CRIMINAL LAW AND MULTIPLE

PERSONALITY DISORDER:
THE VEXING PROBLEMS OF
PERSONHOOD AND RESPONSIBILITY
WALTER SINNOTT-ARMSTRONG∗ & STEPHEN BEHNKE**

I. INTRODUCTION
Multiple Personality Disorder (“MPD”), now also known as
Dissociative Identity Disorder, presents vexing problems for courts as they
attempt to assess criminal responsibility. This challenge to the legal system
arises because traditional tests for criminal responsibility do not address the
unique—and sometimes bizarre—phenomenology of MPD. While the
criminal justice system looks to a variety of tests for assessing
responsibility, such as whether individuals could understand or appreciate
the consequences of their actions, or whether they were able to conform
their actions to the requirements of the law, the unique symptomatology of
MPD lies outside these traditional measures.
MPD is characterized by a dividedness of mind. Dividedness of mind,
by itself, is not a traditional basis for a finding of criminal
nonresponsibility. Yet, in certain instances, individuals who struggle with
MPD and who commit criminal acts appear to be excellent candidates for
an insanity or involuntariness defense.
This Article has four goals. First, it reviews the positions courts have
taken on the question of MPD and criminal responsibility. Second, it
explores arguments behind one commentator’s theory of why many
individuals with MPD should not be held responsible for their crimes, and
examines the theory of personhood upon which this commentator’s
argument is based. Third, it analyzes theories of personhood and personal
identity that challenge this commentator’s position of general
nonresponsibility. Finally, it discusses the relationship between MPD and
criminal responsibility.

∗
Professor of Philosophy, Dartmouth College.
**
Instructor in Psychiatry, Harvard Medical School.
277
278 Southern California Interdisciplinary Law Journal [Vol. 10:2

II. COURT ANALYSES OF MPD AND CRIMINAL RESPONSIBILITY

According to the Diagnostic and Statistical Manual, Fourth Edition,
(DSM-IV), MPD is characterized by the presence of: two or more distinct
identities or personality states, at least two of which recurrently take control
of the person’s behavior; an inability to recall important personal
information that is too extensive to be explained by ordinary forgetfulness;
and the lack of any direct connection between the disturbance and the
physiological effects of a substance (e.g., blackouts) or a general medical
condition (e.g., complex partial seizures).1
While experts disagree about the number of people who meet these
criteria—some mental health experts doubt that MPD exists at all—courts
have considered MPD as the basis for both insanity and involuntariness
defenses. Courts have used three analyses to assess the criminal
responsibility of an individual with MPD. These three analyses are
represented by the Grimsley2, Denny-Shaffer3, and Rodrigues4 cases.
In the Grimsley case, Robin Grimsley received a report of a lump in her
breast. This psychological trauma, according to Grimsley, caused her
alternate personality, Jennifer, to emerge. Grimsley described Jennifer as
impulsive, angry, fearful, and anxious, and as having a drinking problem.
On one occasion, Grimsley drove after drinking, and subsequently was
charged with driving under the influence of alcohol. In her defense,
Grimsley said that when Jennifer, her alternate personality, is in control,
Robin, her primary personality, is not aware of what is going on and cannot
control Jennifer’s actions. Grimsley had previously been diagnosed with
MPD.
The court held Grimsley responsible. The court reasoned:
[t]here was only one person driving the car and only one person accused
of drunken driving. It is immaterial whether she was in one state of
consciousness or another, so long as in the personality then controlling her
behavior, she was conscious and her actions were a product of her own
volition. The evidence failed to
5
indicate that Jennifer was unconscious or
otherwise acting involuntarily.
According to the Grimsley court, an individual’s criminal responsibility
rests on the mental state of the alter in control at the time of the crime. The
court concluded that because the alter personality—Jennifer—was not

1
AM. PSYCHIATRIC ASS’N, DIAGNOSTIC AND STATISTICAL MANUAL OF MENTAL DISORDERS
th
(American Psychiatric Press, 4 ed., 1994).
2
State v. Grimsley, 444 N.E.2d 1071, 1076 (Ohio App. 1982)
3
United States v. Denny-Shaffer, 2 F.3d 999 (10th Cir. 1993)
4
State v. Rodrigues, 679 P.2d 615 (Haw. 1984), cert. denied, 469 U.S. 1078 (1984)
5
Grimsley, 444 N.E.2d at 1075–76.
2001] Criminal Law and Multiple Personality Disorder 279

