Natural and para-natural kinds in psychiatry
Nigel Sabbarton-Leary, Lisa Bortolotti, and Matthew R. Broome
The aim of this chapter is to investigate how we should understand kinds in psychiatry. That psychiatry has kinds of some sort is uncontentious; the ever-growing list of mental disorders is testament to that. The problem with the classificatory categories set out in manuals like the Diagnostic and Statistical Manual of Mental Disorders (DSM) is to articulate precisely what mental disorders amount to. The question that arises is: Are the classificatory categories of psychiatry natural kinds?
Psychiatrists concerned with classification are interested in a number of interconnected issues: 1) whether, or not, a particular mental disorder is real or valid—that is, whether it is a bona fide piece of the furniture of reality; 2) whether the disorder has inductive potential—that is, whether the symptoms cluster uniformly across all instances of the disorder and the category supports inferences about treatment; 3) what the causal history of the disorder is; and 4) just how informative the classificatory category is.
Philosophers discussing whether kinds are natural kinds are interested in the same set of issues (see Beebee and Sabbarton-Leary 2010 for a discussion of contemporary issues). On most standard definitions a natural kind is a discrete, mind-independent entity marking a real division in nature. These natural divisions constitute the classificatory units of the sciences (e.g., physics, chemistry, biology, and so on). The periodic table is the exemplar par excellence of science’s endeavor to discover and demarcate nature’s joints. Furthermore, the discovery of such natural divisions by a particular branch of empirical enquiry seemingly validates that branch of enquiry, elevating it to the status of a science.
The classification of the objects of experience—the things we see, hear, smell, feel, and taste—into natural kinds also brings with it a number of fringe (epistemic) benefits. Knowing that the animal one happens across whilst on an Asian safari is a member of the species Panthera tigris, for instance, allows us to make useful and potentially life-saving inductions about the likely behavior of the said animal: When it issues a low and throaty growl as it stalks us we know it is time to make a hasty retreat. Similarly, since kind concepts are projectible,1 knowing that the substance in front of us is water allows us to induce it is (likely) potable, since typical members of the kind Water are potable.
Of course, not all objects of experience can be classified into natural kinds. Many of the items found in a kitchen, for instance, form kinds of some sort—pots, pans, utensils, ingredients, etc.—but these kinds are not natural. Rather, they are artificial kinds. The contrast is fairly straightforward, at least at the level of intuition. Consider the subatomic particle electron on the one hand and the kitchen utensil spatula on the other. Electrons are negatively charged particles that are responsible for the substance-forming bonds between atoms of different elements. Covalent bonds between atoms of chlorine, for instance, are created when a pair of atoms “share” a pair of electrons—each atom contributing one electron—giving each atom eight electrons on its outer electron shell. Electrons have definite characteristics; their mass, charge, velocity, and so on are uniform between each member of the kind. Put more generally, they have key determinate properties that all and only members of the electron-kind possess; they are natural kinds.
Spatulas, in contrast, are not uniform. They do not have a determinate material or dimension. They are characterized by their function: Roughly, for flipping, turning, stirring, and scraping foodstuffs. But many objects can be employed to fulfill the function of a spatula. A ruler, for instance, could be used to flip a frying pancake, stir a cooking stew, or turn a poaching fish. This interchangeability of function between items of disparate design typifies the intuitive natural/artificial distinction. We take that distinction to boil down to something like the following: There are many reasonable ways to group and divide the objects of our experience. Some groupings and divisions latch onto real distinctions in nature. Their investigation is subject to the highest degree of methodological scrutiny, that is, scientific investigation, and they survive such scrutiny. They are natural kinds. Other groupings and divisions do not survive such scrutiny. They do not latch onto real distinctions in nature but capture different functions the objects of experience might have.
Within the philosophical literature one of the perceived challenges for psychiatry is to justify that its kinds are natural. The critique of the 1960s and 1970s has left psychiatry with vestigial questions concerning its scientific credentials (Murphy 2006, 2009; Cooper 2007, 2009; Pickard 2009. The anti-psychiatry movement went even further and questioned whether mental illness represents the medicalization of socially deviant behavior, and the pathologizing of normal problems of living (Bolton 2008: 163; see also Bortolotti 2013. The thought was that the label of mental illness not only fails to denote a natural kind, but does not denote anything at all. Szasz argued, for instance, that mental illness is just a “convenient myth” (Szasz 1961: 113). If entities classified as mental disorders could be shown to be natural kinds, then progress could be made with many of the controversies surrounding the very notion of mental illness.
There are two challenges faced by psychiatry. The first root pressure is metaphysical: Are mental disorders natural kinds? The second pressure is normative: What behaviors should receive medical treatment?
The first challenge has significant contemporary bite when we consider the ever-increasing list of so-called mental disorders. The list of mental disorders stood at 374 in DSM-IV-TR compared to 106 in DSM-I (Grob 1991: 421–31). Since the DSM has only been running since 1952, this trend indicates a staggering increase in i) our ability to recognize and diagnose mental disorders, ii) our propensity to classify behaviors as mental disorders, or iii) the scientific, clinical, or political need for a more fine-grained taxonomy of disorders, with previous disorders being split into multiple disorders.
The second challenge faced by psychiatry is to some extent a more general problem in medicine.
