You are on page 1of 16

ZACHAR / PSYCHIATRIC DISORDERS ARE NOT NATURAL KINDS ■ 167

Psychiatric
Disorders Are Not
Natural Kinds
Peter Zachar

ABSTRACT: I critique the essentialistic notion that psy- figure in scientific laws and have defining condi-
chiatric disorders should be conceptualized as natural tions”. Defining conditions refer to necessary
kinds, that is, exhaustively defined with reference to and sufficient properties that are inherent to the
inherent properties. Biomedical model thinkers believe thing in question. For example, any element that
that psychiatric natural kinds can best be isolated by
studying underlying biopathological processes, while
has an atomic number of 79 is gold. Having
research-oriented clinical psychologists think they can seventy-nine protons is the essence of what it
be identified statistically. Both groups assume that if a means to be gold, in all possible worlds. Any
category can’t be conceptualized as a natural kind, it is object that looks like gold but is not made out of
an arbitrary category. I argue that conceptualizing psy- atoms having seventy-nine protons is not gold.
chiatric disorders as bounded entities in nature is in- Thinkers who believe in the widespread exist-
consistent both with medicine’s understanding of dis- ence of natural kinds are called essentialists.
ease and evolutionary biology’s understanding of
Although mental health professionals do not
species. In contrast to natural kinds, I introduce the
concept of practical kinds, which are stable patterns use the term natural kind, they have used related
that can be identified with varying levels of reliability concepts to evaluate classification systems. As
and validity. I claim that thinking anti-essentialistically framed in psychiatry and psychology, the goal of
and conceptualizing psychiatric disorders as practical any nosological system is to carve nature at her
kinds is more consistent with a scientific view of the joints. According to Kendell (1975), “[i]n terms
world. of the familiar aphorism that classification is the
KEYWORDS: categorization, diagnosis, DSM, essential- art of carving nature at the joints, it should
ism, evolution, internalism-externalism, natural kinds, indeed imply that there is a joint there, that one
nosology, prototypes, psychiatry is not sawing through bone” (65).
Related to the concept of natural kinds, Blash-
field (1986) applies the traditional philosophical

A
NATURAL KIND IS an entity that is regular distinction between intensional and extensional
(nonrandom) and internally consistent definitions to the analysis of psychiatric catego-
from one instance to the next. Elements ries. An intensional definition of depression would
such as carbon, gold, or a species of animal such be a list of necessary and sufficient conditions
as tigers are common examples. Bechtel (1988, that define the inherent meaning of depression.
57) defines natural kinds as “sets of objects which The extensional definition of depression would

© 2001 by The Johns Hopkins University Press


168 ■ PPP / VOL. 7, NO. 3 / SEPTEMBER 2000

be the set of all people who are depressed. This natural kind is a pure kind, and once you have
model is essentialistic. There is a specific kind of defined its essence, errors of identification are
thing with an inherent meaning (intension), and eliminated.
all members of that set of things can be listed From an essentialistic perspective, any catego-
(extension). ry that cannot be defined with respect to fixed
In this article, I argue that it is a mistake to internal properties is an artificial (or arbitrary)
think of psychiatric syndromes as natural kinds, category. In psychiatry, Thomas Szasz (1961) is
meaning bounded categories that have necessary famous for arguing that schizophrenia is an arti-
and sufficient internal conditions for their diag- ficial kind. Artificial kinds are supposedly pseu-
nosis. This is important because thinking about do-kinds; that is, they don’t really exist. What
something as a natural kind suggests that there is limited reliability they have capitalizes on chance.
a God’s-eye view of that thing, a single accurate Some psychologists, especially those who favor
description of what it really is—independent of dimensional models such as Robert McCrae
any particular way that we may conceptualize it. (1994), also claim that the personality disorder
Because phenomena such as diseases and spe- categories in the American diagnostic system are
cies (as currently understood) are not natural arbitrary. Because both schizophrenia and the
kinds, mental health professionals should not personality disorders are not natural kinds in the
think of psychiatric disorders as natural kinds way that Alzheimer’s disease is supposedly a nat-
either. This holds whether they are using the ural kind, an assumption is made that they must
medical model favored by physicians or the psy- be artificial kinds. Both Szasz and McCrae fail to
chometric model favored by scientific psycholo- consider the continuum of practical kinds.
gists. Like species and diseases, psychiatric syn- Gorenstein’s (1992) description of the kind
dromes are best considered as belonging on the drug is a good example of a practical kind. Ac-
continuum of practical kinds. Thinkers who pre- cording to Gorenstein, drug is a superordinate
fer to think in terms of practical kinds are called category that describes the role of a diverse set of
pragmatists. Pragmatists are committed anti- es- chemical compounds used in medical practice.
sentialists. Drugs include “throat lozenges, cholesterol re-
Thinking of psychiatric disorders as practical ducers, nasal sprays, muscle relaxants, antibiot-
kinds makes it possible to ask whether someone ics and diaper rash relievers” (15). Many differ-
has a generalized anxiety disorder or an anxious ent kinds of compounds can be drugs. Being a
personality disorder without believing that this is drug is not an inherent property of any chemical;
a fixed either/or question. There may be sound it is a relational property. Gorenstein thinks that
reasons for preferring one category over the oth- mental illness is a practical kind of this type.
er in terms of the consequences that the labels Practical kinds are fuzzier than natural kinds,
have, but choosing between these two categories but they are not arbitrary. Psychometrically de-
is not a question of diagnosing the “real” disor- fined, classification of practical kinds requires
der. balancing criteria that change their values in
different contexts. As a result, practical kinds do
Kinds of Kinds not have perfect reliability. They can be thought
of as existing on a continuum, with some of them
Psychometrically defined, since natural kinds having higher reliability than others. To illus-
have fixed internal properties that make them be trate, deciding whether or not a particular in-
what they are, they can potentially be identified stance of behavior is altruistic requires consider-
with perfect reliability. Natural kinds can be iden- ing several factors, and there are no fixed rules
tified as the same kind of thing every time. To telling us which factors are most important. A
illustrate, once you know what the essence of decision about whether a mother lion’s self-sacri-
gold is, you can decide whether a particular ele- fice for the sake of her cubs is altruistic cannot be
ment is or is not gold with perfect accuracy. A made with perfect reliability.
ZACHAR / PSYCHIATRIC DISORDERS ARE NOT NATURAL KINDS ■ 169

