Jerome C. Wakefield
Ever since Aristotle attempted to explain the design-like nature of organismic traits in terms of “final causes” that somehow regulated development, the concept of function and related “teleological” concepts (from the Greek telos, “end” or “purpose”), such as “design,” “purpose,” “adaptation,” and “end,” have been central to biological theory. As the preeminent evolutionary theorist George Williams explained, “the use of such concepts as biological means and ends…is the essence of the science of biology” (1966, p. 11).
Yet, what meaning can such terms have in a universe of mechanical causation? Must teleological attributions to biological mechanisms implicitly rely on illegitimate assumptions of intelligent design?
Darwin's theory of natural selection is the culmination of two millennia of grappling with how to understand biological teleological notions within a nonmystical scientific perspective. Contrary to the common claim that Darwin's theory of natural selection provides a way to eliminate teleology from biology in favor of mechanical causation, Darwin himself suggested that his theory had explained teleology rather than eliminated it (Gotthelf, 1999; Lennox, 1993), noting in a letter that it is “difficult for any one who tries to make out the use of a structure to avoid the word purpose” (Buller, 1999, p. 6). In fact, Darwin's conception of function places “the material principle of natural selection in place of the Aristotelian final cause” (Williams, 1966, p. 258).
Natural selection is only one force shaping organisms. There are many other influences on the nature and form of organisms, including developmental and structural constraints, genetic drift, various kinds of linkages between traits, and chance historical contingencies (Williams, 1992). The relative overall influence of such constraints versus natural selection remains a debated question (Orzack & Sober, 1994). Those who believe that such nonselective influences have a major impact sometime inflate this belief into an “anti-adaptationist” argument (Lewontin, 1979). However, the teleological tradition is exclusively concerned with explaining those striking aspects of organisms that are adaptive and apparently design-like, and no set of developmental constraints, historical conditions, or other nonselective processes can adequately account for this overall domain. Disputes about the specific type of explanation appropriate to a given feature can be addressed empirically (Buss, Haselton, Shackelford, Bleske, & Wakefield, 1999), but the necessity for invoking teleological explanation in a vast number of cases remains.
Anti-adaptationism did, however, fuel methodological concerns about the ease with which one can construct selectionist hypotheses lacking empirical warrant. Such hypotheses were famously demeaned by Stephen J. Gould as “just-so stories” (1991; Gould & Lewontin, 1979). These criticisms were ultimately useful in pointing to the need for greater methodological rigor (Griffiths, 1996), and evolutionary psychological science has dramatically improved its empirical grounding in recent years. But even if objections to specific teleological hypotheses are correct, teleological explanation in some form remains essential for understanding most adaptive traits.
Darwin's contribution to the teleological tradition can only be appreciated in the context of the challenging conceptual puzzles that perennially afflict discussions of biological function, also known in the philosophical literature as “natural” or “proper” function. To take a simple example: The spider's web enables the spider to catch insects, and we believe that this benefit is not just a happy accident but the function of the spider's web. We also believe that the spider's ability to create webs is not a happy accident but the function of a variety of mechanisms, some known and some unknown, within the spider's body and brain. But what exactly do such function statements add to the descriptive facts that certain internal mechanisms have the effect of enabling the spider to create webs, and webs have the effect of catching insects? They must add something, because the overhang of a house's roof may provide a particularly fruitful location to place the web and to catch insects, but nevertheless we don't think that the function of the overhang is to enable the spider to catch insects.
One thing we seem to be adding in citing the function of a trait is a partial explanation of the trait; we are saying that catching insects is part of the explanation for why spiders have webs. Thus, function statements imply what has been called functional explanations. For example, the heart's effect of pumping the blood is part of the heart's explanation because, in virtue of natural selection, one can legitimately partially answer a question like “Why do we have hearts?” or “Why do hearts exist?” with “Because hearts pump the blood.” Thus, pumping the blood is a function of the heart.
It has sometimes been argued that the explanatory dimension of function claims is not essential and that the function of a trait is simply whatever causal role the trait has in conferring various capacities on the larger organismic system (Cummins, 1975). No doubt scientists sometimes talk this way. However, such a “causal role” view allows accidental effects to count as functions. Even pathologies have “functions” according to this account, because they confer distinctive effects and capacities on the organism. Thus, such views by themselves cannot provide an adequate conceptualization either of biological function in the evolutionary nonaccidental sense or pathological dysfunction that is without any biological function. Something needs to be added to causal role to account for function and dysfunction.
However, the requirement that a function must be explanatory raises a classic problem of seeming backward causation. Given that catching insects is an effect of spiders' webs, and thus comes after the webs, how can the effect of catching insects explain its own cause, the web-building, while still observing the scientific prohibition on backward causation and rejecting Aristotelian notions of final causes? An adequate account of biological functions must dispel the notion that backward causation is being invoked. Moreover, even those who do not believe in a divine creator sometimes say colloquially that catching insects is the purpose of the spider's web, but how can a natural object have a purpose in a mechanical universe? Thus, a common concern is that biological function statements are somehow reading human intentionality into physical processes (Cummins, 1975), and an adequate account must make clear that this is not so.