Psychiatry’s crisis revolves around the question of whether the categories of human distress with which it is concerned are properly considered “disease” as currently conceptualized and whether exercise of the traditional authority of the physician is appropriate for their help functions. Medicine’s crisis stems from the logical inference that since “disease” is defined in terms of somatic parameters, physicians need not be concerned with psychosocial issues which lie outside medicine’s responsibility and authority. (Engel 1977: 129).
Engel’s characterization is useful since he identifies important issues that are interconnected, from a practitioner’s point of view, with the justification of psychiatric illnesses as natural kinds. How far does the “traditional authority” of the physician extend? Should all behaviors associated with mental disorder be medically treated? If so, how far along the continuum of human behavior should this treatment extend? In the context of mental disorders, it is also important to ask what “medical treatment” amounts to. Medical treatment is not synonymous with pharmacological treatment. For instance, it has been argued recently in the context of personality disorders that it is appropriate for medical treatment to take the form of a “retraining of habits,” which can be achieved in a community environment with the assistance of different therapies (Pickard 2009). In addition, as science advances, the scope of what can be treated also expands, and what may have been considered an “enhancement” by one generation could be considered a therapy by the next.
There is a range of human behavior that can, more or less, be quantified across a range of variables. For instance, human responses to aggression could, in principle, be statistically mapped. Those responses might include behaviors including: a) answering aggression with aggression; b) answering aggression with violence; c) answering aggression with passivity; or d) answering aggression with panic. Now, imagine that a particular individual, call her Hannah, reacted to aggression in a harmful fashion—say, physical violence, impairing her ability to form healthy, meaningful social relationships, and to function effectively within typical social environments. Does Hannah’s behavior qualify as medically treatable? If so, why is her particular behavior medically treatable as opposed to, say, Peter’s response to aggression—say, total passivity, that also impairs his formation of healthy and meaningful social relationships? What should Hannah’s medical treatment amount to? Does society’s view of Hannah’s response—as one which is inherently problematic—make it more likely that Hannah, rather than Peter, will receive treatment?
Here, the concern is that, if mental disorders are not genuine disorders, and the boundaries of mental illness are drawn arbitrarily, then it may not be justified to treat people diagnosed with those disorders medically. Engel’s characterization of the crisis in medicine now comes into play. On Engel’s account, a wholly somatic characterization of medicine rules out some psychosocial harms that impinge on people’s quality of life. Thus, these impingements either go untreated, leaving those individuals to suffer, or are re-conceptualized as criminal acts and punished. Such considerations do not necessarily set psychiatry apart from the rest of medicine. Physical health has arbitrary boundaries, too. How many bacteria need to be found in a lung for the case to be a case of pneumonia? Chronic illness, earlier detection of illness, and the expanding boundaries of our conception of “ill health” all lead to a need for doctors to manage impaired glucose tolerance as well as diabetes, polyps, and tumors. There is a sense in which the whole of medicine, including psychiatry, deals with fuzzy boundaries. Medicine needs to manage less severe illnesses, more illnesses, and more chronic illnesses than in the past due to our longer lifespan.
One strategy for accounting for kinds in psychiatry we would like to explore in this chapter is to argue that the category “mental disorder” should be restricted to all and only those disorders that have biological causes. This strategy comes with the following caveat: There are mental harms that should be treated medically, if we assume a broad conception of “medical treatment,” but do not count as mental disorders. We shall examine this proposal in some detail, and highlight its benefits and limitations.
There have been different strategies to resist the challenge that mental disorders are not natural kinds. They include: i) justifying the category of mental illness as a category of natural kinds; ii) sketching out the different philosophical notions of natural kind, and illustrating how mental illness falls within the scope of one such notion; and iii) presenting an alternative concept to that of a natural kind, such as a practical kind, that is scientifically respectable and accounts for mental illness.
In response to the general challenge formulated in the 1960s, reinvigorated by debate in the 1970s, the task was clearly set to define mental disorders, to make clear the basis of why conditions were in the manuals or excluded, and clear that this basis was medical—scientific and objective—not a matter of social rules of normal behaviour. (Bolton 2008: 164)
The view we want to explore is broadly naturalist. Hence our investigation of the metaphysical challenge begins with a brief discussion of medical nosology, since the concepts of mental illness, disease, disorder, and impairment are clearly intended to be analogous to established physical counterparts. At this stage it is worth noting that nothing in particular hangs on the distinction between mental illness, disease, disorder, or impairment as we employ those terms. Toward the end of the chapter, we shall draw a terminological distinction between disorders and harms, but until then we will use the current terminology of psychiatry somewhat interchangeably.
What are diseases according to medicine? The Hippocratic tradition construes diseases as entities. The presence of a disease-entity within a host is indicated by a defined cluster of signs or symptoms. Consider for example Transmissive Spongiform Encephalopathy (or TSE). The presence of TSE in a host is indicated by two physical changes: The degeneration of the brain and the spinal column, and the formation of amyloid plaques in the extracellular matrix. There are also behavioral symptoms. Hosts exhibit a general deterioration in behavior, including rapidly progressive dementia and psychotic symptoms, ataxia (an unsteady gait), and myoclonus (sudden jerky movements). The cause of TSE is attributed to a protein called a prion. The prion protein can be found within the brain and spinal column. During standard cell replication the prion protein, specifically in the brain and spinal tissues of the host, misfolds. The misfolding converts the normal prion protein (or PrPc) into a toxic counterpart (PrPSc), which is ultimately responsible for the physical and behavioral symptoms we observe.
