This example is illustrated in Figure 9.3 and can be described as follows. Imagine an organism A, whose behavior R is the function of I, A and E. For instance, A shows behavior R when A detects that E = z and 1= y. Note, however, that because of the mutual and reciprocal union of A and E (denoted by the solid lined arrows), E is also influenced by I, such that E = z is only subsequent to I = w. Now imagine that there are two observers (scientists) of A, both attempting to understand the cause of behavior R. For the first observer, E is not observed or assumed to be of little consequence. Thus, to the confusion of Observer 1, A and I do not predict R directly. I is sometimes y and sometimes some other state (z, w, v, etc.). As a result, Observer 1 discerns (after a while) that R results when I passes successively through states wand y and hypothesizes that R = I + A + (A's memory of I). In other words, Observer 1 endows A with other causal structure. In contrast, Observer 2 does observe E in addition to A and I. Thus, Observer 2 discovers (after only a short period of time) that R occurs when E = z and I = y. As a result, Observer 2 concludes that R is a direct result of the total system, R = I + A + E. That is, Observer 2 makes no hypothesis about "other" cause, internal to A, as such cause is not required. This example, though somewhat obvious in its simplicity, is by no means trivial, nor is its facetious criticism of traditional theory unjustified. To be blunt, when organism is considered separate from environment, and the partial system (organism) deputizes for the whole system (organism and environment), there is a tendency to fashion explanation through variables that are beyond immediate observation. Gratuitous appeals to internal states as explanations of everyday behaviors exemplify this tendency. (Richardson, Shockley, Fajen, Riley, & Turvey, 2008, pp. 165-6).Now, I guess I don't care so much about the fact that the argument is a priori or that it is a thought experiment. I do think this is a caricature of, say, the cognitivist view, so it would be nice to have some references to cognitivists who have actually made this mistake. They could be out there. I haven't read everything.
Moreover, what is wrong with explanations through variables that are beyond immediate observation? Doesn't science regularly appeal to the unobservable to explain the observable?
And, won't everyone agree that gratuitous appeals to internal states as explanations of everyday behaviors are bad? They are, after all, gratuitous appeals. In fact, isn't the issue about internal states really about whether the appeals to them are gratuitous in the first place?
Richardson, M., Shockley, K., Fajen, B., Riley, M., & Turvey, M. (2008). Ecological psychology: Six principles for an embodied-embedded approach to behavior. Elsevier hand book of new directions in cognitive science. New York: Elsevier