insane at the time of the drunken driving, Grimsley as a whole should be

held criminally accountable.
The first test for criminal responsibility, set forth by the Grimsley case,
focuses on the alter personality in control at the time of the crime.
According to Grimsley, a court should assess that alter’s mental state in
order to determine whether the individual should be held responsible. The
alter personality in control at the time of the crime is the touchstone for
accountability.
In the Denny-Shaffer case, Denny-Shaffer disguised herself as a
medical student, entered a hospital nursery, took a newborn baby, and
transported the baby to another state. Upon arriving at her destination, the
home of an ex-boyfriend, she called her ex-boyfriend and told him that she
had given birth to his baby and wanted him to come home from work. She
had taken blood and a placenta from the hospital in order to convince him
that she really did give birth to the baby there. When told to leave his home,
she took the baby to her family, several hundred miles away. After several
weeks with her family, she decided to return to her state of origin. The
police, who had been alerted that she was traveling with a child not her
own, apprehended her while she was in transit. Despite her attempts to
conceal the child from the police, the baby was found and returned to its
parents. Denny-Shaffer was charged with kidnapping and transporting the
infant across state lines. Her defense was that her host personality,
“Gidget,” was unconscious and not aware of the kidnapping for all or part
of those weeks. At trial Denny-Shaffer was found guilty.
Denny-Shaffer appealed her conviction. The appeals court held that the
trial court had been incorrect “in restricting the focus of the court and jury
narrowly to the alter or alters cognizant of the offense, and ignoring proof
that the dominant or host personality was not aware of the wrongful
conduct.”6 The court concluded that there was sufficient evidence at trial to
support critical inferences that would have helped Denny-Shaffer’s insanity
defense. These inferences were the following:
that Denny-Shaffer suffers from a severe mental defect or disease [MPD];
that at the time of the abduction, her dominant or host personality was not
in control so as to cause commission of the offense, and was not aware
that an alter personality or personalities were the cognizant parties
controlling the physical actions; that as a result of the defendant’s severe
mental disease or defect, the host or dominant personality was unable to
appreciate the nature and quality or wrongfulness of the conduct which
the alter or alters controlled; and
7
that defendant had proven these facts by
clear and convincing evidence.

6
Denny-Shaffer, 2 F.3d at 1014.
7
Id. at 1016 (brackets in original).
280 Southern California Interdisciplinary Law Journal [Vol. 10:2

The court remanded the case to the trial court to determine whether the
host personality should be held criminally accountable.
The second test for criminal responsibility, set forth by the Denny-
Shaffer case, focuses on the host personality. According to Denny-Shaffer, a
court should assess the host’s mental state at the time of the crime in order
to determine whether the individual should be held responsible. The host
personality—rather than the alter in control—is the touchstone for
accountability.
The third and final test for criminal responsibility in cases of MPD
arises from the Rodrigues case, in which the court reasoned that “[s]ince
each personality may or may not be criminally responsible for its acts, each
one must be examined under the American Law Institute (ALI)-Model
Penal Code (MPC) competency test.”8 While the court did not explain what
to do when some alters are innocent and others complicit in the crime, the
import of the Rodrigues holding is clear: courts must assess the mental
state of each and every personality in order to determine whether a
defendant should be held criminally responsible.
The Grimsley, Denny-Shaffer, and Rodrigues decisions provide the
three tests for assessing the criminal responsibility of an individual with
MPD. No one test dominates American legal analyses, although the
Grimsley and Denny-Shaffer analyses have held sway over courts and
commentators alike.

III. A THEORY OF GENERAL NONRESPONSIBILITY

Elyn Saks has proposed a theory of general nonresponsibility of
individuals with MPD.9 On Saks’ theory, many persons with MPD will not
be held criminally responsible. According to Saks, persons with MPD are
nonresponsible unless all of their alters have acquiesced to the crime.
Acquiescence occurs when an alter either was complicit in the crime, or
could have prevented the crime but did not. Saks supports her position by
arguing that alters can be understood in one of three ways, each of which
leads to her final conclusion that in many instances, individuals with MPD
should not be held criminally responsible.
According to Saks, the first way of understanding alter personalities is
as separate persons. Saks believes that this way of viewing alters is
consistent with certain philosophical views of personhood and personal
identity. Saks looks to empirical research to bolster the argument that alters
are persons in their own right. If, as Saks points out, alter personalities are
8
Rodrigues, 679 P.2d at 618.
9
ELYN R. SAKS, WITH STEPHEN H. BEHNKE, JEKYLL ON TRIAL: MULTIPLE PERSONALITY
DISORDER AND CRIMINAL LAW (1997).
2001] Criminal Law and Multiple Personality Disorder 281

persons, then the presence of any alter personality innocent of wrongdoing

is sufficient to vitiate responsibility. This way of assessing criminal
responsibility has a strong foundation in our system of jurisprudence,
which holds that ten guilty people should go free rather than one innocent
person be jailed.
The second way of understanding alter personalities is as personlike
centers of consciousness. This second way does not hold that alters are
persons, but that they are sufficiently like persons so that the law should
exonerate an individual with MPD if an alter was not complicit in
wrongdoing. The argument for general nonresponsibility based on the
second way of considering alters closely tracks Saks’ argument based on
her theory of alters as separate persons; personlike centers of
consciousness, like persons, are entitled to legal recognition and protection
by criminal law.
The final way of understanding alter personalities is that they represent
fragments of a deeply divided person. On this third view, alter personalities
are thought of as split-off or never-integrated aspects of a person that
crystallize into seemingly separate personalities. Because of such
“crystallization,” the individual often makes decisions in which significant
aspects of her person are left out. As a consequence, a criminal act may not
represent the person as a whole. Saks points out that there is precedent in
law for deeming deeply divided individuals not responsible for their crimes.
For example, sleepwalking and post-hypnotic suggestion have been said to
render an individual criminally nonresponsible. While each of these three
views of alter personalities can be used to argue for Saks’ theory of general
nonresponsibility, her argument for the personhood of alters is perhaps
most challenging.
According to Saks, the ontological status of alter personalities hinges
on two questions: that of personhood, and that of identification as a
person.10 To answer the first question, Saks looks to Daniel Dennett, who
sets out six criteria for personhood. According to Dennett, a person is
rational, the subject of intentional predicates, a moral object, a moral
subject or agent, a user of language, and an entity possessing a special kind
of consciousness, such as self-consciousness. To illustrate the application
of Dennett’s criteria, animals, which may well be moral subjects, could not
be considered moral agents. Hence, animals are not persons.
Saks believes that some alter personalities meet all six of Dennett’s
criteria. To further her argument that alters within a single body may be
separate and distinct persons, she points to other features of alter
personalities that she sees as consistent with their ontological status as