TSE conforms neatly with the disease-as-entity model espoused by modern medicine. The toxic prion protein PrPSc is a natural kind; it is a mind-independent, objective ontological entity. Moreover, it is causally responsible for a particular set of symptoms used to diagnose TSE in a host subject. Moreover, the identification of PrPSc allows the cause of TSE to be empirically investigated, analyzed, and eventually treated.
The construal of diseases-as-entities appears to be a metaphysical condition of medical nosology. It is not accompanied by an epistemic counterpart: Diseases are classified despite the entity responsible for the symptoms being unknown. For instance, Caffey disease (or Infantile Cortical Hyperostosis) is a disease of unknown cause. Affected host symptoms include bone lesions, soft-tissue swelling, hyperesthesia, tenderness, and irritability. In extreme cases paired bones (e.g., tibia and fibula) can become fused. Recent research claims both that “an underlying viral aetiology has been implicated” (Hall 2005: 2). A genetic mutation within a collagen-encoding gene (COL1A1) has been identified, during a genome-wide screen of affected individuals (Gensure et al. 2005: 1250). Despite this, the existence of the disease is (fallibly) indicated by the existence and stable clustering of the symptoms in repeated cases. The category Caffey disease has the hallmarks of a natural kind: It is mind-independent, apparently causally stable, projectible, and epistemically useful. This, it seems, is sufficient to give medical science confidence that there is a cause of Caffey disease, consequently construing it as a natural kind.
This does not imply that there are no extant questions concerning the kind category. For example, it may be that there are two forms of Caffey disease that do not share a common causal basis. A pre-natal form of the disease has been identified, occurring before week 35 of the gestation cycle. The contraction of pre-natal Caffey disease results, typically, in perinatal death. However, the mutated COL1A1 gene is absent in pre-natal Caffey disease. Thus the relationship between pre- and post-natal versions of the disease is unclear. But this need not threaten the inclusion of Caffey disease in medical ontology—the disease has been successfully identified. Rather, investigation into the precise mechanisms that produce the indicative symptoms continues.
Medical taxonomy’s outstanding challenge is to determine whether pre-natal Caffey disease—which is symptomatically identical but etiologically distinct (with no COL1A1 gene mutation)—is a bona fide instance of the disease. There are two options: i) expand the category of Caffey disease to include both pre- and post-natal variants, and expand our conception beyond the mutation of COL1A1; or ii) maintain that Caffey disease is the mutation of COL1A1 and carve out a new category of disease corresponding to the pre-natal form, given its distinct cause.
The example elucidates philosophically interesting features of physical disease classification. Most pertinent is the appeal to a “disease-as-entity model” that underlies medical nosology (Caplan et al. 2004: Part 1). The example demonstrates that the metaphysics of medical nosology is one of kinds that aspire to be natural kinds. Diseases are predominantly classified as discrete units of our ontology; they are independently existing, objective entities. As members of the natural kind club they enjoy membership perks: They are law-like in their interactions with other kinds, they are projectible (in the inductive sense), and their readily observable features are a product of a unified internal nature or structure. As Georges Canguilhem notes, in his discussion of Louis Pasteur’s germ theory of contagion, what makes the position philosophically attractive is that “it embodies an ontological representation of sickness . . . a germ can be seen” (Canguilhem 2004: 41).
Philosophically, this is an enticing prospect for psychiatry. If a disease or illness is a bona fide ontological object, then it can be reliably investigated without (ontological) reservation. Moreover, even if our understanding of it is incomplete, the existence of the illness is not at issue. The symptoms of Infantile Cortical Hyperostosis, for instance, are observable and empirically investigable. The bone lesions, soft-tissue swelling, and irritability of the host cluster together reliably. Granted, the cause is currently unclear, but the evidence suggests that the category is a natural kind.
However, mistakes may occur. We may be mistaken, for instance, about some of the symptoms of TSE. It could be that the aggressiveness of many humans is, in actual fact, caused by some other disease that tends to accompany, but is not actually a feature of, TSE. But mistakes do not undermine the utility of the thesis. Rather, they are a reminder of the requirement for methodological rigor—hypotheses are subject to revision in light of confounding evidence.
The ontological attraction of the disease-as-entity model is obvious. The challenge for psychiatry is to justify its categories as ontologically respectable by showing why mental disorders should be construed as a bona fide feature of the furniture of reality, defined by their objective features and subject to the highest standards of scientific scrutiny. Naturalism attempts to answer this challenge by situating psychiatric illnesses firmly within biology. What follows is a brief sketch of two versions of naturalism to illustrate the sort of position that would offer a viable account of mental disorders as natural kinds.
The central tenet of naturalism in psychiatry is that mental disorders can be identified with natural facts. Perhaps the two most famous views are those developed by Christopher Boorse (1975) and Jerome Wakefield (1992). According to Boorse, the human body is made up of sub-systems that have natural functions. The function of the amygdala sub-system, for instance, is to process memories and emotions. It is also thought to be responsible for the regulation of memories associated with both negative and appetitive conditioning (see for instance Killcross et al. 1997: 377–80). The idea is that sometimes systems become dysfunctional. Liver dysfunction, for example, can lead to a variety of illnesses and diseases. When the liver becomes infected or damaged by, say, alcohol, it no longer effectively removes bilirubin from the blood. The resultant increase in bilirubin levels in the blood results in jaundice.