Evolutionary psychologists have a particular need to address such issues and to be as clear as possible about the concept of biological function. Claims about how people's minds are designed and about the functions of known or hypothesized mental modules constitute evolutionary psychology's most distinctive contribution to psychology. These functional claims are often highly controversial and undergo unusually intense critical scrutiny. It is therefore important for evolutionary psychologists to have a nuanced understanding of what function claims mean, what kinds of evidence might count for or against them, and what ambiguities or misunderstandings might arise in making such claims. Fortunately, the growing importance of these issues is paralleled by a developing philosophical literature on the concept of function, which is already so large and diverse that only a few strands can be considered here.
One area in which evolutionary psychology is having a major impact is clinical psychology. An account of “function” implies a corresponding account of “dysfunction” or “malfunction,” the concept on which theories of psychopathology are based. Thus, one major test for a theory of function is whether it provides an account of dysfunction that offers an adequate evolutionary-psychological foundation for theories of psychopathology. The ways in which the analysis of function presented here illuminates the foundations of clinical psychology will be considered later in this chapter.
In this chapter, I first examine recent developments in the philosophical analysis of the concept of biological function and propose what I believe is the most adequate approach. I focus on “etiological” or “historical” accounts that interpret functions primarily as naturally selected effects. Then, in the second part, drawing on my “harmful dysfunction” analysis of disorder (Wakefield, 1992a, 1999a, 1999b), I illustrate how the etiological analysis of function offers a conception of dysfunction that can illuminate the concept of mental disorder and the nature of psychopathology.
In the 1960s and 1970s, at about the same time within the literatures of evolutionary theory and philosophy of biology, there were critical turning points in the analysis of the concept of function. In biology, George Williams (1966) complained that “biologists have no logically sound and generally accepted set of principles and procedures for answering the question: ‘What is its function?’” (p. 252). He corrected this situation by publishing his powerful analysis of the field's foundations, including his definition of natural function as naturally selected effect. In philosophy, Larry Wright (1973, 1976) established the etiological account of “function,” which, after it was revised to deal with counterexamples, converged to a natural-selection-based account as well.
Both Wright's and Williams's analyses were aimed at distinguishing functions in their explanatory sense from accidental benefits and other extraneous factors that often are confused with functions. The philosopher Carl Hempel (1965) had posed the challenge of drawing such a distinction in his classic account of functional explanation. The heart has many effects, observed Hempel, included among which is pumping the blood and making a sound in the chest. Indeed, both of these are beneficial effects, due to the medical uses of heart sounds. Yet among these effects, only some are biological functions of the heart. The challenge for a theory of function is to explain how to distinguish those effects of a trait that are its functions from those that are not. And the primary constraint on the account is that it must explain how attribution of a function offers a functional explanation. That is, the analysis must show, in a naturalistic and scientifically acceptable way, without invoking backward causation or divine design, how the effect that is a function can explain the presence of the very trait of which it is an effect. Thus, for example, the analysis must show how it is that, when asked “why do kangaroos have pouches,” one can correctly answer, “to protect their developing young.” This challenge of distinguishing functions from other effects has been the standard framework for recent inquiries into function.
Traditionally, this explanatory sense of function is known as the “strong” sense of function, versus some “weak” senses to be discussed shortly. To describe the function of a trait in the strong sense has been considered a way of outlining a partial explanation of the trait's existence and/or structure and/or maintenance in the species. It has always seemed apparent that the degree of design-likeness of organisms' traits cannot be accidental and that, for example, the eyes must somehow be the way they are because they enable us to see. The challenge has been to provide a scientifically adequate understanding of this fact. Replacing “final cause” and theistic accounts, Darwin's theory of natural selection offers the only scientifically adequate explanation we have of how such nonaccidental effects can exist and can explain the traits that cause them; literally, the effects must be cited in an explanation of the cause. Thus, those analyses of function, known as “etiological” or “historical” analyses, that analyze the strong sense of function by appealing in one way or another to natural selection, best address the traditional conundrum of effects explaining their causes.
The analysis of the concept of biological function is made considerably more difficult by the fact that the term “function” is used in a great variety of ways, most of which have nothing to do with biological functions in the strong sense but which are often confused with this concept. We often metaphorically extend function talk to just about any cause that contributes to any salient or valued outcome, as if it were designed to do so, as in “heart sounds function to alert doctors to medical problems,” or “gravity functions to hold the solar system together.” Moreover, there are a variety of colloquial uses that are essentially value judgments, such as when one says that one has a functional or dysfunctional marriage or work situation. None of these uses are explanatory. Some writers on function mistake these metaphorical extensions for the real thing, hopelessly confusing the account of biological functions. Rather than taking the time to disentangle all these variant uses, I simply confine myself to the “strong” use noted above. Note that one way the strong sense is often marked is by the locution “the function of X is to Y,” as opposed to locutions such as “X functioned to Y,” which are often used in the weak sense. So, for example, the fact that “the pocket Bible functioned to stop a bullet” does not imply that “the function of the pocket Bibles is to stop bullets.”
A watershed in the philosophical analysis of function occurred with Larry Wright's (1973, 1976) “etiological” analysis, from which a variety of competing descendants have sprung. Wright's analysis relies on the insights noted above regarding the effect-explanatory nature of artifact functions. Wright asserted, as a general account of “function,” the following:
The function of X is Z means
X is there because it does Z,
Z is a consequence (or result) of X's being there. (1976, p. 81)
The basic idea is that functions refer to explanatory effects, that is, effects of an entity that explain why that entity exists or is present. As an approach to biological function, Wright's analysis suffered from some technical flaws. One cannot say that a specific instance of a mechanism X is there because it does Z, because that implies backward causation; rather, X is an instance of a type of mechanism and is there because past instances of that type did Z. Similarly, one cannot say that the function Z is a consequence of X's being there, because X could be defective, damaged, or diseased and thus malfunctioning; rather, Z was the consequence of past instances of Xs.
