't know how "easily" our studies
very often opposite in direction to that elicited by the uncondi- of conditioningfitinto current immunological dogma, but
tioned stimulus. Thus, presenting the CS in the absence of the we agree with Rosenberg to the extent that interactions
UCS often results in a response in the direction opposite to that between CNS and immune system processes do not
elicited by the UCS. Applying this concept to the effects of a conflict with existing dogma. In part, the studies of
drug that is immunosuppressive, one might very well expect conditioned alterations in immune function carried out in
that the consequence of presenting a CS that has been paired the 1920s and '30s were rejected because they denied or
with an immunosuppressive agent would be to activate rather minimized the essential role of antigens and argued that
than suppress immune responses so as to maintain homeostasis. the brain was capable of evoking an immune response de
In introducing this question, I do not mean to imply that a
theoretical deduction ought to supplant the data; the data do
novo. Be that as it may, the fact that an exogenous antigen
indeed indicate that the conditioned effect is immunosup- may be necessary for eliciting an immune response is not
pressive. Nevertheless, given that the effect is immunosup- inconsistent with a role for the brain in immunoregulato-
pressive, it becomes a challenge to integrate this type of effect ry processes. The special characteristics and functions of
into an overall concept of how classical conditioning works. This the immune system do not dictate a dualistic approach to
and other challenges remain for this intriguing area of investiga- an understanding of immunity. However sloppy our
tion. language may have been (Engel), we do not believe that it
conveys a dualistic message. It must be recognized (in
contrast to what might be inferred from Engel's com-
ments) that specific immune responses cannot be elicited
by a variety of stimuli; they depend, sooner or later, on
the presence of antigen. We are arguing, as Engel and
Authors' Response other commentators (e.g., Fox, Green) note, that the
homeostatic functions of the immune system are not
independent of other homeostatic mechanisms and that a
complete understanding of immunological function re-
The brain and the immune system: quires the study of the interactions involved.
Conditional responses to commentator Our studies of conditioning provide one avenue for the
stimuli study of CNS-immune system interactions. These stud-
ies have already raised questions about the modulation of
immune responses - and about behavioral processes. The
Robert Adera and Nicholas Cohen"
commentators have focused on several of these issues,
'Division of Behavioral and Psychosocial Medicine, Department of
Psychiatry and "Division of Immunology, Department of Microbiology,
and we have attempted to address these issues below.
University of Rochester, School of Medicine and Dentistry, Rochester, N.Y.
14642 Magnitude and variability of conditioned responses.
Having adopted a rather conservative posture, several
The growth and development of immunology as an inde- commentators (e.g., Elkins, Fox, Krank, Revusky,
pendent field of study has been remarkably rapid. Ad- Roszman, Weiss) remarked further (with different per-
vances in the behavioral and brain sciences have been spectives) on the magnitude of conditioned alterations in
equally rapid. Whatever the historical reasons, these two immune function. The net effect of conditioning on anti-
SRBC (sheep red blood cell) antibody responses has been velop symptoms of this disease significantly sooner than
relatively small, but we should emphasize that the effects males. In BXSB mice, autoimmune disease develops
of conditioning in modifying a graft-versus-host (GvH) rapidly in males and slowly in females. In Mrl-lpr/lpr
reaction were comparable to those of multiple low-dose mice, lupus develops relatively early in life and there is
injections of an immunosuppressive drug. Also, the bio- only a slight sexual dimorphism. Mrl-lpr/lpr mice also
logical impact of conditioning, as reflected by a delay in develop manifest signs of lymphoproliferative disease
the development of proteinuria and mortality in mice early in life (about 12 weeks), whereas congenic Mrl+/+
prone to the development of autoimmune disease, was mice do not develop disease until very late in life (more
quite large. than 12 months). These are but a few of the experimental
While some of the effects may have been small, it models that offer considerable potential for the study of
would be unwarranted to assume, as Anisman & Zachar- interactions between behavior and "naturally" changing
ko do, that the effects of conditioning in modifying im- immune function.
mune function are also variable. Anisman & Zacharko's While studies in young and old animals or congenic
example of the variability obtained by Gamzu et al. (1984) strains that differ in their level of immunocompetence
in a study of stress effects has no direct bearing on the would address the interaction between behavioral manip-
reliability of conditioning data. Indeed, conditioning par- ulations and immune state, they will not suffice to address
adigms would seem to offer the greater opportunities for the issue of whether there is an interaction between the
the definition and precise control of experimental vari- amount of "stress" experienced by the organism and the
ables. There is, of course, variability in our data. If, for level of immunocompetence upon which the stress is
example, we were to delete those control animals that superimposed. Claser & Kiecolt-Glaser take issue with
showed no detectable antibody response following immu- our suggestion that "relatively mild stress" may be capa-
nization with SRBC, the differences between experimen- ble of exerting its effects only in immunocompromised
tal and control groups would be magnified. Since we hosts. They acknowledge that our suggestion is consistent
could not identify the variable(s) responsible for the with some of the animal data; we acknowledge that the
failure of some animals to respond, we retained the data suggestion is apparently inconsistent with their observa-
from all such animals in the statistical analyses of our tions that relatively commonplace stressful events can
modify immunological reactivity in healthy human sub-
results. As noted in our review, Gorczynski, Macrae, and
jects. The inconsistency, however, may be more appar-
Kennedy (1982) cleverly turned the variability among ent than real. Whereas it would seem that the events they
their animals to their advantage and provided a convinc- mention are commonplace, school examinations (and the
ing demonstration of the effects of unreinforced condi- prospect of examinations), for example, may not be
tioned stimulus (CS) exposures on extinction of a condi- "mild" stressors and could perhaps be characterized as
tioned immune response. "chronic" rather than "acute" ones. The definition of
Individual differences in conditionability could, as acute and chronic, of course, is problematical. As noted in
Elkins suggests, be a function of the strain of animals our review, the effects of stress appear to be a function of
used. If we used animals selectively bred for their capaci- several interacting variables, including the intensity and
ty to acquire a taste aversion, we might reduce variability duration (i.e., chronicity) of the stress. Glaser & Kiecolt-
or increase the magnitude of conditioning effects. Thus Glaser also point out that there are species differences in
far, however, neither we nor others (Wayner, Flannery & the sensitivity of the immune system to adrenocortical
Singer 1978) have observed any relation between the steroids and to the nature of stressful stimulation. Thus,
magnitude of the aversive response and the degree of to the extent that one can construe conditioning as
attenuation of immunological reactivity. As a matter of stressful, it is not unreasonable to assume that there are
fact, the available data (Bovbjerg, Ader & Cohen 1984; major differences between a single conditioning trial - or
Wayner et al. 1978) suggest that there may be a dissocia- a series of conditioning trials, for that matter - and the
tion between these conditioned responses. We continue rather massive electric shock stimulation that has been
to use unconditional stimuli (UCSs) that induce taste used to elicit a change in immunological reactivity in
aversions and we continue to measure behavioral as well animals (e.g., Keller, Weiss, Schleifer, Miller & Stein
as immune responses simply to provide independent 1981; Shavit, Lewis, Terman, Gale & Liebeskind 1984).
verification that some association between CS and UCS These apparent inconsistencies point up the need for
has actually occurred (Weiss). We do not know for a fact parametric analyses of the effects of stress on immune
that a conditioned alteration in immunological reactivity function.
cannot occur in the absence of an observable conditioned
modification of behavior.
The level of immunocompetence upon which condi- Methodological Issues. Several methodological issues
tioning (or any other manipulation) is superimposed may were raised by Klosterhalfen & Klosterhalfen. The first
be the most significant source of variability. Engel's concerns the effects of saccharin perse on immunological
suggestion that one could assess the effects of this source reactivity. In some studies, placebo-treated animals re-
of variability by taking advantage of the apparent natural ceived saccharin (instead of plain water) when experi-
decline in immunocompetence during aging is an excel- mental animals were provided with saccharin; in other
lent one. Immunomodulating drugs used as UCSs might experiments, placebo-treated animals drank plain water
well have different effects on young and old animals. throughout the experiment. We have not noted any
Other circumstances in which altered immune respon- difference between these conditions, but that is probably
sivity occurs "naturally" also provide valuable model beside the point. Given that cyclophosphamide (CY) has
systems. New Zealand (NZBxNZW)Fl mice develop au- residual immunosuppressive effects that are reflected in
toimmune systemic lupus erythematosus. Females de- an attenuated immune response in Group CSo and Group

NC relative to placebo-treated animals, Groups CSo and all of which were obtained from "immunocompromised"
NC become the relevant comparison groups for assessing groups.
the effects of conditioning. The fact that Group CSo does Klosterhalfen & Klosterhalfen recommend the para-
not receive as many exposures to saccharin as Group CS digm in which animals are conditioned by pairing differ-
(or NC) is by design; Group CSo is included to control for ent CSs with drugs that do and do not affect immune
the effects of the experience of conditioning per se. As responses. This procedure was used effectively by Rus-
such, it is a stringent control condition. It cannot, as sell, Dark, Cummins, Ellman, Callaway, and Peeke
Klosterhalfen & Klosterhalfen suggest, be excluded from (1984) to condition the release of histamine. There is
the experiment without leaving open the possibility that considerable merit in this approach as long as one keeps
the combination of saccharin and CY somehow acts syn- in mind the distinction between how conditioned altera-
ergistically to depress immunity. Also, in the experiment tions in immunological reactivity can and do come about.
(Ader, Cohen & Bovbjerg 1982) singled out by Kloster- Such a design, however, does not obviate the need for a
halfen & Klosterhalfen there was no NC group (only a CSo group to assess the effects of conditioning per se.
CSo group), and there was no difference in the number of
exposures to saccharin between Groups CS and P. Stress and adrenocortical effects. The attempt to in-
Klosterhalfen & Klosterhalfen also assert that in ex- terpret conditioned changes in immunological reactivity
periments in which a preference testing procedure was as a stress-related phenomenon is based on the inferred
used - and in the study of autoimmunity - different adrenocortical effects of manipulating animals and the
amounts of handling and numbers of intraperitoneal (ip) assumption that an elevation in adrenocortical steroids
injections were administered to different groups. Per- results in a suppression of immune function (which, based
haps we did not detail sufficiently the rationale for these upon the literature cited in our review, is not an appropri-
procedures in the review. Group CSo did receive less ate generalization). Although the understanding of the
manipulation than the experimental group because an ip effects of stress is exaggerated, the concept that "stress
injection (and the attendant handling) would have con- affects immunity" is as well known and acceptable to
stituted reexposure to a part of the CS and thereby immunologists as the concept that "behavioral condition-
violated the definition of Group CSo. In the study of ing effects changes in immunity" is foreign. It is hardly
autoimmune disease, the NC and P groups received the surprising, then, that many immunologists would accept
same number of ip injections as the experimental groups, stress-induced corticosteroid elevations as "the explana-
but they received more handling because the CS and tion" of conditioned immunosuppressive effects (Rosz-
UCS were presented in a noncontingent manner. If one man). Clearly, the possibility that the effects of condition-
would argue, first, that the handling and ip injections ing are mediated by a conditioned release of
constituted a stress that elevated adrenocortical steroids adrenocortical steroids (in contrast to the adrenocortical
and, second, that these acute elevations in corticosteroids effects of some nonspecific stress) needs to be addressed.
were immunosuppressive, then we could only be operat- It is interesting that different commentators adopt differ-
ing against ourselves; that is, the development of lupus ent perspectives on this issue. From a clinical point of
would have been retarded in these NC animals and any view, Cunningham believes that if we are conditioning a
effects of conditioning might have been masked. nonspecific stress response (in contrast to the "specific"
Based on the study of Bovbjerg et al. (1984), we effects of an immunosuppressive drug) the importance
suggested that conditioned immunosuppression might be and significance of our findings are broadened. From an
dissociated from the conditioned taste aversion. Accord- experimental point of view, Weiss argues that the phe-
ing to Klosterhalfen & Klosterhalfen, however, stress nomenon would merely illustrate the physiological ef-
effects (ip injections) could have been acting differently in fects of conditioned emotional responses.
experimental (immunocompromised) and control ani- These are different perspectives rather than inconsis-
mals. We fail to see the connection between these issues. tent or incompatible arguments. Indeed, if the immuno-
The dissociation between the conditioned behavioral and logical effects of conditioning were mediated "simply" by
immune responses is suggested by the difference in the conditioned changes in the release of steroids our results
inflection point of the curves describing the extinction of would not be especially remarkable as a conditioning
these two responses. Experimental extinction involves phenomenon. Our results would be noteworthy nonethe-
unreinforced presentations of the CS (saccharin plus ip less, considering how little data actually exist on the
injections), and varying the number of extinction trials conditioning of endocrine responses. It is not unlikely
involves varying the number of ip injections. Again, if ip that at least some of the effects of conditioning on immu-
injections are stressful, the greater the number of extinc- nological reactivity are influenced by conditioned neu-
tion trials, the greater the number of acute elevations in roendocrine responses, but that is not an explanation of
corticosterone. The more these elevations in steroid level the phenomenon. We are no more able to elaborate the
act to suppress imunological reactivity, the less likely one causal chain of events that elicits a given physiological
would be to observe extinction effects. Furthermore, ifip change in response to a symbolic stimulus than we are
injections are having any effect at all, one would have to able to specify how immunocompetence is influenced by
predict some difference between the placebo groups that conditioning. The fact that the physiological responses
were given 18 ip injections and those that remained subject to conditioning are, in turn, capable of influenc-
unmanipulated. No such effect was seen. One could still ing immunocompetence could, as Cunningham argues,
hypothesize (as these commentators do) that ip injections have important implications for the pathogenesis of im-
act differently in animals that receive CY 7 weeks before munologically mediated disease processes. In any case,
these events, but there is no evidence that this is so - and the issue of the mediation of conditioned alterations in
such an interaction would not bear on the extinction data, immunocompetence is not likely to be resolved until we