Analogous examples can be found within sub-systems in the brain. For instance, the International Classification of Diseases (ICD) entry F07.0 Organic Personality Disorder captures so-called Klüver–Bucy syndrome, where an affected individual may suffer from various symptoms including polyphagia—excessive hunger. The cause of this behavior is attributed to lesions of the anterior temporal lobe (Klüver and Bucy 1939: 979–1000). These lesions are causally responsible for a general dysfunction of the amygdala sub-system, resulting in Kluver–Bucy syndrome. Central to Boorse’s account is the notion of “function.” For Boorse the functionality of a system is determined via its normal role within an organism, where that role contributes to biological success (i.e., survival and reproduction). As such, “dysfunction” can be construed as a disruption to the normal role of a system hampering biological success. Wakefield’s naturalism similarly appeals to the notion of function. On Wakefield’s view there are various mental mechanisms at play in human cognition. When the normal function of a mental mechanism is disrupted, becoming dysfunctional, we have a mental disorder. The key difference between Boorse’s account and Wakefield’s is that for Wakefield the notion of a normal function is determined in relation to the evolutionary design of the mechanism. Wakefield is clear that what he has in mind when he talks of normal function is natural function.
A disorder is different from a failure to function in a socially preferred manner precisely because a dysfunction exists only when an organ cannot perform as it is naturally (i.e., independently of human interactions) supposed to perform. Presumably, the functions that are relevant are natural functions. (Wakefield 1992: 381)
The general feature of Boorse’s and Wakefield’s respective account that is relevant to our discussion of the medical model is that both appeal to biological features of organisms to provide an account of mental dysfunction. Appealing to the natural functions of the organism to explain mental disorders coheres well with the medical model. Illnesses are etiologically connected to empirically investigable, mind-independent physical entities. In other words, they are natural kinds.
There have been various objections to both Boorse’s and Wakefield’s respective versions of naturalism. We do not discuss those objections here, but instead direct readers to Rachel Cooper (2007) for a useful survey of these positions. Our discussion is intended to be higher-level in so far as we are not proposing a particular account of naturalism per se, but endorsing a naturalist approach to the classification of mental disorders.
How does the disease-as-entity model apply to a paradigmatic psychiatric disorder? To illustrate the sort of position we have in mind let us consider a mental disorder that has been revised for the most recent iteration of the DSM, namely Autism.
In DSM-5 the various species of autism have been unified with the introduction of a single disorder Autism Spectrum Disorder (ASD). ASD merges four discrete disorders (which were categorized separately in DSM-IV): 1) Autistic Disorder; 2) Asperger’s Disorder; 3) Childhood Disintegrative Disorder; and 4) Pervasive Developmental Disorder. Like the ICD, the DSM sketches out the symptoms affected individuals will likely exhibit. According to DSM-5, people with ASD typically exhibit a communication deficit. What this amounts to is a general propensity to misread non-verbal interactions and an inability to build age-appropriate friendships. Individuals can be highly routine-dependent, and extremely sensitive to change. They may also be overly attached to inappropriate items.
In the context of the current debate concerning the classification of mental disorder, and the methodology of medical nosology, we might first ask: What is, or what causes, ASD? Underlying that question is a philosophical one: Is there some ontologically respectable entity responsible for the symptoms exhibited (on a spectrum) by subjects? Unfortunately, the causes of ASD remain largely unknown, although there are four causal categories that are typically mentioned as possible candidates. These are: 1) genetic factors; 2) environmental factors; 3) psychological factors; and 4) neurological factors.
Each category postulates its own cause of ASD, sometimes in combination with factors from other categories. For instance, genetic factors are thought to be responsible for an individual’s susceptibility to ASD, due to the correlation between the development of ASD in identical twins. According to the National Health Service (NHS), where one twin has ASD there is a 60 percent chance that the other twin will also develop ASD. In addition, it is also thought that to develop ASD an individual requires an environmental trigger.
The psychological factor thought to be causally responsible for ASD is that the subject lacks a comprehensive theory of mind. The idea, roughly, is that individuals with ASD do not possess an adequate theory of mind, thereby failing to attribute mental states to other individuals. This theoretical gap, or so the hypothesis goes, accounts for their inability to deal adequately with social interactions and to form appropriate relationships.
Finally, neurologists cite two possible explanations of ASD. The first is the connection between the amygdala, the limbic system, and the cerebral cortex. According to this hypothesis, in individuals with ASD the connections between these systems have somehow become jumbled. The empirical basis for the hypothesis comes from neural imaging studies that appear to show signaling differences between individuals with and without ASD. This neural cross-wiring is thought to be responsible for the abnormal responses ASD sufferers exhibit toward trivial events. The second attributes ASD to a difference in the function of mirror neurons found in an individual’s brain. Mirror neurons, which allow us to mimic the behavior(s) of others, are thought to function differently in people with ASD. Since mirror neurons are responsible for language acquisition and recognition of emotion, and furnish us with an ability to learn from others, the deficits ASD sufferers experience in these areas may be attributable to dysfunctional mirror neurons.
