Evolved cognitive mechanisms

The empirical core of evolutionary psychology is the study of evolved information processing mechanisms. This chapter reviews principles of research on evolved cognitive mechanisms and discusses examples of empirical work on mechanisms of face recognition, intentional inference, kin recognition, kin-based social interaction, and social exchange. It is argued that there is no general-purpose method for revealing evolved mechanisms, but rather, methods should be adjusted to fit hypotheses on a case-bycase basis. The explanatory role of mechanisms in evolutionary psychology The goal of the behavioral sciences is to explain behavior in causal terms. This is one of the most difficult problems in science because the causes of human behavior are complex and operate interactively over many scales of space and time. Some approaches to human behavior attempt to gloss this problem by treating humans like elementary particles whose behavior is governed by relatively simple laws. Economic theories, for example, sometimes treat humans as utility maximizers, assuming that humans will act as if they are maximizing utility when viewed in the aggregate even though the proximate mechanisms that cause this behavior are unspecified. 1 To appear in: Crawford, C. & Krebs, D. eds. Foundations of evolutionary psychology: Ideas, issues, applications and findings. (2nd Ed.) Mahwah, NJ: Erlbaum Associates. H. Clark Barrett Center for Behavior, Evolution and Culture and Center for Culture, Brain and Development UCLA Department of Anthropology 341 Haines Hall, Box 951553 Los Angeles, CA 90095-1553 [email protected] Evolved cognitive mechanisms 2 Evolutionary psychology attempts to move past “as if” models by identifying the proximate causal mechanisms of human behavior in the brain and linking these to ultimate evolutionary causes. The principle that guides evolutionary psychology research is that evolutionary processes shape brain mechanisms, and brain mechanisms shape behavior. Evolved mechanisms are the units of explanation that distinguish evolutionary psychological accounts from other approaches, which tend either to attempt to link ultimate causes directly to behavior or to focus on proximate causes only (Cosmides & Tooby, 1987). Because of their ambitious nature, evolutionary psychological approaches have been criticized on several grounds, including that ultimate causal events occurred in the past, and so cannot be directly observed (Buller, 2005). This reflects a misunderstanding of the role of evolutionary theorizing in evolutionary psychology. Evolutionary principles rarely lead to deductive certainties. Instead, they are a heuristic for the generation of hypotheses about the possible design features of mechanisms. These hypotheses are then tested empirically, and it is ultimately the combination of data and theory that weigh for or against a particular evolutionary hypothesis, as illustrated in the examples below. Critics have also attacked the notion that the mind contains many specialized mechanisms that are closely linked to adaptive problems that recurred over evolutionary time, as opposed to a few general mechanisms that are not specialized to solve specific problems (see Barrett & Kurzban, 2006, for a review). This debate is largely unnecessary. Few would argue that there are no mechanisms that can solve a wide range of problems. However, the explanatory burden faced by theories of “general purpose” mechanisms is the same as that faced by theories of specialized mechanisms: namely, what are the information processing features that allow the mechanism to perform the tasks that it is invoked to account for, and what are the evolutionary processes that shaped those features? Presumably, few would postulate mechanisms that have no function at all, or that solve problems without any particular features that allow them to do so. Moreover, arguments about the degree to which the mind contains many, as opposed to few, specialized mechanisms cannot be resolved a priori. That question is an empirical one. Here, I will review research on evolved cognitive mechanisms to show that the evidence for specialized mechanisms is in fact substantial. This research shows how evolutionary reasoning can play a useful heuristic role in the search for the design features of cognitive mechanisms. The form-function fit, design features, and domain specificity A cognitive mechanism is anything that plays a causal role in guiding behavior on the basis of neurally coded information. Evolutionary psychologists Evolved cognitive mechanisms 3 view specialized cognitive mechanisms as synonymous with cognitive modules, but the notion of modules in evolutionary psychology differs substantially from the conventional view in cognitive psychology (Fodor, 1983, 2000). For example, while evolutionary psychologists expect the mind to be multimodular, the modularity that evolutionary psychologists have in mind is interactive, not rigid and isolated, as many psychologists suggest (Barrett, 2005b). Evolved cognitive modules are not expected to operate in isolation from other systems, because a key value of specialization is that it leads to flexibility and computational power when modules interact. Nor are features such as