Understanding more about the role of practi- As opposed to classical categories, categories
cal kinds is important, primarily, because of a in the prototype model have “fuzzy boundaries,”
tendency among some scientifically minded psy- so it is not always clear who is and who is not a
chiatrists and psychologists to think about psy- member of the category. Some members are bet-
chiatric disorders as natural kinds. For example, ter examples of the category than others; for
thinking that schizophrenia or bipolar disorder example, a robin is more prototypical of the
can be fully understood as broken brains or think- category bird than is an ostrich, and the King’s
ing that diagnostic overlap between categories throne is more prototypical of the category chair
such as anxiety, depression, borderline personal- than is a bean bag. There are prototypical (a
ity, and histrionic personality disorder indicates robin is a bird), atypical (a whale is a mammal),
that these categories are arbitrarily defined. Those and borderline (a bookend is a piece of furniture)
who believe in natural kinds claim that the goal examples of any particular category.
of the scientist is to isolate the “real” categories. In addition, prototype categories do not have
In contrast, I argue, similar to Paul Meehl, that necessary and sufficient conditions that define
this kind of essentialistic thinking is scientifically membership. A picture of a horse without legs
malignant. would still be correctly identified as a horse by
most people. Members of a category do not need
The DSM Does Not Assume to share all properties in common; instead, they
Natural Kinds share a family resemblance. This means there
can be alternative criteria for being placed in a
The advantage of rejecting the notion of psy- category.
chiatric natural kinds is reflected in the Diagnos- For example, although an Aristotelian would
tic and Statistical Manual of Mental Disorders take “rational animal” to be a necessary and
(DSM) conceptualization of categories, which is sufficient criterion for being human, one could
based on the prototype model of categories rath- also uniquely define humans as “featherless bi-
er than the classical model of categories. The peds.” Furthermore, those who are brain dead
prototype model is an attempt to define how may not be rational, but we would still think of
human beings actually categorize objects and them as human beings. Nor do we consider a
concepts. It is based on the work of psychologist chimpanzee who can both use tools and rational-
Eleanor Rosch and her colleagues, and it is ro- ly generate sentences in sign language to be hu-
bustly anti-essentialistic (Rosch 1981; Rosch and man. There are not always clear and distinct sets
Mervis 1975). of properties that define category membership.
Classical categories are the kinds of categories As the DSM-IV states, classification is not a
an essentialist would believe in; they are natural- cookbook affair; it sometimes requires clinical
kind categories. According to Lakoff (1987), clas- judgment that takes context into account.
sical categories have distinct boundaries, so some- In the DSM, patients are diagnosed according
one either is or is not a member of the category. to how well they match the criteria set, but no
For example, Mars either is or is not a planet, or one criterion or group of criteria is necessary and
a figure can be a triangle or a square, but not sufficient. Diagnosticians call this the “polythet-
both. Classical categories also have a set of nec- ic” criterion strategy, and it is a version of the
essary and sufficient properties that define them. prototype model. Polythetic criteria sets are or-
Aristotle’s definition of human beings as “ratio- ganized so that the most prototypical criteria are
nal animals” is an example of classical categori- listed first. As Widiger and Francis (1994) note,
zation. Thus, the term rational animal expresses there are 93 different ways to meet criteria for
the essence of what it means to be human. In this being diagnosed with borderline personality dis-
view, anyone who could read the book of nature, order in the DSM-III-R and 848 different ways
God’s cookbook, could know what something to meet criteria for antisocial personality disor-
really is. der. A syndrome such as antisocial personality
170 ■ PPP / VOL. 7, NO. 3 / SEPTEMBER 2000

disorder is a family of personality types rather derline personality are best considered prototyp-
than a single discrete type. ical in structure where there are clear cases and
The manual also provides differential diag- borderline cases for each, there is still a differ-
nostic suggestions to help with the issue of fuzzy ence between schizophrenia and borderline per-
boundaries between categories. For example, sonality. Even though diagnosticians cannot pro-
someone can be depressed or schizophrenic, or vide a single set of conditions that are both
depressed and schizophrenic, or schizoaffective. necessary and sufficient for the diagnosis of
For the diagnostician to distinguish schizoaffec- schizophrenia, schizophrenia can still be distin-
tive disorder from depression with psychotic fea- guished from other disorders. Furthermore, we
tures, the affected person must have experienced can give plenty of reasons for saying that “racist
sustained hallucinations and delusions in the ab- personality disorder” is not a legitimate psychi-
sence of a mood disturbance. To distinguish atric disorder without having to think that we
schizoaffective disorder from schizophrenia and are carving nature at the joints. “Whatever goes”
depression, a mood disturbance must be present is not an option.
for a substantial portion of the active psychotic From the standpoint of pragmatism, theories
phase. The biggest fuzzy boundary categories are and models are instruments that help us navigate
called V-codes, where the category psychiatric through the world. Their validity is in their use-
disorder itself has borderline cases. V-codes refer fulness.1 Similar to natural scientists, pragmatists
to conditions such as marital discord and spiritu- are open to the possibility that a better model
al crises, problems that may deserve clinical at- can always be developed. This cautions them
tention but are not labeled as mental illnesses. from too easily believing that their categories
Scientific realists influenced by Meehl, such as directly correspond to how things really are.
Grove and Tellegen (1991), claim that prototype Models are best considered prescriptions—pos-
models confuse clinicians’ cognitive processes sible tools for understanding the world, rather
with the nature of reality. In their view, we may than descriptions—meaning authoritative state-
naturally think in terms of prototypes, but that ments about what the world is really like. This
does not mean the world is really organized into kind of pragmatic anti-essentialism is also con-
prototype categories. Unfortunately, Paul Meehl sistent with the following three propositions from
overemphasizes the project of discovering reali- Lakoff (1987):
ty, jumping from the legitimate goal of trying to
(a) There is a world external to human beings.
find out how the world is to the questionable (b) The world is somehow the cause of our knowl-
suggestion that we can find out how the world edge.
really is. As a good falsificationist, he knows that (c) Some belief systems are better than other belief
claims about finding out about The Truth are systems.
wrong. He is so good at pointing out what is
false that he sometimes talks as if robust state- Two Classification
ments that survive logical critique can be thought
Approaches, Both Arguably
of as having been confirmed—a possibility that
Popper rejects. His University of Minnesota col- Anti-Essentialistic
leagues are even looser with talk about “carving In the next two sections, I argue that the
nature at the joints.” medical model approach to identifying psychiat-
To calm the fears of my “tough-minded” read- ric categories and the psychometric approach to
ers who may be less familiar with the details of identifying psychiatric categories are both con-
pragmatism, the absence of absolute criteria does sistent with anti-essentialism.
not leave professionals with “anything goes.”
There are criteria and those criteria are a func- Approach 1: The Medical Model
tion of more than personal whims. For example, Classification has been and continues to be
even if the categories of schizophrenia and bor- one of the most important problems in psychia-
ZACHAR / PSYCHIATRIC DISORDERS ARE NOT NATURAL KINDS ■ 171