have considerably more information about which param- fact that the timing of CY in relation to sensitization and
eters of immune function are and are not modifiable by challenge is a critical factor in determining the increase or
conditioning. decrease of a DTH response (Schwartz, Askenase &
Without repeating the arguments presented in our Gershon 1978). Drug treatment after sensitization and
review, we have adopted the position that the data that before challenge suppresses the response, and other
have been collected together with the available literature immunosuppressive drugs or agents that stimulate the
do not support the hypothesis that conditioned altera- release of corticosterone could act the same way. On
tions in imnuinological reactivity are attributable to these grounds alone, it would be impossible to draw any
(stress-induced) adrenocortical steroid elevations. conclusions about mediation of the conditioned immu-
Roszman tends to agree with our position that an eleva- nosuppressive effects of CY based on the effects of LiCl.
tion in corticosterone is not likely to be the sole or even Even if we assume that LiCl is an immunologically
the primary mechanism through which conditioning ex- neutral stimulus, there are problems with the interpreta-
erts its effects. In contrast, Cunningham believes that our tion of the Kelley et al. data. These investigators argue
arguments against the "stress hypothesis" stretch the that "if this phenomenon truly represents a conditioned
point. As he points out, it is difficult to rule out the response, it should theoretically not be possible to induce
possibility based on reconstructive experiments. Both he a significant suppression in SRBC-DTH using a non-
and the Klosterhalfens argue that injections of LiCl or immunosuppressive drug" (p. 126). The fact that one can
corticosterone do not necessarily reproduce precisely the observe a conditioned immunosuppressive response via
stress effects of reexposing conditioned animals to the CS. conditioned elevations in corticosterone level, which do
One could hardly disagree with the point; we raised it occur using LiCl as the UCS (Ader 1976), does not
ourselves (Ader, Cohen & Grota 1979). Nevertheless, we necessarily mean that all conditioned immunosup-
wonder what would have happened had our reconstitu- pressive responses are the result of a conditioned eleva-
tion experiments yielded immunosuppression. Would tion in steroid levels. Using a steroid-sensitive DTH
the argument that the exogenous elevation of steroid system, it is hardly surprising that elevations of steroid
levels did not mimic the effects of CS reexposure be level induced by conditioning (or by any other procedure,
considered seriously - or as stretching the point? for that matter) could influence the immune response.
Roszman and the Klosterhalfens refer to a paper by The observations that still remain to be explained include
Kelley, Dantzer, Mormede, Salmon, and Ayaud (1984) our inability to condition suppression of an antibody
published since the preparation of the target article. In response using LiCl as the UCS, and the ability of
apparent conflict with our data (Ader & Cohen 1975; Ader Kusnecov, Sivyer, King, Husband, Cripps and Clancy
et al. 1979), these investigators reported that LiCl as well (1983) to condition suppression of a different cell-medi-
as CY was an effective UCS for conditioned suppression of ated response in the presumed absence of an elevation in
delayed-type hypersensitivity (DTH), a cell-mediated steroid level. Thus, while the study by Kelley et al.
response. Actually, there is no conflict with our data since confirms that an elevation in "stress hormones" can be
we used LiCl in conjunction with suppression of an immunosuppressive, the data have no direct bearing on
antibody response. The study by Kelley et al. may not the mediation of conditioned alterations in immunologi-
have any bearing on conditioned modifications of immu- cal reactivity in general. Considering the difficulties enu-
nological reactivity at all. merated above, referring to the Kelley et al. study to keep
Kelley et al. deliberately selected a T-cell response that afloat the "stress hypothesis" is, in our view, grasping for
was sensitive to "stress hormones." Since they observed straws.
suppression following conditioning with either LiCl or If, as Cunningham maintains, it is difficult to rule out
CY as the UCS, they concluded that an immunosup- the stress hypothesis by reconstructive experiments, it is
pressive drug is not necessary for the conditioning of an equally difficult - if not more so - to rule out the stress
immunosuppressive response. However, Kelley et al. hypothesis by extirpation experiments. Weiss points out
did not consider that LiCl can have immunomodulating the difficulties that would be encountered in interpreting
properties (e.g., Gelfand, Cheung, Hastings & Dosch the results of a study of conditioned immunosuppression
1980; Shenkman, Wadler, Borkowsky & Shopsin 1981). in adrenalectomized animals. In brief, adrenalectomizing
In our study, we showed, first, that LiCl did not influence and maintaining animals accomplish considerably more
the anti-SRBC response by including a conditioned group than eliminating the corticosteroid response to stress; it
that was again exposed to LiCl at the time of immuniza- would seriously disrupt neuroendocrine function in gen-
tion. There was no comparable group in the study by eral. If conditioning effects were observed following adre-
Kelley et al. to establish that LiCl did not unconditionally nalectomy, it would indicate that adrenocortical steroids
suppress the delayed-type hypersensitivity (DTH) reac- were not a critical mediating factor. There is already
tion. Second, two conditioning trials with LiCl were used reasonably good, albeit not definitive, evidence that this
and the unconditioned animals in the Kelley et al. study is the case. Besides, as Weiss argues, it would not rule out
did not receive noncontingent presentations of saccharin the stress hypothesis; it would only rule out the most
and LiCl. In addition, there was no CSo group to control conspicuous manifestation of stress effects. If condition-
for the conditioning experience per se. Finally, in the ing did not occur in adrenalectomized animals, one could
experiment with CY, animals were immunized with conclude only that an intact, normally functioning endo-
SRBC 11 days after conditioning and challenged 5 days crine system was necessary for the conditioning of
later; in the experiment with LiCl, animals were immu- changes in immunological reactivity.
nized 3 days before conditioning and challenged 2 days (as Roszman refers to our having conducted a study in
opposed to 16 days) after. Leaving aside the design of the adrenalectomized animals, a study that the reader will
conditioning aspect of the study, we have referenced the not find in our review. We did conduct such a study,

alluded to by Ader et al. (1979), but we have not pub- provide no support for the hypothesis that the effects of
lished it. Adrenalectomized animals did display a condi- conditioning can be attributed to elevated cor-
tioned attenuation of the antibody response to SRBC. ticosteroids. Why, in the absence of supporting data,
However, because the conditioned attenuation of anti- does the argument persist? To some degree, this may
body titer in sham-operated animals was, in this particu- result from the misguided notions that: (a) there is a
lar experiment, not statistically significant, we cannot constellation of responses that can be labelled as "the
claim to have reliable data documenting immunosuppres- stress response"; (b) we understand and agree upon what
sion in adrenalectomized animals. constitutes "stress"; and (c) the concept of stress can serve
Based on the aforementioned point that extirpation as an explanatory device. The attempt to explain the
experiments, like replacement experiments, are vulnera- effects of conditioning on the basis of elevated cor-
ble to a variety of criticisms, we believe that the question ticosteroid levels is, in this context, an illustration of a
of how conditioned changes in immunological reactivity need to simplify what appears to be an extremely complex
come about might better be approached by trying to phenomenon. Our current inability to specify the neural
determine what factors are involved rather than what and endocrine pathways by which symbolic stimuli come
factors are not involved. More specific and precise inter- to alter immunological reactivity contributes to the skep-
ventions will be possible as we learn more about the ticism with which our research is viewed (Melnechuk).
immunomodulating effects of neuroendocrine and neuro- There is, too, an implied equivalence between stress and
transmitter substances. adrenal activation, its most prominent manifestation.
If exogenous stimulation of adrenocortical activity on Whether the hypothesis that our conditioning effects
the one hand and adrenalectomy on the other are of were really stress effects and thus mediated by elevated
limited value, one can still monitor the effects of condi- adrenocortical steroid levels, or whether the immu-
tioning on these (and other) parameters of endocrine nomodulating effects of corticosteroids led to the in-
function. Such studies are not without their own prob- terpretation of our conditioning effects as a reflection of
lems. For example, Roszman suggests that the temporal stress, does not matter. The fact is, these are really
measurement of plasma corticosterone levels in condi- separate issues; one could adopt a stress hypothesis with-
tioned animals would prove useful, as indeed it might. In out assuming a direct involvement of adrenal steroids,
making such measurements, however, one would have to and one could hypothesize that our effects were mediated
take into consideration the fact that antigenic stimulation by changes in corticosteroids without invoking the con-
evokes neuroendocrine responses (e.g., Besedovsky, del cept of stress.
Rey, Sorkin, Da Prada, Burri & Honegger 1983). What Whatever pattern of responses occurs as a result of
one would actually be measuring would be the interaction (stressful) environmental circumstances, including condi-
of the effects of conditioning, conditioned responses, and tioning, must encompass more than a change in cor-
unconditioned responses elicited by antigenic stimula- ticosteroid levels. Thus, we completely agree with the
tion. explicit or implicit references to the need for a more
Repeated measurement of corticosterone levels would complete assessment of the neural and endocrine changes
provide data relevant to Smotherman's suggestion that a affected by conditioning that could alter immunological
decrease in steroid level might have some influence on reactivity (Anisman & Zacharko, Cunningham, Fox,
the conditioned alteration in immunological reactivity. Roszman, Solomon, Weiss). We already know that we
His speculation is based on the drinking behavior and can condition an elevation in corticosteroid level (Ader
steroid levels of animals maintained under water-depri- 1976), and there is no reason to believe that this is the
vation schedules. Based on the data described by only endocrine response that can be so affected. Indeed,
Smotherman, we would expect our experimental pro- as we have pointed out, there is no shortage of candidates
cedures to have contributed to a homogeneity among our for the mediation of conditioned alterations in immuno-
experimental and control animals with respect to levels of logical reactivity. One need only keep in mind the limita-
circulating steroids. A difference might occur, however, tion of correlational data and the inability to generalize
when conditioned animals were presented with a single from one parameter of immune function to another.
bottle containing the CS solution but did not drink.
Conditioned animals presented with two drinking bot- Underlying mechanisms. An elaboration of the neu-
tles, one containing plain water and the other the CS roanatomical, neurochemical, and endocrine pathways to
solution, however, do drink. If the duration of any condi- and from the immune system and how they are changed
tioned suppression of steroid levels - or of a return to an by conditioning will provide only a partial analysis of the
elevated steroid level - was a result of exposing these mechanisms underlying conditioned modulations of im-
animals to one or two drinking bottles, it was not reflected mune function. We must also assess which functions and
in the conditioned suppression of immunological reac- components of the immune system are subject to the
tivity. Different procedures were used in different ex- influence of conditioning, how these alterations are af-
periments and conditioned immunosuppression was ob-
fected, and how they in turn modify the net immune
served under both conditions.
response. Considering its complexity, it seems unlikely
As the reader has no doubt surmised, we think that the that different components of the immune system are
so-called stress hypothesis and the adrenocortical media- equally susceptible to conditioning or that there is a
tion of conditioned alterations in immunological reac- single mechanism by which conditioning exerts its influ-
tivity have been given undue emphasis. We have never ence. The question of how conditioning effects come
dogmatically claimed that elevated adrenal steroids were about (Anisman & Zacharko, Ballieux & Heijnen, Cun-
not involved in the conditioned alteration of immune ningham, Fox, Melnechuk, Rosenberg, Roszman,
function. Rather, we have argued that the available data Weiss) will most certainly require more precise measure-