The genetic, environmental, psychological, and neurological causes of ASD cited in the literature are not necessarily distinct. For example, neurological changes could underpin psychological and cognitive changes, which may have been caused by environmental or genetic effects. To complicate things further, the propensities of such an organism may further lead to that individual being exposed to other risks, such as bullying or social isolation, that may compound the issue, leading to further genetic and neurological changes.
The ASD example is interesting for a number of reasons. There are different but potentially complementary causal stories we can tell (more than one type of factor may contribute to a given case of ASD and different cases of ASD may have different causal etiological structures). These alternative causal explanations reveal our classificatory intentions, but need not undermine our confidence in the conceptualization of ASD as a natural kind—remember the Caffey disease example. However, not all the causal accounts of ASD are as ontologically respectable. Compare and contrast the ontological attraction of a broadly neurological account, citing genetic and environmental triggers, with purely psychological accounts—the former, but not the latter, has clear parallels with the disease-as-entity model that dominates medical nosology. Tracking ASD back to either a particular way that electric and chemical signals are transmitted between the amygdala, limbic system, and cerebral cortex, or to dysfunctional mirror neurons, has clear parallels with Louis Pasteur’s germ theory of contagion, and coheres well with the disease-as-entity biomedical model. The potential of such an account is illustrated by one of the inaugural stories of biological psychiatry—namely, that the myriad psychiatric manifestations of syphilis, including general paralysis of the insane, has been causally traced back to an infection by Treponema pallidum, and the discovery of spirochetes in the brain of those affected (Pearce 2012).
Issues surrounding the etiology of ASD (and other mental disorders) need to be considered next to analogous cases in medical nosology such as Caffey disease, where the cause is uncertain, but the stable cluster of symptoms is well established, and the classificatory category appears to have significant empirical utility.
The position we want to describe and test is that there are objective, mind-independent facts that are the causes of mental disorders. Once these causes have been identified an additional piece of furniture is added to our theory of the natural kind structure of the world. The precise shape of that piece of furniture may take some time to become fully defined, but this is an epistemic rather than a metaphysical problem, affecting all branches of classification across the sciences. The state of our knowledge may be such that we are unable to determine precisely what the cause of such-and-such a mental disorder is.
One unique problem psychiatric classification faces is that the main symptom of a disorder is often a behavior or a first-person account of a particular behavioral episode. Because there are numerous causes of behavior, the accurate diagnosis of a particular set as disorderly is particularly challenging. The key conceptual point is that we should restrict the category of mental disorders to those with biological causes, aligning the main conceptual category of psychiatry to the prevailing strong version of the medical model. The advantage of the naturalist position is that it is less metaphysically controversial and more empirically respectable. Moreover, the construal of mental disorders as natural kinds brings with it a host of fringe benefits. The move alleviates pressure from the (in)famous critique of the 1960s and 1970s; the condition on the classification of mental disorders—that they must be natural kinds—along with convincing examples like ASD reduces the critique to battles over the reality of particular illnesses, not the category “disorder” simpliciter. And this is a benefit, since individual battles over particular disorders will assist psychiatric classification by promoting methodological rigor.
The difficulty is that psychiatry often starts from a psychological phenomenon deemed “disordered” on the basis of whether psychological characteristics fit normative frameworks of good functioning, rationality, social acceptability, etc. When it is possible to identify causes of the disordered phenomenon, these may not be entire pathological mechanisms but deviations resulting in slight qualitative changes to normal function. As such, not all the kinds classified in DSM-5 will qualify as natural kinds, and hence as disorders, within the framework. To put it another way, some conditions that are presently treated by psychiatrists do not qualify as mental disorders. To account for these kinds that are not disorders we introduce the notion of a para-natural kind.
Roy Sorensen has recently introduced a novel notion that may assist current efforts to conceptualize psychiatric kinds, namely the notion of a para-natural kind. According to Sorensen, a para-natural kind is parasitic upon some natural kind or other; it is “an absence defined by a natural kind” (2011: 113). Sorensen explicates his idea with two intuitive examples. Consider the natural kind heat and its counterpart cold. Where heat is defined (very) roughly as molecular motion, cold is simply the absence of heat (i.e., the absence of molecular motion). Hence cold is parasitic upon heat, and a para-natural kind. Similarly one can distinguish the contrasting pair shadow and light. Where light is a natural kind (i.e., the presence of photons), shadow can be defined as the absence of light and thus, again, a para-natural kind.
So, parasitic kinds are not natural kinds. They are absences of actual natural kinds. However, they do appear to have some of the characteristics of natural kinds in so far as they “inherit the lawfulness and projectibility of the natural kinds that shape them” (Sorensen 2011). To illustrate, consider, for instance, the electron hole postulated by Paul Dirac to explain the “intense chemical activity of chlorine” (Sorensen 2011: 119). The idea is that since the outer shell of a chlorine atom has seven electrons, but space for eight electrons, there is a kind of energy hole, a gap in the electron shell that has the opposite charge to an electron (i.e., it has a positive charge). This electron-gap attracts atoms of other substances which, in turn, can result in intense chemical activity.