A more fundamental problem was that Wright's claim that sheer effect-explanation is sufficient for the existence of a function cannot be correct. Such effect-explanations are everywhere; for example, a rock resting against another rock is in a state of pressure equilibrium in which its position is explained by its effect in pressing against the other rock in a way that causes the other rock to press back with equal force. Another common example is that, in certain meteorological phenomena, there arise feedback-loop systems that sustain themselves by their own effects; for example, whirlpools and storm systems often move water or air in a circular path that forms a set of pressures that cause continued stable movement in the same way. Yet it is not the function of the whirlpool's water movement or the storm's air movement to maintain the whirlpool or storm.
More problematically for the analysis of “biological function,” these examples can have analogs within the organism. For example, near the heart valves, the turbulent blood flow may regularly form small whirlpools of blood with effect-explanatory structures that cause stable continuation of the whirlpool pattern. Yet the existence of such a stable vortex might be merely a mechanical oddity with no fitness implications, thus no functions.
The reaction to the failures of Wright's analysis was to conclude that the analysis simply was not specific enough, and that biological functions, at least, must involve selected effects, not merely any explanatory effects; the reason the effects are explanatory must be that they were selected. Ruth Millikan (1984) attempted to address this problem by building into the analysis of “function” abstract analogs of the critical features of natural selection.
It is generally held that natural selection in the general sense occurs if and only if four conditions are met: (1) reproduction, in which a family of entities is such that one generation gives rise to another of their kind (“offspring”); (2) variation among the traits of the members of the population; (3) inheritable traits, so that offspring tend to be like their “parents”; and (4) differential reproductive success, in which different variants leave different numbers of offspring. Or, as Hull (1990) has characterized natural selective processes, they consist of the activities of replicators, “an entity that passes on its structure largely intact in successive replications,” and interactors, “an entity that interacts as a cohesive whole with its environment in such a way that this interaction causes replication to be differential” (p. 96). Millikan (1984) builds all these standard conditions into her analysis. She requires a “reproductively established family” in which new members are produced by some kind of copying procedure, and a selection process that explains changed proportions of family members bearing given traits over time by greater reproductive success in the copying process due to the possession of the trait.
However, Bedau (1993), borrowing an example from Dawkins (1986), describes a process occurring in inorganic clay silicates in which chemical processes mimic all the elements of natural selection captured in Millikan's analysis, yet no function attributions seem warranted. Moreover, there are many structures within organisms that are “selected” in Millikan's sense but appear not to have functions. For example, parasitic DNA builds linkages to other genes such that it replicates when the others do, yet its linkage-building, though an effect that via selection explains the parasitic DNA's presence, has no organismic biological function. A further persuasive example is that of the segregation distorter gene, which has special mechanisms that manipulate the cell division (meiosis) which produces eggs and sperm (gametes) in a way that gets the gene into more than its usual share of half of the gametes. But, as Godfrey-Smith (1999b) explains, “disrupting meiosis is something that segregation distorter genes do, that explains their survival.…Further, this explanation appeals to natural selection, at the gametic level.…Disrupting meiosis is not generally claimed to be the genes' function though” (p. 204).
Segregation distorter genes manage to increase their representation in future generations and are thus “selected,” but via a causal pathway that does not increase the individual's fitness. Consequently, these genes are not considered to have biological functions for the individual. Functions need not be at the organismic level, and some traits of parasitic DNA and segregation distorter genes possess specifiable biological functions relative to these genes (e.g., certain segregation distorter gene features have the function of causing sperm carrying the rival chromosome to self-destruct, thus increasing the frequency of the segregation distorter gene). But the segregation distorter genes themselves, although selected in Millikan's abstract sense, possess no biological function within the organism because they do not contribute to the organism's greater selective advantage. Consequently, if something goes wrong with a segregation distorter gene's mechanisms, that is not in itself a dysfunction in the sense that yields disorder attributions.
Thus, sheer selection of a feature in Millikan's abstract sense is insufficient for function. The selection must occur because the feature contributes to the organism's fitness (Brandon, 1990; Godfrey-Smith, 1999b). Neander (1991) puts the point this way:
It is the/a proper function of an item (X) of an organism (O) to do that which items of X's type did to contribute to the inclusive fitness of O's ancestors, and which caused the genotype, of which X is the phenotypic expression, to be selected by natural selection. (p. 174)
Note that Neander's and other etiological analyses of “function” cite not simply the trait's fitness value but the causal contribution that the trait makes to fitness. For example, co-occurring traits, such as the weight and warmth of a polar bear's coat, have the same fitness values, but of course it is the warmth of the coat and not its weight that contributes to fitness. Crucial here is Sober's (1984) now-classic distinction between “selection for” versus “selection of” a trait. For example, imagine a machine that separates balls by sifting them for size through various-sized holes and selects for retention only one size that does not fit through any of the holes. The machine may also happen to separate the balls by color and retain only one color, if size and color correlate. In such a case, the machine's selection process results in the “selection of” one color, but the machine's process “selects for” size because the latter property is the one that has the causal impact on the selection process. Similarly, it is the warmth of the polar bear's coat that is selected for, even though there is selection of weight. It is only those features that are selected for, and thus have a causal impact, that are considered relevant to the etiological account of function.