ments of a variety of immune processes. Such research is far, our studies of conditioned immunosuppression have
now going on in several laboratories, including our own. capitalized on the immunosuppressive effects of a phar-
A major issue in attempting to elaborate the underlying macological agent (CY) and the antigen SRBC, neither of
mechanisms of conditioned alterations of immunity con- which is likely to be encountered in the real world.
cerns the extent to which conditioning can influence Leaving aside the paradoxical nature of the conditioned
antigen-dependent responses. According to Melnechuk, response (discussed below), these studies do indicate the
it has been argued that CNS-immune system interac- potential of the organism for such adaptive responses.
tions could not include antigen-specific immunoregulato- This potential has clinical relevance as illustrated by the
ry processes because neurons lack receptors for discrimi- anticipatory reactions observed in allergic individuals
nating among antigen conformational specificities. exposed to symbolic, nonallergenic stimuli (Hill 1930;
Indeed, we have no reason to believe that the condi- MacKenzie 1886). If there is a learning component to the
tioned immunopharmacological effects that we have ob- manifestation of allergic symptoms, one might be able to
served are antigen-specific. Even in the case where exert some therapeutic control by interventions that take
antigen itself was used as the UCS and an apparent such learning into consideration (Dekker, Pelser &
antigen-specific conditioned response was observed Groen 1957).
(Corczynski et al. 1982), Ballieux & Heijnen point out
that the specificity of the response could have occurred as Direction of conditioned responses. Because the majority
a result of a nonspecific mechanism such as a CNS signal of studies have involved conditioned immunosuppressive
to lymphoid tissues that increased the number of specific responses, the issue of adaptive significance becomes
alloreactive lymphocytes in the circulation.l Implicit in particularly salient since the ability to condition immu-
this argument, of course, is the need for the original nosuppression would appear to be counterintuitive func-
alloantigenic stimulus to expand or alter the relevant tionally as well as theoretically. Quite appropriately,
clones (memory cells?) and to set the stage, so to speak, several commentators (Dworkin, Green, Hinson, Krank,
for a nonspecific signal (the CS) to implement what is Revusky, Siegel & Scoles, Weiss) have raised the ques-
measured as an antigen-specific change. In any antigen- tion: Why does the conditioned response mimic the
specific response, antigen must play a selective and unconditioned response, immunosuppression, rather
interactive role with appropriate receptors on T and B than induce an anticipatory, defensive reaction to com-
lymphocytes. That such antigen-specific changes could pensate for the immunosuppressive effects of the UCS?
appear to be "specifically" conditioned seems quite rea- In part, this argument presumes that the immunosup-
sonable in view of the data on sympathetic and parasym- pressive effects of CY are "maladaptive." In fact, low-
pathetic innervation of lymphoid tissues and the evidence dose injections of CY can be immunoenhancing (Turk &
that lymphocytes and accessory cells bear receptors for a Parker 1982); treatment with CY before sensitization
variety of peptide hormones and neurotransmitter sub- (rather than before challenge) will potentiate certain T-
stances (referenced in our review). Without directly af- cell-mediated reactions (Schwartz et al. 1978); and the
fecting certain interactions between antigens and specific disordered immunoregulation that characterizes autoim-
lymphocytes, then, there are a variety of means by which mune disease is counteracted by treatment with CY (Ader
the CNS (and thus conditioning) might influence immu- & Cohen 1982), an example of the beneficial effects of an
nity. To deny the influence of these CNS processes in the immunosuppressive drug. These illustrations highlight
regulation or modulation of immunity is to adopt a seem- the problem of attempting to define the adaptive signifi-
ingly restrictive definition of what constitutes an immune cance of immunosuppression. However, because we ob-
response and what constitutes regulation. That is, per- served a conditioned response that was similar in direc-
haps, a defensible posture. Being concerned with the net tion to the unconditioned response, the question of why
adaptive response of the immune system, however, we we did not observe a compensatory response remains.
are viewing the immune system as part of an integrated We cannot answer the question. We can, however,
network of defensive processes. If we entertain the pos- suggest several reasons why the "expected" result was
sibility that antigenic stimulation is a necessary but insuf- not obtained. We could speculate, to begin with, that we
ficient condition for evoking an antibody response of a may in fact have conditioned compensatory responses. In
given magnitude, there is much that needs to be learned titrating serum antibodies one week after animals were
about the role of the CNS in the process of immunoregu- immunized, for example, we were actually measuring the
lation. net effect of a complex chain of events triggered by an
Because we do not yet know how or to what extent CNS antigenic stimulus. Presumably, the depression of the
processes influence immunity, it is difficult to define the antibody response effected by reexposing conditioned
role of conditioning in the normal, overall (everyday) animals to the CS was the result of changes that occurred
functioning of the immune system. This issue was raised at the time of exposure to the CS several days earlier. One
explicitly by several commentators (Cunningham, could hypothesize, then, that the question of the direc-
Dworkin, Hinson, Riley, Siegel & Scoles, Weiss). Bovb- tion of the conditioned immune response(s) can be an-
jerg et al. (1982) have described our conditioning effects swered only by measuring the immediate effects of reex-
in terms of "feedforward" regulatory mechanisms. Pre- posure to the CS, for example, changes in number or
sumably, conditioned alterations in immunological reac- function of different lymphocyte populations or subpopu-
tivity could function as preparatory responses. Except for lations, the interactions among which ultimately deter-
the poorly defined studies conducted several decades ago mine net antibody production. It is not inconceivable that
(reviewed by Ader 1981c), the study in which Gorczynski (neuroendocrine) changes elicited by the CS might have a
et al. (1982) used an antigen as the UCS provides the only differential effect on suppressor or helper T-cells, for
available illustration of such an anticipatory effect. Thus example, or that the immunological effects of reexposure

to the CS would depend on the temporal relationship ostasis, it may require more than a single CS-UCS
between CS reexposure and antigenic stimulation. As pairing to effect such a response (or it may at least be more
Roszman suggests, a study of the effects of conditioning likely to be observed if more than one trial is used).
on lymphocyte populations is clearly needed. Moreover, we could infer from the commentators' re-
There is a particularly noteworthy difference between marks that the development of such responses would vary
our studies of conditioned physiological effects induced as a function of the CS, the UCS, and (we would add) the
by drugs and most other studies of conditioned phar- nature and function of the system that one is attempting
macological effects that may be relevant to the issue of to modulate. A conditioning protocol that would permit
compensatory responses. As noted above, the elicitation variation of the number of conditioning trials would be
of antibody responses necessitated an antigen to activate only the first step in approaching this problem. It is
the system that was being monitored. As a consequence, unlikely that we will resolve these issues until additional
one could speculate that the unconditioned effects of data are obtained from a systematic analysis of the effects
antigenic stimulation (or manifestations of a preexisting of different conditioning paradigms (using interoceptive
immunological dysfunction) are read out as an interaction and exteroceptive CSs and UCSs) on different parameters
between the effects of antigen and the conditioned re- of immune function in immunized and nonimmunized
sponses elicited by reexposure to a stimulus previously subjects.
paired with an immunomodulating substance. Simply
stated, the effects of the stimulation required to activate Other conditioning paradigms. Several commentators
the system being monitored may add another level of have emphasized the need to assess the effect of other
complexity to an analysis of the nature of the processes conditioning paradigms in modulating immunological re-
that are involved in conditioning changes in immune activity (Anisman & Zacharko, Cunningham, Roszman,
function. On one level, there is an obvious strategy that Weiss). This would, we agree, establish the generality of
could obviate these problems - although not without the phenomenon and provide an answer to some of the
some cost. Instead of measuring antibody production, for issues addressed above. Some such data are already
example, one could measure a lymphocyte activity that is available (Gorczynski et al. 1982; Sato, Flood & Makino-
not dependent on specific antigen or not restricted to a dan 1984; Smith & McDaniels 1983), so the phenomenon
small number of clones of cells (e.g., mitogen responses, cannot be characterized as being limited to the special
lymphokine production, natural killer cell activity). features of taste aversion learning (Riley). In a study
These responses can be measured in vitro without intro- referred to by Veldhuis & De Wied (Heijnen, Croiset,
ducing antigen. Under these circumstances, it would be Bevers, Veldhuis, De Wied & Ballieux, in press), immu-
quite reasonable to hypothesize that the repeated pairing nosuppression was observed in a one-trial passive avoid-
of an interoceptive or exteroceptive CS and UCS would ance conditioning situation without the use of immu-
eventually elicit a compensatory conditioned response. nopharmacological agents and thus without the potential
The fact that we have not observed a conditioned confound of an immunocompromised host. These authors
compensatory response does not mean that a conditioned were kind enough to send us their abstract and to present
compensatory response in immunological reactivity is not their data during a visit to Rochester. Placing rats on an
observable. That is, we are not searching for an explana- elevated platform from which they had previously en-
tion of a documented failure to observe such conditioning tered a dark compartment and received a single electric
under circumstances that have been hypothesized, at shock was sufficient to reduce immunological reactivity as
least, to provide the latitude to observe such effects. We measured by mitogen responsiveness.
do not know that the salient effect of CY is not a direct Considering that previous studies (e.g., Keller et al.
effect on the nervous system (Revusky), but on the other 1981, 1983; Laudenslager, Ryan, Drugan, Hyson &
hand we do not know which of the myriad effects of CY do Maier 1983; Sato et al. 1984; Shavit et al. 1984) had used
contribute to the conditioning (Green). Other factors that (required?) considerably more electric shock stimulation
might have contributed to the fact that we did not observe to influence immunological reactivity, we attempted to
compensatory responses would include the issue raised repeat these findings. Actually, we simply attempted to
by Krank and, indirectly, by Revusky; namely, other determine whether immunological reactivity could be
contextual cues might exert competing effects or other suppressed using a single shock trial - or if multiple trials
cues might be more appropriate for conditioning changes might be necessary or would produce even larger effects.
in immunological reactivity in the first place. Krank also Whether we used individually or group-housed rats or
notes that our studies represent a departure from the mice and whether we measured mitogen responsiveness,
more typical use of exteroceptive cues in the conditioning natural killer cell activity, or the PFC response to SRBC,
of drug-induced physiological responses. Although these we observed no suppression of immunological reactivity,
possibilities are speculative, they are reasonable and even after 15 electric shocks administered at the rate of
require examination. one per day.
In terms of our experimental procedures, one could ask The apparent discrepancy between our results and
whether compensatory conditioned responses can be those of Heijnen et al. do not represent a failure to
acquired on the basis of a single conditioning trial. We are replicate their results. We simply asked whether a single,
not aware of any report of conditioned compensatory brief electric shock would unconditionally suppress im-
responses after a single conditioning trial (which is not to munological reactivity, and whether a series of such
say that such an effect cannot occur). If compensatory shocks would magnify the effects. Our data indicated that
conditioned responses or opponent processes (Solomon it would not. How, then, do we interpret the results
1980) develop in anticipation of the need to counteract obtained by Heijnen et al. ? We suggest that the suppres-
the effects of UCSs that elicit deviations from home- sion of immunological reactivity observed by Heijnen et