10
Id. at 41–51.
282 Southern California Interdisciplinary Law Journal [Vol. 10:2

persons. According to Saks, alter personalites differ from one another

characterologically, have distinct senses of self, view their fellow alter
personalities to be separate people, see the world from a first-person
perspective, have their own unique life-histories and memories, and show
distinct and separate physiological responses. As persuasive as these
arguments in favor of personhood may be, Saks is mindful that the feature
many consider most central to personhood—the possession of a body—is
something alter personalities lack.
To examine the role that a body plays in thinking about personhood,
Saks turns to the philosophers John Locke, Bernard Williams, William
Carter, and Derek Parfit. Saks examines the writings of these philosophers,
such as Locke’s famous question of what would happen were a prince and a
cobbler to exchange psychological characteristics, and responds to
arguments in favor of a bodily criterion for personality identity. Saks
concludes that a psychological criterion for personal identity is at least
plausible, and perhaps more persuasive, than a bodily criterion. She
concludes that alter personalities may be—perhaps should be—considered
separate and distinct persons in the eyes of the criminal law.
Given the premise that alter personalities may be persons, Saks
concludes that many individuals with MPD should not be held responsible
for their crimes. Indeed, because certain alters probably will not be
complicit in a criminal act, it would be unfair to hold them accountable for
criminal wrongdoing. Thus, unless all alters are criminally involved, or at
least aware of and do nothing to prevent a crime, the individual should not
be found guilty by a criminal court.

IV. PERSONHOOD AND RESPONSIBILITY IN MPD: A RESPONSE

TO SAKS
Saks theory of the personhood of alter personalities, upon which in part
she bases her argument for criminal nonresponsibility, raises challenging
and provocative questions. What persuasive arguments could be made for
the multiple-persons view?
One common strategy to argue for the multiplicity of persons is to
apply a paradigm like Daniel Dennett’s criteria of personhood.11 Dennett
claims that a person is rational, the subject of intentional predicates, a
moral object, a moral subject or agent, a language user, and the possessor
of a special kind of consciousness (or self-consciousness). Each alter has
these characteristics in at least some cases of MPD. In the Grimsley case,

11
Daniel Dennett, Conditions of Personhood, in THE IDENTITIES OF PERSONS 177–78 (Amélie
Oksenberg Rorty ed., 1976), cited in Elyn Saks, Multiple Personality Disorder and Criminal
Responsibility, 10 S. CAL. INTERDISC. L.J. 185, 190 (2001).
2001] Criminal Law and Multiple Personality Disorder 283

for example, both Robin and Jennifer use language, are self-conscious,
make rational choices and form rational beliefs (which is an intentional
predicate), have rights (including not to be hurt), and have duties as moral
agents (including not to lie). One might therefore conclude that each alter is
a separate person.
This argument fails because all of these observations can be explained
by someone who denies that Robin and Jennifer are separate persons.
Imagine a normal person without MPD named Norm. That part of Norm
that consists of Norm’s life on a certain day, or when he is in a certain
mood, can be called NormPart. NormPart uses language and is self-
conscious and rational, insofar as Norm uses language and is self-conscious
and rational on that particular day, or when he is in that particular mood.
NormPart also has a right not to be hurt and a duty not to lie insofar as
Norm has those rights and duties on that particular day (or when he is in
that particular mood). All of these characteristics of persons can be
ascribed to NormPart, because they can be ascribed to Norm during the
time when that part of Norm exists.
Now compare Norm with alters in a case of MPD. If alters are parts of
a person, instead of separate persons, then they can meet all of Dennett’s
criteria of persons simply because the persons of whom they are parts have
those properties during times when those alters are in control of the person.
Even if different alters exist at the same time, they perform different
individual acts, and each act can be ascribed to a particular alter only
because the person of whom that alter is a part does that act. In this way,
those theorists who claim that different alters are parts of the same person
can explain why we ascribe Dennett’s criteria when we do. Thus, Dennett’s
criteria do not favor one view of individuation over the other.
Technically, the problem with this argument is that it confuses different
contrast classes. Dennett’s criteria are intended to distinguish a person from
a non-person, such as an object, plant, or lower animal. Dennett never
intended his criteria to be used to distinguish one person from another to
individuate persons. Moreover, Dennett’s criteria are ill-suited for
individuation. Suppose all we know is that Mark Twain is a rational, self-
conscious language user and a moral object and subject; and so is Samuel
Clemens. This tells us that Mark is a person and that Sam is a person, but it
does not tell us whether Mark and Sam are the same person. “Mark” might
be a nom-de-plum for “Sam,” in which case Mark is the same person as
Sam. This person might go by the name “Mark” in some contexts at some
times, such as when writing, but by the name “Sam” in other contexts at
other times, such as when dining. Similarly, if “Robin” and “Jennifer” are
simply different names for different parts of a single person’s life, then each
can fully meet Dennett’s criteria but still be the same person. Thus, even if
284 Southern California Interdisciplinary Law Journal [Vol. 10:2

