Reason and Representation
Reconstructing Reason and Representation brings evolutionary psychology to bear on philosophy of mind and epistemology. Murray Clarke argues for an adaptive, naturalistic, modular construction of reason and representation. Evolution, he declares, based largely on the work of Leda Cosmides and John Tooby, has created viable, though not perfect, systems of inference. Though he offers modifications where appropriate, Clarke's project is largely the application of the Cosmides and Tooby massive modularity hypothesis to naturalist epistemology, thus uniting wrongly disparate disciplines in a broader search for a proper understanding of cognition and knowledge. What Tooby and Cosmides (and thus Clarke) call transtheoretical consistency might be more simply termed interdisciplinary collusion, where the results of science and social science serve as guides and barriers for further inquiry.
The human mind is composed of what all three of these authors term Darwinian modules, which essentially form the binding agents of universal human nature. They have evolved, not surprisingly, through natural selection, though the comparatively slow process of shaping the human mind has left us "adapted to the way of life of Pleistocene hunter-gatherers, not necessarily to our modern circumstances" . These modules are domain-specific computers—"a confederation of hundreds of thousands of functionally dedicated computers" —that use formal operations to respond to individual representations. The working of these Darwinian modules is, for the three, massive modularity, a hypothesis based on the experimentation of empirical psychologists. And it is precisely those psychologists who have been responsible for "the deconstruction of the notion that there is an underlying Russellian psycho-logic that girds thought and language" .
We are, in effect, wildly irrational when it comes to everyday choices. Our actions do not match what philosophy tells us to do. Clarke here uses the example of the Wason tests of the late 1960s and early 1970s, which "demonstrated that reasoning performance on distinct tasks that require the use of a single rule of deductive inference varied as a function of the content plugged into the inference rule" . In other words, people made far better choices when the variables were familiar and real, despite the fact that each Wason test had the same structure, the same logic, and the same conditions for what constituted a "right" answer. Cosmides, Tooby, and Clarke read the phenomena not as a consequence of familiarity, however, but as a consequence of a "social contract." The modern mind is essentially the mind of that Pleistocene hunter-gatherer, and thus chooses more wisely when there is more at stake, even if what is at stake is vague or unfamiliar to the chooser.
In one of his most convincing and engaging discussions, Clarke applies the massive modularity hypothesis to the disjunction in accounts of misrepresentation. To use Clarke's initial and most elementary example, "D" (what might be thought by any random layperson to represent entity-D) is given validity by its existence as representation, not because of its happenstance resemblance to D itself. When that random layperson incorrectly assumes that "D" is caused by D, it creates a situation where the condition of representation itself is equivalent to truth, which necessarily leaves no room for the possibility of falsity. "One might think that D-caused "D" tokenings are false," writes Clarke. "But this will not work" . "D"'s validity as representation, however, makes E-caused "D"s true as well, meaning that misrepresentation disappears. By demonstrating the possible selection process for and function of modules designed for specific tasks, and those modules' relationship with their functional domain, Clarke finds the ability to judge the space between the two as an indication of misrepresentation—the gap between intent and actuality. "A module is reliable," argues Clarke, "if and only if it produces successful results in the actual domain (AD = <RF1, L1, T1>) and would usually produce successful results in possible domains (PD = <RF1 + ARF, L1, T1 + N>) that are relevantly similar to the actual domain" . A large enough gap between the two creates malfunction, and thus false beliefs. Clarke tweaks the theories of Quine, Dretske, and others by viewing the problem within the framework of natural selection. The creation of false beliefs leads to "the misidentification of significant adaptive problems" , so the reliable modules, through whatever ups and downs, continue to win out.
And so human regeneration has been inherently dependent on the proper survival of accurate representation, which leads Clarke to his next treatment of the conduit of that survival—Darwinian algorithms. These algorithms are "a plethora of specialized, domain-specific inference rules designed to solve specific, recurrent, adaptive problems in social exchange contexts" . True beliefs are beneficial to human survival, and those beliefs are grounded by reliable cognitive processes, and those processes were selected during that same Pleistocene hunter-gatherer period. Here Clarke's account verges on the pre-pragmatism of Chauncey Wright more than on the post-pragmatism of Davidson or Quine. The "cognitive mechanisms" are for "specific, recurrent, adaptive problems [ ] in specific, social exchange contexts," not through any "propositional calculus" . There is here a hint of Wright's cosmical weather. "Humans reason not logically," Clarke writes, "but adaptively" . His case is well-made, but partial. Surely, even from this brief account, the absence of sensory data such as sight and sound—the data that act as a mechanism for any adaptation for cognitive processes that are inherent yet malleable—seems significant. Clarke allows that his treatment is not all-consuming. His principal purpose is to validate Leda Cosmides's account of the selection of the "special-purpose mechanisms"  that aid in the solution of adaptive problems.
Clarke's is a defense of an epistemic naturalism that is reliabilist and externalist. The group of naturalist epistemologists, however, often finds itself caught somewhere between analytic epistemologists and philosophers of science. The analytic philosophers accuse the naturalists of not being enough like them—naturalists are, essentially, asking and answering the wrong kinds of questions. Philosophers of science fail to see the move away from analyticity, arguing that the naturalist project follows too closely the questions of analytic philosophers whose overwhelming concern for the intricate and elaborate responses to the rigorous doubts of the skeptics as to viable forms of knowledge only tortures their ability to get any meaningful work done. Clarke argues, however, that the two groups cannot see the forest through the trees. Epistemic naturalism acts as a bridge between the two in any useable theory of knowledge.