Response I Ader & Cohen:CNS-immune system interactions
al. was not the result of an association between being & Scoles). Theoretically, the institution of a schedule of
placed on an elevated platform and electric shock stimula- partial reinforcement could apply to a number of clinical
tion. Electric shock was experienced in a different en- situations in which it would be advantageous to reduce
vironment and dependent upon entering a distinctive the maintenance levels of prescribed drugs. From a
chamber. Rather, suppression may have been a result of practical point of view, it should not prove difficult to
the physiological concomitants of an emotional response. arrange for the reliable repetition of contingent rela-
The physiological responses to a brief electric shock are tionships in a medical setting in which medication is
evidently insufficient in magnitude or duration to influ- administered (Riley). More theoretically, Siegel & Scoles
ence immunological reactivity. However, the pattern of raise the question of what might be expected from the
the physiological concomitants of an emotional response imposition of extinction procedures. There are indeed
induced by returning these conditioned animals to the conflicting reports concerning the effects of partial rein-
elevated platform (and leaving them there for 5 min.) was forcement schedules on resistance to extinction in classi-
qualitatively or quantitatively sufficient to influence im- cal conditioning paradigms. Mackintosh (1974) and
munological reactivity. Whether this depression of im- Schwartz (1978), however, argue that there are really few
munological reactivity was a conditioned response re- grounds for distinguishing between classical and instru-
mains to be determined. The observation that psycholog- mental conditioning, even with respect to resistance to
ical events can be more effective than physical events in extinction. We do not know whether the application of a
eliciting hormonal responses (including adrenocortical taste aversion type of conditioning in the pharmacother-
steroid elevations) has been amply documented (e.g., apy of murine autoimmune disease can be classified as a
Mason 1968). It will be of interest to determine whether pure instance of either classical or instrumental condi-
such a sensitivity to environmental influences also applies tioning. There is certainly reason to believe that if condi-
to the immune system. tioning processes influence the response to drugs and are
It seems clear from the above that there are a variety of a component of drug therapy in humans, there are instru-
experimental paradigms that can be applied to the study mental components to such conditioning. Thus, we are
of CNS-immune system interactions, including animal not discussing the effects of reinforcement schedule on
models of disease. There are some advantages to studying the extinction of a drug-induced physiological response
conditioning effects in animals that are not immu- but the effects of reinforcement schedule on the extinc-
nocompromised in any way (Veldhuis & De Wied), as tion of a pharmacotherapeutic response. There is the
discussed earlier in connection with compensatory re- reasonable possibility, then, that pharmacotherapeutic
sponses. But whether this is a better strategy remains to effects maintained by a partial reinforcement schedule
be determined. Considering the nature and function of
would be more resistant to extinction than those main-
the immune system and the fact that antigen is required
to initiate many of the responses with which one would be tained by continuous reinforcement. The experimental
concerned, we can only say that the use of immu- studies required to address this issue are worth pursuing.
nomodulating drugs or antigenic stimulation is different Dworkin addresses the same general issue in question-
from using untreated or nonimmunized animals. One ing whether the data on conditioned alterations in im-
particular advantage of using a disease model is that it mune function can be interpreted in terms of instrumen-
dictates relevant measures of immune function. More- tal (type II) learning. Actually, Dworkin's comments are
over, we have observed relatively large conditioning directed to our speculation that deviations from home-
effects using an animal model of autoimmune disease. ostasis at the level of the immune system might be
sufficient to motivate behavior. We are indeed entertain-
ing the possibility of type II processes and the implication
Clinical and theoretical considerations. Although Rosen- that the nervous system is capable of sensing as well as
berg has expressed reservations about the clinical impact influencing immunocompetence. Granted, we are cur-
of conditioning effects, we believe that conditioning ef- rently unable to specify the mechanisms by which the
fects are ubiquitous in clinical situations and that a sys- CNS and immune system communicate, but we pre-
tematic examination of these effects would yield data of viously referred to the accumulating research indicating
interest and direct clinical relevance (e.g., Ader, in press; that (a) there is autonomic innervation of lymphoid
Siegel 1983). Several commentators concur with this tissues; (b) lymphocytes bear receptors for hormones and
view. It may be some time, however, before the study of neurotransmitters; (c) changes in neuroendcrine function
CNS-immune system interactions has practical conse- influence immune function; and (d) elicitation of an im-
quences for the practicing clinician (Friedman). Al- mune response results in changes in neural and endo-
though some clinical studies have, as Solomon points out, crine activity. To this we might add recent data indicating
provided presumptive evidence for CNS-immune sys- that lymphocytes are capable of producing ACTH-like
tem interactions, there is a paucity of such data. Several immunoreactive substances (Smith, Meyer & Blalock
studies have been conducted to examine the link between 1982) and Blalock's (1984) speculative analysis of the
stress and cancer, for example. Based on currently avail- immune system as a sensory organ. The pieces of the
able data, however, we agree with Fox that thus far the puzzle have not been completely assembled - and there
involvement of the immune system in studies of the are still pieces missing - but there are enough pieces
relationship between cancer and stress or personality available to envision CNS-immune system interactions.
factors is for the most part speculative. A demonstration that a dysregulation of immune function
A specific extrapolation from our data on autoimmunity could be detected in or alleviated by behavior would
in mice to clinical situations concerns the adoption of provide in vivo evidence that a mechanism exists within
conditioning strategies in some pharmacotherapeutic the CNS for monitoring and influencing immune func-
protocols (Anisman & Zacharko, Kimmel, Krank, Siegel tion.

References I Ader & Cohen: CNS-immune system interactions
Concluding remarks. In concluding our response we wish phenomena. In: Psychoneuroimmiitwlofiy, ed. R. Ader. Academic
to thank all the commentators for their remarks. They Press. [taRA]
Alexander, F. (1950) Psychosomatic medicine. Norton. [SBF]
have, in a very real way, sharpened and focused attention
Alvarez-Buylla, R. & de Alvarvz-Biiylla, E. R. (1975) Hypoglyccinic
on many unresolved issues that will influence our own conditioned reflex in rats: Preliminary study of its mechanisms. Journal of
research and, we hope, the research of others. We have Comparative and Physiological Psychology 88:155-60. [LG|
tried to respond to what we viewed as the most outstand- Ambrose, C. T. (1970) The essential role of corlicosteroids in the induction of
ing concerns. If, for lack of space or the time constraints of the immune response in vitro. In: Hormones and the immune response.
ed. G. E. W. Wolstenholme & J. Knight. Ciba Foundation. [taRA]
preparing this response, we have omitted a particularly Anisman, H. (1984) Vulnerability to depression: Contribution of stress. In:
salient issue - or if additional issues arise as a result of Neurobiology of mood disorders, ed. R. M. Post & J. C. Ballenger.
these comments - we hope that they will be raised as Williams and Wilkins. [HA]
"continuing commentary." The emerging field of psycho- Anisman, H. & Sklar, L. S. (1979) Catecholamine depletion upon lecxpnsure
neuroimmunology, as illustrated by Solomon's com- to stress: Mediation of the escape deficits produced by inescapable shock.
Journal of Comparative and Physiological Psychology 93:610-25. I HA]
ments, represents a fertile field of research for the behav- Arrenbrecht, S. (1974) Specific binding of growth hormone to thymocytes.
ioral and brain sciences. Our target article was a rather Nature 252:255-57. [taRA]
lengthy open invitation for increased participation in this Bach, J. F. (1975) The mode of action of immunostippressive agents. Elsevier
venture. North-Holland. [taRA]
Behan, P. O. & Geschwind, N. (1985) Hemispheric laterally and immunity.
In: Neural modulation of immunity, ed. R. Guilleniin, M. Cohn, & T.
NOTE Melnechuk. Raven Press. |TM]
1. Specificity and nonspecificity are used here and in our Berczi, I., Nagy, E., Kovacs, K. & Horvath. E. (1981) Regulation of humoral
target article (see note 2) to refer to antigen-dependent or immunity in rats by pituitary hormones. Ada Endocrinologica 98:506-
-independent responses. These terms do not refer to our knowl- 13. [taRA]
edge of underlying mechanisms (Engel). Bereznykh, D. V. (1955) On the question of conditioned reflex restoration of
immunogenesis. Byulletin Eksperimcntalnoi Biologii i Meditsiny 40:49-
52. [taRA]
Besedovsky, H. O., del Rey, A. E. & Sorkin, E. (1983) What do the immune
system and the brain know about each other? Immunology Today 4:342-
References 46. |taRA]
Abraham, A. D. & Bug, G. (1976) 3 H-testosterone distribution and binding in Besedovsky, H. O., del Rey, A. E., Sorkin, E., Da Prada, M., Biirri. R. &
rat thymus cell in vivo. Molecular and Cellular Biochemistry 13:157— Honegger, C. (1983) The immune response evokes changes in brain
63. jtaRA] noradrenergic neurons. Science 221:564-66. [tarRA, HA, BHF]
Ader, R. (1974) Letter to the editor. Psychosomatic Medicine 36:183- Besedovsky, H. O., del Rey, A. E., Sorkin, E., Da Prada, M. & Keller, H.
84. [taRA] (1979) Immune regulation mediated by the sympathetic nervous system.
(1976) Conditioned adrenocortical steroid elevations in the rat. Journal of Cellular Immunology 48:346-55.' [taRA]
Comparative and Physiological Psychology 90:1156-63. [tarRA.WK] Besedovsky, H. O., and Sorkin, E. (1981) Immunologic-neuroendocrine
(1980) Presidential address: Psychosomatic and psychoimmunologic circuits: Physiologic approaches. In: Psychoneuroimmunology, ed. R.
research. Psychosomatic Medicine 42:307-22. [taRA] Ader. Academic Press. [taRA]
(1981a) Animal models in the study of brain, behavior, and bodily disease. Besedovsky, H. O., Sorkin, R., Felix, D. & Haas, H. (1977) Hypothalamic
In: Brain, behavior, and bodily disease, ed. H. Weiner, M. A. Hofer & changes during the immune response. European Journal of Immunology
A. J. Stunkard. Raven Press. [taRA] 7:325-28. [taRA, HA]
(1981b) The central nervous system and immune responses: Conditioned Biziere, K., Guillaumin, J. M., Degenne, D., Bardos, P., Renoux, M. 6t
immunopharmacologic effects. Advances in Immunopharmacology 1:427- Renoux, C. (1985) Lateralized ueocortical modulation of the T-cell
34. |taRA] lineage. In: Neural modulation of immunity, ed. R. Cuillcmin, M. Cohn
(1981c) A historical account of conditioned immunobiologic responses. In: & T. Melnechuk. Raven Press. [TM]
Psychoneuroiinmunology, ed. R. Ader. Academic Press. [tarRA] Black, S., Humphrey, J. H. & Niven, J. S. (1963) Inhibition of Mantoux
ed. (1981d) Psychoneuroimtnunology. Academic Press. (taRA] reaction by direct suggestion under hypnosis. British Medical Journal
(1983) Developmental psychoneuroiinmunology. Developmental 1:1649-52. [taRA]
Psychobiology 16:251-67. [taRA] Blalock, J. E. (1984) The immune system as a sensory organ. Journal of
(1985) Conditioned immunopharmacologic effects in animals: Implications Immunology 132:1067-70. [tarRA]
for a conditioning model of pharmacotherapy. In: Placebo: Clinical Blecha, F., Barry, R. A. & Kelley, K. W. (1982) Stress-induced alterations in
phenomena and new iiuights, ed. L. White, B. Tursky or G. E. delayed-type hypersensitivity to SRBC and contact sensitivity to DNFB
Schwartz. Cuilford Press. [tarRA] in mice. Proceedings of the Society for Experimental Biology and
Ader, R. & Cohen, H. (1975) Behaviorally conditioned immunosuppression. Medicine 169:239-46. [taRA]
Psychosomatic Medicine 37:333-40. [tarRA, RLE, WK] Blecha, F., Kelley, K. W. & Satterlee, D. G. (1982) Adrenal involvement in
(1981) Conditioned immunopharmacologic responses. In the expression of delayed-type hypersensitivity to SRBC and contact
Psychoneuroiinmunology. ed. R. Ader. Academic Press. [taRA, RLE, sensitivity to DNFB in stressed mice. Proceedings of the Society for
TR] Experimental Biology and Medicine 169:247-52. [taRA]
(1982) Behaviorally conditioned immunosuppression and murine systemic Bourne, H. R., Lichtenstein, L. M., Melmon, K. L., Henney, C. S.,
lupus erythematosus. Science 215:1534-36. [tarRA, BHF, LG, WK, Weinstein, Y. & Shearer, G. M. (1974) Modulation of inflammation and
GFS, SS] immunity by cyclic AMP. Science 184:19-28. |taRA]
(1984) Behavior and the immune system. In Handbook of behavioral Bovbjerg, D. (1983) Classically conditioned alterations in two cell-mediated
medicine, ed. D. Gentry. Cuilford Press. [taRA] immune responses. Ph.D. dissertation, Univ. of Rochester. [taRA]
Ader, R., Cohen, N. & Bovbjerg, D. (1982) Conditioned suppression of Bovbjerg, D., Ader, R. & Cohen, N. (1982) Behaviorally conditioned
humoral immunity in the rat. Journal of Comparative and Physiological immunosuppression of a graft-vs-host response. Proceedings of the
Psychology 96:517-21. [tarRA, WK] National Academy of Sciences, USA 79:583-85. [taRA, WK, GFS]
Ader, R., Cohen, N. & Grata, L. J. (1979) Adrenal involvement in (1984) Acquisition and extinction of conditioned suppression of a graft-vs-
conditioned immunosuppression. International Journal of host response in the rat. Journal of Immunology 132:111-13. [tarRA,
Immunopharmacology 1:141-45. [tarRA] WK]
Ader, R. & Friedman, S. B. (1968) Plasma corticosterone response to Bovbjerg, D., Cohen, N. & Ader, R. (1982) The central nervous system and
environmental stimulation: Effects of duration of stimulation and the 24- learning: A strategy for immune regulation. Immunology Today 3:287-
hour adrenocortical rhythm. Neuroendocrinology 3:378-86. [taRA] 91. [taRA]
Ader, R., Weijnen, J. A. W. M. & Moleman, P. (1982) Retention of a passive (1983) Conditioning immune responses (letter). Immunology Today 4:63-
avoidance response as a function of the intensity and duration of electric 64. [taRA]
shock. Psychonomic Science 26:125-28. [HDV] Boyse, E. A., Beauchamp, G. K. 6e Yamazaki, K. (1983) Critical review: The
Ahlqvist, J. (1981) Hormonal influences on iinmunologic and related sensory perception of genotypic polymorphism of the major