Dennett’s tests show that each alter is or is part of a person, as opposed to a

plant, they cannot show that these alters are different persons, as opposed to
parts of the same person.
To argue for or against personal identity, a different kind of criterion is
necessary. In particular, one needs a necessary condition of personal
identity in order to argue that two alters are or are parts of different persons.
A necessary condition of personal identity is simply some relation between
X and Y such that, whenever X and Y do not stand in that relation, X is not
the same person as Y. In contrast, a sufficient condition of personal identity
is a relation between X and Y such that, whenever X and Y stand in that
relation, X is the same person as Y.
We propose a sufficient condition of personal identity to show that
different alters in cases of MPD are or are parts of the same person. This
strategy will not settle the philosophical issues surrounding personal
identity or MPD. Indeed, part of our point is that one need not settle these
issues to defend criminal punishment in cases of MPD against objections
based on personal identity.
Many criteria of personal identity have been proposed. We will discuss
criteria that focus on the body, brain, memory, personality, projects,
awareness, and control. These criteria could be conjoined and disjoined.
One particular combination of criteria will be crucial to our argument.
In real life, we usually determine whether the person next to us is the
same person as the one we married by looking at the shape of her body or
at least her face. Of course, the shape of a person’s body and face can
change over the years or suddenly as a result of some injury. Large parts of
the body can be amputated or transplanted. The possibility of such changes
in one’s body show that complete bodily identity is not necessary for
personal identity, but identity between bodies as wholes, which does not
require identity of all body parts, might still be necessary. Complete bodily
identity and even a continuous chain of stages linking the body of the
person we married to the person next to us still seems sufficient for
personal identity in the real world.
Science fiction raises doubts. Since John Locke,12 the bodily criterion
of personal identity has been rejected by most philosophers on the basis of
examples in which minds exchange bodies. Suppose a Prince and a Cobbler
go to sleep in the same castle. In the morning, the Prince’s mind is in the
Cobbler’s body, and the Cobbler’s mind is in the Prince’s body. Locke and
his followers claim that the person who was the Prince is now identical
with the Prince’s mind in the Cobbler’s body, not the Cobbler’s mind in the
Prince’s body. This example intended to demonstrate that a person’s
12
JOHN LOCKE, AN ESSAY CONCERNING HUMAN UNDERSTANDING 457 (1959).
2001] Criminal Law and Multiple Personality Disorder 285

identity depends on the mind rather than the body. Saks uses Locke’s
example in this way.
Such examples seem more realistic when one considers the possibility
of a brain transplant. In Shoemaker’s example,13 while performing two
brain transplants at once, a surgeon inadvertently puts Brown’s brain in
Robinson’s body, thereby producing a combination called Brownson, which
has all of Brown’s memories and personality traits. The surgeon also puts
Robinson’s brain in Brown’s body, but that combination dies. Shoemaker
and his followers claim that Brown is the same person as Brownson. Again,
this example is intended to demonstrate that personal identity follows the
mind rather than the body.
It is crucial, however, that Brownson has Brown’s brain, even though
Brownson no longer has Brown’s body. As Bernard Gert argues,14 if
Robinson were hypnotized in such a way that Robinson’s body assumed all
of Brown’s memories, personality traits, and other mental features, then
nobody would say that the result was identical to Brown, especially if
Brown still existed with all of the same memories, personality traits, and
other mental features. Thus, brain identity still seems necessary for
personal identity.
It also seems sufficient. This point is clearest when brain transfer
follows hypnosis:
[P]rior to brain transfer between Brown and Robinson, Brown had been
hypnotized and told that he was Robinson. He then makes memory claims
to be Robinson and assumes the personality and character of Robinson.
Thus, prior to the brain transfer, Brown has the psychological features of
Robinson. If, while in this state, Brown’s brain is put in Robinson’s body,
we would have a person who would have Robinson’s body, Brown’s brain,
and Robinson’s psychological
15
features. Yet I should want to say that this
person is really Brown.
This person would not be Brown if any mental features were necessary
for personal identity, since the body with Brown’s brain has none of
Brown’s mental features. Thus, if the resulting person really is Brown, then
mental features are not necessary, and brain identity is sufficient for
personal identity.
Why is it sufficient? The answer appears to be that the brain is the part
of the body associated with psychological features. (It would be more
precise to refer to only part of the brain, but we will speak of the brain as a
whole for simplicity.) If the heart were the seat of the mind, then heart
13
SYDNEY SHOEMAKER, SELF-KNOWLEDGE AND SELF-IDENTITY 23 (1963).
14
Bernard Gert, Personal Identity and the Body, X DIALOGUE 458, 472 (1971). For a more recent
defense of a bodily criterion, see ERIC OLSON, THE HUMAN ANIMAL; PERSONAL IDENTITY WITHOUT
PSYCHOLOGY (1997).
15
Gert, supra note 14, at 472.
286 Southern California Interdisciplinary Law Journal [Vol. 10:2