Neander's phrase that a trait's function caused it “to be selected by natural selection” seems to refer to the original selection of the trait, during which the trait initially spread through and became stabilized in the population. However, natural selection also works to maintain traits once they are selected. Without continued selective force acting to preserve a trait and eliminate alternatives, there generally would be eventual erosion of the trait in the population. It is tempting to assume that selective forces stay roughly constant so original selective forces are more or less identical to maintaining forces. However, recent developments in evolutionary theory, especially Gould's influential criticisms of functional reasoning, have underscored the potential complexity of the history of natural selection of a trait and focused attention on divergences between original and maintaining selection.
Problems in addressing maintenance in accounts of function date back at least to Williams's (1966) seminal analysis, in which he distinguished the function of a mechanism from its other effects using an etiological analysis: “The designation of something as the means or mechanism for a certain goal or function or purpose will imply that the machinery involved was fashioned by selection for the goal attributed to it. When I do not believe that such a relationship exists I will avoid such terms and use words appropriate to fortuitous relationships such as cause and effect” (p. 9). Williams intended to exclude from the “function” category any benefits not specifically the product of design for that benefit: “One should never imply that an effect is a function unless he can show that it is produced by design and not by happenstance” (Williams, 1966, p. 261). Williams's language (e.g., “fashioned by selection for the goal attributed to it,” “produced by design”) is most plausibly interpreted as referring to the original selection of a trait. Williams was quite aware that in the search for functions what one initially sees in a population is the ongoing maintenance of design-like selected traits, and the “next task would be to explain why the mechanism in question is maintained as a normal characteristic of the species and not allowed to degenerate” (1966, p. 259). Nonetheless, his definition of function appears to presuppose that original and maintaining selective forces must be the same for a function attribution, because only then can one infer from current maintenance why a trait was originally fashioned.
Gould (1991; Gould & Vrba, 1982) coined the term “exaptation” for novel effects of a trait that increase fitness and thus maintain the trait but were not involved in the trait's original selection. Exploiting the flaw in Williams's analysis, Gould argued that exaptations are neither functions nor explained by natural selection because the feature in question was not originally fashioned for its current purpose. He thus constructed a critique of explanation by natural selection squarely on Williams's definition. Indeed, Williams's definition continues to be cited as justification for Gould's claims about exaptation and function: “By convention (see Williams [1966] for a brief history), the term ‘function’ applies to the beneficial effect that explains the alteration of a trait through positive selection, a usage that Gould and Vrba (1982) merely adopted” (Andrews, Gangestad, & Matthews, 2002, p. 539; emphasis in original).
Thus, in trying to clarify the definition of “function,” Williams inadvertently provided ammunition for an attack on evolutionary functional explanation. The snowballing confusion offers a cautionary tale about the importance of getting clear about conceptual issues. Williams (personal communication, 1995) agreed that changing his definition to include maintenance was necessary, and in later years emphasized in talks the necessity of including maintenance in the definition of function (Randolph Nesse, personal communication, 2002).
Gould is correct that the benefits of a trait can change over time; thus, one cannot blithely equate the current benefit with the original benefit for which the trait was naturally selected. For example, although penguins' wings were selected in nonaquatic ancestors for the function of enabling flight, in penguins' more recent evolutionary history the wings have been maintained exclusively as swimming appendages for enabling penguins to propel and steer themselves in the water, and not for flight, so the fitness-enhancing effects of wings have changed. However, contrary to both Williams's definition and Gould's views, “function” attributions do not require original selection for the claimed function but only selective pressure, whether original or maintaining. Natural selection and functions exist whenever there is maintaining selection because maintaining selection offers genuine causal explanation by selected effects of the (continued) existence of the trait. Consequently, it is considered that the function of penguins' wings is swimming, not flight. Maintaining selection is just as much “natural selection” as original selection; thus, so-called exaptations are simply one type of naturally selected function.
Consider, for example, a species of moth in which white coloration has been originally selected for its effect of camouflage against white bark. Imagine further that due to habitat destruction the species migrates to a new forest in which the bark is dark, yet over generations it maintains its white coloration due to a novel selective advantage in the new environment, namely, mimicry of a toxic white species of moth that inhabits the same forest and is avoided by predators. In the first generation in which the moths arrive in their new habitat, it is a lucky accident that their whiteness has the novel benefit of mimicry; however, after generations of maintaining selection for mimicry and no maintaining selection for camouflage, would anyone resist labeling mimicry as the naturally selected function of the moths' coloration? Original selection with alteration of a trait is not a necessary condition for natural selection or function. Consequently, the entire notion of exaptation as Gould deployed it is of questionable usefulness, other than as a marker for changes in selective pressures and thus in function (Wakefield, 1999b).
A further challenge for natural selection accounts of function concerns exactly when natural selection must have taken place to warrant a current attribution of function. Not any selection, no matter how transient or remote in time, qualifies. Obviously, in prototypical cases in which a trait is originally designed and continuously maintained through to the present for the same effect, timeline issues do not arise. However, especially in light of Gould's arguments, evolutionary thinkers have become sensitive to the ways in which selective pressures for the same trait can vary over time. Questions about the evolutionary timeline can enter into disputes about what is normal versus pathological and so are not purely theoretical issues.
