Barresi & Moore provide an impressive account of how the coordination of first and third person information about the self and other could produce an account of intentional relations. They are less explicit as to how the child comes to understand the basic epistemic relation between experience and knowledge, that is, how informational access causes belief. We suggest one route.

Although Barresi & Moore could have grounded their framework more explicitly in existing models, they offer a provocative testbed for the assumptions of symbolic interactionism and further thinking about self-regulation, especially in autistics.

Barresi & Moore assume an equivalence between ontogenetic and evolutionaiy transformations of social understanding. The mechanisms of evolution allow for novel structures to arise, both through terminal addition and through the onset of novel pathways at time points that precede the end points of ancestral pathways. Terminal addition may not be the appropriate model for the evolution of human object-directed imitation, intermodal equivalence, or joint attention.

Barresi & Moore's “intentional schema” will not do the work of “theory of mind.” Their model will account neither for fundamental facts of social competence, such as the social attributions of the 10-month-old infant, nor the possibility that, though having a theory of mind, the chimpanzee's theory is “weaker” than the human's.

Barresi & Moore do not consider information about intentional relations available within emotional engagement with others and do not see that others are perceived in the second as well as the third person. Recognising second person information forces recognition of similarities and connections not otherwise available. A developmental framework built on the assumption of the complete separateness of self and other is inevitably flawed.

Barresi & Moore's account predicts that infants deprived of visual input will be delayed in achieving social understanding, a hypothesis that receives some support from studies of language use. by blind children. It is proposed that recently developed false belief and appearance/reality tasks be used to explore this issue further. Three possibly distracting conceptual issues are also discussed.

We consider the various criticisms and requests for clarification made by the commentators of our framework for understanding intentional relations. Our response is organized according to the main themes in the target article: general theory, phylogeny, development, and autism. We also add some discussion of further issues, such as simulation and moral theory, that were not addressed in the target article.

The driving force in the evolution of language and the human mind was the advantage gained in courtship by efficient communicators. Ontogeny recapitulating phylogeny may offer an alternative to Donald's picture of the relative origins of language and mimesis. Very recent evidence is pertinent to dating the origin of language.

Loring Brace's assertion that “Neanderthals were just as capable of articulate speech as we are” reflects ignorance of speech anatomy and physiology. Metrical analyses of hyoid bone morphology cannot predict supralaryngeal vocal tract (SVT) shape. Houghton's (1993) “modern” Neanderthal SVT reconstruction yields an impossible creature who had a larynx positioned in his chest. The reconstructed modem SVTs of early fossil Homo sapiens indicate brains that can regulate speech, consistent with Merlin Donalds timetable for the evolution of language.

Writing is important not because it “extends memory” but because it invites new concepts. The basic cognitive resources, including memory, have remained unchanged for perhaps a million years but the objects on which it has to work have changed significantly partly because of writing which has tended to turn speech into an object of reflection and analysis.

Donald presents a three-stage theory of the evolution of cognition – episodic, mimetic, symbolic. This commentary is chiefly concerned with his use of the first term “episodic” which conflicts with standard usage in the memory literature. This conflict of usage has more than terminological implications for Donald's theory.

The human vocal apparatus is part of a vertically integrated system, and I agree with Licbennan that modern high-speed phonology co-evolved with our capacity for grammar. Olson and I agree that some distinctly human thought skills appear to be fairly recent cultural acquisitions related to the introduction of new symbolic technologies and external (that is, nonbiological) memory storage. Stenning's concern with my use of the term “episodic” can be resolved by distinguishing between episodic storage and retrieval. Baum's suggestions regarding courtship and cognitive evolution seem to apply better to mimetic expression than to language.