identity might determine personal identity. What determines personal

identity is the identity of whatever organ, or set of organs, or part of an
organ, is associated with mental properties. Since this organ in humans is
the brain, we will talk simply about brain identity.
Admittedly, if the brain is frozen until it ceases to function, then there
is no person at all, so there cannot be the same person. Nonetheless, if the
same brain is functioning, then it is the same person, even if the brain is not
functioning in the same way. Since our brains function very differently
when we are asleep or drunk or hypnotized or suffering from Alzheimer’s,
yet we are still the same persons, personal identity cannot require the same
brain function, even if it does require some brain function.
What about Locke’s case? Now that we realize that it matters how the
cobbler’s body came by the prince’s mind, we need to ask what happened
overnight. If the cobbler was hypnotized, and so was the prince, then we
would not say that the prince is now the cobbler with the prince’s mind. If
an evil surgeon switched the brains of the prince and the cobbler, then we
would say that the cobbler with the prince’s mind was the prince, but that
would not affect the brain criterion. The only case that would affect the
brain criterion would be where the mind switched bodies without the brain
and the personal identity followed the mind. But according to many views
of mind, including all varieties of physicalism, this is impossible. Locke’s
example is possible only if one thinks, as Locke did, that the mind is a
separate entity, independent of the body and brain. Since such views of the
mind are questionable, so is Locke’s argument against the brain criterion.
This criterion is easy to apply to cases of MPD. Different alters in a
case of MPD inhabit the same body completely. That body has a single
brain. It is not even split, as in split-brain patients. Some evidence suggests
that different alters use different parts of the brain or that the brain displays
different patterns of activity when different alters are in control. There are
also some differences, however, between the patterns of activity when a
single person is in different moods or stages of life, or when asleep or
drunk. These normal differences might not be as extreme as in cases of
MPD, but they still show that, even if the brain functions differently when
different alters are in control, this fact alone cannot undermine the
conclusion that these alters must be parts of the same person, as long as
brain identity is sufficient for personal identity.
Many philosophers reject the brain criterion on the basis of examples
like Locke’s, and focus instead on alternative criteria, such as memory. A
memory criterion has some basis in everyday life. Even if you cannot
recognize your old friend, Joe, because he was burned beyond recognition,
you might still be able to identify him by recalling certain memories.
Admittedly, I might be able to recall that Joe jumped off a diving board
2001] Criminal Law and Multiple Personality Disorder 287

with you on September 17 twenty years ago, even if I am not Joe. What I
cannot do, unless I am Joe, is remember having the experience of jumping
off the diving board with you that time. Similarly, only you can remember
having the experience of jumping off the board with Joe. We will call such
memories of having experiences “experiential memories.”
Experiential memories can fade just like other memories. I might not
be able to remember jumping off the board. A person can suffer complete
amnesia and lose all previous experiential memories. This would not show
that he was not the same person as he was before the amnesia. He would
still have the same parents, children, social security number, and so on,
even if his character or personality changed. Consequently, experiential
memories are not necessary for personal identity, and no lack of
experiential memory can be used to argue against personal identity.
This point is important because the experiential memories of people
with MPD are radically disconnected. If a connected chain of experiential
memories were necessary for personal identity, then alters in a case of
MPD would be different people, as would people with amnesia,
dissociative fugue, and blackouts during intoxication or after hypnosis.
Since this conclusion is implausible, disconnection among experiential
memories cannot be used to show lack of personal identity in cases of MPD
or anywhere else.
Experiential memories still might provide a sufficient condition of
personal identity. Suppose that on Wednesday, Ron remembers having his
experience of eating breakfast on Monday, although he does not remember
any experience on Tuesday. On Thursday he also remembers having the
same experience on Monday, although he does not remember having any
experience on Tuesday or Wednesday. Despite the disconnections among
his memories, the fact that his experiential memories on Wednesday and
Thursday are memories of having the same experience on Monday is
enough to show that the person who had that experiential memory on
Wednesday is the same as the person who had the experiential memory on
Thursday. More generally, when X and Y have memories of having at least
one common experience, then their memories are said to converge. A single
pair of convergent experiential memories is sufficient for personal identity.
This criterion can be extended if identity is transitive, so that, if X is
identical with Y, and Y is identical with Z, then X is identical with Z.
Suppose that each day Ron remembers the preceding day but nothing
before. On Friday, Ron remembers eating breakfast on Thursday, but not on
Wednesday, on Thursday, Ron remembers eating breakfast on Wednesday,
but not on Tuesday, and so on. Ron’s experiential memory chain shows that
Ron on Friday is the same person as Ron on Thursday, Ron on Thursday is
the same person as Ron on Wednesday, and so on. By transitivity, it follows
288 Southern California Interdisciplinary Law Journal [Vol. 10:2

that Ron on Friday is the same person as Ron on Wednesday, and so on.
These identities hold even if no experiential memory is shared on Friday
and Tuesday. More generally, we can say that two experiential memory
chains converge when each chain contains one of a pair of convergent
experiential memories. Such convergence between experiential memory
chains is sufficient for personal identity, even between times when there are
no current convergent memories.
It is not important whether my experiential memory is distorted in
some ways, but it is not enough to have a similar experiential memory of
eating toast for breakfast on Monday at 8:00 a.m., since different people
can eat the same thing at the same time and later remember having a similar
experience of eating toast for breakfast on Monday at 8:00 a.m. On this
theory, different people cannot remember, or have memory chains that
converge on a memory of, having exactly the same experience of eating
breakfast on Monday. Thus, this test of personal identity rests on some test
of experience identity.
This reliance creates problems when we distinguish numerical identity
from exact phenomenological similarity. Two experiences are exactly
similar phenomenologically when they do not differ in any detail that is
observable or introspectable by the person having the experiences. Such
experiences still might not be numerically identical. Possibly through
hypnosis, Ron can have an experience of eating breakfast on Monday that
is exactly like Don’s in all detectable respects. Each sees the same people,
dishes, and decorations from the same angles, so each of their heads would
seem to have been in the same place at the same time looking in the same
direction. Each also experiences the same flavors, smells, sounds, and
feelings in exactly the same order. Later Ron and Don can have experiential
memories that are exactly similar phenomenologically, and they can
reliably describe them to me. Nonetheless, if Ron and Don are now
standing next to each other with different brains and personalities, then they
cannot be the same person. Thus, exact phenomenological similarity is not
enough for personal identity. What would be sufficient is that the
remembered experiences be numerically identical, which seems to require
that the same person had them. In this case, we need to know whether Ron
and Don were the same person while they were having the experience
before we can know whether they remember having the same experience
numerically and before we can use memories to determine personal identity
at a later time.
This problem of circularity has led many philosophers to abandon
memory as a criterion of personal identity. It still suffices for personal
identity that a chain of experiential memories converges on the same
experience numerically. This phenomenon, however, cannot provide a
2001] Criminal Law and Multiple Personality Disorder 289

