Evolutionary accounts of deductive reasoning have been enjoying a fair amount of popularity in the last decades. Some of those who have defended views of this kind are Cooper, Maddy, and more recently Joshua Schechter. The basic idea is that an explanation for why we have developed the ability to reason deductively (if indeed we have developed this ability!) is that it conferred a survival advantage to those individuals who possessed it among our ancestors, who in turn were reproductively more successful than those individuals in the ancestral population who did not possess this ability. In other words, deductive reasoning would have arisen as an adaptation in humans (and possibly in non-human animals too, but I will leave this question aside). Attractive though it may seem at first sight (and I confess having had a fair amount of sympathy for it for a while), this approach faces a number of difficulties, and in my opinion is ultimately untenable. (Some readers will not be surprised to hear this, if they recall a previous post where I argue that deductive reasoning is best seen as a cultural product, not as a biological, genetically encoded endowment in humans.)
In this post, I will spell out what I take to be the main flaw of such accounts, namely the fact that they seem incompatible with the empirical evidence on deductive reasoning in human reasoners as produced by experimental psychology. In this sense, these accounts fall prey to the same mistake that plagues many evolutionary accounts of female orgasm, in particular those according to which female orgasm has arisen as an adaptation in the human species. To draw the parallel between the case for deductive reasoning and the case for the female orgasm, I will rely on Elisabeth Lloyd’s fantastic book The Case of the Female Orgasm (which, as it so happens, I had the pleasure of re-reading during my vacation last week).
As spelled out by Lloyd, a number of adaptationist accounts of female orgasm are based on the idea that having orgasms during intercourse would have conferred reproductive advantages to those female members of the population who were lucky enough to have this ability. (What exactly the nature of this reproductive advantage would be varies per author.) And thus, the story goes, human females having this trait would have been selected for over those who did not, which then would eventually have led to the current pervasiveness of female humans having this trait – again, the ability to orgasm during intercourse.
But wait: is this what comes out of the large (though not unproblematic) body of empirical research on sex? Is it really the case that, in the current population, the ability to orgasm during intercourse is widespread? Of course not! Though the numbers vary from study to study, in each of them what emerges is that the number of females in the current human population who systematically and reliably have orgasms during intercourse is very small, in the order of 30% to 10%. So how can anyone claim that a given trait is an adaptation, if it is in fact so scarcely represented in the population in question? As Lloyd puts it:
Orgasm with intercourse is a problematic subset of female orgasm in itself. When explanations are given for female orgasm only during intercourse, they are not addressing the phenomenon they purport to explain: the existence of female orgasm itself. This fact is usually obscured by the erroneous assumption that intercourse ensures orgasm for the female. Most of the evolutionary accounts neglect the orgasm/intercourse discrepancy, and this omission constitutes a fundamental evidential problem, because a basic assumption of their explanation is false. (Lloyd 2006, 105)
In other words, such accounts assume the prevalence of a trait (the ability to orgasm systematically during intercourse) that is simply scarcely present in the female population in question. As such, to describe this trait as an adaptation is a mistake, given that for a trait to be an adaptation it must actually be widely present in a population as a result of natural selection.
Adaptationist accounts of deductive reasoning seem to suffer from the same kind of evidential problem. For deductive reasoning to be an adaptation, first of all it would have to be a trait widely present in the population in question, namely humans. It would be necessary not only that humans have the capacity for deductive reasoning, but also that this capacity be regularly and systematically deployed. Do human reasoners indeed exhibit the trait of successfully performing deductive reasoning, that is the kind of reasoning that respects the criterion of necessary truth-preservation? According to a large body of literature in experimental psychology, they do not, as can be seen from results in well-known reasoning tasks such as the Wason selection task, syllogistic reasoning, conditionals and so forth. Much of the research on the psychology of reasoning since the 1960s has documented a discrepancy between the deductive canons and the performance of participants during reasoning experiments. Moreover, some robust patterns of deviation have been identified, and were then described as reasoning biases. (For a review of these results, see this 2002 survey article by Jonathan Evans, informally known among psychologists of reasoning as the final blow against the deductive paradigm that had dominated research in the field for decades: the “death of deduction”.)
How can an evolutionary account of deductive reasoning as an adaptation even get off the ground given the empirical evidence that human reasoners do not systematically engage in deductive reasoning? Again, to be an adaptation, a trait has to be widely present in the population in question. Joshua Schechter seems vaguely aware of the conflict between his account and these experimental results, but relegates a discussion thereof to footnotes. (Symptomatically, all of the psychological studies Schechter cites date back to the 1990s, and since then much has happened in the empirical study of reasoning.) For example, he writes:
Some philosophers and psychologists have claimed that we do not reason deductively but rather employ only inductive or abductive patterns of reasoning. I find this view implausible, but cannot argue against it here. (Footnote 8)
There are well-known experimental results showing that humans are prone to errors in deductive reasoning. Perhaps the most striking results concerns variants of the Wason selection test. So the claim that we are reliable in our deductive reasoning should not be overstated. But these errors seem largely to be performance errors. The rules built into our deductive competence are reliable. (Footnote 14)
That he finds the view that we humans do not systematically reason deductive ‘implausible’ seems like a rather inefficient argument against a large body of empirical research. He then suggests that this discrepancy still does not affect his position because it is a performancy/competence discrepancy. However, even if human reasoners have the capacity to reason deductively but in fact rarely deploy it, it is still unclear how a capacity that is rarely used could confer an evolutionary advantage over individuals in the ancestral population not having this capacity so as to be selected for. (One could suggest perhaps that our ancestors made much more extensive use of deductive reasoning than we now do, and thus that we still have an ability which was selected for but is now rarely used. But why would such a change have occurred?)
All things considered, I truly do not see how an account of deductive reasoning as an evolutionary adaptation can be reconciled with the experimental observation that human reasoners do not seem very familiar with the canons of deductive reasoning, judging from the discrepancy between these canons and their performance in experiments (which does not mean that human reasoners cannot learn to reason deductively). Just as the evolutionary accounts of female orgasm which focus on orgasm during intercourse attempt to explain a trait scarcely present in the relevant population, an evolutionary account of deductive reasoning that assumes that the trait is widely present in the relevant population seems to make a fundamental evidential mistake. A trait scarcely present in a population cannot be an adaptation.
Recent Comments