histocompatihility complex and other gi'iu'S: Sonic physiological and Dunbar, H. F. (1935) Emotions and bodily changes. Huuber. [TM|
phylogenetic implications. Human Immunology 6:177-83. [taKA] (1943) Psychosomatic diagnosis. Harper (Hoeber). [SBF]
Braveman, N. S. (1979) The role of blocking and compensatory conditioning Dworkin, B. (1984) Operant mechanisms in physiological regulation. In: Self-
in the treatment preexposure effect. Psychopharmacology 61:177- regulation of the brain and liehavior, ed. T. Elbert, B. Rockstroh, W.
89. [MDK] Lutzenberger & Birbaumer. Springer-Verlag. [BD|
Bulloch, K. & Moore, K. Y. (1980) Nucleus ainhiguous projections to the (in press) Learning and long-term physiological regulation. In:
thymus gland: Possible pathways for regulation of the immune response Consciousness and self-regulation, IV. ed. D. Shapiro, G. Schwartz & R.
and the neuroendocrine network. Anatomical Record 196:25A. ItaKA] Davidson. Plenum Press. |BD]
(1981) Innervation of the thymus gland by brainstcm and spinal cord in Dwyer, D. S. (1983) Conditioning immune responses (letter). Immunology
mouse and rat. American journal of Anatomy 162:157-66. [taKA| Today 4:63. [taRA]
Bulloch, K . Roth, E. & Cullen, M. H. (1984) Nodose and superior cervical Eikelboom, R. & Stewart, J. (1982) Conditioning of drug-induced
ganglia projections to the rat thymus gland. Society for Neuroscience physiological responses. Psychological Review 89:507-28. itaRA, LG,
Abstracts 10(2):723. [TM] REH. MDK, SR, SS]
Bykov, K. M., ed. (1960) Text-book of physiology. Foreign Languages Elkins, R. (1973) Individual differences in bait-shyness: Effects of drug dose
Publishing House. [SS] and measurement technique. Psychological Record 23:349-58. (RLE]
Cake, M. H. & Litwak, G. (1975) The glucocorticoid receptors. In: (1980) Covert sensitization treatment of alcoholism: Contributions of
Biochemical action of hormones, ed. G. Litwak. Academic Press. [taHA] successful conditioning to subsequent abstinence maintenance. Addictive
Cassens, G., Roffman, M., Kuriic, A., Orsulak, P. J. & Schildkraut. J. J. Behaviors 5:67-89. [RLE]
(1980) Alterations of brain norepinephrine metabolism induced by (1984) Taste aversion retention: An animal experiment with implications for
environmental stimuli previously paired with inescapable shock. Science eonsuminatory aversion alcoholism treatments. Behavioral Research and
209:1138-40. [HA| Therapy 72:179-86. [RLE]
Chance, W. J., White, A. C , Krynock, G. M. & Rosecrans, J. A. (1978) (in press) Separation of taste aversion prone and resistant rats through
Conditional fear-induced antinociception and decreased binding of H-N- selective breeding: Implications for individual differences in
leuenkephalin to rat brain. Brain Research 141:371-74. (HA] conditionability and aversion therapy alcoholism treatment. Behavioral
Clanian, H. N. (1972) Corticosteroids and lymphoid cells. New England Neuroscience. [RLE]
Journal of Medicine 287:388-97. [RG] Elkins, R. & Boudewyns, P. (1984) Separation of learned eonsummatory and
Coe, C. L., Wiener, S. G., Rosenberg, L. T. & Levine, S. (in press) environmental aversions through selective breeding. Poster presentation
Endocrine and immune responses to separation and maternal loss in at the annual meeting of the American Psychological Association,
nonhuman primates. In: The biology of social attachment, ed. M. Reite Toronto, Canada. [RLE]
& T. Field. Academic Press. [RG] Elkins, R. & Harrison, W. (1983a) Effects of CS flavor substitution on
Cohen, F. (1979) Personality, stress and the development of physical illness. selectively bred taste aversion prone and taste aversion resistant rats.
In: Health psychology, ed. G. C. Stone, F. Cohen & N. E. Adler. Paper presented at the annual meeting of the Southern Society for
Jossey-Bass. [BHF] Philosophy and Psychology, Atlanta, Ceorgia. [RLE]
Cohen, J. J. & Clanian, H. N. (1971) Hydrocortisone resistance of activated (1983b) Rotation-induced taste aversions in strains of rats selectively bred
initiator cells in graft-versus-host reactions. Nature 229:274-75. [taRA] for strong or weak acquisition of drug-induced taste aversions. Bulletin of
Cohen, N., Ader, R., Green, N. & Bovbjerg, D. (1979) Conditioned the Psychonomic Society 21:57-60. |RLE]
suppression of a thymus-independent antibody response. Psychosomatic (1984) Effects of a new CS flavor substitution on taste aversion learning of
Medicine 41:487-91. [taRA, WK] selectively bred taste aversion prone and taste aversion resistant rats.
Cohn, M. (1968) The molecular biology of expectation. In: Nucleic acids in Paper presented at the annual meeting of the Southeni Society for
immunology, ed. O. J. Plescia & W. Braun. Springer. [taRA] Philosophy and Psychology, Columbia, South Carolina. [RLE]
Comsa, J., Leonhardt, H. & Wekerle, H. (1982) Hormonal coordination of Elkins, R., Harrison, W., Murphy, J. & Hobbs, S. (1984) Open field and
the immune response. Reviews of Physiology, Biochemistry and enclosed compartment reactivity in two strains of rats selectively bred for
Pharmacology 92:115-89. [taRA] proneness or resistance to taste aversion learning. Paper presented at the
Cooper, E. L., ed. (1984) Stress, immunity, and aging. Marcel annual meeting of the Southern Society for Philosophy and Psychology,
Dekker. [TM] Columbia, South Carolina. [RLE]
Coover, G. D., Satterfield, G. T., Smothennan, W. P., Steinke, D. & Dorsa, Elkins, R. & Hobbs, S. (1982) Taste aversion proneness: A modulator of
D. M. (1983) Endocrine activation, but not tail-flick latency alteration, conditioned eonsummatory aversions in rats. Bulletin of the Psychonomic
induced by consumption of concentrated sucrose solutions. Society for Society 20:257-60. [RLE]
Neuroscience Abstracts 9:1123 [WPS] Engel, G. L. (1954) Selection of clinical material in psychosomatic medicine.
Cousins, N. (1979) Anatomy of an illness as perceived by the patient. Psychosomatic Medicine 16:368-73. [SBF]
Norton. [TM] (1977) The need for a new medical model: A challenge for biomedicine.
Cross, R. J., Markesberry, W. R., Brooks, W. H. & Roszman. T. L. (1980) Science 196:129-36. [SBF]
Hypothalamic-immune interactions: 1. The acute effect of anterior Fanselow, M. S. & German, C. (1982) Explicitly unpaired delivery of
hypothalamic lesions on the response. Brain Research 196:79— morphine and the test situation: Extinction and retardation of tolerance to
87. [taRA] the suppressing effects of morphine on locomotor activity. Behavioral and
Crowell, C. R., Hinson, R. E. & Siegel, S. (1981) The role of conditional Neural Biology 35:231-41. [SS]
drug responses in tolerance to the hypothennic effects of ethanol. Felten, D. L., Overhage, J. M., Felten, S. Y. & Schmedtje, J. F. (1981)
Psychopharmacology 73:51-54. [REH] Noradrenergic sympathetic innervation of lymphoid tissue in the rabbit
Dann, J. A., Wachtel, S. S. & Rubin, A. L. (1979) Possible involvement of appendix: Further evidence for a link between the nervous and immune
the central nervous system in graft rejection. Transplantation 27:223- systems. Brain Research Bulletin 7:595-612. [taRA]
26. [taRA] Flaherty, C. F., Uzwiak, A. J., Levine, J., Smith, M., Hall, P. & Schuler, R.
Dekker, E., Pelser, H. E. & Groen, J. (1957) Conditioning as a cause of (1980) Apparent hyperglycemic and hypoglyceinic conditioned responses
asthmatic attacks: A laboratory study. Journal of Psychosomatic Research with exogenous insulin as the unconditioned stimulus. Animal Learning
2:97-108. [tarRA] and Behavior 8:382-86. [LG]
Dixon, B. (1978) Beyond the magic bullet: The real story of medicine. Harper Fox, B. H. (1978) Premorbid psychological factors as related to cancer
& Row. [taRA] incidence. Journal of Behavioral Medicine 1:45-117. ]BHF]
Dolin, A. O. & Krylov, V. N. (1952) The role of the cerebral cortex in (1983) Current theory of psychogenic aspects of cancer incidence and
immune reactions of the organism. Zhurnal Vysshei Nervnoi Deyatclnosti prognosis. Journal of Psychosocial Oncology 1:17-31. |BHF]
Imeni I. P. Pavlova 2:547-60. [taRA] Friedberger, V. E. & Gurwitz, I. (1931) Sind bedingte Reflexe im Shine von
Dolin, A. O., Krylov, V. N., Luk'ianenko, V. I. & Flerov, B. A. (1960) New Pawlow befahigt, die Bildung von Immunantikorpern anzuregen? Ein
experimental data on the conditioned reflex production and suppression Beitrag zur Frage der Bildung spezifischer Antikorper. Zeitschrift fur
of immune and allergic reactions. Zhurnal Vysshei Nercnoi Deyatelnosti Immunitdtatsforschung und Experimentelle Therapie 72:173-79.
Imeni I. P. Pavlova 10:832-41. [taRA] |taRA]
Doroshkevich, A. A. (1954) Origination of immunological reactions from the Friedman, S. B., Ader, R. & Glasgow, L. A. (1965) Effects of psychological
effect of a conditioned stimulant. Zhurnal Vysshei Nercnoi Deyatelnosti stress in adult mice inoculated with Coxsackie B viruses. Psychosomatic
Imeni I. P. Pavlova 4:108-115. [taRA] Medicine 27:361-68. ]SBF]
Dowling, St. J. (1984) Lourdes cures and their medical assessment. Journal of Friedman, S. B. & Glasgow, L. A. (1966) Psychologic factors and resistance to
the Royal Society of Medicine 77:634-38. [TM] infectious disease. Pediatric Clinics of North America 13:315-35. |SBF|
Dubos, R. (1959) Mirage of health. Harper. |taRA] Gamazu, E., Vincent, G., Tare, N., Benjamin, W., & Farrar, J. 6c Sullivan.