general criterion of personal identity because it cannot be applied without

making assumptions about personal identity at the time of the remembered
experience.
Luckily, these assumptions seem plausible in cases of MPD. Suppose
Mel is born in 1990, goes to day care for the first time in 1994, and
subsequently develops MPD. If each of Mel’s alters has an experiential
memory of something that happened to him during the first day of day care,
or of any other experience before developing MPD, there should be no
doubt that these experiential memories are memories of the same
experience numerically. There was clearly only one person at the time of
the remembered experiences. Since numerical identity of persons and
experiences is not in doubt at that earlier time, the later memories seem
sufficient for personal identity at the later time, when personal identity is at
issue.
Moreover, even if such memories are rejected as a criterion of personal
identity, they can still be used as evidence. Only in science fiction do
different people have exactly similar phenomenologically experiential
memories. Consequently, the fact that Mel’s alters share some exactly
similar phenomenologically experiential memories in 1994 is at least very
strong evidence that these alters are parts of the same person.
Admittedly, Mel is fictional. To apply the memory test in real cases is
harder. In the Grimsley case, for example, it is not obvious whether Robin
and Jennifer remember having any common experience. Fortunately, no
common experiential memory is necessary. It would be sufficient if their
chains of experiential memories converge to include some memory of
having the same experience. Unfortunately, this convergence is difficult to
prove. We cannot rely on current reports, even if sincere, since the shared
experiential memory in the chain need not be accessible at the current time.
Thus, personal identity will be difficult to prove from memories on the
record.
Nonetheless, it is likely that Robin’s chain of experiential memories
and Jennifer’s chain of experiential memories converge on at least one
shared memory of the same experience. Each remembers being told about
the lump in her breast. In addition, each is probably linked to many
experiential memories, possibly experiences in childhood before the onset
of MPD, when there was definitely only one person. Most likely, each alter
can remember going to school or to a friend’s house, or having been scared
when her parents fought or angry when a sibling hurt her, or something
else. Although, it is possible that no such convergence ever existed, this
seems unlikely.
290 Southern California Interdisciplinary Law Journal [Vol. 10:2

Moreover, even if none of their current memories are links in

convergent chains, it is even more likely that that Robin and Jennifer would
have such memories after receiving therapy. If they would then be the same
person, and if each is the same person then as they are now, then they must
be the same person now. It seems very unlikely that they are not the same
person according to the memory criterion, even if their identity is difficult
to prove.
Similar analysis applies to every actual case of MPD. It is possible that,
when MPD arises, a new person appears with no memories of the past and
no actual or potential convergence with the other person who shares the
same body. There is little or no evidence for this phenomenon in actual
cases and the opposite seems much more plausible, at least in the cases
above.
If no converging chain exists, then we cannot use this sufficient
condition of personal identity to argue that the alters are the same person.
This result would not imply that the alters are different people, because
convergent memory chains are not necessary for personal identity, as we
argued above. In any case, there is reason to believe that such convergent
memory chains exist in many cases of MPD, which provides reason to
believe that these alters are parts of the same person.
Doubts remain about whether either brain identity or a convergent
experiential memory chain is sufficient for personal identity. One might
still wonder whether a convergent memory chain is sufficient when the
memories in the chain occur in or are associated with separate bodies. One
also might wonder whether complete brain identity or even whole body
identity is sufficient when memories are not linked, as in Locke’s prince
and cobbler example.
Neither criterion alone need be defended by opponents of the multiple
persons view of MPD. Even if neither brain nor memory alone suffices, it is
clearly sufficient for personal identity. For example, a person named “Liz”
and a person named “Beth” may share complete bodily identity, including
complete brain identity, in conjunction with a multiple convergent
experiential memory chain, and no one else, now or ever, will have exactly
phenomenologically similar experiential memories. When all of these
factors come together, there is no reason to doubt that Liz and Beth are the
same person.
These factors come together in cases of MPD. In the Grimsley case,
Robin and Jennifer share the same body, including the same brain. It is very
likely that their experiential memory chains converge at many points. No
one else has any experiential memories that are phenomenologically
exactly similar to theirs. This situation appears sufficient to establish their
2001] Criminal Law and Multiple Personality Disorder 291