try. It involves deciding what syndromes psychi- as an underlying pathological process. Explain-
atrists should diagnose and treat. Proponents of ing schizophrenia and depression just as general
the biomedical model would define syndromes paresis was explained is an important goal for
just as other syndromes are defined in medicine. the biomedical model.
The first step involves a clinical analysis where Although the notion that syndromes concep-
various signs and symptoms are seen to co-occur tualized as diseases will help mental health pro-
in a way that suggests that more than chance is fessionals discover psychiatric natural kinds, dis-
operating. So the co-occurrence of a sore throat, eases cannot be conceptualized only as discrete
runny nose, and head and chest congestion would physical entities. Following an examination of
suggest an integrated syndrome, classified as the this issue, I will examine the concept of the spe-
“common cold.” This is called the clinical pre- cies, showing that evolutionary biologists reject
sentation of the illness. The next step is to de- the idea of absolute boundaries between species.
scribe the course of the syndrome, so we find out If diseases and species are not considered to be
the sore throat may come first and disappear, natural kinds, psychiatric disorders should not
then sinus congestion, manifested in a yellow be considered to be natural kinds either.
discharge, followed by a clear discharge as the
person becomes noninfectious. At some point in Diseases Are Not Natural Kinds
the process, chest congestion develops, and it can
linger on for weeks. Recovery is spontaneous. In According to Blashfield (1984), disease liter-
this model, syndromes have a common etiology ally once meant dis-ease, but advancing medical
and therefore a common cure. Describing the knowledge led to the discovery of conditions
physical mechanisms that produce the syndrome where people could have a disease without any
is the heart of the biomedical model. Once it is subjective discomfort, for example, high blood
clear that there are such mechanisms, syndromes pressure. The meaning of the term disease evolved
are called diseases and traditionally conceptual- and continues to do so. Some diseases are fatal,
ized as natural kinds. others are inconveniences. Some diseases repre-
If psychiatric syndromes are biological natural sent high or low ends on a normal continuum of
kinds, biological variables such as genetic codes biological processes, while others represent qual-
and drug response may help us isolate their under- itative deviations in biological processes (Guze
lying reality. Some psychiatrists and psychologists 1992).
assume that, because physical diseases are the Our inability to point to one thing and say
bedrock reality of medical science, in order to be this is the disease is summarized by Roth and
scientifically valid, depression and schizophrenia Kroll (1986):
have to be understood as physical diseases. As
Thus for example, not everyone exposed to tuberculo-
diseases, they must have underlying pathological sis develops the disease in its complete form. The state
processes. For example, in the one-time epidemic of the immune system depends on the host’s genetic
psychiatric disorder general paresis, the symp- constitution, nutritional status, viral infections that
toms or the presentation can vary from a para- may produce an immunodeficiency syndrome, previ-
noid syndrome to a depressive syndrome to a ous exposure to similar microbial pathogens, state of
grandiose syndrome, but the underlying patho- fatigue, state of anxiety, level or morale, presence of
depression, recent major life changes and other ‘psy-
logical process is the same (Blashfield 1984). The
chological’ factors. (63)
underlying pathological process is untreated syph-
ilis. The presence of the spirochete is necessary The bacillus is necessary but not sufficient for
and sufficient for a diagnosis of syphilis. tuberculosis. Infections exist in hosts, and dis-
According to Staats (1991), one characteristic ease processes result from the interaction be-
of a mature and unified science is being able to tween infection and host. They are relational
see how superficially diverse phenomena are re- rather than inherent properties. Wallace (1994)
ally manifestations of the same phenomena, such shows that anti-essentialism pertains to treat-
172 ■ PPP / VOL. 7, NO. 3 / SEPTEMBER 2000

ment as well. He notes that although infectious chiatric Association’s reclassification of homo-
diseases are the most physiologically based dis- sexuality from a pathological sexual perversion
eases in medical science: (a) the same infection in to a normal variation in sexual orientation. We
two different patients may not respond to the still think of homosexuality as a type of behavior
same antibiotic or antiviral medication, and (b) with a biological basis, but we do not think of it
different microorganismic infections may respond as a maladaptive disease. The gay-lesbian com-
to the same medication. Wallace’s observations munity has even used theories about a biological
should lead us to be skeptical of some biological basis for homosexuality to support the idea that
psychiatrists’ claims that every disorder that re- it is a normal variation. Thirty years ago, an
sponds to anti-depressant medication must be a identified biological basis would have been con-
variation of the same disorder. sidered confirmation of the real existence of a
Infectious diseases are prototype diseases. If disease. If the community is unwilling to label
someone has the TB bacillus, he or she will be homosexuality as an illness that needs to be treat-
diagnosed as having tuberculosis and treated. ed, its biological basis will not be called a disease.
The same is true for syphilis. If the spirochete is If we separate the question “What is schizo-
present, the person will be treated for syphilis phrenia?” from the question “Is schizophrenia a
before any symptoms appear. Such problems can disease?” it might be possible to define schizo-
be reliably diagnosed, and accurate diagnosis phrenia essentialistically. If natural kinds are de-
helps physicians make predictions about what is fined only with respect to their causal mecha-
going to happen to the person if they are not nisms, the essence of schizophrenia would be its
treated. Once a category is defined, understand- causal mechanisms. However, schizophrenia
ing its causal mechanisms becomes an important would still not be a classical category. Defining a
clue to deciding whether or not it is present. The disease with respect to its causal mechanisms
high reliability and predictive validity of infec- without considering that disease to be a classical
tious diseases makes them the most practical of category might be called a “soft” natural kind.
practical kinds. This raises the thorny question of what level is
Because practical kinds cannot be fully de- going to be considered the essential causal level
fined with respect to inherent properties, exter- and the problem of overdetermination (or multi-
nal criteria play a role in their definition. Goren- ple causality). Also, it may not be so easy to
stein (1992) notes that we mistakenly confuse separate the question “What is paranoid person-
the issue of the biological basis of syndromes ality disorder?” from the question “Is paranoid
such as schizophrenia and borderline personality personality disorder a disorder?”
with the question of whether they are diseases. One promising strategy for rescuing the clas-
Showing that they have a biological basis does sical category model has been to substitute the
not demonstrate that they are diseases any more concept of disease for the concept of a disorder.
than showing that extroversion has a biological Wakefield (1992a, 1992b, 1993) specifically de-
basis demonstrates that it is a disease. Like Ful- fines mental disorder as “harmful dysfunction.”
ford (1991), Gorenstein thinks the disease con- The term harm refers to the fact that the condi-
cept also involves a social evaluation of mal- tion has negative consequences for the person. It
adaptiveness, which is a different problem from involves a reduction in well being, defined by
the problem of deciding if schizophrenia exists. social values and meanings. The term dysfunc-
As Kendell (1975) notes, in practice, claiming tion refers to the fact that something has gone
that a person has a disease really means that wrong with an internal mechanism; it is not
there is something wrong that needs to be treat- operating the way it was designed to operate.
ed. Like all practical kinds, diseases cannot be The concept of dysfunction helps distinguish a
fully defined with respect to inherent properties. mental disorder from normal responses that also
An example of the role of external criteria in have negative consequences, such as grief and
identifying practical kinds is the American Psy- trauma. So disorder refers to harm to the person
ZACHAR / PSYCHIATRIC DISORDERS ARE NOT NATURAL KINDS ■ 173