It must be kept in mind that to be a function, an effect has to explain the current presence of the trait via its effect on fitness. To explain the current presence of a trait, one must cite forces operating earlier than the present that led to the present trait. Neither initial shaping per se nor current selective pressure per se provide such an explanation. Current pressures do not suffice because, for example, there can be sudden changes in selective pressures due to alterations in social circumstances (e.g., oppression of a minority) that do not imply functions because they are too brief to influence the presence of distinguishing features. In attributing a function, we must be assuming that there has been some recent period of explanatory selective pressures that may or may not be the same as original or current pressures and that explain the current trait.
How, then, do changing selective pressures over the history of the trait influence current function? The answer, according to recent proposals, is that what is relevant is relatively recent evolutionary periods of time leading up to the present in which significant selection did take place or might have taken place. This analysis has come to be known as the “modern history” etiological view (Godfrey-Smith, 1999a; Griffiths, 1999): The function of X is Y only if selection for Y has been responsible for maintaining X in the recent past (see Kitcher, 1999, for a discussion of this and other timeline options).
Obviously, this leaves unresolved exactly what is “recent.” Griffiths (1999) attempts to explicitly define “recent” in terms of periods during which regressive evolution might be expected to take place:
An evolutionarily significant time period for trait T is a period such that, given the mutation rate at the loci controlling T, and the population size, we would expect sufficient variants for T to have occurred to allow significant regressive evolution if the trait was making no contribution to fitness. A trait is a vestige relative to some past function F if it has not contributed to fitness by performing F for an evolutionarily significant period. (p. 155)
An interesting feature of this definition is that it recognizes that for a variety of reasons actual variation in a trait, and thus actual selective processes, may not occur during an evolutionarily relevant period. The analysis allows for judgments of function based on what we believe would have happened (“we would expect”) if a trait had sufficiently varied during such a period. The assumption that in principle all traits could vary, and they could do so independently even of other traits to which they may be currently linked, is an idealization that seems to be assumed in some function judgments we make in certain kinds of cases in which actual selection has not occurred. For example, a gene that has two vital effects will be judged to have both as its functions. Yet one role might occur during early development, so that actual variations in the gene may always be selected out due to the loss of this developmentally early and critical role, and never because of the developmentally later but equally vital role, simply because variations never allow the opportunity for the organism to develop to the point where there can be selection against the gene based on the later phenotypic variation. In such cases, based on the idealizing assumption that the roles could be independently controlled in principle and that such variations would occur over adequate time spans, we judge that the second role is a function, despite lack of any actual process of natural selection.
Williams (1966) notes that many function attributions can be confidently made independently of any particular theory and without reference to the history of the organism, on the basis of careful observation of design-like adaptive qualities. These judgments would be the same whether the observer is Darwinian, Lamarckian, creationist, or Aristotelian, and thus are entirely independent of evolutionary theory.
Thus, Williams really presents two concepts of function: the intuitive widely shared concept defined by direct observation of design-likeness, and the concept defined by the theory of natural selection. The first concept has existed since antiquity. For example, without knowing anything about evolution but just working within the intuitive concept of function, Harvey discovered that the function of the heart is to pump the blood. Presumably the concept of biological function transfers some basic understanding from the sphere of human-designed artifacts to the biological realm, but this does not explain what the basic common feature is that allows such a transfer of conceptual structure because many theorists from earliest times assumed no actual intelligent designer of biological features and their functions. Presumably this basic notion common to artifact and biological explanation is that the effect of a feature must be invoked in explaining why the feature that has the effect is present. Both artifact and biological function explanations have this unique explanatory structure.
The second concept represents the modern scientific account of the essential nature of functions. Analogously, “water” has a traditional intuitive meaning of “the stuff in the rivers and lakes,” but we now have a scientific theory of the essential nature of that stuff, namely, molecular structure H2O. Indeed, for clarity, it might be better not to call the Darwinian account a concept of function at all, but rather a theory of what biological functions (in the intuitive sense) essentially are and how they came to exist. If the meaning of “function” were “effect responsible for the natural selection of a trait,” then it would be an empty tautology that the functions of biological traits are naturally selected effects. However, the reality is that Darwin's discovery that natural selection explains why traits have their traditionally understood functions is a momentous scientific discovery, not a tautology.
What, then, is the intuitive concept of biological function that is shared by Darwinians and non-Darwinians and provides the target of Darwin's analysis? The formation of the concept starts from obvious facts. Organisms are unique in the way their features are complexly related, hierarchically organized, strikingly beneficial for survival and reproduction, and produce a viable overall pattern of life. There is no other natural context where causal processes are so seemingly miraculously interwoven. It is not difficult to imagine how a storm or whirlpool might accidentally occur and, via causal feedback processes, remain stable; it is much more difficult to imagine how organisms' design-like traits might have come about, and this cries out for explanation. However, the specific nature of the process long remained unknown. The concept “biological function,” I suggest, was a placeholder to refer to the hypothesized unique kind of biological process, whatever they are, that explained apparent functions.
Based on philosophers' insights into natural kind concepts (Putnam, 1975; Searle, 1983) as well as psychologists' insights into essence placeholders (Medin & Ortony, 1989), I call this kind of concept a black box essentialist concept (Wakefield, 1999a, 2000a, 2004). Such concepts postulate and allow one to talk about a hidden unknown “essence”—that is, an underlying theoretical process or structure—that explains some initial prototypical set of phenomena. The concept remains agnostic on the specific identity of the underlying essence until scientific research provides an answer. In the case of “function,” the prototypical instances would consist of such clear explanatory-effect functions as eyes seeing, hands grasping, feet walking, teeth chewing, fearing danger, thirsting for needed water, and so on.