automaticity, or other features suggested by Fodor (1983), necessary features of evolved modules (Barrett, Frederick, Haselton, & Kurzban, in press). Instead, evolutionary psychologists regard the key feature of modularity to be functional specialization (Barrett, 2005b; Barrett & Kurzban, 2006; Carruthers, 2005; Sperber, 1994, 2002, 2005). Functional specialization refers to the fit between form and function that is characteristic of biological adaptations. For morphological adaptations like fins or wings, the meaning of “form” is clear. In the case of cognitive mechanisms, form refers to information-processing features of the mechanism. These can be thought of as the mechanism’s design features (where “design” refers not to design by an intelligent agent, but by evolutionary processes). Typically, a list of a mechanism’s design features would include a specification of the kinds of inputs the mechanism accepts, and the operations that it performs on those inputs. Of necessity, all mechanisms will operate on information only of a particular format. The format requirements of a mechanism delineate the mechanism’s domain (Barrett & Kurzban, 2006; Sperber, 1994). Many authors use the term “domain” in a more narrow sense, to refer to “content” or “meaning” domains. However, from an evolutionary perspective, there is no reason to restrict the concept of domain specificity just to content domains (Barrett and Kurzban, 2006). For example, the hypothesized phonological loop in working memory (Baddeley, 2002) has a clear input domain in that it accepts only representations of sound, yet the content of the sounds it handles is not restricted. Nevertheless, the set of inputs handled by the phonological loop and the visuospatial sketchpad, another hypothesized component of working memory (Baddeley, 2002), are well-defined and distinct. The domains of these information-processing mechanisms are specific and do not overlap. A useful distinction can be made between a mechanism’s proper domain – the range of inputs that the mechanism evolved to process – and its actual domain, the range of inputs that the mechanism actually accepts, whether or not they influenced the evolution of the mechanism (Sperber, 1994). For example, a mechanism for detecting biological motion might be triggered by computergenerated animations of dinosaurs, even though these animations clearly played Evolved cognitive mechanisms 4 no role in the evolution of the mechanism. Together, the notions of specialized function, input conditions, operations on inputs, and the distinction between proper and actual domains provide the theoretical basis for the study of evolved information-processing mechanisms. The empirical study of specialized cognitive mechanisms Evidence for specialized mechanisms comes in the form of signatures of specialization that can be observed empirically. For example, evidence that information of one kind is processed differently from information of another kind suggests either that multiple mechanisms are involved, or a single mechanism that is structured to handle particular information types differently. Another kind of signature can be observed in neuropsychological dissociations: the differential loss of information processing abilities following brain damage or developmental disruption (Shallice, 1988). However, just as it is the case that no single set of features is general to all specialized mechanisms, it is also the case that no single method or set of methods can be used across the board to diagnose the presence of specialized mechanisms. For example, mechanisms will vary in the extent to which their operations share resources with, or are influenced by, other systems. Therefore, although evidence that manipulating one system or mechanism (e.g., occupying working memory with a string of digits) affects some other mechanism might bear on hypotheses about how such systems interact, it does not falsify that specialized systems are operating (Barrett, Frederick, Haselton, & Kurzban, in press). The same goes for neuropsychological dissociations. Brain damage won’t necessarily affect all of and only one mechanism, nor will developmental damage necessarily affect all of and only one system, because development is interactive (Shallice, 1988). Because causation in the brain is complex, it can be difficult to disentangle the effects of distinct mechanisms, and multiple sources of evidence are usually necessary. A general heuristic for empirical studies of cognitive mechanisms is that the methods should fit the hypotheses about the design feature under investigation. If rapid speed is an expected design feature of the system in question based on evolutionary reasoning – as in the case, for example, of a perceptual mechanism for detecting snakes – then methods such as reaction time might be appropriate. For other systems, such as mate choice, for which there is reason to expect slow processing integrating much information, rather than speed, such methods might reveal little. With these principles in mind, I will now review a few examples of specialized information-processing mechanisms, focusing on evidence of functional specialization and how it relates to hypotheses about evolved function. Evolved cognitive mechanisms 5