ings did he suggest that the per- function by moderating, through a variety of REFERENCES formed by an needs to be biologi- themes, our evolutionarily functional fear of . Thus, recognition of our own mortali- Buss, D. M., Haselton, M. G., Shackelford, T. cal. As Buss et al. reported, Gould (1991) K., Bleske, A. L., & Wakefield, J. C. (1998). offered two definitions of exaptation: "a fea- ty is a spandrel, and the practice of religion is , , and spandrels. ture, now useful to an organism, that did not an exaptation. American Psychologist, 53, 533-548. arise as an for its present role, but Moreover, it is a priori unlikely that Gould, S. J. (1987). Freudian slip. Natural was subsequently coopted for its current func- such complex psychological phenomena can History, 96, 14-21. tion" (p. 43) and "features that now enhance meet the standards of evidence proposed by Gould, S. J. (1991). Exaptation: A crucial tool fitness, but were not built by Buss et al. (1998). Specifically, they suggest- for an . Journal of for their current role" (p. 47). Thus, although ed that evidence of special design for a hy- Social Issues, 47, 43-65. exaptations always provide a current func- Gould, S. J. (1997). The exaptive excellence pothesized function be demonstrated before of spandrels as a term and prototype. Pro- tion, that function need not be biological. In concluding that any structure or behavior is fact, Gould's (1991) thesis is geared toward ceedings of the National Academy of Sci- adaptive. As an example, they presented the ences, 94, 10750-10755. exapted functions performed by the brain to hypothesis that female serves the serve psychological rather than biological adaptive function of facilitating sperm trans- functions. port, for which evidence is reportedly lack- Correspondence concerning this comment Additionally, Buss et al. (1998) sug- ing. It is thus concluded that female orgasm should be addressed to Theodore Beauchaine, Department of Psychology, State University gested that Gould (1991) used the term exap- does not serve the hypothesized adaptive tation "to cover novel but functionless uses of New York, Stony Brook, NY 11794- function. Although instructive, this example 2500. Electronic mail may be sent to or consequences of existing characteristics" trivializes the difficulty of falsifying, at the tbeaucha@psych 1 .psy.sunysb.edu. (p. 539). My reading of Gould (1987,1991, phenotypic level, hypotheses about exapted 1997) suggests otherwise: Structures or char- psychological functions, such as the practice acteristics for which there are no identifiable of religion. There are at least two reasons for functions, either current or historical, are al- this. First, as outlined by Buss et al., exapta- ways spandrels. Thus, the term spandrel sub- tions "carry the additional evidentiary burdens Noise, Logic, and the sumes what Buss et al. referred to as func- tionless by-products. When a spandrel is of documenting both later co-opted functional- Span of Time coopted (exapted) for a function, it becomes ity, and a distinctive original adaptational an exaptation; there is no need to differentiate functionality" (p. 546). Thus, to confirm Lucian Gideon Conway III between spandrels and functionless by- the exapted function of religious practice, University of British Columbia products. Ironically, Gould (1991) coined one would be required to demonstrate (a) the terms exaptation and spandrel to avoid that enlarged brains were naturally selected Buss, Haselton, Shackelford, Bleske, and this sort of confusion. for reasons independent of religious practice, Wakefield (1998) should be commended for (b) that such brain enlargement resulted in the a thoughtful article that did the important Additionally, although Buss et al. (1998) capacity to practice religion, and (c) that reli- sought to outline the usefulness of the terms work of clarifying and elaborating the theo- gious practices function to assuage the fear of adaptation, exaptation, and spandrel for the retical terminology of evolutionary psychol- death (which would itself be required to meet science of psychology, most of their discus- ogy. However, I was disappointed in their sion was restricted to structural rather than the evidentiary standards of a spandrel). Al- treatment of an issue that, in my opinion, psychological levels of analysis. Although though the first two of these conjectures may needs to be more fully addressed before evo- understandable given their strictly biological be true, they can at best be confirmed only at lutionary psychology can gain further episte- interpretation of exaptations, this approach is the pseudoempirical level. Gould (1991) rec- mological ground. problematic because the authors sought to ognized this and suggested that the term ap- Buss et al. (1998) noted the three major apply their standards of evidence to the list of tation be applied to cases where a lack of conceptual means through which - exaptations proposed by Gould(1991), name- historical evidence precludes the determina- ary theory accounts for our current psycholo- ly language, religion, principles of commerce, tion of whether a characteristic is an adapta- gy: (a) adaptation, (b) exaptation, and (c) warfare, reading, writing, and