According to the black-box-essentialist approach, having observed that prototypical biological functions clearly involve effects that must themselves explain the presence of the mechanisms that give rise to them, but not understanding how such a thing was possible, early biologists defined function as encompassing any effect of a trait that explains the trait's presence by the same underlying process, whatever it is, by which the prototypical instances of function explain the traits of which they are effects. Of course, theorists have differed greatly over the nature of the hypothesized underlying process that explains the observed clear cases of design-likeness. It was Darwin's explanation in terms of natural selection that succeeded in providing the needed explanation and thus provided a scientific theory of function.
Evolutionary psychopathology is a growing subdiscipline with many diverse strands. First, evolutionary psychopathologists put forward specific evolutionary hypotheses about naturally selected mental modules and their normal functions, and offer theories of the dysfunctions of those mental modules that may underlie specific mental disorders; for example, evolutionary hypotheses suggest the ways in which normal emotional reactions may malfunction to yield panic attacks (Klein, 1993; McNally, 1994) or pathological depression (Nesse, 1991). Second, evolutionary psychopathologists attempt to explain how specific debilitating mental disorders have continued to exist in the population despite presumed selective pressures against them, by showing how they might be the direct or indirect results of selective pressures. For example, certain personality disorders may be due to unselected extremes on dimensional traits. Third, evolutionary theorists use an evolutionary framework to distinguish psychopathology from other problematic conditions that might be subject to treatment by mental health professionals (Cosmides & Tooby, 1999), including excessive designed defensive reactions (as in high fever or sadness) and mismatches between naturally selected mechanisms and the current environment (e.g., the appetite for sugar and fat in our food-rich environment).
I focus here on a further and more fundamental issue for clinical psychology, namely, clarifying the concept of mental disorder itself. A mental disorder may be considered a disorder of mental mechanisms and thus is conceptually analogous to disorders of other kinds of mechanisms. Thus, the problem is to define “disorder” in the general sense used in medicine and then apply it to the domain of mental mechanisms. It is here, at the foundation of clinical psychology, that evolutionary psychopathology is crucial and determines an agenda that heavily involves evolutionary considerations in psychopathology research.
The view to be considered here is the harmful dysfunction (HD) analysis of the concept of mental disorder (Wakefield, 1992a, 1992b, 1993, 1996, 1997, 1999b, 2000b, 2006; Wakefield & First, 2003), which asserts that a mental or physical disorder must be (a) harmful, that is, negative as judged by social values, and (b) caused by a dysfunction, that is, by failure of a psychological mechanism to perform its function, in the sense of “biological function” as analyzed earlier in this chapter. This concept is arguably at the root of both psychiatric and lay judgments of disorder versus nondisorder. Dysfunction and function in the relevant sense are theoretically best understood in evolutionary terms, and thus in principle are factual scientific concepts. Thus, disorder in the medical sense is a hybrid value and factual concept. Because dysfunction is a necessary requirement for disorder according to the HD analysis, this offers a way to critique standard diagnostic criteria in the DSM-5 (American Psychiatric Association [APA], 2013) when they are applied to problems of living that are not true disorders, while still accepting the reality of mental disorders as genuine medical disorders and thus rejecting the anti-psychiatric claim that mental disorder does not exist (Szasz, 1974). I focus here on exploring the evolutionary dysfunction component.
The view that the concept of disorder somehow involves dysfunction emerges with remarkable consistency in the remarks of many authors who otherwise differ in their views (e.g., Ausubel, 1971; Boorse, 1975; Kendell, 1975, 1986; Klein, 1978). Spitzer and Endicott (1978) note the seeming necessity and virtual universality of using “dysfunction” to make sense of “disorder”: “Our approach makes explicit an underlying assumption that is present in all discussions of disease or disorder, i.e., the concept of organismic dysfunction” (p. 37). The DSM-5 (2013) also specifies that a disorder exists only when symptoms are due to an underlying dysfunction: “A mental disorder is a syndrome characterized by clinically significant disturbance in an individual's cognition, emotion regulation, or behavior that reflects a dysfunction in the psychological, biological, or developmental processes underlying mental functioning” (p. 20). To understand “disorder,” one requires an adequate analysis of “function” and “dysfunction.”
Supposing that a disorder is a dysfunction, what, then, is a dysfunction? Dysfunction implies unfulfilled function, thus failure of some mechanism in the organism to be able to perform its function. However, not all uses of “function” and “dysfunction” are relevant to disorder judgments. The medically relevant sense of “dysfunction” is clearly not the colloquial sense in which the term refers to failure of an individual to perform well in a social role or in a given environment, as in assertions like “I'm in a dysfunctional relationship” or “discomfort with hierarchical power structures is dysfunctional in today's corporate environment.” These kinds of problems need not be individual disorders. Moreover, the kinds of functions that are relevant are not those that result from social or personal decisions to use a part of the mind or body in a certain way. For example, the nose functions to hold up the glasses, and the sound of the heart performs a useful function in medical diagnosis. Nonetheless, a person whose nose is shaped in such a way that it does not properly support glasses does not thereby have a nasal disorder, and a person whose heart does not make the usual sounds clearly enough to be useful for diagnosis is not thereby suffering from a cardiac disorder. A disorder is different from a failure to function in a socially or personally preferred manner precisely because a dysfunction exists only when a feature cannot perform its naturally (i.e., independently of human intentions) designed function. The functions that are relevant to disorder attributions are precisely the “natural” or “biological” functions analyzed above.
