Peter Slezak
Affiliation: University of New South Wales
Category: Philosophy
Keywords: homunculus error, consciousness, theories of mind, materialism, illusionism
Date: Wednesday 3rd of September
Time: 17:30
Location: Room 161 (161)
View the full session: Subjective Experience
The “Cartesian Theater” is Dennett’s (1991) metaphor for the notorious error of positing a homunculus or “little man” who watches an inner screen on which our thoughts appear – “the most tenacious bad idea bedevilling our attempts to think about consciousness.” This is the pseudo-explanation famously caricatured by Ryle (1949) with “deliberate abusiveness” as the idea of a “ghost in the machine.” However, I suggest Dennett’s celebrated analysis needs to be modified to avoid two significant mistakes.
Dennett (1991, 107) laments that, despite his persistent efforts “the persuasive imagery of the Cartesian Theater keeps coming back to haunt us.” He acknowledges that he had under-estimated the “potency of the visceral resistance” (Dennett 2005, 22) but he remarks that the Theater conception is “simply a mistake” (2001, 3). However, the very “potency” and “visceral resistance” suggest that it is not simply a mistake, nor is it plausibly regarded as “supported by tradition and nothing else” (2005, 16). Dennett (2016, 65, 72) refers to this mistake as a kind of “illusion” by analogy with the effects of stage-magic. However, I argue that the Cartesian Theater or homunculus error is less like a stage magician’s deception and more like genuine perceptual and cognitive illusions which persist even after they have been explained. Accordingly, I suggest Dennett’s celebrated analysis needs to be modified to avoid two significant errors.
First, I show that, contrary to much academic teaching and scholarship since Ryle, Descartes was not guilty of the homunculus or Theater fallacy, as he explains in his physiological writings La Dioptrique and Traité de L’homme neglected by philosophers. In these works Descartes gives an account of the way that movements of nerve filaments transmit the impinging effects of the external world. This is an abstract symbolic encoding in the nervous system that avoids the notorious homunculus which is explicitly rejected by Descartes. The issue has become familiar through Dennett’s (1978a, 56) exposition of the emptiness of virtus dormitiva accounts, parodied by Molière’s doctor in Le Malade imaginaire. However, it is acutely ironic that Molière was following Descartes’ exposé of the pseudo-explanation.
Second, I argue that the Theater metaphor must be understood quite differently from the way it has been conceived and popularized by Dennett (1991). Indeed, following Descartes, I suggest that the pseudo-explanation arises not from what is included in our theory to do the “clever work” but rather from what is missing. The temptation is not “to imagine an inner agent” (Dennett 2013, 91) but, on the contrary, it is the failure to notice the omission of an inner agent needed for the system to work. The problem arises because a model may be incomplete and can’t work without depending on “an intelligent and comprehending reader” (Chomsky (1962, 528) – the theorist. A theory may give the illusion of explanatory completeness because we are intuitively using the very perceptual or other cognitive faculty that is the subject of the theory. In this case, all the “clever work” is being done by us. As Fodor (2007) notes, the question is not what is obvious to the theorist but what follows from the theory. It is the failure to notice that we are the spectators in the Cartesian Theater.
Finally, I argue that, once the error is properly understood in this way, we can see that it is implicated in several controversial theories. I indicate how theories of visual imagery, Searle’s Chinese Room thought-experiment, the Gettier Problem, Davidson’s interpretative semantics and Kripke’s account of proper names inter alia have nothing in common besides the error of depending on an intelligent and comprehending “reader.” For example, as Dennett (1977, 101) recognized, the idea of a symbolic “language of thought” (Fodor 1975) faces what he called “Hume’s Problem” according to which representations require an interpreter whereas, in a sense, they ought to “interpret themselves.” In other cases, too, it seems difficult to avoid the error of positing internal representations which have their meaning because we can “understand” them in a special intuitive sense of this term.