because of the failure of some internal mechanism what kind of environments the organism is adapt-
to operate in the way it was designed to operate. ed for. Learning more about the internal design
Wakefield’s definition conforms to the classical of the life form will always include more specific
category model in that design failure and harm theories about what kind of environments it is
are in combination necessary and sufficient for adapted for. Natural selection acts on interac-
labeling a particular state a disorder—with design tions between an organism and the environment.
failure being the underlying pathological pro- Realizing this, Wakefield (1999) indicates that
cess. Wakefield defines dysfunction like Thomis- identical internal mechanisms may constitute a
tic philosophers defined evil, it is a privation—an design failure for one subspecies and adequate
absence of something that ought to be there. It is design for another subspecies—depending on their
not an entity. The biggest challenge in using this evolutionary history. Design failure is therefore
model is in deciding what ought to be there. not a natural kind, one defined solely with re-
I’m not convinced that Wakefield’s concept of spect to fixed inherent properties.
harmful dysfunction is being proposed as a natu- Contra Wakefield and similarly to Lilienfeld
ral kind because the harm criterion does not and Marino (1999) and Richters and Hinshaw
refer to internal or inherent properties. For Wake- (1999), I would prefer to leave open the possibil-
field, harm means maladaptive. As long as mal- ity that syndromes such as psychopathy and dys-
adaptive is part of the meaning of psychiatric thymia are currently maladaptive variants rather
disorders, defining psychiatric disorders as iden- than dysfunctions imposed on the brain. Evolu-
tical with some fixed internal state will be insuf- tionary theory readily admits the existence of
ficient. This is because rather than there being individual variations that are maladaptive but
inherent, fixed traits that define adaptativeness, consistent with an organism’s “design.”
adaptiveness is defined as whatever confers a For example, after watching enough Nature
competitive advantage. As the local ecology programs on PBS, it seems reasonable to hypoth-
changes, what counts as adaptive changes, so esize that a condition such as narcissistic person-
ality disorder could be labeled alpha male syn-
that traits adaptive in some situations can be
drome—a normal variation within the range of
maladaptive in other situations. For example,
male behavior. If we follow leading evolutionary
Wakefield (1993) notes that being in a psycho-
theorists such as Hamilton (1964), Williams
logical state of hyperalertness because you be-
(1966), Trivers (1971), Dawkins (1976), and
lieve the Mafia is trying to kill you is adaptive if
Cosmides and Tooby (1999) in holding that the
you are a government informant and the Mafia
gene is the unit of selection, narcissistic personal-
actually is trying to kill you. That same psycho-
ity disorder could be an effective strategy for
logical state is maladaptive if you are delusional,
maximizing gene replication and therefore con-
and the Mafia is not trying to kill you. These
sistent with an organism’s design. All of Wake-
kinds of states have what philosophers call nar-
field’s examples of design use an individual-cen-
row content. Their adaptiveness can’t be evalu-
tered concept of evolution. It is possible that from
ated independent of external conditions—espe-
a “gene’s-eye” point of view, Wakefield would
cially social norms and practices. have to bite the bullet and say that narcissistic
Even design failure cannot be understood only personality disorder is not really a disorder. If
with reference to internal properties. According these kinds of counterintuitive cases (false nega-
to Dennett (1987), we can’t understand an or- tives) begin to pile up, we will begin to think that
ganism’s internal design without making refer- the harmful dysfunction model needs to at least
ence to external conditions. Imagine, asks Den- be augmented with some auxiliary propositions.2
nett, finding a heretofore unknown life form that
has been put into a kind of suspended animation.
Species Are Not Natural Kinds
Our job is to study this organism’s design and
figure out how it would behave. We could not If the anti-essentialistic model of disease sounds
proceed without developing some theory about too tender-minded, it may help to know that the
174 ■ PPP / VOL. 7, NO. 3 / SEPTEMBER 2000

existence of natural kinds is also doubtful in cies contradicts the idea of a fixed inherent struc-
zoology and paleontology. For example, Hull ture that defines all members of a species. The
(1989) notes that species are statistical abstrac- facts of evolution suggest that any criteria set
tions rather than essences. will eventually become outdated. This is why
Mayr (1969) claimed that the traditional ap-
At any one time, one can rarely discover a set of
traits which is possessed by all members of a species proach of classifying what species exist, that is,
and by no members of some other species. In addi- classical taxonomy, must be augmented with an
tion, the members of successive generations of the understanding how life in all its diversity fits
same species are usually characterized by a sightly together, that is, systematics.
different set of traits. (147) Biologists do not think of the individual-spe-
Gould (1983) points out that alternative taxo- cies relation as analogous to the member-class
nomic systems provide scientists with different relation where members belong to a class be-
classifications of species. A major debate in zool- cause they share common properties; rather, they
ogy in the past thirty years has occurred between think of it as more analogous to the cell-organ-
proponents of phentic versus cladisitic classifica- ism relation, where individual cells make up a
tion. According to the numerical phenetic model, larger organism. Instead of organism, they fol-
which is based on a mathematical analysis of low Darwin in using the word population. In
outward appearances, mountain zebras, Burch- contrast to Linnaeus, Darwin defined a species in
ell’s zebras, and Grevery’s zebras are all part of terms of populations of unique individuals rath-
the species zebra. According to the cladistic or er than individuals sharing a common essence. A
phylogenetic model, where organisms are grouped population is a genetic, behavioral, and ecologi-
according to their common ancestors, mountain cal system whose members compete with each
zebras are classified as a kind of horse. Gould other and as a whole with the members of other
(1983) also notes that orangutans are cladistical- species. Rather than what an essentialist would
ly more distant from chimps and gorillas than call “imperfections,” “errors,” or “accidents,”
humans are. Another example is given by Ridley individual variation is central for understanding
(1989), who points out that lungfish are more the long-term fluid nature of species.
similar to cows than to salmon according to Mayr’s own compromise between the phentic
cladistic criteria. Neither phenetic nor cladistic and the cladistic models is called the biological
classifications are whimsical; for example, nei- species model. A species is defined as:
ther model would classify orangutans as a type (a) A reproductive community, usually mating only
of reptile. The phenetic and cladistic taxonomies within the group.
also have significant overlap, but what counts as (b) An ecological unit; individuals in the group share
a group is partly model-dependent. Zoologists an environmental niche and relate as a group to
have not achieved what could be called a fixed other species.
“God’s-eye” view of species. (c) A genetic unit; any individual only holds a part of
A compromise between the phenetic model the species’ gene pool.
and cladistic model is offered by one of the cen- Especially for the behavioral criteria (a) and
tury’s most prominent biologists, Ernst Mayr (b), species are defined in terms of relationship
(1988, 1989, 1991), who believes in the reality patterns rather than an internal essence. In con-
of species more than Darwin did but rejects the trast to the phenetic model, Mayr shows that
natural-kind view of species favored by pre-Dar- groups with very different external appearances
winian thinkers such as Linnaeus. Ruse (1988) can interbreed, and groups with similar external
and Mayr (1988) both claim that post-Darwini- appearances cannot interbreed. In contrast to the
an biology has given up on Aristotle’s idea of cladisitic model, even though birds and croco-
species as natural kinds for an understanding of diles are closer together with respect to shared
species defined by polythetic criteria. Mayr notes ancestors than crocodiles are to other reptiles,
that the very possibility of the evolution of spe- ecological and behavioral variables make croco-
ZACHAR / PSYCHIATRIC DISORDERS ARE NOT NATURAL KINDS ■ 175