Thus, disorders involve dysfunctions, and dysfunctions are failures of organismic mechanisms to perform their biologically designed functions. More strictly, to eliminate possible counterexamples in which normal organisms cannot perform their functions because the environment does not allow them to (in which case they are not generally considered disordered), dysfunctions are failures of mechanisms to be capable of performing their functions under environmental circumstances for which the mechanisms were designed to perform such functions. The disruption of the biologically designed “order” is why there is a “disorder,” according to this view.
One might object that sometimes, for example in reading disorders, what goes wrong is a social function that has nothing to do with naturally selected categories. However, even though we are not biologically designed to read, inability to read is only considered indicative of disorder when evidence suggests the cause is a failure of some brain mechanism to perform its natural function.
If one looks down the list of DSM-5 (APA, 2013) disorders, it is apparent that it is a list of the various ways that something can go wrong with the seemingly designed features of the mind. Roughly, psychotic disorders involve failures of thought processes to work as designed, anxiety disorders involve failures of anxiety- and fear-generating mechanisms, depressive disorders involve failures of sadness and loss-response regulating mechanisms, disruptive behavior disorders of children involve failures of socialization processes and processes underlying conscience and social cooperation, sleep disorders involve failure of sleep regulation, sexual dysfunctions involve failures of various mechanism involved in sexual motivation and response, eating disorders involve failures of appetitive mechanisms, and so on. There is also a certain amount of nonsense in the DSM-5. However, in the vast majority of categories, a good case can be made that the category is inspired by conditions that even a layperson would correctly recognize as a failure of designed functioning.
When we distinguish normal grief from pathological depression, or normal delinquent behavior from conduct disorder, or normal criminality from antisocial personality disorder, or illiteracy from reading disorder, or normal lack of empathy for enemies of one's group from sociopathic lack of empathy for anyone, or normal childhood rambunctiousness from attention-deficit/hyperactivity disorder, we are implicitly using the “failure-of-designed-function” criterion. That criterion explains why some of these conditions are considered disorders and others that are quite similar and also negatively evaluated are not.
The HD analysis holds that the intuitive concept of “disorder” requires dysfunction, and dysfunction occurs when an internal mechanism is incapable of (or impaired in) performing one of its natural functions. Until this point in the analysis, natural function is used in the intuitive black-box essentialist sense that has existed for millennia, not in a technical evolutionary sense. The further evolutionary theoretical argument applied in the case of “function” applies to “dysfunction” as well. Given that evolutionary theory explains natural functions, disorders are harmful failures of mechanisms to perform functions for which they were naturally selected.
Thus, evolutionary psychology and the field of psychopathology converge. Indeed, evolutionary psychopathology becomes the discipline of psychopathology. The DSM-5 can be seen as an attempt to list categories that prima facie involve failures of naturally selected mental mechanisms to perform their functions, whether they concern thought, emotion, sexual functions, sleep functions, socialization and moral development functions, and so on.
The HD analysis implies that a society's categories of mental disorder offer two pieces of information. First, they indicate a value judgment that the society considers the condition negative or harmful. Second, they make the factual claim that the harm is due to the mind's failure to work as designed. This claim may be correct or incorrect. The value component cannot be reduced to the evolutionary component because natural selection works ultimately on genes, not people, and sometimes what serves the gene's replication is not what a given culture considers good for the individual, or failure of some function does not matter enough to be labeled a disorder. Natural human functioning and human visions of a good life do not necessarily coincide. Moreover, normal human defensive responses are often biologically designed to over-respond to potential signals of threat and thus confer needless but normal distress for the same reason that smoke detectors often respond to harmless conditions such as the smoke coming from cooking fish, namely, because the cost of missing a true danger is so much greater than the cost of some needless responses to false signals.
The HD analysis is often the framework for arguments that conditions that are considered disorders are in fact naturally selected variants and thus not disorders, contrary to DSM-5 (Nesse & Stein, 2012). For example, intense normal sadness can easily be mistaken for depressive disorder (Horwitz & Wakefield, 2007; Wakefield, 2013); intense anxiety over evolutionarily shaped fears can be mistaken for anxiety disorders (Horwitz & Wakefield, 2012; Stein & Nesse, 2011); adolescent rambunctious behavior can be mistaken for conduct disorder (Wakefield, Pottick, & Kirk, 2002); and so on. The HD analysis has been applied to a great range of diagnostic categories in debates over their diagnostic status as disorders or nondisorders, including hebephilia (the proposed disorder of sexual desire for young teenagers; Rind & Yuill, 2012), psychopathy (which has been argued to be a normal variant; Krupp, Sewall, Lalumiere, Sheriff, & Harris, 2012), and substance dependence (Martin, Chung, & Langenbucher, 2008; Vergés, Steinley, Trull, & Sher, 2010; Wakefield & Schmitz, 2014).
The HD analysis is in tension with arguments in evolutionary psychopathology that attempt to explain apparent disorders (e.g., schizophrenia, major depression) as naturally selected. The HD analysis implies that such claims that disorders are naturally selected are not merely false but incoherent. A disorder is a failure of a natural function and thus cannot itself be the function of a naturally selected trait, according to the HD analysis. A major source of confusion in attempts to explain how disorders were naturally selected is that current disorder categories are not adequately founded on an evolutionary understanding of normality (Nesse & Stein, 2012) and thus tend to run together normal distress and true pathology.
