In his analysis of four identifiable accounts of the relationship between “mindreading” and “metacognition,” Carruthers makes a bold attempt to explain a major inconsistency in three of these and assert the primacy of “mindreading” as a precursor to self-understanding. He does this mainly by highlighting a common key feature in the literature in neuropsychology, reasoning, and psychopathology: why adults with sophisticated skills in accounting for their own and others' thoughts still seem to show flaws in their abilities to demonstrate and reconcile these skills. However, we make three connected points to argue that his account falls short of a clear answer to his main concerns about the nature of the two processes and how they relate.
First, Carruthers treats the terms “metacognition” and “mindreading” as unproblematic, distinct, and clearly grounded within an overarching, internalist cognitive architecture. Yet, the article is unclear about how this architecture fits together into a working psychological system. There is more than a hint of a homunculus in Carruthers' descriptions of how we access and use these skills (Kenny Reference Kenny and Hyman1971/1991). Let's look at the role of the role of language in mindreading as an example.
Early in the target article Carruthers dismisses the importance of language in a note because it seems “plausible” to him “that basic capacities for both mindreading and metacognition are independent of our capacity for natural language” (target article, Note 2). Yet in his analysis of reasoning (sect. 8), he acknowledges its importance in certain circumstances, but does not convincingly show how this contradiction can be reconciled. Then, in the bulk of the article, speech only serves as input for Carruthers' “magical mindreading machine,” revealing a view that thinking is computation and meaning is mechanistic, entailing a code model of language. All of these assumptions have been repeatedly critiqued (e.g., Bickhard Reference Bickhard2001; Goldberg Reference Goldberg and Hyman1991; Heil Reference Heil1981; Proudfoot Reference Proudfoot1997; Tomasello Reference Tomasello2003; Turnbull Reference Turnbull2003; Wittgenstein Reference Wittgenstein1968). For example, problems have been pointed out with how the symbols involved in computation could acquire meaning, and whether understanding can adequately be conceived of as a process of computation (Proudfoot Reference Proudfoot1997). These criticisms have been applied to the theoretical approach that Carruthers advances.
Carruthers claims to have considered four different theories. However, since he starts with such a restricted set of theories to consider, knowing that he declares his version of theory-theory as the winner in this contest tells us nothing about how the many other approaches would fare. The one place where we do agree with Carruthers is that assuming introspection as a source of social understanding is problematic. We find it odd, however, that he ignores Wittgenstein's far stronger arguments against introspection. Alternative accounts based upon a more sophisticated grasp of language deal more easily with the confabulation literature that is so well described in the target article (sect. 6), but not fully explained.
Indeed, second, the theory-theory's response to these criticisms is simply to ignore it (Carpendale & Lewis Reference Carpendale and Lewis2004; Leudar & Costall, in press). Carruthers' perspective is symptomatic of an insularity that has been commented upon repeatedly in the wider literature, but which barely gets a mention in the “mainstream.” The idea of the 1980s that we, or something inside ourselves, can manipulate our own social understanding was criticised by Bruner (Reference Bruner1986; Reference Bruner1990), who rejected a cognitive system that is bereft of serious contact with the outside world. Many commentators have repeated this claim. Carruthers buys into forms of attributing mental states to ourselves and others which assume the theory metaphor so loosely as to be bereft of meaning (see Campbell & Bickhard Reference Campbell and Bickhard1993). The fact that this movement has been as immune to debate in the outside world as the cognitive system that they depict has led its critics (e.g., Leudar & Costall, in press, Ch. 1) to use the term ToMism to refer to the assumption that the individual is trapped behind a mindreading mechanism that filters their interpretation and observations of social interactions.
Third, in his neglect of the growing literature which attempts to explore the interactional basis of social understanding, Carruthers' account is both parasitic upon the evidence from developmental psychology and simultaneously dismissive of its analysis of the nature of the two processes. We suspect that without a clear analysis of development of “mindreading” and/or “metacognition,” Carruthers' definition of these two terms remains grossly underspecified. There is a large body of recent work that attempts to explore the dynamics of interaction in early (Reddy Reference Reddy2008) and later infancy (Moll & Tomasello Reference Moll and Tomasello2007) and in childhood (Carpendale & Lewis Reference Carpendale and Lewis2004), as well as with reference to special populations (Hobson Reference Hobson2002), which describes how social understanding, as opposed to some miraculous yet unspecified mindreading system, gradually emerges in social interaction. In these approaches there is no Cartesian split between the individual in the spectator's role observing others' physical movements and having to attribute mental states. Without a reference to this expanding and exciting literature, and its data and critique of the solipsism of theory-theory, Carruthers' argument is a better description of atypical “mindreading” like autism, not an account of how we understand ourselves and others.