diles more like reptiles than like birds. Both mod- anxiety disorder or to reduce anxiety and depres-
els provide evidence that we can use in a compre- sion to negative emotionality as some people in
hensive conceptualization. psychiatry have suggested. Internal biological
As a confirmed anti-essentialist who explicitly properties, whatever they may be, are important,
rejects nominalism, Mayr believes that species but they tend to be insufficient for understanding
are not just inventions. The distinctions accepted syndromes. Phenomenology still has an impor-
by taxonimists are far from arbitrary. For exam- tant role to play in categorization.
ple, there are nonarbitrary gaps between many Psychologists such as Widiger and Trull (1991)
species. There is a clear gap between primates use terms such as arbitrary to refer to psychiatric
and reptiles, whose inability to interbreed is sta- classification systems which are based on simi-
ble. Whether genetic manipulation could trans- larity of presentation. Arbitrary is a strong word,
form this “law” into an empirical generalization suggesting whims or preferences as opposed to
is an open question (and probably one that should the kind of sustained regularity one would ex-
never be answered). pect in natural kinds. Unfortunately, a pejorative
In terms of populations that are less distinct, word such as arbitrary is merely a rhetorical
however, it can be difficult to clearly see what device used to demean an opposing model.
counts as a genetic or reproductive community, Let me illustrate this by analyzing Corning’s
especially when you have incipient species— (1986) example of classification by similarity of
groups that have acquired some but not all char- presentation. It involves his six-year old son’s
acteristic of a species. With these borderline cas- rearrangement of his office. Corning drew baby-
es, a species as an absolutely fixed type evaporates sitting duty with his six-year old son on an after-
away. With a change of habitat, single groups noon that he had to chair a thesis defense. He
can divide into two or more groups (speciation), arranged it so the boy would be occupied in his
or two groups that were separate under certain office while he was at the meeting. Upon return-
conditions (for example, wolves and dogs) can ing to his office after the meeting, he found that
relate to each other as a species. As Dawkins his son had reorganized his filing system. All his
(1986) notes, the primary reason that we can brown folders were neatly stacked in a pile on
maintain a belief in discrete boundaries between the floor. All his manila folders were stacked in
species is that intermediaries tend to be extinct.3 their own pile. The documents in each file had
A few intermediaries do, however, still exist. For been removed and placed in their appropriate
example, Kendell (1975) notes the platypus is piles. Legal size yellow notepaper was in one
neither cold-blooded like a reptile nor warm- pile, legal size white notepaper was in another
blooded like a mammal, but a mixture of the pile, notebook-size yellow notepaper was in a
two. third and so on. All white paper with typing on it
The process of categorizing species can illumi- was in its own pile. When his father returned, the
nate the process of categorizing psychiatric dis- boy proudly proclaimed that he had “straight-
orders. As the phenetic-cladistic debate shows, ened the office out.” Corning suggests that his
although external appearances do not define the son’s classification was analogous to psychiatric
essence of categories, they still have an impor- classification, where a superficial analysis based
tant role to play in categorization. Appearances on outward appearances takes the place of an
are always part of the evidential basis in any analysis of categories in terms of the information
comprehensive system for defining categories. they contain.
For example, it is impractical to exclusively clas- This amusing story, however, fails to make the
sify lungfish with cows and not with salmon point that proponents of natural kind categori-
because lungfish behavior is so much more con- zation think it does, that is, that it is an example
gruent with salmon behavior. Habitat matters. of arbitrary classification. The boy’s reclassifica-
In the same way, it would be impractical to tion of Corning’s files into folder type and paper
classify borderline personality exclusively as an type was not arbitrary. It was concretely system-
176 ■ PPP / VOL. 7, NO. 3 / SEPTEMBER 2000

atic in its organization. The important issue is Schizophrenia equals a positive symptom presenta-
not what rules we use to develop categories. The tion a negative symptom presentation (Andreasen
important issue is deciding how useful the pro- and Olson 1982).
“Borderline schizophrenia” includes borderline person-
posed categories are. Organizing offices in terms
ality disorder and schizotypal personality disorder;
of files, so that information on attention deficit Schizoid personality in DSM-II equals schizoid and
disorder is in one file, information on lateraliza- avoidant personality in DSM-III (Gabbard 1994).
tion is in another file, and information on cluster
analysis is in another file is more useful to an Each taxonomy would have a different set of
academic than putting all the yellow paper in advantages and disadvantages, but syndromes
one pile. Categories are practical kinds. The cat- would not just be inventions.
egories developed by Corning’s son were imprac- Another version of the same point is made by
tical, not arbitrary. Traditional psychiatric cate- Widiger and Francis (1994). They note that one
gories are not arbitrary either. Rejecting of the most important decisions in developing
essentialism in favor of anti-essentialism does DSM diagnoses was to decide where to place cut-
not require one to adopt nominalism. Practical off points. For example, to be diagnosed as hav-
kinds are more than names, but less than inher- ing antisocial personality disorder, one has to
ent essences. meet at least five of the nine criteria for that
disorder. Widiger and Francis note that any diag-
Addendum: Maximize Taxonomic Advantages, nostic system will be used for many purposes,
Compensate for Disadvantages including “hospitalization, medication, psycho-
therapy, insurance coverage, scientific research,
If philosophers of biology are correct, medical criminal responsibility, disability, and so forth”
model theorists should be able to propose alter- (23). The optimal cut-off point would be slightly
native taxonomic systems for defining syndromes, different for each of these purposes. No diagnos-
and each system may have different sets of syn- tic system can be considered “The System.”
dromes. Classification schemes developed at dif-
ferent level of analysis, for example, (including Approach 2: Psychological
but not limited to) the genetic level, the neuro- Measurement
chemical level, the anatomical level, the phenom- One problem with clinical observation is that
enological level, or the sociocultural level may human judgment may not always be sensitive
not be perfectly isomorphic with each other. Each
enough to detect subtle patterns. For example,
taxonomy would have validity for certain pur-
AIDS as a syndrome existed in patient popula-
poses, but no one could be called the real taxon-
tions long before physicians noticed it. In psychi-
omy.
atry, ever since Kraepelin proposed the label de-
All but the most partisan thinkers agree that mentia-praecox, students of what came to be
this is true with respect to the categorical versus called schizophrenia have believed that is not a
the dimensional classification of psychiatric dis-
unitary syndrome but a group of related disor-
orders. Under certain conditions, disorders con-
ders. Unfortunately clinicians have not been able
sidered unique can be seen to be a variants of the
to intuit the different syndromes that make up
same disorder:
what we call “schizophrenia.”
Alcoholism, attention deficit hyperactivity disorder, The traditional medical model is not the only
and psychopathy equal the same disorder (Wender approach to classification. Clinical psychologists,
and Klein 1981). who are trained in scientific methodology rather
Comorbidity between personality disorders is a func-
than applied medicine, claim that co-occurrence
tion of shared pathogenic factors (McCrae 1994).
is a synonym for correlation. Correlation has an
Or a single disorder can be more usefully exact mathematical definition—it is an index of
studied by being fragmented into several differ- the proportion of total variance that is due to
ent types: scores that co-vary in a predictable manner. The
ZACHAR / PSYCHIATRIC DISORDERS ARE NOT NATURAL KINDS ■ 177