A. C. (1984) Effects of stress un immune function in mice and rats. Herman, J. P., Guillnnncau, D., Dantzer, R., Scatton, B., Seincrdjian-
Society of Neuroscieiice Abstracts 10:724. [rRA, HA| Rouquiur, L. & Le Moal, M. (1982) Differential effects of inescapable
Cantt, W. H. (1953) Principles of nervous breakdown in schizokincsis and footshock and stimuli previously paired with inescapable footshocks on
uutokinesis. Annals of the New York Academy of Sciences 56:143- dopamine turnover in cortical and limbic areas of the rat. Life Sciences
63. ItaHA] 30:2207-14. [HA]
Garcia. J. & Ervin, F. R. (1968) Gustatory-visceral and telcreccptor- Hilderman, J. H. 6c Strom, T. B. (1978) Specific insulin binding site on T and
cutaneous conditioning: Adaptation in internal and external milieus. B lymphocytes as a marker of a cell activation. Nature 274:62-63. [taRA]
Communications in Behavioral Biology 1:389-415. |ALR] Hill, D. L. (1975) A review of cyclophosphamide. Charles C.
Garcia. J. 6r Koelling, R. A. (1967) The relationship of cue to consequence in Thomas. [taRA]
avoidance learning. Psychonomic Science 4:123-24. [MDK] Hill, L. E. (1930) Philosophy of a biologist. Arnold. ItarRA]
Gelfand, E. W., Cheung, R., Hastings, D. & Dosch, H. M. (1980) Hingtgen, J. N., Smith, J. E., Shea, P. A., Aprison, M. H. 6c Caff, T. M.
Characterization of lithium effects on two aspects of T-cell function. (1976) Cholinergic changes during conditioned suppression in rats.
Advances in Experimental Medicine and Biology 127:429-46. [rRA] Science 193: 332-34. |HA]
Cershon, R. K. (1974) T-cell control of antibody responses. In: Contemporary Hinson, R. E. & Potilos, C. X. (1981) Sensitization to the behavioral effects of
topics in immunobiology, vol. 3, ed., M. D. Cooper & N. L. Warner. cocaine: Modulation by Pavlovian conditioning. Pharmacology,
Plenum Press. |TR| Biochemistry and Behavior 15:559-62. |REH]
Gillette, S. 6e Gillette, R. W. (1979) Changes in thymic estrogen receptor Hobbs, S. 6c Elkins, R. (1983) Operant performance of rats selectively bred
expression follciwing orchidectomy. Cellular Immunology 42:194- for strong or weak acquisition of conditioned taste aversions. Bulletin of
96. [taRA] the Psychonomic Society 21:303-306. [RLE]
Giron, L. T., Crutcher, K. A. & Davis, J. N. (1980) Lymph nodes - A Hodgson, R. 6c Rachman, S. (1974) 2. Desynchrony in measures of fear.
possible site for sympathetic neuronal regulation of immune responses. Behavior Research and Therapy 12:319-26. [JMW]
Annals of Neurology 8:520-25. [taRA) Hollenberg, M. D. 6c Cuatrecasas, T. (1974) Hormone receptors and
Claser. R. 6r Gottlieb-Stematsky. T., Eds. (1982) Human herpes-virus membrane glycoproteins during in pirro transformation of lymphocytes.
infections: Clinical aspects. Marcel Dekker. [RG] In: Control and proliferation of animals cells, ed. S. B. Clarkson it R.
Glaser, R., Kiecolt-Glaser, J. K., Speicher, C. E , 6t Holliday, J. E. (in press) Baserga. Cold Spring Harbor. [taRA]
Stress, loneliness, and changes in herpes-virus latency. Journal of Hollis, K. L. (1982) Pavlovian conditioning of signal-centered action patterns
Behavioral Medicine. [RC] and autonomic behavior: A biological analysis of function. In: Advances in
Goldman, L., Coover, C. D. 6c Levine, S. (1973) Bidirectional effects of the study of behavior, ed. J. S. Rosenblatt, R. A. Hinde & C. Beer.
reinforcement shifts on pituitary adrenal activity. Physiology and Academic Press. [REH, MDK]
Behavior 10:209-14. |WPS| Huber, B., Devinsky, O., Gershon, R. K. 6c Cantor, H. (1976) Cell-mediated
Corczynski, R., Kennedy, M., 6c Ciampi (1985) Cimetidine reverses tumor immunity: Delayed-type hypersensitivity and cytotoxic responses are
growth enhancement of plasmacytoma tumors in mice demonstrating mediated by different T-cell subclasses. Journal of Experimental Medicine
conditioned immunosuppression. Journal of Immunology 134:4261- 143:1534-39. [taRA]
66. [taRA] Hull, C. L. (1934) Learning: 2. The factor of conditioned reflex. In: A
Gorczynski, R. M., Macrae, S. 6t Kennedy, M. (1982) Conditioned immune handbook of general experimental psychology, ed. C. Murchison. Clark
response associated with allogeneic skin grafts in mice. Journal of University Press. (taRA]
Immunology 129:704-9. (tarRA, REB, WK. GFS, SS] Hutton, R. A., Woods, S. C. 6c Makous, W. L. (1970) Conditioned
(1984) Factors involved in the classical conditioning of antibody responses in hyperglycemia: Pseudoconditioning controls. Journal of Comparative and
mice. In: Breakdown in human adaptation to stress: Towards a Physiological Psychology 71:198-201. [taRA]
multidisciplinary approach, ed. R. Ballieux, J. Fielding & A. L'Abbatte. Il'enko, V. 1. 6c Kovaleva, C. A. (1960) The conditioned reflex regulation of
Martinus Nijhoff. [taRA, AJC, REH, MDK, JMW] immunological reactions. Zhurnal Mikrobiologii, Epidemiohgii i
Gordienko, A. N. (1957) A few words regarding the reflex agency in Immunobiologii 31:108-13. [taRA]
immunogenesis. Zhurnal Mikrobiologii, Epidemiohgii I Immunobiologii Jacobs, S. & Ostfeld, A. (1977) An epidemiological review of the mortality of
28:138-42. [taRA] bereavement. Psychosomatic Medicine 39:344-57. (BHF]
Creer, S., Morris, T. 6t Pettingale, K. W. (1979) Psychological response to James, R. 4c Bradshaw, R. A. (1984) Polypeptide growth factors. Annual
breast cancer: Effect on outcome. Lancet 2:785-87. [BHF] Review of Biochemistry 53:259-92. [TM]
Cuillemin, R., Cohn, M. & Melnechuk, T., eds. (1985) Neural modulation of Jankovic, B. D. & Isakovic, K. (1973) Neuro-endocrine correlates of immune
immunity. Raven Press. [TM] response. International Archives of Allergy and Applied Immunology
Guyton, A. C. (1977) An overall analysis of cardiovascular regulation. 45:360-72. [taRA]
Anesthesia and Analgesia 56:761-68. [BD] Jemmott, J. B., Borysenko, J. Z., Borysenko, M., McClelland, D. C ,
Hadden, J. W., Hadden, E. M. 6t Middleton, E. (1970) Lymphocytoblast Chapman, R., Meyer, D. 8c Benson, H. (1983) Academic stress, power
transformation: 1. Demonstration of adrenergic receptors in human motivation, and decrease in salivary secretory immunoglobulin A
peripheral lymphocytes. Cellular Immunology 1:583-95. [taRA] secretion rate. Lancet 1:1400-1402. [RG]
Hahn, B. D., Knotts, L., Ng., M. 6c Hamilton, T. R. (1975) Influence of Jemmott, J. B. 6c Locke, S. E. (1984) Psychosocial factors, immunologic
cyclophosphamide and other immunosuppressive drugs on immune mediation, and human susceptibility to infectious disease: How much do
disorders and neoplasia in NZB/NZW mice. Arthritis and Rheumatism we know? Psychological Bulletin 95:78-108. (taRA, BHF]
18:145-52. itaRA] Jenkins, P. E., Chadwick, R. A. 6c Nevin, J. A. (1983) Classically conditioned
Hall, N. R. 6t Coldstein, A. L. (1981) Neurotransmitters and the immune enhancement of antibody production. Bulletin of the Psychonomic Society
system. In: Psychoneuroimmunology, ed. R. Ader. Academic 21:485-87. [taRA|
Press. [taRA, TR] Jerne, N. K. (1974) Towards a theory of the immune system. Annals of
(1983) The thymus-brain connection: Interactions between thymosin and Immunology I25C:373-89. [TR]
the neuroendocrine system. Lymphokine Research 2:1-6. [taRA] Keller, S. E., Weiss, J., Schleifer, S. J., Miller, N. E. 6c Stein, M. (1981)
Harris, A. H. & Brady, J. V. (1974) Animal learning - Visceral and autonomie Suppression of immunity by stress: Effects of a graded series of stressors
conditioning. Annual Review of Psychology 25:107-33. [taRA] on lymphocyte stimulation in the rat. Science 213:1397-1400. [tarRA,
Hart, R. W., Coover, C. D., Shnerson, A. 6t Smotherman, W. P. (1980) JMW]
Elevations in plasma corticosterone in ruts in response to concentrated (1983) Stress-induced suppression of immunity in adrenalectomized rats.
sugar solutions. Journal of Comparative and Physiological Psychology Science 221:1301-4. [tarRA, JMW]
94:337-45. [WPS] Kelley, K. W. (1980) Stress and immune function: A bibliographic review.
Heijnen, C. J., Croiset, C., Bevers, C , Veldhuis, H. D., De Wied, D. & Annales de Recherches Vcterinaires 11:445-78. [taRA]
Ballieux, R. E. (in preparation) Modulation of the immune response by Kelley, K. W., Dantzer, R., Mormede, P., Salmon, H. & Aynaud, J. M.
emotional stress. [rRA, HDV] (1984) Conditioned taste aversion induces immunosuppression in the
Henkiu, R. I. (1970) The effects of cortieosteroids and ACTH on sensory absence of an immunosuppressive drug. Comptes Rendus des Seances de
systems. Progress in Brain Research 32:270-94. [taRA] I Academic des Sciences, serie III, 299:123-26. [rRA, WK, TR]
Hennessy, J. W. & Levine, S. (1979) Stress, arousal, and the pituitary- Kiecolt-Glaser, J. K.. Garner, W., Speicher, C , Penn, C. M., Holliday, J. E.
adrenal system: A psychoendoerine hypothesis. In: Progress in 6c Glaser, R. (1984) Psychosocial modifiers of immunocoinpetcncc in
psychobiology and physiological psychology, vol. 8, ed. J. M. Sprague & medical students. Psychosomatic Medicine 46:7-14 [RG]
A. N. Epstein. Academic Press. IWPS] Kiecolt-Glaser, J. K., Glaser, R., Strain, E. C , Stout, J. C , Tarr, K. L.,
Herberman, R. B. & Holden, H. T. (1978) Natural cell-mediated immunity. Holliday, J. E. 6c Speicher, C. E. (in press) Modulation of cellular
Advances in Cancer Research 27:305-77. [taRA] immunity in medical students. Journal of Behavioral Medicine. [RC]