Moreover, these interactions are law-like and projectible. We can and do make reliable inductions about chlorine’s chemical activity on the basis of the electron-gap, and knowing that other atoms have similar electron gaps helps us to induce information about them. Ontologically speaking, since we have a prior commitment to the existence of electrons—they are bona fide natural kinds, and part of the furniture of reality—the explanatory power of the electron-gap, which is to say the absence of an electron, comes with no ontological costs. We do not have to postulate some further mind-independent entity to explain an observable phenomenon, but we nevertheless are able to provide an accurate and true explanation, which is consistent with our current theory.
How might this idea extend to psychiatry? Well, first we need to identify a natural kind to work with, and be ontologically committed to, around which we can then note the absence(s). One starting point would be to construe human cognition as such a kind, albeit of a more general nature. Nancy Andreasen argues that schizophrenia can be conceived of as a deficit in certain information processing systems (see for example Andreasen 2000). Inspired by her account, we could say that, when the brain as a complex set of information processing systems and sub-systems is operating sub-optimally, what causes the sub-optimal functioning can be divided into two broad categories: Natural kinds and para-natural kinds.
The natural kinds will be those sorts of causes that have a biological basis, which are independent items of our ontology. The etiology of these sub-optimal states will be pathophysiological, and the disorder should be identified with that cause. For instance, brain lesions are known to cause behavioral changes, mental confusion, and loss of memory, amongst other things. As such, any behavioral changes that symptomatically match a mental disorder, which can be causally attributed to a lesion in a particular region of the brain, genetic mutations, or infective agents would conform to this sort of model. On this view, sub-optimal states with the correct type of etiology are natural kinds.
The para-natural kind category will capture the other causes of sub-optimally functioning processes that are not attributable to an independent natural kind, but rather to the absence of the fully functional information-processing system.2 The para-psychiatric kinds will be those mental harms that are not grounded in specific biological causes per se (although the causes may well have a pathophysiological realizer, in the sense that all behaviors must be physically realized), but are nevertheless sub-optimal states of the information-processing system. Consider, for instance, depression, whose changed diagnostic criteria in DSM-5 has caused a stir. Imagine an agent that qualifies for a diagnosis of a major depressive episode under the auspices of DSM-IV-TR (satisfying five of nine qualifying criteria over a two-week period). Now, imagine that the cause of that agent’s depression is bereavement, and that the qualifying factor leading to the diagnosis—the straw that broke the camel’s back, so to speak—stems from the fact that they now fall outside the arbitrary grief time-limit DSM-IV places upon patients. Whilst the symptoms the agent exhibits will be physically realized if depression is construed as a para-natural kind, there need be no biological etiology for the episode.
According to the proposal, only natural kinds are appropriately labeled “disorders,” whilst para-natural kinds are not. However, the scope of behavior that can be treated by psychiatrists is broader than the category of mental disorders, and can extend to para-natural kinds provided treatment decisions are subject to robust scrutiny.
Consider major depressive episodes, which appear to lack a biological etiology. On our account they are a species of para-natural kind. This, we claim, coheres well with recent work by Horwitz and Wakefield (2007), discussed by Dominic Murphy (2009: 110), who identify a challenge for the contemporary category “depression.” They claim that it is applied without common sense, using symptoms that capture far too much normal human behavior, and that as such there is a “needless alarmism about an epidemic of depression” based upon erroneous diagnosis (Murphy 2009: 111). The root of the challenge can be illustrated via an appeal to one of philosophy’s oldest puzzles concerning vagueness, the Sorites paradox.
Consider a heap of sand consisting of 10,000 grains. Does the heap remain a heap if we remove one grain of sand? Since removing one grain of sand does not make much difference to the heap, and since 10,000 grains of sand constitute a heap, so too should 9,999 grains. However, repeating this process eventually leads to one grain of sand, which definitely is not a heap. Where, then, is the boundary between heaps and non-heaps? The point is that at each step of the process (of removing grains of sand) we are inclined to say that no significant change has occurred. Hence the boundary is a vague one. Consider a man, Andrew, who has lost his wife, and now experiences symptoms associated with a major depressive episode, including: A depressive mood, diminished interest in life (e.g., work, his house, and so forth), diminished appetite leading to weight loss, insomnia, and fatigue. For the sake of argument let us imagine that Andrew experiences these symptoms for 30 days. On day 1 Andrew is grieving within normal human limits. The loss of a loved one is a significant negative event that can have a serious impact upon cognitive equilibrium. Since there is no significant difference between day 1 and day 2, on day 2 Andrew is grieving within normal human limits, and so on until we reach day 30. However, according to DSM-IV-TR, if the symptoms continue beyond 14 days then the agent in question is experiencing a major depressive episode. So, on day 15 Andrew’s grief has developed into a major depressive episode. But if there is no cognitive change in Andrew between days 14 and 15, and the only difference is the length of time he has experienced these symptoms, then the diagnosis appears as arbitrary as, say, stipulating that anything less than 5,000 grains of sand is not a heap, whilst anything more than 5,000 is a heap.3
The problem is that arbitrarily stipulated boundaries for mental disorders seem to undermine a) the disorder’s scientific respectability, and b) a fortiori the scientific credentials of psychiatry. The purported “disorder” cannot satisfy some of the more conservative ontological credentials of natural kind-hood, namely an ontologically discrete entity with a decisive boundary separating it from all other kinds of entity. In the proposal we are examining, such kinds should be construed as candidate para-natural kinds. Horwitz and Wakefield, we claim, are correct about the erroneous diagnosis of major depressive episodes based upon the way we have conceptualized it—as a natural kind. The arbitrarily drawn boundaries of a major depressive episode, in conjunction with the lack of a unified causal etiology, disqualify such episodes from being natural kinds. But are such episodes para-natural kinds?
