Dual System Theories of Cognition: Some Issues

Issues discussed in this paper relate to the age of evolution of the systems, consciousness and control, domain specificity, individual differences methods and the question of how many distinct systems there might be in the mind.

Abstract

Dual System Theories of Cognition: Some Issues Jonathan St B T Evans ([email protected]) Centre for Thinking and Language, School of Psychology, University of Plymouth, Plymouth PL4 8AA, UK than answers. There are a number of significant issues that need to be addressed if a fully coherent dual system theory of cognition is to be developed that is consistent with the available evidence. The purpose of this paper is to highlight these issues. Abstract Theories positing dual cognitive systems have become popular in cognitive and social psychology. Although these theories have a lot of common features, close inspection of the literature reveals a number of difficult and unresolved theoretical issues. Issues discussed in this paper relate to the age of evolution of the systems, consciousness and control, domain specificity, individual differences methods and the question of how many distinct systems there might be in the mind. Issue 1: Old system/new system The idea that System 1 cognition is ancient and System 2 cognition is modern, in evolutionary terms, is a recurring theme in dual-process theories. This is often linked to the assertion that while System 1 cognition is shared with other animals, System 2 cognition is uniquely human. The last idea arises from its association with uniquely human processes such as language and reflective consciousness and the apparent ability to perform cognitive acts (such as hypothetical simulation of future and counterfactual possibilities) that are assumed to be beyond animals. However, little if any direct reference is generally made to literature on animal cognition. In fact, there is evidence also for distinct cognitive systems in animals. For example, in a recent survey of cognition in a wide range of animal species, the biologist Toates (in press) has claimed that there is a widespread division between cognition that is stimulus-bound on the one hand and involving higher order control on the other. Stimulus-bound cognition includes conditioning and the application of instinctively programmed behavior of a fixed nature. However, he shows that higher-order cognition is also present in many species and suggests that this developed into consciousness in humans. If he is right, then the assumption that the dual system distinction is unique to humans may be wrong. At best we could say that System 2 was better developed in humans. The description of System 1 as ancient may be oversimplified also, as it is likely that System 1 really includes a number of different forms of implicit cognitive processing that evolved at different times (see also Issue 5). What does appear to be ancient is associative learning, at least in its most basic forms, since humans share this facility with most other animals, including birds, reptiles and fish. However, if our implicit system also includes cognitive modules, predisposing us to develop cognitive abilities that are encapsulated in specific domains, there is good reason to believe that these evolved much later. The cognitive archaeologist Mithen (1996) has argued from data in the archaeological and anthropological record that ancient hominids (in common with apes) relied heavily on general learning, whereas early humans (such as Neanderthals and Dual processing accounts of cognition have been developed in a range of areas including learning (e.g. Reber, 1993), attention (Schneider & Shiffrin, 1977), reasoning (Evans, 2003), decision making (Kahneman & Frederick, 2002) and social cognition (Chaiken & Trope, 1999). In spite of a considerable degree of independence in the formulation of these accounts, they include a number of striking similarities. Processes that are rapid, automatic and effortless on the one hand are contrasted with those that are slow, sequential and controlled on the other. These theories typically characterize the two processes as independent sources of control for behavior that may come into conflict and competition. A number of theorists have mapped these dual processes on to two distinct cognitive systems. These systems have been given various names including experiential-rational (Epstein, 1994), heuristic-analytic (Evans, 1989; in press), heuristic-systematic (Chen & Chaiken, 1999), implicit- explicit (Evans & Over, 1996; Reber, 1993), associative and rule-based (Sloman, 1996) and the neutral System 1 and System 2 (Stanovich, 1999). Again the characteristics attributed to these underlying systems show quite a large degree of consensus across theories and domains of application. For example, System 1 (using this as a generic label for the fast, automatic system) is often described as evolutionarily old, shared with other animals and independent of individual differences in general intelligence, whereas System 2 is by contrast evolutionarily recent, uniquely human and related to heritable differences in intelligence and working memory capacity. Given this degree of consensus in dual system theories developed across different cognitive domains, it is tempting to conclude that the brain must indeed contain two systems broadly as described. However, closer inspection of these theories and the relevant literature reveals more questions