T H E

B E H A V I O R

A N A L Y S T

T O D A Y

V O L U M E

4 ,

I S S U E

1 ,

2 0 0 3

THE MOLAR VIEW OF BEHAVIOR AND ITS USEFULNESS IN BEHAVIOR

ANALYSIS

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

William M. Baum
University of California, Davis
The molar view of behavior contrasts with the older, molecular view. The difference is
paradigmatic, not theoretical. No experiment can decide between them, because they interpret all
the same phenomena, but in different terms. The molecular view relies on the concepts of discrete,
momentary events and contiguity between them, whereas the molar view relies on the concepts of
temporally extended patterns of activity and correlations. When dealing with phenomena such as
avoidance, rule-governed behavior, and choice, the molar view has the advantage that it requires no
appeal to hypothetical constructs. The molecular view always appeals to hypothetical constructs to
provide immediate reinforcers and stimuli when none are apparent. As a result, the explanations
offered by the molar view are straightforward and concrete, whereas those offered by the molecular
view are awkward and implausible. The usefulness of the molar view for applied behavior analysis
lies in the flexibility and conceptual power it provides for talking about behavior and contingencies
over time.

momentary events and causes when none are

apparent, whereas the molar view relies on
extended activities and extended causation,
avoiding postulation of hypothetical constructs.
Replacing the concept of momentary
response with the concept of extended activity
requires one to become familiar with thinking in
more continuous termsthat is, in terms of
extended patterns that cannot be seen at a
moment in time. A familiar example is the
concept of probability. An unbiased coin, when
flipped, comes up heads with a probability of
.50. What does this mean? On any particular
flip, the coin comes up heads or tails; nothing
more can be observed. Only for a long series of
flips can one observe the probability of .50. If
one says that on a particular flip the probability
is .50, all one means is that in a long series of
such flips about half would show heads. The
same is true of response rate. At any particular
moment, an activity (lever pressing) is occurring
or not. One can only observe the response rate
over some substantial time period. A response
that occurs 60 times per minute cannot occur 60
times per minute at a moment.
Although Skinner advocated the use of
response rate as a dependent variable, he was a
molecularist. In his well-known paper on
superstition, Skinner (1948) proposed a
snapshot view of reinforcement, in which
delivery of a reinforcer strengthens whatever
behavior happens to be occurring at the moment.
The molecularity of his approach is perhaps
nowhere clearer than in a short piece he wrote
called Farewell, My Lovely! in which he
deplored the absence of cumulative records in
the pages of JEAB and extolled the virtues of

The molar view of behavior is relatively

new. Although its origins may be traced back
earlier, its first partial articulation was by Baum
and Rachlin (1969), in a paper called Choice as
time allocation. It was presented more fully in
a paper by Baum (1973), The correlation-based
law of effect. Rachlin (1994) offered a booklength presentation, and Baum (2002) elaborated
on his 1973 paper in another paper, From
molecular to molar: A paradigm shift in
behavior analysis and some papers in-between
(Baum, 1995a; 1997).
The molar view contrasts with an older view that
behavior analysis inherited from nineteenthcentury psychology. I call this older view
molecular, because it is based on the notion
explanations of behavior may be constructed by
thinking of small discrete units being joined
together into larger units, like the joining
together of atoms into molecules in chemistry.
The difference between the molecular
and molar views of behavior is paradigmatic, not
theoretical. No data, no experiment can decide
between the two views, because no matter what
behavioral phenomenon one chooses, a
proponent of either view is able to construct an
account of it. The difference between the two
lies in the concepts each brings to bear in such
an account. The molecular view relies on
momentary events and momentary causation,
which leads to postulating hypothetical
Authors Note:
Correspondence concerning this article may be addressed to:
William M. Baum, 611 Mason, #504, San Francisco,
CA 94108 or 415-345-0050. Email address:

[email protected]