personal identity. At the very least, it is strong evidence that Robin and
Jennifer are parts of the same person. The same argument could be made in
all other real cases of MPD, refuting any multiple persons view of MPD.
Defenders of a multiple persons view respond that something more is
necessary to establish personal identity. If so, the conjunctive criterion is
not sufficient. If the added necessary condition were not met in cases of
MPD, then alters in cases of MPD would not be parts of the same person.
One condition that is often upheld as necessary is continuity or
similarity of personality or character. In this context, one’s personality is
the set of values or dispositions that constitutes or explains why one acts,
feels, and thinks in the general ways that one does. According to this view,
X is the same person as Y only when X’s personality is sufficiently similar
to that of Y. This theory inspires much common conversation, such as the
comment to one’s degenerate spouse, “You are not the person I married,”
meaning that the spouse’s personality has changed dramatically.
This criterion will not function in law. Suppose Joe is a heroine addict
who undergoes a religious conversion to fundamentalism, changing his
character or personality changes almost completely. We would and should
allow him to inherit what his grandfather left to him in a valid will. We
would and should find him guilty for crimes that he committed before his
conversion, even if we reduced or commuted his sentence. The same
applies to people with manic depression or bipolar disorder, who change
dramatically, unprompted by external factors. If a continuous personality
were required for personal identity, and if personal identity were required
for punishment, then one could escape punishment by changing one’s
personality after committing a crime. Criminals who plan such an escape
should not be excused. Moreover, one main purpose of criminal law is
deterrence. I can fear future punishment even if I know that I will have a
different personality by the time I am punished. Personality can vary in
degrees, but criminal law needs an on-off criterion as a necessary condition
of guilt, as courts cannot declare anyone partly guilty. For all of these
reasons, continuity of personality does not seem necessary for the kind of
personal identity that is necessary for criminal responsibility.
The vague idea of personality could be made more precise if one’s
personality is constituted or determined by one’s projects and how they
should fit together. That idea might lead to a test like that suggested by
Carol Rovane:
an individual person exists when there is a set of intentional episodes such
that: (1) these intentional episodes stand in suitable rational relations so as
to afford the possibility of carrying out sustained coordinated activities;
(2) the set includes commitment to particular unifying projects whose
execution involves the very sorts of sustained coordinated activities made
292 Southern California Interdisciplinary Law Journal [Vol. 10:2

possible by (1); and (3) the commitment to carrying out these unifying
projects brings16 in train a commitment to achieving overall rational unity
within the set.
This analysis implies that an alter is an individual person if and only if
its intentional episodes meet conditions (1)–(3). It also suggests that two
alters are not parts of an individual person if and only if either suitable
relations do not hold between the intentional episodes of the alters or they
do not share any unifying project or they are not committed to achieving
rational unity between their projects.
This test does not seem sufficient to establish personal identity. My
wife and I execute sustained coordinated activities that make it possible for
us to execute our shared projects of raising our children and maintaining
our personal relationship. We are committed to achieving overall rational
unity within and between our projects. Nonetheless, my wife and I are still
separate persons. Rovane countenances group persons, but my wife and I
are not parts of a group person, at least in any way relevant to criminal law.
Moreover, the part of Norm that exists on Monday and the part of Norm
that exists on Tuesday might each meet Rovane’s three conditions if Norm
needs both days to complete a certain project. However, neither the part of
Norm that exists on Monday nor the conjunctive part of Norm that exists
through Monday and Tuesday is an individual person who is distinct from
Norm. If Rovane’s conditions were sufficient, they would yield so many
persons that we wouldn’t know which one to hold responsible.
More importantly, Rovane’s three conditions are not necessary for
personal identity over time. Even if no shared project or commitment
unifies me at age six with me now, I am still the same person now as I was
at age six. Rovane’s criteria might seem to work better within a short time
period, but consider the example of a man with extreme manic depression.
Sometimes he is suicidal, and sometimes he thinks that he is a great artist.
When he is manic, he sees no value in suicide. When he is depressed, he
sees no value in his art. In neither phase is he actually committed to rational
unity within this set of projects, nor is he able to sustain coordinated
activities through both phases. Still, neither phase by itself is a person, and
the man is the same person in both phases, because he meets the body and
memory conditions. If he commits a crime while manic, then we may
legitimately hold him responsible and punish him while he is depressed.
Thus, unifying projects cannot be necessary for any kind of personal
identity that is crucial to criminal responsibility.
Rovane might respond that such objections misinterpret commitment as
a psychological state, whereas her notion of commitment is normative.17 In
16
CAROLE ROVANE, THE BOUNDS OF AGENCY 174 (1998).
17
We are indebted to Isaac Levi for pressing this response.
2001] Criminal Law and Multiple Personality Disorder 293

the normative sense, intentional episodes are committed to overall rational

unity if and only if they are subject to criticism as irrational to the extent
that they fail to achieve overall rational unity among their projects,
regardless of whether they actually accept this commitment. Manic-
depressive moods are not distinct persons because they need to unify
projects in different moods in order to be rational, but alters in MPD might
still be distinct persons if they need not unify their projects to be rational.
However, this move to normative commitments cannot undermine our
claim that alters in a case of MPD are parts of the same person. Alters in a
case of MPD are subject to criticism for lacking rational unity. If they were
not, then it would be difficult to understand why they need to be cured or
may be required to undergo treatment. Perhaps unification should not
always be sought, as Rovane argues. We do not always hold alters to their
commitments by criticizing them for lacking unity. Nonetheless, in the
above cases and others that involve crimes, it seems irrational for one alter
to interfere with the projects of another alter in some situations where such
interference need not be irrational for a distinct person. The alters still share
the same body and brain, as well as convergent memories. Consequently,
alters appear to be parts of the same person even on this normative
interpretation of Rovane’s view.
A lack of awareness between phases might distinguish MPD from
manic depression. Alters might be different people if X is the same person
as Y only when X is aware of what Y does. Awareness is relevant to
criminal law insofar as X cannot control what Y does or prevent Y from
committing a crime if X is not aware that Y is committing a crime. One
could also address the issue of control by holding that X is the same person
as Y only when X is able to control what Y does.
When we apply such tests to our cases, the alters become different
persons. In the Grimsley case, Jennifer is not always aware of what Robin
does, and vice-versa. Moreover, Jennifer cannot stop Robin from doing
certain things, even when she is aware of what Robin is doing. In the
Denny-Shaffer case, Bridget is not aware of or able to stop the other alter’s
crime. These different alters are different persons if either awareness or
control is necessary for personal identity.
Alters are sometimes reported to have a kind of coconsciousness. One
alter might be aware of the other’s thoughts even if nothing is said out loud.
Awareness, however, is not essential to or happening at the time of the
crimes in our cases. Such coconsciousness also does not enable one alter to
control actions by another. Each alter lacks the kind of awareness and
control that might seem necessary for criminal responsibility.
Neither awareness nor control, however, is necessary for personal
identity over time. I cannot control what I did yesterday, since it is in the
294 Southern California Interdisciplinary Law Journal [Vol. 10:2