He defends the position by positing the dual notions of meliorative and nonmeliorative justification. Nonmeliorative justification is necessary for knowledge, involving a third-person assessment of pre-established beliefs. Meliorative justification, not necessary for knowledge, is a first person method of obtaining seemingly reliable advice for acquiring or changing beliefs. "The first sets out a paradigmatically objective standpoint concerning actual justifiedness; the second is a subjectivist notion that provides advice that makes justifiedness likely, but does not ensure justifiedness" . So the nonmeliorative form is necessary for knowledge, its counterpart is necessary for inquiry. It may appear at first glance that meliorative justification is not really justification at all. Clarke, however, writes that "one must be able to tell from the first person perspective that one has satisfied a set of epistemic constraints" .
The concern of philosophers of science is the meliorative project. What they fail to see, for Clarke, is the connection between both types of justification. "Having a sense of what one is pursing, that is, nonmeliorative justification, even if one admits that all one can be sure of is that one has satisfied meliorative constraints on justifiedness, is crucial. It is akin to the distinction between knowing that and knowing how" . Nonmeliorative justification acts as a method of clarification, meliorative as a method of inquiry. The latter presupposes the former. When carried out properly and successfully, this is a fancy way of describing learning. After establishing the justification models, and a picture of human reason as adaptive rather than logical (in the traditional content-neutral form it is usually ascribed), Clarke understandably moves to probability—or, to the evolved mind as applied to providing the best meliorative advice for gaining truth. To match the subjective meliorative justification project, Clarke argues for a notion of subjective probability by devising probability estimates to use as a base reference point to match up with contemporary information, thus "conditionalizing" the data . This is the same domain-specific model applied to Darwinian modules writ large, and the consistency of the formula provided throughout is its greatest asset. Knowledge, for Clarke, is natural and empirical, not conceptual, rendered so (if by nothing else) by its adaptive makeup. "If epistemology is ever to regain its intellectual credibility," he writes, it must "go modular" . It must reflect "the modular, adapticist natural capacities of our species in our native environments" .
Obviously, Clarke finds benefit in the participation of science in the program of analytic epistemology and, further, that "a properly naturalized epistemology must be squared with the latest results from empirical psychology" . After a brief but effective critique of nonnaturalist epistemology, he develops a modular version, beginning with a twelve-point analytic argument justifying his claim that knowledge is a natural, rather than conceptual, kind. The formula is convincing. His second premise, "If knowledge exists, it is either a conceptual kind or a natural kind," seems to leave no room for alternatives (such as, say, that knowledge is not a kind), but Clarke anticipates the criticism and argues that even the large majority of the most ingrained and hardened skeptics do not eliminate the notion of knowledge, which is the functional equivalent of disclaiming it as a kind. The author, too, posits strong arguments in anticipation of criticism of his seventh proposition, "If knowledge is a conceptual kind, then it is a social construct." Despite the referent, a concept is independent of its corresponding object and thus socially constructed. Knowledge is a series of natural kinds. Natural kinds are, in the words of Hilary Kornblith, "homeostatic clusters of properties" . In the given example, water is a kind because it is formed by two hydrogens and an oxygen joined by a chemical bond. "The epistemic bond that holds true beliefs together," argues Clarke, "is justification" . That justification is measured by its probability for producing truth, the significance of probability trumping that of truth. "Justified beliefs, on this account, are such that they can, though they need not, be false. Clearly, it must be possible to have justified though false beliefs" . Certainly they will never disappear, but Clarke's version of knowledge remains firmly attached to the rigors of daily life. "The true beliefs that are ingredient for knowledge," he argues, "are instrumentally useful for the satisfaction of our biological needs and are, ipso facto, bound up with our evolutionary history" .
Justification supports belief. Justified belief presents what amounts to the straightest path to truth. These three elements combine to form the aforementioned homeostatic cluster that constitutes a kind of knowledge. Knowledge, taken en masse, is a series of these kinds. Of course, justification, truth, and belief are not physical. There is no strictly quantifiable way to measure them with any degree of accuracy, which would seem a problem for an author so intent on empirical validity. Clarke, though, calls for a biological account of knowledge that could act as a companion to his account, further demonstrating his argument for the necessity of a "transtheoretical consistency" in the philosophical and scientific projects.
Until that biological account appears, Clarke could possibly be building his castles on sand. The widespread credence given to evolutionary psychology in the cognitive sciences and the legitimate disconnect that the author describes between those and the philosophical community, however, make the attempt necessary. Throughout Clarke's text is an analysis of the relevant literature, as you might expect, but the work is unique in its combination of both the evolutionary psychology historiography and its epistemological counterpart. Rudolph Carnap engages in dialogue with Jerry Fodor. Alvin Goldman and Steve Stich appear alongside Willard Quine and Bertrand Russell. The juxtaposition is unique, but soon appears natural and (even) necessary. The massive modularity hypothesis is, and must be, just that. And Darwinian modules themselves could legitimately be called constructs, but Clarke's argument is convincing for its consistent rigor and demand for interdisciplinary collusion for the benefit of reconciling evolutionary psychology and naturalist epistemology.