78

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

B A U M

The likeliest way to overcome problem drinking

is with local reinforcers for abstinent behavior.
Thus, the molar view, like the molecular view,
says that, in practice, the one who would shape
behavior needs to be swift with the reinforcers.
The molecular view has one point in its
favor: It coincides with a prejudice toward
immediate causes. The notion that the events
that affect behavior occur either immediately
before or immediately after a response lends
simplicity to analysis. One knows just where to
look for the antecedents and consequences that
control the response. That simplicity, however,
comes at a high price: the necessity of inventing
immediate antecedents and consequences when
none are apparent. Perhaps the best example is
explaining avoidance.
To explain avoidance, in which success
means that nothing happens following a
response, molecularists turn to two-factor
theory. Since a reinforcer must follow the
avoidance response, even if none is apparent,
one has to be invented. Suppose that the
stimulus preceding the response becomes a
Pavlovian conditional stimulus, eliciting fear.
Then, when the response turns off the stimulus,
the reduction in fear reinforces the response.
Avoidance responding occurs, however, even if
no stimulus precedes or is terminated by the
activity (Herrnstein & Hineline, 1966;
Herrnstein, 1969). Having already invented the
fear-reduction reinforcer, the molecularist now
also invents the stimulus. Dinsmoor (2001), for
example, argued that response-produced stimuli,
paired with a lower frequency of electric shock
than their absence, become safety signals. The
cost of maintaining the molecular view here is
that one must appeal to hypothetical reinforcers
and stimuli when none are observable. The
result is a theory that cannot be refuted.
The molar view of avoidance is
arguably simpler, but requires one to think in
terms of temporally extended patterns.
Avoidance activity is acquired and maintained
because when that activity is present the rate of
noxious events is lower than when it is absent.
People avoid sensitive topics in conversation to
lower the likelihood of embarrassment to
themselves and others. People buy insurance to
lower the likelihood of financial hardship.
Much apparently dysfunctional behavior may be
understood as avoidance. If working and failing
would be too hard an outcome, one may avoid it
by being ill.

being able to observe molecular, moment-tomoment changes in behavior (Skinner, 1976). A

cumulative recorder, however, is an averaging
machine; it only produces smooth curves
because the chart moves slowly and the pen
moves in small steps. At any particular moment,
either a response is occurring or not. The local
changes in response rate are changes from one
interval to another. If, however, one were to fit
a truly continuous curve to a cumulative record,
then one might think of momentary rate as the
slope of the curve at a particular point. This,
however, requires abstracting the continuous
function.
In the molecular view, each response is
taken as a concrete particular (i.e., the basic
observation), and response rate is a derived
measure (i.e., an abstraction) summarizing
behavior over a period of time. The molar view
turns this distinction around, making the
extended pattern the concrete particular and the
momentary response the abstraction. A response
rate or activity exists as a pattern through time.
Any attempt to infer activity at a moment
depends on abstraction, as in the example of the
cumulative record. In fact, no behavior can be
observed at a moment, because even the
simplest unit of behaviorlever press, key peck,
button pushtakes up time and must unfold
from beginning to end before it can be recorded
with certainty (for further discussion, see Baum,
1997; 2002). Because every activity takes up
time, the concept of behavior at a moment is an
abstraction, an inference made after the fact.
Although it has little use for momentary
events, the molar view supports analysis in more
and less extended time frames (Baum, 1995;
1997; 2002). That patterns take up time in no
way precludes them from being brief. A
pigeons key peck, for example, is an extended
pattern that takes a fraction of a second.
Analysis may be as local or as extended as suits
ones purpose. When trying to change behavior,
one should make sure that reinforcers are closely
coordinated with the activity one is trying to
increase. The molecularist insists reinforcers
must immediately follow the responses they are
to strengthen; the molarist says reinforcers
should coincide closely with the activity to be
increased. Such local relations often have
powerful effects, sometimes to our grief, when
they override more extended relations (Rachlin,
2000). Each additional drink might seem
harmless, but in the long run they add up to ruin.

79

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

T H E

B E H A V I O R

A N A L Y S T

T O D A Y

V O L U M E

4 ,

I S S U E

1 ,

2 0 0 3

view, such a pattern constitutes a concrete

particular. The molecular view, focusing on a
moment, immediately moves to hypothetical
constructs. Each alternative has a certain
strength, unobservable at the moment but
existing at the moment. The extended pattern of
allocation is thought to reveal the relative
strengths of the alternatives. If a pigeon pecks
twice as often at the left key than the right key,
the strength of left pecking is considered twice
that of right pecking. If a child spends twice as
much time disrupting classroom activities as the
child spends doing schoolwork, the strength of
disrupting is twice that of remaining on task. In
the molar view, no hypothetical strength enters
in, because these patterns of allocation are what
the science is about.
Even if the molar view seems to allow
such phenomena as avoidance, rule-governed
behavior, and choice to be understood more
readily, the question arises as to whether the
molar view has any implications for applied
behavior analysis. It makes for the same sort of
rule of thumb as the molecular view when one is
trying to change behavior: reinforcement must
be frequent and quick. Beyond this, however, I
think the molar view might have some
advantages for applications. First, it offers
flexibility in thinking about goals and
treatments. No need arises to define some
artificial discrete response for reinforcement.
One needs only to make sure that reinforcers
accompany appropriate activity. For example,
in school settings applied behavior analysts
already often talk about time on task as a
reinforceable activity. The molar view allows
this kind of flexible thinking about
reinforcement of activities to be extended
indefinitely. Second, it frees one to think about
time spent instead of response rate. Without
artificial discrete responses, activities like
reading, playing, grooming, and the like can be
measured by timing them. Time spent should be
no harder to measure than counting responses
and often will be less ambiguous, because one
may be able start and stop timing more easily
than decide whether exactly the right response
occurred. Once applied behavior analysts grow
accustomed to the molar way of talking, they
will find it more congenial for communicating
with one another about behavior and
contingencies, because it is more flexible and
more concrete.