Kiccolt-Glaser, J. K.. Glascr, R., Williger, D., Stout, J., Messick, C Mason, J. D , ed. (1968) Organization of psychoundocrine mechanisms.
Sheppard, S., Ricker, D., Romisher, S. C , Briner, W., Bunncll. G., & Psychosomatic Medicine 30:565-808. [rRA]
Donnerl>erg, K. (1985) Psychosocial enhancement of inumiiiDcompctcnce Matysiak, J. 6c Green, L. (1984) On the directionality of classically
in a geriatric population. Health Psychology. [HG| conditioned glyecmic responses. Physiology and Behavior 32:5-9. [LG]
Kiecolt-Glaser, J. K., Speicher, C. E., Hollklay, J. E. & Claser, K. (1984) Meluechuk, T. (1983) Neuroimmtinomodulation: University of Kentucky,
Stress and the transformation of lymphocytes by Epstcin-Barr virus. Lexington; May 15-17, 1983. Advances l(Introductory):29-31. |TM|
;«iiriifl/ of Behavioral Medicine 7:1-12. [RC| (1984a) Member's report on "How might positive emotions affect physical
Kiecolt-Claser, J. K., Stephens, R. E., Lipetz, P. D., Speieher, C. E. & health?" Advances l(3):4-8. ITM]
Clascr, R. (in press) Distress and DNA repair in human lymphocytes. (1984b) Neural modulation of immunity: University Foundation, Brussels;
Journal of Behavioral Medicine. |RG] Ovtolwr 27-28, 1983. Advances l(4):56-59. [TM]
Klostcrhalfen, S. & Klosterhalfen, W. (1984a) Conditioned Mctal'nikov, S. & Chorine, V. (1926) Role des reflexes conditionnels dans
immunopharmacologic effects and adjuvant arthritis: Further results. I'lmmunitc. Annales de I'lnstitut Pasteur 40:893-900. [taRA, CFS|
Paper presented at the First International Workshop on (1928) Role des reflexes couditionnels dans la formation des anticorps.
Neuroiminunomodulation, Bethesda, Maryland. |WK| Comptes Rendus de la Societe Biologic 102:133-34. [taRA]
(1984b) Personal communication. [taRA] Miller, N. E. (1978) Biofeedback and visceral learning. Annual Review of
Klosterhalfen, W. & Klosterhalfen, S. (1983a) A critical analysis of the animal Psychology 29:373-404. [taRA]
experiments cited in support of learned helplessness. Psychologischc Miller, N. E., DiCara, L. V. & Wolf, G. (1968) Hoineostasis and reward: T-
Beitrage 25:436-58. |WK| inaze learning induced by manipulating antidiuretic hormone. American
(1983b) Pavloviau conditioning of iintnuuosupprcssion modifies adjuvant Journal of Physiology 215:684-86. [taRA, BD|
arthritis in rats. Behavioral Neurosciencc 97:663-66. [taRA, BHF, WK, Moiijau, A. A. (1981) Stress and immunologic competence: Studies in animals.
GFS) In: Psychonetimimmunology. ed. R. Ader. Academic Press. (taRA,
Kopeloff, L. M., Kopeloff, N. & Posselt, E. (1935) Agglutiuins and the RLE]
conditioned reflex. Journal of Immunology 29:359—66. [taRA] Monjan, A. A. & Collector, M. I. (1976) Stress-induced modulation of the
Kopeloff, L. M., Kopeloff, N. & Raney, M. (1933) The nervous system and immune response. Science 196:307-8. [taRA]
antibody production. Psychiatric Quarterly 7:84-106. |taRA] Myer, R. J. & Haggerty, R. (1962) Streptococcal infections in families: Factors
Korneva, E. A. & Khai, L. M. (1963) Effect of destruction of hypothalamic altering individual susceptibility. Pediatrics 29:539-49. |SBF]
areas on iniimiuogenesis. Fiziologicheskii Zhurnal, SSSR 49:42- Nagel, J. E. (1983) Immunology. Review of Biological Research in Aging
48. (tuRAl 1:103-60. [BTE]
Krank, M. D., Hinson, R. E. & Siegel, S. (1984) Effect of partial Nagy, E. Ac Berczi, I. (1978) Immunodeficiency in hypophysectomized rats.
reinforcement on tolerance to morphine-induced analgesia and weight Ada Endocrinologica 89:530-37.
loss in the rat. Behavioral Neuroscience 98:72-78. [MDK, SS] Newberry, B. H. (1981) Effects of presumably stressful stimulation (PSS) on
Kusnecov, A. W., Sivyer, M., King, M. G., Husband, A. J., Cripps, A. W. the development of animal tumors: Some issues. In: Perspectives on
& Clancy, R. L. (1983) Behaviorally conditioned suppression of the behavioral medicine, ed. S. E. Weiss, J. A. Herd & B. H. Fox.
immune response by antilymphocyte serum. Journal of Immunology Academic Press. (BHF]
130:2117-20. [tarRA] Nicolau, I. it Antinescu-Dimitriu, O. (1929a) Role des reflexes conditionnels
Laudenslager, M. U, Ryan, S. M., Drugan, R. C , Hyson, R. L. 6c Maier, S. dans la formation des anticorps. Comptes Rendus de la Societe Biologie
F. (1983) Coping and immunosuppression: Inescapable but not eseapable 102:133-34. [taRA]
shock suppresses lymphocyte proliferation. Science 221:568-70. [tarRA] (1929b) L'influence des reflexes conditionnels sur I'exsudat peritoneal.
Lehman, D. H., Wilson, C. B. & Dixon, F. J. (1976) Increased survival times Comptes Rendus de la Societe Biologie 102:144-45. [taRA]
of New Zealand hybrid mice immunosuppressed by graft-versus-host Nieburgs, H. E., Weiss, J., Navarrete, M., Strax, P., Teirstein, A., Grillione,
reactions. Clinical and Experimental Immunology 25:297-302. [taRA] G. & Siedlecki, B. (1979) The role of stress in human and experimental
Lett, B. T. (1983) Pavlovian drug-sickness pairings result in the conditioning oncogenesis. Cancer Detection and Prevention 2:307-36. [taRA]
of an antisickness response. Behavioral Neuroscience 97:779-84. [SR] Obal, F. (1966) The fundamentals of the central nervous control of vegetative
Levine, S. & Coover, G. D. (1976) Environmental control of suppression of homeostastis. Ada Physiologica Academy of Science, Hungary 30:15-
the pituitary-adrenal system. Physiology and Behavior 17:35-37. [WPS] 29. [REH]
Levy, S., ed. (1982) Biological mediators of behavior and disease: Neoplasia. Ostravskaya, O. A. (1930) Le reflex conditionnel et les reactions de
Elsevier/North Holland. [TM] I'immunite. Annales de I'lnstitut Pasteur 44:340-45. [taRA]
Locke, S. E. & Hornig-Rohan, M., eds. (1983) Mind and immunity: Ottenberg, P., Stein, M., Lewis, J. & Hamilton, C. (1958) Learned asthma in
Behavioral immunology: An annotated bibliography 1976-1982. Institute the guinea pig. Psychosomatic Medicine 20:395-400. [taRA]
for the Advancement of Health. [TM] Pavlov, I. P. (1927) Conditional reflexes, transl. G. V. Anrep. Oxford
Luk'ianenko, V. I. (1959) The conditioned reflex regulation of imiminologica) University Press. [REH]
reactions. Zhurnal Mikrobiologii, Epidemiologii i Immunobiologii 30:53- (1928) Lectures on conditioned reflexes. Liverwright. [taRA]
59. [taRA] Peacock, L. & Fuller, G. (1982) Radiation-induced taste aversion in two
(1961) The problem of conditioned reflex regulation of iinmunobiologic strains of rats. Paper presented at the annual meeting of the Southern
reactions. Uspekhi Sovremennoi Biologii 51:170-87. [taRA] Society for Philosophy and Psychology, Fort Worth, Texas. [RLE]
McConnell, I., Munro, A. & Waldman, H. (1981) The immune system. Pearson, C. M. & Wood, F. D. (1959) Studies of polyarthritis and other
Blackwell. [taRA] lesions induced in rats by injection of mycobacterial adjuvant. 1. General
McDevitt, H. O. (1980) Regulation of the immune response by the major clinical and pathologic characteristics and some modifying factors.
histocompatibility system. New England Journal Medicine 303:1514- Arthritis and Rheumatism 2:440-59. (taRA]
17. [TR] Persinger, M. A., Carrey, N. J., Lafreniere, G. F. & Mazzuchin, A. (1978)
McFarland, D. J. (1971) Feedback mechanisms in animal behavior. Academic Step-like DRL schedule change effects on blood chemistry, leukocytes
Press. [BD] and tissue in rats. Physiology and Behavior 21:899-904. [taRA]
McFarland, D. J. it Hotchin, J. (1983) Host genetics and the behavioral Pierpaoli, W., Kopp, H. G., Muller, J. & Keller, M. (1977) Interdependence
sequelae to herpes encephalitis in mice. Physiology and Behavior 30:881- between neiiraeiidocrim* programming and the generation of immune
84. [taRA] recognition in ontogeny. Cellular Immunology 29:16-27. [taRA]
McKenzie, I. F. C. & Potter, T. (1979) Murine lymphocyte surface antigens. Plant, S. M. & Friedman, S. B. (1981) Psychosocial factors in infectious
Advances in Immunology 27:179-338. [taRA] disease. In: Psijchoneuroimmunology, ed. R. Ader. Academic
MacKenzie, J. N. (1886) The production of the so-called "rose cold" by means Press. [taRA, SBF]
of an artificial rose. American Journal of the Medical Sciences 91:45- Pletsityi, D. F. (1957) The role of the nervous system in resistance to
57. (tarRA] infectious disease. Zhurnal Mikrobiologii, Epidemiologii i Immunobiologii
Mackintosh, N. J. (1974) The psychology of animal learning. Academic 28:131-37. [taRA]
Press. [rRA, ALR, SS] Podkopaeff, N. A. & Saatchian, R. L. (1929) Conditioned reflexes for
(1983) Conditioning and associative learning. Oxford University immunity: 1. Conditioned reflexes in rabbits for cellular reaction of
Press. [ALR] peritoneal fluid. Bulletin of the Battle Creek Sanitarium and Hospital
Maddox, J. (1984) Psychoimmunology before its time. Nature 309:400. [TM] Clinic 24:375-78. |taRA]
Maestroni, C. J. M. & Pierpaoli, W. (1981) Pharmacologic control of the Polettini, B. (1929) Importance des reflexes dit conditionnuls sur certaines
hormonally-mediated immune response. In: Psijchoneuroimmunology, ed. phenomenes iinrmmitaircs. Bollctino delta Sezionc Italiana delta Societa
R. Ader. Academic Press. [TR] Internationale di Microbiologica 1:84-87. [taRA]