the fine arts. tion or an exaptation. Furthermore, confirm- random noise (surviving mechanisms that All of the items on this list are psychological ing or disconfirming the third conjecture is neither aided nor inhibited gene survival). phenomena borne of a structural complexity far more difficult than establishing the utility The question troubling me is this: Might the (i.e., the brain) that is not well under- of female orgasm as a sperm transport mecha- possibility of random noise explanations in- stood and that was not sufficiently consid- nism, because the functional level of analysis hibit the potential acceptance of explanations ered by the authors. Here again, Buss et al. is psychological, not structural. based on adaptations or exaptations? may have misinterpreted Gould, who did not All of the proposed exaptations listed According to the logic of scientific in- use human brain size merely as an "example by Gould (1991), because they are specified vestigation, unless one rules out plausible of an exaptation" (p. 539), as the authors at the psychological level, are similarly pre- alternative theoretical explanations for one's stated. Rather, Gould suggested that the hu- cluded from meeting the strict evidentiary findings, then the proposed theoretical ex- man brain, by virtue of both complexity and standards set forth by Buss et al. (1998). Yet, planation for those findings cannot fully be flexibility, is "the best available case for pre- it would be unfortunate indeed if psycholo- accepted (e.g., Cook & Campbell, 1979). dominant exaptation—in other words, for a Given that evolutionary theory provides log- near certainty that exaptations must greatly gists, in an effort to meet such standards, were to reject Gould's distinctions and con- ical room for random noise, a real difficulty exceed adaptations in number and impor- for evolutionary psychologists is how to rule tance" (p. 55). Gould offered the practice of tinue in adaptationist practices. We should recognize the error in logic of inferring evo- out such noise as an alternative explanation religion as an example and suggested that our for psychological mechanisms that are pro- enlarged brains force us to confront our own lutionary cause from current consequence, posed as adaptations. Buss et al. (1998) sug- mortality. Because it is quite unlikely that whether or not we can empirically demon- gested that "as more and more functional brain enlargement evolved to serve this end, strate the existence of psychological exapta- features suggesting special design are docu- the practice of religion performs an exapted tions.

440 June 1999 • American Psychologist mented ... the alternative hypotheses of for any given psychological mechanism that, tive on human mating. Psychological Re- chance and incidental by-product become in- in lieu of direct facts, are as logically sound as view, 100, 204-232. creasingly improbable" (p. 537). adaptationist explanations. Buss, D. M., Haselton, M. G., Shackelford, T. Special design, however, seems a vague The logical difficulties of evolutionary K., Bleske, A. L., & Wakefield, J. C. (1998). standard highly dependent on intuition. Dem- psychological explanations would be of no Adaptations, exaptations, and spandrels. American Psychologist, 53, 533-548. onstrating special design requires understand- great concern except for this additional trou- Cook, T. D., & Campbell, D. T. (1979J. Quasi- ing exactly how a psychological mechanism bling fact: It seems difficult (at least until a experimentation: Design and analysis is- would have been functional to the survival of time machine is invented) to provide even an sues for field settings. Boston: Houghton the genetic code in the distant reaches of approximate empirical test of whether a par- Mifflin. human history—for it was survival in that ticular psychological mechanism is the result Dawkins, R. (1976). The selfish gene. New history for which the mechanism was pur- of an adaptation, an exaptation, or random York: Oxford University Press. portedly designed. However, it has been point- noise. Gould, S. J. (1991). Exaptation: A crucial tool ed out by other scientists that the task of One distinction should be noted: Evi- for evolutionary psychology. Journal of logically estimating what psychological traits dence for the existence of psychological mech- Social Issues, 47, 43-65. would have promoted genetic survival in a Ornstein, R. (1991). The evolution of con- anisms themselves is not evidence for the sciousness. New York: Prentice Hall. given historical context is an immensely dif- proposed adaptive qualities of those mecha- ficult one—and one that by no means should nisms. Finding mate-preference differences be taken lightly. As (1976) between men and women across 37 cultures, Correspondence concerning this comment cautioned in The Selfish Gene, "it is a very for example, is powerful evidence of biolog- should be addressed to Lucian Gideon complicated business to demonstrate the ef- ical differences between men and women. Conway III, Department of Psychology, Uni- fects of behavior on long-term survival pros- However, how those biological differences versity of British Columbia, 2136 West Mall, pects" (p. 5). got there (in a Darwinian framework) is a Vancouver, British Columbia Canada V6T separate question altogether, and we are real- 1Z4. Electronic mail may be sent to One of the reasons it is difficult to logi- [email protected]. cally argue for one particular type of evolu- ly no further toward answering that question tionary explanation (over another) is that Dar- for having discovered the phenomenon itself. winism is a very flexible framework and Thus, evidence of such biological differences offers a nearly endless variety of ways that certainly implicitly supports the functionality mechanisms could have been adaptations, to psychology of evolutionary perspectives Spandrels in the exaptations, or random noise (see Ornstein, (in that these perspectives generated produc- Consulting Room 1991). To be sure, explanations proposing tive research), but does not support the pro- adaptations or exaptations can be logical, but posed adaptivity of the mechanisms. because there are potentially so many logical- How then does one claim empirical sup- Shawnee L. W. Cuzzillo ly plausible explanations for a given psycho- port for adaptiveness? Consider Gould's The Wright Institute, Berkeley, California logical mechanism within an evolutionary (1991) hypothesis that religion is an exapta- framework, logic alone becomes a hollow tion. Buss et al. (1998) suggested that pro- Instead of embracing the complexity of evo- argument. If all explanations are logical, then viding evidence for this involves demonstrat- lutionary theory, Buss, Haselton, Shackel- logic itself is useless as a means of determin- ing (a) what original adaptation was coopted ford, Bleske, and Wakefield (1998) coopted ing which one of those explanations is the for religion, (b) what caused the coopting, Gould's (see Gould & Vrba, 1982) notion of exaptation for their own purposes. Giving right one. and (c) the new biological adaptiveness of only lip service to Gould's critique, Buss et religion. Buss et al. then alarmingly stated To pick an example from sexual strate- al. ultimately reiterated the need for adaptive that "these predictions can then be subjected gies theory (e.g., Buss & Schmitt, 1993), it is explanations for all aspects of human exist- certainly logical to suggest that a male prefer- to evidentiary standards of empirical testing ence and behavior. Buss et al. criticized Gould ence for young (and thus fertile) women and potential falsification" (p. 542). for not reducing language, religion, and so on would lead to a greater chance of passing on The trouble is, how does one empirical- to clear adaptive stories. They revealed their that male's genes, but random noise explana- ly test hypotheses that are about processes own reductionist agenda in stating that "pa- tions can be equally logical. Consider, as one that occurred millions of years ago? Does rental care—investing in one's own children— example, the following logical progression: even extensive testing on current human be- is merely [italics added] a special case of (a) The preference for young women would ings really get you much closer to the adaptive- caring for kin who carry copies of one's have caused men to get into more fatal fights ness in the distant past of the psychological genes in their bodies" (p. 535). We lose so (over the "rights" to the young women)—an processes discovered now? (As evolutionary much in this description of human experi- activity surely not conducive to the passing psychologists often emphasize, those things ence, both good and bad. What about the on of genes, (b) Such a disadvantage could that were adaptive then are not necessarily the knowledge that we have of our mortality, our potentially offset any advantages of a prefer- things that are adaptive now.) In addition, desire to be connected to a chain of ancestors ence for fertile women, (c) Although not because most psychological mechanisms— and descendants, and our delight in knowing aiding survival, the preference could have within a Darwinian framework—can logical- the stories that connect us to our forebears? survived as random noise, first because sur- ly be noise as easily as they can be adapta- How do the experiences of adoption and egg vival is a complex thing involving thousands tions or exaptations, how, given this state of donation fit into this scheme? On the darker of different trait combinations, and second affairs, does one ever rule out with even side of human experience, the rosy implica- because the random operation of genetics did moderate confidence the probability mat some- tions of kin selection theory deny the reality not provide a truly adaptive mechanism to thing is simply evolutionary noise? of child abuse and neglect, as well as the compete with it. relatively high proportion of homicides com- mitted against kin. The point here is not to argue that ran- REFERENCES dom noise explanations are better than adap- When complex human behavior is in tationist explanations. Rather, the point is that Buss, D. M., & Schmitt, D. P. (1993). Sexual effect reduced to determinants in the genes, one can construct random noise explanations strategies theory: An evolutionary perspec- many important intervening variables are ig-

June 1999 • American Psychologist 441