Kilograms Matter: Rational
Analysis, Ecological Rationality,
and Closed-Loop Modeling of
Interactive Cognition and Behavior
Michael D. Byrne, Alex Kirlik, and Chris S. Fick
It should be noted that from now on “the system” means not the nervous system but the
whole complex of the organism and the environment. Thus, if it should be shown that 'the
system' has some property, it must not be assumed that this property is attributed to the
nervous system: it belongs to the whole; and detailed examination may be necessary to
ascertain the contributions of the separate parts.
W. Ross Ashby, Design for a Brain (1952, p. 34)
More than 50 years have passed since cybernetics pioneer Ross Ashby so clearly articulated the nature and difficulty of the challenge imposed in making correct causal attributions to the multiple human and environmental components participating in closed-loop interaction with the world. In psychology, it has been more than 100 years since John Dewey tried to caution psychologists about the dangers inherent in the tendency, however seductive, of basing a theory of human behavior on a causality of stimulus–response relations, in his classic (1896) article “The Reflex Arc Concept in Psychology.” Dewey emphasized that such a framing was to a large extent arbitrary because behavior itself nearly always plays a role in creating the available environmental stimulation, with the result that there is no clear logical, conceptual, or epistemological basis for drawing the S–R causal arrow in either direction and that perhaps it should be drawn in both directions simultaneously. These ideas have always found resonance in the study of perception and action, where the intimacy and interdependency between the human and external world are most obvious. Perhaps the best examples of such closed-loop thinking are Powers (1973), the field of manual control modeling (Jagacinski & Flach, 2002), Gibson's studies of active touch (1962), and dynamical systems approaches to modeling human–environment interaction (e.g., Beer, 1995).
Although Kirlik (chapter 1) has noted that creating integrated, closed-loop models of both the human and the task environment has a long tradition in the field of human–machine systems (albeit perhaps with overly primitive models of the human), the computational cognitive modeling community appears to be learning this lesson only somewhat recently. Why? There are almost certainly many factors involved. Among them, as Bently (1954) put it, is that people have a pronounced tendency to identify their own boundaries physically, even though adaptive behavior is often best understood in terms of a functional interaction between a person and the external world (see Clark, 2003, for a more recent treatment of this same point in the context of cognitive prostheses and “embodiment”). Viewing cognition and behavior in this functional light, implying that the organism–environment system should be the appropriate unit of analysis in psychology, provided the common foundation for both