The second in a series of excerpts from Minds, Models and Milieux: Commemorating the Centennial of the Birth of Herbert Simon.
Robert D. Rupert
One can hardly underestimate Herbert Simon’s influence on contemporary cognitive science and empirically oriented philosophy of mind. Working with collaborators at Carnegie Mellon and the Rand Corporation, he wrote Logic Theorist and General Problem Solver (GPS) and thereby helped to set the agenda for early work in Artificial Intelligence (AI) (Simon and Newell, 1971). These projects also provided AI with some of its fundamental tools: in their work in the 1950s on Logic Theorist and other programs, Simon and colleagues invented list processing (which, in John McCarthy’s hands, became LISP – McCarthy, 1960), while the conceptual framework of GPS gave birth to production systems (and became SOAR – Rosenbloom et al., 1991). In 1981, John Haugeland included Newell and Simon’s computationalist manifesto (“Computer science as empirical enquiry: Symbols and search” – Newell and Simon 1976) in his widely read anthology Mind Design (Haugeland, 1981). As a result, the names of Newell and Simon became, in the philosophical world, nearly synonymous with the computational theory of the mind – at the top of the list with Fodor’s (1975) and Pylyshyn’s (1984). Moreover, in their early work, Simon and the Carnegie-Rand group emphasized the relative independence of an information-processing-based characterization of thought from the material components of the system so engaged (Newell, Shaw, and Simon, 1958a, p. 51; Newell, Shaw, and Simon, 1958b, p. 163, cf. Vera and Simon, 1993, p. 9). Prominent philosophers, most notably Putnam (1960, 1963, 1967) and Fodor (1974), reified this distinctively functional level of description, thereby formulating the late-twentieth century’s dominant metaphysics of mind, functionalism, according to which mental states are, by their nature, multiply realizable functional states.
The list of Simon’s influential ideas extends much further than this, however. Arguably, Simon’s work on hierarchical structure and the near-decomposability of complex systems (1996, Ch. 8) planted the seeds of modularity-based thinking, which blossomed in the work of Marr (1982), Fodor (1983), and evolutionary psychologists (Barkow, Cosmides, and Tooby, 1992; Pinker, 1997). In addition, Simon’s (1996) emphasis on satisficing and the bounded nature of rationality set the stage for the fascinating research programs of such varied figures Gerd Gigerenzer (2000), Christopher Cherniak (1986), and Ron McClamrock (1995). In fact, in conjunction with Kahneman and Tversky’s heuristics and biases program (Kahneman, Slovic, and Tversky, 1982), Simon’s observations about the bounded nature of rationality yielded what is now the leading view of human cognition: it’s the work of a complex but nonoptimized system that manages surprisingly well by deploying its limited resources in a context-specific way – that is, in a way that exploits a grab-bag of relatively domain-specific shortcuts that are reasonably reliable given the environments in which they’re typically employed. At the same time, Simon sometimes emphasized adaptive rationality – the idea that, under a wide range of circumstances, intelligent human behavior is, given the subject’s goals, a straightforward function of the structure of the task at hand – a theme appearing in the work of such luminaries as Dan Dennett (1987) and John Anderson (1990). Simon also articulated a vision of intelligent behavior as the product of simple, internal mechanisms interacting with a complex external environment, and in doing so, inspired nearly two generations of philosophers and cognitive scientists who take intelligence to be the by-product of, or to emerge from, bodily interaction with the environment (Clark, 1997; Brooks, 1999). In addition, Simon (1996, pp. 88, 94, 110) was perhaps the earliest working cognitive scientist explicitly to place aspects of internal processing on cognitive par with aspects of the environment external to the organism, a central theme in Andy Clark and David Chalmers’s enormously influential essay “The Extended Mind” (Clark and Chalmers, 1998). There could scarcely be a more humbling list of one thinker’s achievements, and little has been said about Simon’s contributions to our understanding of search algorithms, economics, design, or management!
As impressive as this list is, one might nevertheless wonder whether Simon’s views can be integrated into a single overarching vision of human cognition – of its structure, workings, and relation to the environment. Motivated by this kind of concern, I focus, in what follows, on an apparent inconsistency in Simon’s thinking, one that stands out especially clearly against the backdrop of decades of accumulated empirical results in what is sometimes known as ‘embodied cognitive science’ (Varela, Thompson, and Rosch, 1991; Clark, 1997, 2008b; Lakoff and Johnson, 1999; Rowlands, 1999; Gallagher, 2005; Gibbs, 2006; Rupert, 2006, 2009; Shapiro, 2010; Wilson and Foglia, 2011). At first blush, embodied results seem to support certain strands of Simon’s thought at the expense of others. I argue, however, that even in cases in which Simon’s pronouncements about the mind were premature or his emphasis out of balance (e.g., not sufficiently oriented toward the body-based contributions to human cognition), many of his own views about the mind provide the necessary corrective and allow him to accommodate – even anticipate – embodiment-oriented insights. Simon wears an early, and perhaps slightly misshapen, version of the coat we should all gladly wear today, that of the embodied functionalist.