Post, H. M., Luclcluld. A., Squillace, K. M. & Contcl, N. R. (1980). Drug- (1975a) Conditioning insulin effects. Journal of Comparative and
environment interaction: Context dependency of cocuinc-induecd Physiological Psychology 89:189-99. |taRA, BD, LG]
behavioral sensitization. Life Sciences 28:755-60. |REH| (1975b) Evidence from rats that morphine tolerance is a learned response.
Rabin, B. M., Hunt, W. A. & Lee, J. (1983) Acquisition of lithium chloride Journal of Comparative and Physiological Psychology 89:498-506. [SR]
and radiation-induced taste aversions in hypophyseetomized rats. (1976) Morphine analgesic tolerance: Its situation specificity supports a
Pharmacology. Biochemistry and Behavior 18:463-65. [tuRA] Pavlovian conditioning model. Science 193:323-25. [BD, LG]
Reilly, F. D., McCuskey, P. A., Miller, M. L., McCuskey, R. S. & Meineke, (1978) Tolerance to the hyperthermic effect of morphine in the rat is a
H. A. (1979) Innervation of the periarteriolar lymphatic sheath of the learned response. Journal of Comparative and Physiological Psychology
spleen. Tissue and Cell 11:121-26. (taRA) 92:1137-49. [BD]
Rcscorla, R. A. (1967) Pavlovian conditioning and its proper control (1983) Classical conditioning, drug tolerance, and drug dependence. In:
procedures. Psychological Review 74:71-80. [WK] Research advances in alcohol and drug problems, vol. 7, ed. Y. Israel, F.
Rescorla, R. A. & Holland, P. C. (1976) Some behavioral approaches to the B. Glaser, H. Kalent, R. E. Popham, W. Schmidt & R. G. Smart.
study of learning. In: Neural mechanisms of learning and memory, ed. R. Plenum Press. [rRA, REH, MDK, SS]
A. Rosenzweig & P. C. Bennett. MIT Press. |WK| (1984) Pavlovian conditioning and heroin overdose: Reports by overdose
Revusky, S. (1984) Associative predispositions. In: Dahlem workshop on the victims. Bulletin of the Psychonomic Society 22:428-30. [ALR]
biology of learning, ed. P. Marler & H. Terrace. Springer Verlag. [SR] (in press) Conditional drug responses and drug tolerance.
Revusky, S., Coombes, S. & Pohl, R. W. (1982) Drug states as discriminative Psychophannacology Bulletin. [taRA]
stimuli in a flavor aversion learning experiment. Journal of Comparative Siegel, S., Hinson, R. E. & Krank, M. D. (1978) The role of predrug signals
and Physiological Psychology 96:200-11. [SRI in morphine analgesic tolerance: Support for a Pavlovian conditioning
Revusky, S., Taukulis, H. K., Parker, L. A. & Coombes, S. (1979) Chemical model of tolerance. Journal of Experimental Psychology: Animal Behavior
aversion therapy: Rat data suggest it may be countertherapeutic to pair Processes 4:188-96. [MDK, JMW]
an addictive drug state with sickness. Behaviour Research and Therapy (1981) Morphine-induced attenuation of morphine tolerance. Science
17:177-88. [SR] 212:1533-34. [MDK]
Riley, A. L., & Tuck, D. L. (in press) Conditioned food aversions: A Siegel, S., Hinson, R. E., Krank, M. D. & McCully, J. (1982) Heroin
bibliography. In: Experimental assessments and clinical of conditioned "overdose" death: Contribution of drug-associated environmental cues.
food aversions, ed. N. S. Braveman & P. Bronstein, Annals of the New Science 216:436-37. [BD, MDK, SS]
York Academy of Sciences. [RLE, ALR] Singh, U., Milson, D. S., Smith, P. A. & Owen, J. J. T. (1979) Identification
Riley, V. (1981) Psychoendocrine influences on immunocompetence and of B adrenoceptors during thyinocyte ontogeny in mice. European
neoplasia. Science 212:1100-09. [BHF] Journal of Immunology 9:31-35. [taRA]
Rogers, M. P., Reich, P., Strom, T. B. & Carpenter, C. B. (1976) Sklar, L. S. & Anisman, H. (1981) Stress and cancer. Psychological Bulletin
Behavioratly conditioned immunosuppression: Replication of a recent 89:369-406. [BHF]
study. Psychosomatic Medicine 38:447-51. [taRA, RLE] Smith, E. M., Meyer, W. J. & Blalock, J. E. (1982) Virus-induced
Roszkowski, W., Plaut, M. & Lichtenstein, L. M. (1977) Selective display of corticosterone in hypophyseetomized mice: A possible lymphoid adrenal
histamine receptors on lymphocytes. Science 195:683-85. |taRA] axis. Science 218:1311-12. [tarRA]
Roszman, T. L., Cross, R. J., Brooks, W. H. & Markesberry, W. R. (1982) Smith, G. R. & McDaniels, S. M. (1983) Psychologically mediated effect on
Hypothalamic-immune interactions: 2. The effect of hypothalamic lesions the delayed hypersensitivity reaction to tuberculin in humans.
on the ability of adherent spleen cells to limit blastogeuesis. Immunology Psychosomatic Medicine. 45:65-70. [tarRA]
45:737-42. [taRA] Smith, H. R. & Steinberg, A. D. (1983) Autoimmiinity - A perspective.
Rozin, P. N. & Mayer, J. (1961) Thermal reinforcement and thermoregulatory Annual Review of Immunology 1:175-210. [taRA]
behavior in the goldfish, Carassius auratus. Science 134:942-43. [BD] Smotherman, W. P., Hennessey, J. W. & Levine, S. (1976) Plasma
Rusiniak, K. W., Garcia, J. & Hankins, W. G. (1976) Bait shyness: Avoidance corticosterone levels during recovery from LiCI produced taste aversions.
of the taste without escape from the sickness in rats. Journal of Behavioral Biology 16:401-12. [taRA]
Comparative and Physiological Psychology 90:460-67. (SR] Smotherman, W. P., Margolis, A. & Levine, S. (1980) Flavor preexposures in
Russell, M., Dark, K. A., Cummins, R. W., Ellman, G., Callaway, E. & a conditioned taste aversion situation: A dissociation of behavioral and
Peeke, H. V. S. (1984) Learned histamine release. Science 225:733- endocrine effects in rats. Journal of Comparative and Physiological
34. ItarRA, WK] Psychology 94:24-35. [taRA]
Russell, P. J. & Hicks, J. D. (1968) Cyclophosphamide treatment of renal Solomon, G. F. (1969) Stress and antibody response in rats. International
disease in (NZBxNZW)F, hybrid mice. Lancet i:440-6. [taRA] Archives of Allergy 35:97-104. [GFS]
Sakanjan, S. & Kostanjan, A. (1957) The role of the cerebral cortex in the (1981a) Emotional and personality factors in the onset and course of
formation of postvaccination immunity. Izvestiya Akademii Nauk autoimmune disease, particularly rheumatoid arthritis. In:
Annianskoi SSR Biologiceskoi Nauki 12:9-16. [taRA] Psychoneuroimmunology, ed. R. Ader. Academic Press. [taRA]
Sato, K., Flood, J. F. St Makinodan, T. (1984) Influence of conditioned (1981b) Immunologic abnormalities in mental illness. In:
psychological stress on immunological recovery in mice exposed to low- Psychoneuroimmunology. ed. R. Ader. Academic Press. [taRA]
dose X-irradiation. Radiation Research 98:381-88. [tarRA, WK] (in press) The emerging field of psychoneuroimmunology. Hypotheses,
Savchuk, O. Ye. (1958) The reflex mechanism of the formation of antibodies. supporting evidence, and new directions. Advances. [GFS]
Zhurnal Mikrobiologii, Epidemiologii i Immunobiologii 29:304-5. Solomon, G. F. & Amkraut, A. A. (1981) Psychoneuroendocrinological effects
[taRA] on the immune response. Annual Review of Microbiology 35:155-
Schwartz, A., Askenase, P. W. & Cershon, R. K. (1978) Regulation of 84. [taRA]
delayed-type hypersensitivity reactions by cyclophosphamide-sensitive T Solomon, G. F., Levine, S. & Kraft, J. K. (1968) Early experience and
cells. Journal of Immunology 121:1573-77. [tarRA] immunity. Nature 220:821-22. [GFS]
Schwartz, B. (1978) Psychology of learning and behavior. Norton. [taRA] Solomon, G. F. & Moos, R. H. (1964) Emotions, immunity, and disease: A
Seligman, M. E. P. (1969) Control group and conditioning: A comment on speculative theoretical integration. Archives of Ceneral Psychiatry
operationism. Psychological Review 76:484-91. [WK] 11:657-74. [CFS]
Shand, F. L. (1979) The iinmunopharmacology of cyclophosphamide. (1965) The relationship of personality to the presence of rheumatoid factor
International Journal of Iinmunopharmacology 1:165-71. [taRA] in asymptomatic relatives of patients with rheumatoid arthritis.
Shavit, Y., Lewis, J. W., Terman, G. W., Gale, R. P. & Liebeskind, J. C. Psychosomatic Medicine 1:350-60. [CFS]
(1984) Opioid peptides mediate the suppressive effect of stress on natural Solomon, R. L. (1980) The opponent-process theory of acquired motivation:
killer cell cytotoxicity. Science 223:188-90. [tarRA] The costs of pleasure and the benefits of pain. American Psychologist
Shek, P. N. & Sabiston, B. H. (1983) Neuroendocrine regulation of immune 35:691-712. (rRA, REH, JMW]
processes: Change in circulating cortiscosterone levels induced by the Solomon, R. L. & Corbit, J. D. (1974) An opponent-process theory of
primary antibody response in mice. International Journal of motivation. Psychological Review 81:119-45. [SS]
Iinmunopharmacology 5:23-33. [taRA] Solvason, H. B., Ghanta, V., Hiramoto, R. & Spector, N. H. (1984) Natural
Shenkman, L., Wadler, S. Borkowsky, W. & Shopsin, B. (1981) Adjuvant killer cell activity augmented by classical Pavlovian conditioning.
effects of lithium chloride on human inononuclear cells in suppressor- Proceedings of the First International Workshop on
enriched and suppressor-depleted systems. Iinmunopharmacology 3:1- Neuroimmunomodulation. In press. [CFS]
8. [rTA] Spector, N. H. (1980) The central state of hypothalamus in health and disease:
Siegel, S. (1972) Conditioning of insulin-induced glycemia. Journal of Old and new concepts. In: Physiology of the hypothalamus, vol. 2, ed. P.
Comparative and Physiological Psychology 78:233-41. [BD] J. Morgane & J. Panksepp. Marcell Dekker. |taRA]

Stein, M , Schleifer, S. J. & Keller, S. E. (1981) Hypothalamic influences on Wayner, E. A., Flannery, G. R. & Singer, G. (1978) Effects of taste aversion
immune responses. In: Psychoneuroimmuttology, ed. H. Ader. Academic conditioning on the primary antibody response to sheep red IIIIKKI cells
Press. | HA. taHA| and Brucella abortus in the albino rat. Physiology and Behavior 21:995-
Steinlwrg, A. D.. Gelfand. M. C , Hardin, J. A. «t Lowenthal, D. T. (1975) 1000. itarRA. HLE]
Therapeutic studies in NZB/W mice: 3. Relationship between renal status Weiss, B. & Laties, V. C. (1960) Behavioral thermoregulation. Science
and efficacy of imimmostippressive drug therapy. Arthritis and 133:1338-44. [BD]
Rheumatism 18:9-14. [taRA] Whitehouse, M. W., Levy, L. & Beek, F. J. (1973) Effect of
Steinlicrg, A. D., Huston, D. P., Taurog, J. D., Cowdery, J. S. & Kaveche, cyclophosphamide on a local graft-versus-host reaction in the rat influence
E. S. (1981) The cellular and genetic basis ofmuriiie lupus. Immunology of sex disease and different dosage regimens. Agents and Actions 3:53-
Reviews 55:121-54. |taKA| 60. [taRAj
Strom, T. B., Sytkowsky, H. A., Carpenter, C. B. & Merrill, J. P. (1974) Wigzell, H. (1985) Properties of the specific immune system in relation to its
Cholinergic augmentation of lymphocyte-mediated cytotoxicity: A study of possible regulation by the central nervous system. In: Neural nunUdation
the eholinergic receptor of cytotoxic T-lymphocytes. Proceedings of the of immunity, ed. R. Guillemin, M. Cohn & T. Melnechuk. Raven
National Academy of Sciences of the United States of America 69:1330- Press. [TM]
33. [taRA] Wikler, A. (1973a) Conditioning of successive adaptive responses to the initial
Talal, N. (1976) Disordered iinmunologic regulation and autoiininunity. effects of drugs. Conditional Reflex 8:193-210. [ALR]
Transplantation Reviews 31:240-63. |taRA] (1973b) Dynamics of drug dependence. Implications of a conditioning
Teinoshok, L., Van Dyke, C. & Zegans, L. S., eds. (1983) Emotions in health theory for research and treatment. Archives of General Psychiatry
and illness: Theoretical and research foundations. Crune and 28:611-16. |ALR|
Stratton. [TM] Williams, J. M., Peterson, R. G., Shea, P. A., Sehmedtje, J. F., Bauer, D.
Theofilopoulos, A. N. & Dixon, F. J. (1981) Etiopathogenesis of murinc Sl.E. C & Felten, D. L. (1981) Sympathetic innervation of murine thyinus and
Immunology Reviews 55:179-216. [taRA] spleen: Evidence for a functional link between the nervous and immune
Turk, J. L. & Parker, D. (1979) Tile effect of cyclophosphamide on the systems. Brain Research Bulletin 6:83-94. |taRA, TR]
immune response. Journal of lmmunophaniMCulogy 1:127-37. [taRA] Woods, S. C (1976) Conditioned hypoglyceinia. Journal of Comparative and
Turk, J. L. & Parker, D. (1982) Effect of cyclophosphamide on immunological Physiological Psychology 90:1164-68. (LG]
control mechanisms. Immunological Reviews 65:99-113. [rRA] Zdrodovskii, P. F. (1956) Current status of theoretical immunology and its
van Sommers, P. (1962) Oxygen-motivated behavior in the goldfish. Carassius immediate task. Vestnik Akademii Meditsinkikh Nauk SSSR 11:43-
auratus. Science 137:678-79. [BD] 57. [taRA]
Vygodchikov, G. V. (1955) Certain controversial questions in the theory of Zeitlenok, N. A. & Bychkova, E. N. (1954) On the study of the role of higher
immunity. Zhurnal Mikrobiologii, Epidemiologii i Immunobiologii 26:5- nervous activity in infection and immunity. Zhurnal Vysshei Nervnoi
14. [taRA] Deyatelnosti Imeni I. P. Pavlova 4:267-81. [taRA]
(1957) Results cf a discussion on the basic problems of the study of Zellner, D. A., Dacanay, R. J. & Riley, A. L. (1984) Opiate withdrawal: The
immunity. Zhurnal Mikrobiologii, Epidemiologii i Immunobiologii result of conditioning or physiological mechanisms? Pharmacology,
28:623-28. [taRA] Biochemistry and Behavior 20:175-80. [ALR]
Vygodchikov, G. V. & Barykini, O. (1927) The conditioned reflex and
protective cell reactions. Zhurnal Biologii i Meditsiny Eksperimentalnoi
6:538-41. [taRA]

Ader 1985

Uploaded by

Copyright:

Available Formats

Ader 1985

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ader 1985

Uploaded by

Copyright:

Available Formats

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8, 379-394

Printed in the United States of America

CNS-immune system interactions:

The purpose of this paper is to direct attention to an function-and behavior-psychoneuroimmunology may

© 1985 Cambridge University Press OUO-525XI85IO3O379-16I$O6.OO 379

380 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

1. Historical perspectives trials in which a CS was paired with injection of a

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 381

382 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

cs0 cs, cs, us

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 383

384 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

6:30 a.m. 6:30 p.m 6 30 a.i

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 385

386 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

the mechanisms, the experimental demonstration of the

these results also suggest that the conditioned immu-

received 4, 9, or 18 extinction trials did not differ among

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 387

cytes from the CBA recipients were analyzed to assess the

388 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 389

390 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 391

392 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 393

394 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 395

396 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 397

398 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 • 399

400 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 401

402 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 403

404 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 405

406 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

Questions about conditioned

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 407

408 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 409

410 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 411

412 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

414 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 415

416 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 417

418 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 419

420 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 421

422 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 423

424 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3

THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3 425

426 THE BEHAVIORAL AND BRAIN SCIENCES (1985) 8:3