Another example of paying a high price to

retain a molecular view is in accounting for rulegoverned behavior. Rules present a problem for
the molecular view because they are invariably
associated with behavior that has important
consequences in the long run (Baum, 1994;
1995). Since long-delayed effects must be
ineffective to the molecularist, if rule-governed
behavior is maintained, some immediate
(effective) consequences must be found. Why
would someone eat vegetables instead of candy
when no one else is present to observe? Why
would someone save a piece of trash until a trash
can appears, when it might have been dropped
on the street with impunity? Mallott (2001), in a
paper about moral and legal control, provides
the molecularists answer: thoughts and selfpunishment. He argues, For moral control to
work, society must have established a special,
learned aversive conditionthe thought of the
wrath of ones God or the thought of the wrath
of ones parents. And those thoughts must be
aversive, even when no one is looking (p. 4).
Again the molecular view leads directly into the
realm of the hypothetical and unverifiable.
The molar view of rule-governed
behavior allows that any contingency, no matter
how extended, may control behavior, even
though more local contingencies may be more
powerful than more extended ones (Baum, 1994;
1995; Rachlin, 1994; 1995; 2000). Rules exist,
however, because extended contingencies are
weak. A rule is a discriminative stimulus
produced by one person that induces in another
person behavior that is reinforced socially in the
short run (and reinforced in some major way in
the long run). The behavior may come under the
control of the long-term contingencyfor
example, the relationship between diet and
health. Although people often say that then the
rule has been internalized, from the molar
point of view, it actually is further externalized,
because the control is exerted by a more
extended contingency. In looking at rulegoverned behavior this way, the molar view
introduces no hypothetical events and no new
terms.
Perhaps the strongest area of application
of the molar view is to choice, the allocation of
behavior among alternatives. At any moment,
behavior is assigned to only one alternative.
Over time, however, one sees a pattern of
allocation among alternatives. In the molar

80

B A U M

Baum, W. M., & Rachlin, H. C. (1969). Choice as time allocation.

Journal of the Experimental Analysis of Behavior, 12, 861874.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

In conclusion, two points might be

made. First, although the molecular view was
useful early in the development of behavior
analysis, the science has outgrown it, and the
molar view supplies the conceptual power
required for the new developments. Second, the
molar view may be recommended for the
flexibility and power that it allows both applied
and basic researchers in talking about behavior
and contingencies.

Dinsmoor, J. A. (2001). Stimuli inevitably generated by behavior

that avoids electric shock are inherently reinforcing. Journal
of the Experimental Analysis of Behavior, 75, 311-333.
Herrnstein, R. J. (1969). Method and theory in the study of
avoidance. Psychological Review, 76, 49-69.
Herrnstein, R. J., & Hineline, P. N. (1966). Negative reinforcement
as shock-frequency reduction. Journal of the Experimental
Analysis of Behavior, 9, 421-430.

References

Mallott, R. W. (2001). Moral and legal control. Behavioral

Development Bulletin, 1, 1-7.

Baum, W. M. (1973). The correlation-based law of effect. Journal

of the Experimental Analysis of Behavior, 20, 137-153.

Rachlin, H. (1994). Behavior and mind: The roots of modern

psychology. New York: Oxford University Press.

Baum, W. M. (1994). Understanding behaviorism: Science,

behavior, and culture. New York: HarperCollins.

Rachlin, H. (1995). Self-control: Beyond commitment. Behavioral

and Brain Sciences, 18, 109-159.

Baum, W. M. (1995a). Introduction to molar behavior analysis.

Mexican Journal of Behavior Analysis, 21, 7-25.

Rachlin, H. (2000). The science of self-control. Cambridge, MA:

Harvard University Press.

Baum, W. M. (1995b). Rules, culture, and fitness. The Behavior

Analyst, 18, 1-21.

Skinner, B. F. (1948). "Superstition" in the pigeon. Journal of

Experimental Psychology, 38, 168-172.

Baum, W. M. (1997). The trouble with time. In L. J. Hayes & P.

M. Ghezzi (Eds.), Investigations in behavioral epistemology
(pp. 47-59). Reno, NV: Context Press.

Skinner, B. F. (1976). Farewell, my lovely! Journal of the

Experimental Analysis of Behavior, 25, 218.

Baum, W. M. (2002). From molecular to molar: A paradigm shift

in behavior analysis. Journal of the Experimental Analysis of
Behavior, 78, 95-116.

81