best way to decide what variables are correlated ing what strategy the investigator uses to form
is by statistical analysis. Therefore multivariate clusters. All these decisions can influence the
methods such as factor analysis and cluster anal- results. Therefore, the derived patterns are partly
ysis, which can objectively determine the pres- method-dependent.
ence or absence of patterns in the data, may be
more scientifically sound than clinical observa- Mathematically Structured Folk Constructs Are
tion. Because the observer-independent patterns Not Natural Kinds
found by statistical analysis are also by defini-
Widiger and Corbitt (1994), Widiger and Trull
tion nonrandom, they could even be called natu-
(1991), and McCrae (1994), who favor psycho-
ral kinds.
metrically discovered dimensional models, cri-
tique clinically discovered categorical models be-
Statistics Require Assumptions
cause they are “hypothetical constructs,”
Unfortunately for psychological science, sta- “arbitrary,” and “not naturally occurring cate-
tistical methods are not so artless. Skinner (1981) gories.” They contend that the DSM’s comorbid-
and Blashfield (1980) have both criticized the ity problem is unacceptable for a system that is
naive empiricism adopted by some proponents supposed to be composed of discrete entities.
of statistically guided classification. For exam- This essentialistic critique of categorical models,
ple, in Monte Carlo studies, where artificial data which conforms to the nineteenth-century view
sets are generated to test statistical assumptions, of disease, suggests that their own dimensions
even with a random pattern of data, factor ana- based on the five-factor model of personality will
lytic procedures will extract factors. They capi- somehow carve nature at her joints. Their pre-
talize on chance variance in the data to find a ferred measure of the five-factor model is called
pattern. The same can be true for cluster analy- the NEO-PI.
sis. Even if no patterns exist, statistical analysis As noted, there are good reasons for claiming
may find them. This means that there is a poten- that psychometric methods for discovering psy-
tial false-positive problem with multivariate sta- chological dimensions do not carve nature at the
tistical methods. Such methods may find pat- joints. It is true that once certain parameters are
terns that are not really there. Furthermore, if defined, relatively stable solutions emerge, but
the appropriate variables are not entered into the change the parameters and different solutions
analysis, multivariate methods can also fail to may appear. In carving, we should not find dif-
detect patterns, that is, yield false negatives. ferent joints if we switch knives. When the issue
Another problem with multivariate statistical is considered under the rubric of scientific real-
methods is that their calculations are not as ob- ism versus instrumentalism, dimensional models
jective as some of their proponents claim. As do not themselves meet the standards that their
Skinner (1981), Blashfield (1981), and Tinsley proponents use to reject categorical models.
and Tinsley (1987) note, various theoretical and Furthermore, considerable subjective judgment
mathematical decisions have to be made before is required to name factors. For example, the
analyses can be run, and these decisions can Conscientious factor on the NEO-PI could also
affect the results. These decisions include decid- be called dependable, responsible, scrupulous,
ing whether variables or persons should be cor- conformist, or, as a rebellious young client of
related; deciding whether factors should be cor- mine once stated, “ass-kisser.” These are not
related or uncorrelated (orthogonal); deciding mere synonyms. It is like describing someone as
whether all the variance to be partitioned is con- flexible versus describing them as flaccid or as
sidered common variance; deciding whether items rigid versus rigorous. Different terms have differ-
should load highly onto one and only one factor ent connotations. These are not natural kinds.
(simple structure); deciding how distances be- The biggest misinterpretation of latent mathe-
tween variables are to be computed; and decid- matical categories is to confuse a factor with a
178 ■ PPP / VOL. 7, NO. 3 / SEPTEMBER 2000

factor-derived scale. As Cattell (1978) notes, a guages adequately. For example, their exclusion
factor is a latent entity that accounts for a pro- criteria eliminated both evaluative terms such as
portion of variance in a correlation matrix. Each pretentious or charismatic, and mood states such
item in the correlation matrix usually loads onto as happy and fearful. Almagor, Tellegen, and
the factor. Psychometricians often take the items Waller (1995) claim that the initial narrowness
on a factor with the highest loadings and put of the McCrae and Costa item pool explains why
them on a single scale, but the scale is not the John (1989) could not classify traits such as
same thing as the factor. The factor usually con- independent, peculiar, and conservative on the
tains some of the variance from all the items, not Big Five dimensions.
just the items with the highest loadings. Further- Tellegen also argues that the a priori elimina-
more, the items with high loadings still contain tion of evaluative and mood terms from what is
variance that is not associated with the factor. supposed to be a comprehensive index of folk
Scales are not factors. They are constructs of personality descriptions weakens the NEO’s ap-
convenience. plication to be the framework for the DSM’s
During the development of the DSM-IV in the assessment of maladaptive personality styles. For
early 1990s, the NEO-PI dimensions were pro- example, removing evaluative terms eliminates
posed as alternatives to the personality disorder any consideration of self-esteem as a source of
categories. The NEO-PI measures five traits, Neu- individual differences. Ben-Porath and Waller
roticism, Extroversion, Openness, Conscientious- (1992) astutely claim that NEO-PI still needs to
ness, and Agreeableness. These are called the demonstrate incremental validity above and be-
“Big Five” because they have emerged in several yond standard clinical measures such as the Min-
major factor-analytic research programs over the nesota Multiphasic Personality Inventory even to
years. In their review of its conceptual beginnings, be included in the assessment of psychopatholo-
McCrae and Costa (1990) claim that the NEO-PI gy, let alone to define it.
is based on the descriptions of personality found Supporting Tellegen’s claim is the fact that the
in natural languages, which they refer to as “folk Big Five used to be the Big Three! Costa and
wisdom” (30). They also claim that, over the McCrae’s (1985) first attempt to measure the
centuries, all important traits have been encoded structure of personality was called the NEO in-
in natural languages. In their view, because fac- ventory. It had three factors, Neuroticism, Ex-
tor analysis can find latent dimensions underly- troversion, and Openness. Later, McCrae and
ing these person-in-the-street labels for traits, we Costa decided that the structure of personality
can, in principle, isolate real categories for the changed. They added two more factors, Consci-
psychology of personality. McCrae and Costa entiousness and Agreeableness. It is not unrea-
(1997) use evidence for the cross-cultural consis- sonable to expect that, if someone improves it,
tency of the NEO-PI to claim that they have the model will change again. Many clinical and
discovered the universal structure of personality. counseling psychologists have ignored these is-
Not all philosophers and psychologists would sues. They have ignored them because McCrae
agree with McCrae and Costa’s assessment. For and Costa have succeeded in making traits scien-
example, in what he calls the “fallacy of the tifically respectable again, and because the NEO-
perfect dictionary,” Alfred North Whitehead PI has had great heuristic value with respect to
(1938) criticized the pernicious idea that humans topics for theses, dissertations, and tenure-track
have “consciously entertained all the fundamen- publications. The popularity of McCrae and Cos-
tal ideas that are applicable to [their] experi- ta’s model has, unfortunately, circumvented need-
ence” and encoded them in language (173). With ed criticism.
respect to the NEO-PI, Tellegen (1993) convinc- Ironically, dimensional classification is very
ingly argues that McCrae and Costa’s exclusion similar to the anti-diagnostic model of Karl Men-
criteria led to an item pool that failed to sample ninger, who held that the discrete categories de-
the personality descriptions used in natural lan- scribed in textbooks cannot help us truly under-
ZACHAR / PSYCHIATRIC DISORDERS ARE NOT NATURAL KINDS ■ 179