past. If I do not now remember doing it, then I might not be aware of it.
This does not mean that I am not the person who did it, or that I cannot be
punished for doing it. The same theory applies to control over the future.
Even if I am not able on Monday to control what I eat for dinner on
Tuesday, and even if I am not aware of what I will eat, much less aware of
eating it, that I am still the same person on both days. Thus, personal
identity can hold over time without neither awareness nor control.
Alters in a case of MPD are analogous in relevant respects to different
parts of a single person’s consciousness at different times. If each alter is
aware of and can control what it does at a certain time, but is unaware of
and cannot control what the other alter does at a different time, then the
alters are related much like temporal parts of a single person when one
temporal part is aware of and can control what it does at a certain time, but
is unaware of and cannot control what another temporal part does at a
different time. Thus, if awareness and control are not necessary for personal
identity over time, then awareness and control are also not necessary for
personal identity between alters.
One might respond that this analogy is irrelevant, because it is
important that alters can exist consciously at the same time in some cases.
It is not clear, however, why this fact destroys the analogy to temporal parts
of a single person. It is true that we do not hold individuals responsible for
an action that they cannot control at the time, but alters who commits
crimes in a case of MPD are often able to control what they do at the time.
In the Denny-Shaffer case, the alter who took the baby could have stopped
herself from taking the baby. The other alter, Bridget, who became
conscious later, could not then change the past, but none of us can change
what our past selves did. If lack of control over past selves removed
criminal responsibility, then all criminals would go free. Thus, lack of
control over a past alter also should not remove criminal responsibility.
Opponents might still object that one alter should not be held
responsible for what another alter did when the former could not control
the latter, any more than one person should be held responsible for what
another person does when the former cannot control the latter. But this
response blatantly assumes that different alters are or are like different
persons. That begs the question addressed herein, which is precisely
whether alters are or are like different persons. It makes sense to deny such
criminal responsibility in cases of MPD if one assumes that different alters
are different persons, but we need an independent test to show that they in
fact are different persons. No criterion of awareness and control can show
the lack of personal identity among alters, because that criterion itself
remains suspect at best.
2001] Criminal Law and Multiple Personality Disorder 295

One final position is skepticism. Under the influence of powerful

arguments by Hume18 and Parfit19, many contemporary philosophers believe
that the concept of absolute personal identity is incoherent, so we should
not ask whether X and Y are or are not the same person. We will not assess
this claim. If the notion of absolute personal identity is as incoherent as
skeptics claim, then it cannot be used to argue against punishing people
with MPD when one alter commits a crime. Incoherence in a concept
makes it useless for arguments and shows that it is not necessary for
criminal responsibility.
Moreover, even if no natural bundles of states make a person, we still
might need to bundle some alters or other mental states together artificially
for our purposes. As skeptics admit, the law constructs artificial categories
in many areas. If skeptics are correct, criminal law’s concept of a person
would be an additional category and one that is unavoidably artificial. Since
in a case of MPD we cannot punish or even involuntarily commit one alter
without the others, the bundles to be used in criminal law seem obvious.
For practical reasons, we must treat all alters that share the same body as if
they are parts or aspects of the same person. This treatment would be
objectionable if the alters really were different people, but no objection on
that basis could be raised if absolute personal identity is incoherent. Thus,
skepticism about personal identity cannot stand in the way of punishing a
person with MPD. It might even support it.

V. CONCLUSION
We have argued that alter personalities are parts of the same person in
any way that is required for criminal responsibility. Thus, we believe there
is reason to reject Saks’ argument for nonresponsibility, insofar as it is
based on assigning the status of personhood to alter personalities. Our
rejection of the personhood of alters, however, may be relevant to other of
Saks’ arguments for nonresponsibility, even though these arguments do not
depend on the personhood of alters.
If alter personalities are not persons, then the alternative left to the
criminal law is to view alter personalities as mental states. This way of
viewing alter personalities is consistent with traditional concepts of insanity
and involuntariness, which look to a person’s mental state at the time of the
crime in order to determine whether that person should be held criminally
accountable. Thus, assessing the responsibility of an individual with MPD
becomes very similar to assessing the responsibility of any other defendant.

18
DAVID HUME, A TREATISE OF HUMAN NATURE bk. I, pt. IV, sec. VI (L.A. Selby-Bigge ed.
1973).
19
DEREK PARFIT, REASONS AND PERSONS (1984).
296 Southern California Interdisciplinary Law Journal [Vol. 10:2

The court attempts to determine whether the individual met the relevant
standard at the time the crime took place. The Grimsley court came to
precisely this conclusion.
Saks argues that the dividedness of individuals with MPD will often
serve to vitiate responsibility. Yet, if the personhood question is foreclosed,
then the question is not how divided the person was, but whether, at the
time of the crime, the person knew, appreciated, understood, could control,
or whatever the jurisdiction calls for. This view of assessing responsibility
seems consistent with philosophy, psychology, and precedent in criminal
law. Given the complexity of MPD, however, our position will hardly
foreclose the debate.