Proponents of natural-selection explanations of disorders generally attempt to explain away the apparent paradox of a seemingly debilitating disorder being naturally selected in one of two ways: either the disorder has hidden fitness benefits that offset its seeming disadvantages, or the disorder increased fitness in past environments but has become problematic in the current environment. However, if the apparent negative effects of a condition are just the evolutionary price paid for the condition's positive effects, then the condition should not be labeled a disorder. For example, the immense pain associated with childbirth is not judged a disorder because birth is obviously a designed process. Similarly, mismatches between human nature and our changing social environment are not necessarily disorders. For example, sexual desire for individuals other than one's spouse and desire for sugar and fat are aspects of human nature that are mismatched to our monogamous and food-rich environment and are thus often harmful, but neither is a disorder.
In advancing explanations of why seeming disorders are in fact naturally selected, evolutionary psychologists often ignore the more plausible hypothesis that the mechanisms underlying the condition are malfunctioning, yielding a condition that was never selected for. There are many pitfalls that can lead a theoretician astray when making claims that disorders have been naturally selected.
First, an evolutionary psychologist might rely on standard DSM-5 (APA, 2013) diagnostic categories and criteria as the targets of the analysis, without independently assessing the conceptual validity of the criteria. Thus, what are claimed to be natural-selection explanations of disorders in fact may be explanations of nondisorders. Ample evidence suggests that DSM-5 criteria often mistakenly diagnose nondisorders as disorders. For example, DSM-5 symptom-based criteria likely erroneously classify many normal states of sadness and anxiety as disorders (Horwitz & Wakefield, 2007, 2012). Nondisordered states of sadness might indeed be subject to evolutionary explanation in terms of, say, retreat during periods of threat, that erroneously is thought to explain depressive disorder. However, true depressive disorders that chronically debilitate cannot be explained as fitness enhancing in this way and are better explained as malfunctions in sadness-generating mechanisms. Because of the invalidity of DSM-5 diagnostic criteria, the theorist may not adequately distinguish cases within a category that are disorders from other relatively mild and possibly adaptive versions of the same condition.
Second, DSM-5 criteria aside, an evolutionary psychologist might unreflectively assume that any harmful condition that is treated or treatable can be considered a disorder. Again, this can lead to the theoretician evolutionarily explaining conditions that he or she labels disorders but are not true disorders. Obviously, there are many life problems that are treated but are not disorders. For example, normal shortness is sometimes treated with growth hormones, and normal grief with antidepressants, and birth control pills and abortion treat normal reproductive conditions.
Third, an evolutionary psychologist might embrace a natural-selection explanation of a condition that is an apparent disorder, and overlook a more plausible indirect evolutionary explanation of why the condition has been retained in the population (e.g., retention of a homozygous condition such as sickle cell anemia due to selective advantage of heterozygous condition). However, the indirect explanation may imply that the condition has no function at the organismic level.
Finally, evolutionary analysis may succeed in explaining a condition currently labeled a disorder, and the theorist may erroneously assume that the explanation itself has no implications for the condition's classification. That is, the theorist may fail to appreciate that demonstrating that a condition is part of human design inevitably casts doubt on the condition's disorder classification and triggers its reclassification as normal. This is exactly what happened, for example, with fever, which was once considered a pathology involving a breakdown in temperature regulation, but was reclassified as normal when it turned out that fever is a highly regulated process designed to aid recovery from illness. Treatment of fever is still undertaken because this problematic defense often is not needed for recovery, although it remains possible that reduction of fever may have some negative consequences (Earn, Andrews, & Bolker, 2014). Such defensive reactions are not considered disorders once they are recognized for what they are.
Conceptual analysis leads to the conclusion that at the roots of “function,” “dysfunction,” and “disorder” are notions of design and failure of designed functioning. These notions in turn are built on the idea of effects that explain why the mechanisms that produce them are there, and, in the case of “dysfunction,” on the failure of such mechanisms to be capable of producing such explanatory effects. Darwin's scientific discoveries revealed that the only plausible way to understand such explanatory effects in biology is in terms of natural selection. Consequently, judgments of psychological normality and disorder are in fact judgments about evolutionary design. Whether the DSM-5 diagnostic criteria used by clinicians and researchers every day are valid, and how to make them more valid, hangs on our understanding of human mental design and on progress in evolutionary psychology.
Fortunately, plausible judgments about design and failure of design often can be made without knowledge of evolutionary history, based on immediate evidence regarding design-like properties. Current DSM-5 categories, which pick out some categories of disorder reasonably well, testify to this fact. So, clinical psychology need not come to a halt while awaiting evolutionary psychology's progress. But in the long run, the DSM-5 must attend to evidence of dysfunction and work towards a functional understanding of mental mechanisms (First & Wakefield, 2013; Nesse & Stein, 2012). The scientific progress of the mental health professions in understanding the etiology, diagnosis, and treatment of mental disorder may well depend partly on progress in evolutionary psychology.
However, there is the potential for intense controversy about function judgments, as issues of what is human nature become politicized. Moreover, judgments of dysfunction and disorder are often highly controversial due to the stakes of different constituencies in pathological versus normal classification of certain conditions. The intersection of these two intensely controversial areas promises that those in the field of evolutionary psychopathology will continue to find themselves living in interesting times.