Here are a few comments born of my extensive interest in SPT.

(1) It appears that basic repetitive behaviors of a mental sort might get inner voices started without TOM and mentalese, and perhaps without TOM or mentalese. Inner voices, ie some set of species of the genus, appear to function by conditioning of the self in ways familiar from behaviorism. Eg after doing something one immediately regrets one might respond ‘No! No! No!’ As written it takes shape as an English word and exclamation point, but the associated conditioning might have developmental origins that are independent of TOM. Don’t mistake my meaning. I doubt it–developmental problems with inner speech have been identified as significant in children with autism, who have TOM-related problems (see Whitehouse 2006, Child Psych & Psych), which I take it supports your hypothesis–but the alternative must be ruled out. So to what implicit empirical support for the claim ‘an inner voice requires theory of mind and language’ are you appealing–or is your argument a conceptual one?

(2) I tried to answer that question myself but not much available empirical literature on the development of inner speech, from what I could quickly find. See Flavell 1997, Development of Children’s Knowledge about Inner Speech. From abstract: “Study 1 showed that, in contrast to 6-to7-year-olds and adults, 4-year-olds usually did not infer that a person silently engaged in such intrinsically verbal mental activities as reading, counting, or recalling items from a shopping list was saying things to herself. They also tended to deny that covert speech is a possible human activity. Study 2 demonstrated that 4-and 5 year-olds are much poorer than adults at detecting their own inner speech.”

I suppose if we knew at what developmental points children freely deployed concepts like SPIRIT or SOUL, we could correlate this with developmental data about the ways children conceive of the metaphysics of their inner voice. Has anyone experimented *directly* on inner voices in ways that fix the intuitive metaphysics of them? Perhaps you could draw on the literature about imaginary friends.

(3) You say that the inner voice is a form of supernatural agent. What is the identity or ‘form of’ claim? Are you saying that the inner voice is conceived by the agent as (numerically?) identical with a spirit insider herself who is not her, is identical with a spirit inside herself who is her, or are you not making any claim about the agent’s intuitive metaphysics being such as to imply that the inner voice is a supernatural agent? Perhaps there’s relevant research here on how individuals manifesting multiple personalities conceive of those inner voices as or not as SNAs. However that works out, perhaps you could explain what you mean by ‘form of’ and whether the relationship that describes has been shown to exist.

(4) You write, “they are not necessarily mutually exclusive—perhaps the evolutionarily ancient cognitive architecture of conscience is the vehicle by which religion gains traction in altering our beliefs and behavior.” Happy to grant the first modal claim (not necessarily mutually exclusive) as true. For the sake of argument, strip the second modal claim of the operator ‘perhaps’ and the claim is interesting. But is it falsifiable? If so, what do you mean by ‘gains traction’–that, in their own developmental origins, religions used the cognitive architecture of conscience to produce cognition of, in particular, SNAs? Intriguing idea.