Since a major depressive episode is an absence of an optimally functioning information-processing state, from day 1 it is an absence, as we would expect from a para-natural kind. Similarly, major depressive episodes are, by definition, uniform; to qualify as an episode the majority of the nine available symptoms listed in DSM-IV-TR must be present. The arbitrariness of stipulating that there is only a major depressive episode after 14 days—given that Andrew feels precisely the same on day 1 as he does on day 15—still presents a residual challenge. But importantly, that challenge is not to the category of mental disorders (via implication), since a major depressive episode is not a mental disorder.
The proposed distinction, then, is as follows: On the one hand, there are natural kinds, which are those illnesses of the mind with a biological etiology, firmly rooted within the ontology of the medical model—that is, bona fide entities that are categorically distinct and properly called mental disorders. They result in a sub-optimally functioning information-processing system, and qualify for medical treatment of a pharmacological nature, since it is the biological cause of the symptoms that is the target of the treatment. On the other hand, there are the para-natural kinds of psychiatry, which are those cognitive states that are harmful to the agent without a clear and distinct biological etiology (although they are, of course, biologically realized). They are characterized as absences of optimally functioning information-processing states, with various causes. They are harmful mental states in which an agent may find herself, where she requires access to medical treatment broadly construed. The harms are biologically realized, in so far as they occur (in some sense) within human minds and brains. However, although pharmacological treatment can impact the acuteness of the symptoms—since symptoms are biologically realized—such treatment may be inappropriate, since there is no biological cause to target and alter.
Exploring the extent and prevalence of para-natural kinds amongst the taxonomy of psychiatry within the latest editions of the DSM is a larger project than we can investigate here. But the benefits of para-natural kinds for psychiatry will include a fairly open set of potential causes of sub-optimal information-processing states, including psychosocial causes, environmental causes, and so on. Para-natural kinds are still a proper feature of the taxonomy of psychiatry (along with natural kinds), but they are not mental disorders.
Para-natural kinds, then, are accompanied by their own unique problem of demarcation, when distinguishing between health and illness. Consider the nosological status of psychotic experiences, such as hearing voices and having unusual beliefs, in young people. When does prodromal psychosis become a first episode? Here, the experiences could be within the range of normal experience and development, but conversely, may also signify the prodromal phase of schizophrenia or even a frank first episode of a psychotic illness. What does seem to be the case is that the nature of the experiences themselves may not help us in demarcating pathology from non-pathology. It is other factors, such as intensity, duration, distress, or additional dysfunctions such as cognitive impairment or depression (Murray and Jones 2012; Broome and Fusar-Poli 2012; Broome et al. 2013) that may help determine whether the prognosis of such experiences is clinically of interest and where intervention may prevent future harms.
None of this implies that the investigation of para-natural kinds is not an appropriate practice for medical science; quite the contrary. The point, rather, is that the current list of mental disorders are not all natural kinds, and that the effective conceptualization of psychiatric kinds is critical to our understanding of both mental harms and mental disorders, and critical to the appropriate treatment(s) available to the agents affected by them.
The natural kind account of mental disorders we have considered in this chapter aims to align the nosology of mental disorder with that of contemporary medicine. The benefit of the account is that it avoids some of the vestigial pressures of the anti-psychiatry movement, being realist, naturalist, and broadly empiricist. The inclusion of the notion of a para-natural kind helps account for those mental harms that are not natural kinds but are, nevertheless, worthy of the attention of clinicians and medical professionals. Both conceptions we have discussed rely upon an underlying notion of a general natural kind of effectively/appropriately functioning human cognition. We have adopted the notion of an optimally functioning information-processing system, but other accounts may be available. An underlying notion of a natural kind of human cognition is vital to the account, to distinguish disorders and harms from orderly mental processes—after all, orderly mental processes such as happiness are certainly physically realized in precisely the way we have envisaged mental disorders to be.
How we conceptualize psychiatric kinds is fundamental to how we understand and treat mental disorders and mental harms. In this chapter we have attempted to demonstrate that psychiatry has natural kinds, called “mental disorders,” but that the various iterations of the DSM have not restricted their nosology to them. The list of mental kinds of psychiatric interest has increased with each issue of the DSM, and includes both natural and para-natural kinds, that is both mental disorders and mental harms. The recognition of two distinct categories of mental kind within the DSM will, we hope, lead to a more fine-grained approach to the way we think of mental disorders.
Finally, the recognition of natural kinds in psychiatry provides the discipline with the sort of theoretical basis enjoyed by other branches of empirical medicine and science more generally. Mental disorders are no more mythical than misfolded prion proteins or mutated collagen-encoding genes. Nevertheless, the precise extent of psychiatry’s natural kinds is something in need of more study and discussion. Our hope is that the framework we have outlined here will contribute to that discussion.
1.The notion of projectibility within philosophy was introduced by Nelson Goodman (1984) and is connected to the notion of induction and inductive inferences. The idea is that a concept is projectible when it supports an inference from premise to conclusion that is ampliative. Or, to put it another way, when the concept within the premise of an argument supports an inference to a conclusion that contains information which is not contained within the premise. Mental disorder concepts would be projectible if they supported inferences about, say, future unobserved instances.