stand people’s problems. Instead, he thought we chological state has some kind of biological ba-
should think in terms of scales or yardsticks. On sis. For example, the reason a rock cannot get
one end of the scale would be “maladjusted” and depressed or do calculus is that it doesn’t have a
on the other end “adjusted.” Once people get biological basis for either depression or calculus.
into a maladjusted range, the mental health pro- Any trait, such as extroversion, or a cognitive-
fessional is supposed to help them figure out emotional state, such as depression, exists be-
how to achieve a greater sense of “normality” cause of a biological predisposition. Every aspect
(Menninger, Mayman, and Pruyser 1963). This of human psychology has a biological basis.
recommendation compares favorably with Widi- Second, to say that something has a genetic
ger’s (1994) proposal that mental health profes- basis does not mean that it is a discrete entity at
sionals first assess degree of maladjustment and the level of DNA. For example, other traits, such
then determine the person’s position on the basic as traditionalism, religiosity, well-being, delin-
dimensions of personality to understand the na- quency, emotional stability, ego-strength, and time
ture of the maladjustment. watching television, have been shown to have a
Rather than using fixed states called disease genetic basis as well (Bouchard, Lykken, McGue,
entities, Menninger claimed that we should think Segal, and Tellegan 1990; Bouchard and McGue
in terms of shifting positions on various yard- 1990; Cattell, Rao, and Schuerger 1985; Pres-
sticks of personality functioning. In an interesting cott, Johnson, and McArdle 1991). No one (I
parallel with neo-Kraepelinianism, Menninger calls hope) suggests that our ancestors evolved a time-
this view neo-Jacksonianism, after J. Hughlings watching-television gene! Dispositions to react
Jackson. By this he means a focus on quantita- to possibilities presented by one’s culture may
tive (dimensional) rather than qualitative (cate- not have evolved with those possibilities in mind.
gorical) distinctions between different kinds of With respect to watching television, there is
mental illness. What modern-day dimensional probably a biological basis, a protein-synthesiz-
proponents propose to add to neo-Jacksoninan ing program creating a nervous system with a
psychiatry is a scientifically based model of per- cognitive-affective predisposition, which initiates
sonality functioning. a process that ends up in a person watching more
television if they are given the opportunity to do
Personality Traits Having a Genetic Basis Are so, but the final point in the process is not the
Not Natural Kinds biological basis. The same is true for personality
traits. They are end products and cannot be re-
Like some physicians, psychometric thinkers duced to inherent necessary and sufficient criteria.
may also be vulnerable to confusing the presence Genetics may be necessary as process initiators,
of a biological basis with something’s being a but they are not sufficient causes of most traits.
natural kind. For example, according to Lykken When we look at a trait such as extroversion
and Tellegen (1996), folk constructs such as neg- or a syndrome such as bipolar disorder, psycho-
ativism and happiness and altruism have a bio- logical, behavioral, and biological covariation
logical/genetic basis. Sandra Scarr (1987) also reflects some kind of coherent organization so
suggests that 24 to 40 percent of the variance in consistently that we can say there is something
personality stems from heredity. Based on these there, we just cannot reduce it only to biological
findings, psychologists might conclude that some covariation. There are no inherent properties that
psychological traits really exist—as natural kinds. make traits and syndromes be what they are.
First, the notion that traits that have a biolog- They are practical kinds.
ical basis are traits that really exist has to be
rejected by anyone claiming to be a materialist. Conclusion
For materialists, no cognitive or emotional states
would exist without brains. According to the There are no necessary and sufficient internal
materialist’s brain-as-substrate thesis, every psy- conditions of objects that makes them be some-
180 ■ PPP / VOL. 7, NO. 3 / SEPTEMBER 2000

thing like a chair. Chairs are not natural kinds. making an authoritative statement about what a disor-
There are many reasons for refusing to believe der really is. Furthermore, it is not unreasonable that
that syndromes, diseases, species, and personali- harmful dysfunction in practice would become a radial
category with prototype cases and borderline cases.
ty traits are natural kinds as well. This is a
3. Dawkins (1986) shows that genes don’t have
property of any categorical system that can also inherent effects either. The effect of a gene is not a
be considered as continuous or dimensional. Nei- simple property of the gene itself. It “is a property of
ther the traditional medical model and its meth- the gene in interaction with the recent history of its
ods for isolating disease nor the psychologist’s local surroundings in the embryo. . . . A gene turned on
mathematical approach to classification have suc- in the cells at the base of the spinal cord in the third
ceeded in isolating what could be called natural week of development will have a totally different effect
from the same gene turned on in the cells of the shoul-
kinds. Both have discovered stable patterns that
der in the sixteenth week of development” (296).
are more than mere inventions, but the idea of an
isolated inherent reality, definable by using only
biological experimentation and sophisticated sta- References
tistical analysis, is mistaken. We need too many Almagor, A., A. Tellegen, and N. G. Waller. 1995. The
other variables and kinds of evidence to individ- big seven model: A cross cultural replication and
uate patterns, and adopting different methods or further exploration of the basic dimensions of nat-
evidential priorities can alter the patterns we ural language trait descriptors. Journal of Person-
find. ality and Social Psychology 69:300–307.
Meehl (1986) discussed the advantages of American Psychiatric Association. 1987. Diagnostic
and statistical manual of mental disorders. 3rd ed.
thinking of diagnostic taxa as open concepts and
rev. Washington, DC: American Psychiatric Press.
considered any other strategy to be “scientifical- American Psychiatric Association. 1994. Diagnostic
ly malignant” (220). Reality is always going to and statistical manual of mental disorders. 4th ed.
be more complex than what is captured by our Washington, DC: American Psychiatric Press.
categories. No matter how specifically we define Andreasen, N. C., and S. Olsen. 1982. Negative ver-
disorders such as schizophrenia, we will always sus positive schizophrenia. Archives of General
have to admit exceptions—cases that do not fit Psychiatry 39:789–94.
the model. The more specific the criteria, the more Bechtel, W. 1988. Philosophy of mind: An overview
for cognitive science. Hillsdale, NJ: Lawrence Erl-
exceptions we can expect. We can avoid the
baum.
problem of exceptions by using broader defini- Ben-Porath, Y. F., and N. G. Waller. 1992. “Normal”
tions but that would lower reliability. Using the personality inventories in clinical assessment: Gen-
terms discussed in this article, thinking of our eral requirements and the potential for using the
categories as natural kinds, as closed absolute NEO Personality Inventory. Psychological Assess-
concepts, is unwarranted. Scientific openness to ment 4:14–19.
evidence is better supported by considering psy- Blashfield, R. K. 1980. Propositions regarding the use
of cluster analysis in clinical research. Journal of
chiatric categories to be practical and not natural
Consulting and Clinical Psychology 48:456–59.
kinds. ———. 1984. The classification of psychopathology:
Neo-Krapelinian and quantitative approaches. New
Notes York: Plenum.
1. From a pragmatic standpoint, what an essential- ———. 1986. Structural approaches to classification.
ist calls an artificial kind is just a practical kind that is In Contemporary directions in psychopathology,
not very useful. ed. T. Millon and G. Klerman, 363–80. New York:
2. As long as Wakefield is defining disorder by Guilford Press.
stipulation, I think his proposal has some merit. By Bouchard, T. J., D. T. Lykken, M. McGue et al. 1990.
stipulation I mean that he is only proposing a model Sources of human psychological differences: The
that should be useful in helping diagnosticians distin- Minnesota study of twins reared apart. Science
guish between disorders and non-disorders, rather than 250:223–28.
ZACHAR / PSYCHIATRIC DISORDERS ARE NOT NATURAL KINDS ■ 181

Bouchard, T. J., and M. McGue. 1990. Genetic and Lakoff, G. 1987. Women, fire, and dangerous things.
rearing environmental influences of adult person- Chicago: University of Chicago Press.
ality. Journal of Personality 58:263–92. Lilienfeld, S. O., and L. Marino. 1999. Essentialism
Cattell, R. B. 1978. The scientific use of factor analy- revisited: Evolutionary theory and the concept of a
sis in behavioral and life sciences. New York: Ple- mental disorder. Journal of Abnormal Psychology
num Press. 108:400–411.
Cattell, R. B., D. C. Rao, and J. M. Schuerger. 1985. Lykken, D., and A. Tellegen. 1996. Happiness is a
Heritability in the personality control system: Ego stochastic phenomena. Psychological Science
strength (C), super ego strength (G), and the self 7:186–89.
sentiment (Q3); by the MAVA model, Q-data, and Mayr, E. 1969. Principles of systematic zoology. New
maximum likelihood analysis. Social Behavior and York: McGraw-Hill.
Personality 13:33–41. ———. 1988. Toward a new philosophy of biology:
Corning, W. C. 1986. Bootstrapping toward a classifi- Observations of an evolutionist. Cambridge: Har-
cation system. In Contemporary directions in psy- vard University Press.
chopathology, ed. T. Millon and G. Klerman, 279– ———. 1989. Species concepts and their application.
303. New York: Guilford Press. In Philosophy of biology, ed. M. Ruse, 136–45.
Cosmides, L, and J. Tooby. 1999. Toward an evolu- New York: Macmillan.
tionary taxonomy of treatable conditions. Journal ———. 1991. One long argument: Charles Darwin
of Abnormal Psychology 108:453–64. and the genesis of modern evolutionary thought.
Costa, P. T., and R. R. McCrae. 1985. The NEO Cambridge: Harvard University Press.
Personality Inventory Manual. Odessa, FL: Psy- McCrae, R. R. 1994. A reformulation of axis II:
chological Assessment Resources. Personality and personality related problems. In.
Dawkins, R. 1996. The blind watchmaker. New York: Personality disorders and the five-factor model of
W. W. Norton. personality, ed. T. Costa and T. A. Widiger, 303–
Dennett, D. C. 1987. The intentional stance. Cam- 10. Washington, DC: American Psychological As-
bridge, MA: MIT Press. sociation.
Fulford, W. 1991. The concept of disease. In Psychiat- McCrae, R. R., and P. T. Costa. 1990. Personality in
adulthood. New York: Guilford Press.
ric ethics, ed. S. Bloch and P. Chodoff, 2d ed., 77–
———. 1997. Personality trait structure as a human
99. New York: Oxford University Press.
universal. American Psychologist 52:509–16.
Gabbard, G. O. 1994. Psychodynamic psychiatry in
Meehl, P. K. 1986. Diagnostic taxa as open concepts:
clinical practice. Washington, DC: American Psy-
Metatheoretical and statistical questions about reli-
chiatric Press.
ability and construct validity in the grand strategy
Gorenstein, E. 1992. The science of mental illness. of nosological revision. In Contemporary directions
San Diego: Academic Press. in psychopathology, ed. T. Millon and G. Klerman,
Gould, S. J. 1983. Hen’s teeth and horse’s toes. New 215–31. New York: Guilford Press.
York: W. W. Norton. Menninger, K. 1948. Changing concepts of disease.
Grove, W. M., and A. Tellegen. 1991. Problems in the Annals of Internal Medicine 29:118–325.
classification of personality disorders. Journal of Menninger, K., M. Mayman, and P. Pruyser. 1963. The
Personality Disorders 5:31–41. vital balance. New York: Viking Press.
Guze, S. 1992. Why psychiatry is a branch of medi- Prescott, C. A., R. C. Johnson, and J. J. McArdle.
cine. New York: Oxford University Press. 1991. Genetic contributions to television viewing.
Hamilton, W. D. 1964. The genetical evolution of Psychological Science 2: 430–31.
social behavior. I and II. Journal of Theoretical Richters, J. E., and S. P. Hinshaw. 1999. The abduction
Biology 7:1–52. of disorder in psychiatry. Journal of Abnormal Psy-
Hull, D. L. 1989. The ontological status of species as chology 108:438–45.
evolutionary units. In Philosophy of biology, ed. Ridley, M. 1989. Principles of classification. In Philos-
M. Ruse, 146–55. New York: Macmillan. ophy of biology, ed. M. Ruse, 167–79. New York:
John, O. P. 1989. Toward a taxonomy of personality Macmillan.
descriptors. In Personality psychology: Recent Rosch, E. 1981. Prototype classification and logical
trends and emerging directions, ed. D. M. Buss and classification: The two systems. In New trends in
N. Cantor, 261–71. New York: Springer. cogntive representation: Challenges to Piaget’s the-
Kendell, R. E. 1975. The role of diagnosis in psychia- ory, ed. E. Schilnick, 73–86. Hillsdale, NJ: Lawrence
try. London: Blackwell Scientific. Erlbaum.
182 ■ PPP / VOL. 7, NO. 3 / SEPTEMBER 2000

Rosch, E., and C. Mervis. 1975. Family resemblances: ———. 1992b. Disorder as harmful dysfunction: A
Studies in the internal structure of categories. Cog- conceptual critique of the DSM-III-R’s definition
nitive Psychology 7:573–605. of mental disorder. Psychological Review 99:232–47.
Roth, M., and J. Kroll. 1986. The reality of mental ———. 1993. Limits of operationalization: A critique
illness. Cambridge: Cambridge University Press. of Spitzer and Endicott’s (1978) proposed opera-
Ruse, M. 1988. Philosophy of biology today. Albany, tional criteria for a mental disorder. Journal of
NY: SUNY Press. Abnormal Psychology 102:160–72.
Scarr, S. 1987. Personality and experience: Individual ———. 1999. Evolutionary versus prototype analyses
encounters with the world. In The emergence of of the concept of disorder. Journal of Abnormal
personality, ed. J. Aronoff., A. I. Rabin, and R. A. Psychology 108:374–99.
Zucker, 49–78. New York: Springer. Wender, P. H., and D. F. Klein. (1981). Mind, mood,
Skinner, H. A. 1981. Toward the integration of classi- and medicine. New York: Farrar, Straus, Giroux.
fication theory and methods. Journal of Abnormal Whitehead, A. N. 1938. Modes of thought. New York:
Psychology 90:68–87. Macmillan.
Staats, A. 1991. Unified positivism and unification Widiger, T. A. 1994. Conceptualizing a disorder of
psychology: Fad or new field? American Psycholo- personality from the five-factor model. In Person-
gist 46:899–912. ality disorders and the five-factor model of person-
Szasz, T. 1961. The myth of mental illness. New York: ality, ed. T. Costa and T. A. Widiger, 311–17.
Harper & Row. Washington, DC: American Psychological Associa-
Tellegen, A. 1993. Folk concepts and psychological tion.
concepts of personality and personality disorder. Widiger, T. A., and E. M. Corbitt. 1994. Normal
Psychological Inquiry 4:122–30. versus abnormal personality from the perspective
Tinsley, H. E. A., and D. J. Tinsley. 1987. Uses of of the DSM. In Differentiating normal and abnor-
factor analysis in counseling psychology research. mal personality, ed. S. Stack and M. Lorr, 158–75.
Journal of Counseling Psychology 34:414–24. New York: Springer Publishing.
Trivers, R. L. 1972. The evolution of reciprocal altru- Widiger, T. A., and A. J. Francis. 1994. Toward a
ism. Quarterly Review of Biology 46:35–57. dimension model for the personality disorders. In
Wallace, E. R. 1994. Psychiatry and its nosology: A Personality disorders: And the five-factor model of
historico-philosophical view. In Philosophical per- personality, ed. T. Costa and T. A. Widiger, 19–39.
spectives on psychiatric diagnostic classification, Washington, DC: American Psychological Associa-
ed. J. Z. Sadler, O. P. Wiggins, and M. A. Schwartz, tion.
16–86. Baltimore: Johns Hopkins University Press. Widiger, T. A., and T. J. Trull. 1991. Diagnosis and
Wakefield, J. C. 1992a. The concept of mental disor- clinical assessment. Annual Review of Psychology
der: On the boundary between biological facts and 42:109–33.
social values. American Psychologist 47:373–88. Williams, G. C. 1966. Adaptation and natural selec-
tion. Princeton, NJ: Princeton University Press.

You might also like