2.Note that one must be committed to “normal mental function” being a natural kind for the account to work. However, the details of what precisely “normal mental function” amounts to can be specified in different ways, provided the cluster of behaviors associated with the category cluster together reliably.
3.The illustration involves a simplification. We are aware that duration of a depressive episode is also used as a treatment filter—those that remit spontaneously are screened out of treatment. The DSM wants to pick out more chronic conditions that are worth treating.
Andreasen, N. (2000). Schizophrenia: the fundamental questions. Brain Research Reviews, 31, 106–12.
Beebee, H., and Sabbarton-Leary, N. (eds) (2010). The Semantics and Metaphysics of Natural Kinds. London, UK: Routledge.
Bolton, D. (2008). What is Mental Disorder: an essay in philosophy, science, and values. Oxford, UK: Oxford University Press.
Boorse, C. (1975). On the distinction between disease and illness. Philosophy and Public Affairs, 5, 49–68.
Bortolotti, L. (2013). Rationality and Sanity: The role of rationality judgements in understanding psychiatric disorders. In K. W. M. Fulford, M. Davies, G. Graham, et al. (eds), Oxford Handbook of Philosophy and Psychiatry. Oxford, UK: Oxford University Press.
Broome, M. R., and Fusar-Poli, P. (2012). Philosophical Issues in the Prodromal Phase of Psychosis. Current Pharmaceutical Design, 18(4), 596–605.
Broome, M. R., Wuyts, P., and Fusar Poli, P. (2013). Philosophical and Ethical Issues in the Prodromal Phase of Psychosis. In K. W. M. Fulford, M. Davies, G. Graham, J. Sadler, G. Stanghellini, and T. Thornton (eds), Oxford Handbook of Philosophy and Psychiatry. Oxford UK: Oxford University Press.
Canguilhem, G. (2004). The Normal and the Pathological—Introduction to the Problem. In A. L. Caplan, J. McCartney, and D. A. Sisti (eds), Health, Disease and Illness: Concepts in Medicine. Georgetown University Press.
Caplan, A. L., McCartney, J., and Sisti, D. A. (eds) (2004). Health, Disease and Illness: Concepts in Medicine. Georgetown University Press.
Cooper, R. (2007). Psychiatry and Philosophy of Science. Stocksfield, UK: Acumen.
Cooper, R. (2009). Is psychiatric research scientific? In M. R. Broome and L. Bortolotti (eds), Psychiatry as Cognitive Neuroscience. Oxford, UK: Oxford University Press.
Engel, G. L. (1997). The need for a new medical model: a challenge for biomedicine. In Science, Vol. 196, No. 4286. (Apr. 8, 1977), pp. 129–136.
Gensure, R. C., Mäkitie, O., Barclay, C., et al. (2005). A novel COL1A1 mutation in infantile cortical hyperostosis (Caffey disease) expands the spectrum of collagen-related disorder. The Journal of Clinical Investigation, 115(5), 1250–7.
Goodman, N. (1984). Of Mind and Other Matters. Cambridge, MA: Harvard University Press.
Grob, G. (1991). Origins of DSM-I: A study in appearance and reality. American Journal for Psychiatry, 148(4), 421–31.
Hall, C. (2005). Caffey Disease. Available at: Orphanet—<https://www.orpha.net/data/patho/Pro/en/Caffey-FRenPro431.pdf> (accessed 01/06/2013).
Horwitz, A., and Wakefield, J. (2007). The Loss of Sadness. New York, NY: Oxford University Press.
Killcross, S., Robbins, T. W., and Everitt, B. J. (1997). Different types of fear-conditioned behaviour mediated by separate nuclei within amygdala. Nature, 388(6640), 377–80.
Klüver, H., and Bucy, P. (1939). Preliminary analysis of function of the temporal lobe in monkeys. Archives of Neurology, 42, 979–1000.
Murphy, D. (2006). Psychiatry in the Scientific Image. Cambridge, MA: MIT Press.
Murphy, D. (2009). Psychiatry and the concept of disease as pathology. In M. R. Broome and L. Bortolotti (eds), Psychiatry as Cognitive Neuroscience. Oxford, UK: Oxford University Press.
Murray, G. K., and Jones, P. B. (2012). Psychotic symptoms in young people without psychotic illness: mechanisms and meaning. British Journal of Psychiatry, 201, 4–6.
Pearce, J. M. S. (2012). Brain Disease Leading to Mental Illness: A Concept Initiated by the Discovery of General Paralysis of the Insane. European Neurology, Vol. 67, No. 5, 272–278.
Pickard, H. (2009). Mental illness is indeed a myth. In M. R. Broome and L. Bortolotti (eds), Psychiatry as Cognitive Neuroscience. Oxford, UK: Oxford University Press.
Sorensen, R. (2011). Para-natural Kinds. In J. K. Campbell, M. O’Rourke, and M. H. Slater (eds), Carving Nature at Its Joints: Natural Kinds in Metaphysics and Science. Cambridge, MA: MIT Press.
Szasz, T. (1961). The myth of mental illness. American Psychologist, 15, 113–18.
Wakefield, J. (1992). The concept of mental disorder: on the boundary between biological facts and social values. American Psychologist, 47, 373–88.