Assuming that the vehicle of imaginal thought is a spatial model may not be quite as egregious an error as assuming it is a two-dimensional picture, but it represents no less a reification error. Because the model is not a literal physical layout, one is still owed an explanation of why spatial properties hold in the model – whether because of architectural constraints or by stipulation. The difference is like the difference between explaining behavior from a principle and predicting it by looking it up in a list. In the latter case no purpose is being served by calling it a mental model.

Carpendale & Lewis (C&L) propose that social knowledge is constructed from triadic interactions. This account generates testable predictions concerning social knowledge in infancy. Current evidence is not entirely consistent with these predictions. Infants possess action knowledge before they engage in triadic interactions, and triadic use of an action does not always precede knowledge about the action.

The analogy between the rules that subtend ventriloquism and bimodal neurons responding suggests a possible neural mechanism for audiovisual interactions in spatial scene analysis. Perinatal data, such as those on synesthesia, sensory deprivation, and sensory surstimulation, as well as neuroanatomical evidence for transitory intersensory connections in the brain support the view that audition and vision are bound together at birth.

Phillips & Silverstein (P&S, 2003) propose that NMDA-receptor dysfunction may be the fundamental neurobiological mechanism underlying and associating impaired holistic perception and cognitive coordination with schizophrenic psychopathology. We discuss how the P&S hypothesis shares different aspects of the weak central coherence account of autism from both theoretical and experimental perspectives. Specifically, we believe that neither those persons with autism nor those with schizophrenia integrate visuo-perceptual information efficiently, resulting in incongruous internal representations of their external world. However, although NMDA-hypofunction may be responsible for perceptual impairments in schizophrenia and possibly autism, we suggest that it is highly unlikely that NMDA-hypofunction is specifically responsible for the autistic behavioral symptomology, as described by P&S in their target article.

Carpendale & Lewis (C&L) stress the importance of social interaction for social understanding, but focus on the adult-child relationship. In the present commentary, we discuss the development of social understanding within early peer relationships. We argue that peer interaction stretches the limits of early social understanding, thereby providing both unique challenges and unique opportunities for constructing an understanding of others' minds.

Phillips & Silverstein (P&S, 2003) have proposed that NMDA-receptor hypofunction is the central reason for impaired cognitive coordination and abnormal gestalt-like perceptual processing in schizophrenia. We suggest that this model may also be applicable to non-pathological (or normal) aging given the compelling evidence of NMDA-receptor involvement during the aging process that results in age-related change in higher-level perceptual performance. Given that such deficits are present in other neurological disorders such as autism, an argument for a systematic assessment of perceptual functioning in these conditions may be posited.

Sensation should indeed be understood globally: some infant behaviors do not make sense on the model of separate senses; neonates of all species lack time to learn about the world by triangulating among different senses. Considerations of natural selection favor a global understanding; and the global interpretation is not as opposed to traditional work on sensation as might seem.

Studies of aging and autism as outlined by Bertone, Mottron, & Faubert (Bertone et al.) and by Faubert & Bertone suggest that disorders of cognitive coordination involving impairments of dynamic gestalt grouping and context-sensitivity may be common to several different disorders. We agree that such studies may shed light on these processes and their neuronal bases. However, we also emphasize that dynamic grouping and context-sensitivity can fail in various ways, and that, although the underlying pathophysiology may often involve NMDA-receptor malfunction, many different malfunctions are possible, and each of these may result from any one of a number of different etiologies.

The existence of sensory convergence does not establish that the senses function as a single unified perceptual system. Reality is fully specified only by a one:many mapping onto the totality of energy arrays, and these provide alternative frames of reference for movement. It is therefore possible that higher order crossmodal relationships are detected by skilled perceivers, but this has not been confirmed empirically.

We explore three types of criticisms of our theory on the development of children's social understanding. We reject suggestions that we offer nothing new to traditional theories of development or recent “social” accounts of “theory of mind.” Second, we take the point that there are grounds for improving our account of dyadic interaction in infancy but reject claims that we have not sufficiently accounted for how we incorporate the notions of criteria and structure into the theory. Third, we accept that the epistemic triangle, as defined, would benefit from an affective dimension and such a formulation could be used to describe the dynamic of developmental change from infancy to beyond early childhood. We still feel that the combination of Wittgenstein, Vygotsky, and Piaget remains as an antidote to the flaws in current “theories of mind” approaches to social understanding.

To convince us that computers cannot have mental states, Searle (1980) imagines a “Chinese room” that simulates a computer that “speaks” Chinese and asks us to find the understanding in the room. It's a trick. There is no understanding in the room, not because computers can't have it, but because the room's computer-simulation is defective. Fix it and understanding appears. Abracadabra!

The metacognitive stance of Smith et al. (2003) risks ignoring sensory consciousness. Although Smith et al. rightly caution against the tendency to preserve the uniqueness of the human mind at all costs, their reasoned stance is undermined by a selective association of consciousness with high-level cognitive operations. Neurobiological evidence may offer a more general, and hence more inclusive, basis for the systematic study of animal consciousness.

The new commentaries raise important issues about the target article (Stoffregen & Bardy 2001). The commentaries also highlight some assumptions, often implicit, that underlie traditional interpretations of perception. We argue that evaluation of the global array and its implications for perception requires both analytical research on specification in the global array and new empirical research on the use of information in the global array for the control of action.

Mealey argued that sociopathy is an evolutionary stable strategy subject to frequency-dependent selection – high levels of sociopathy being advantageous to the individual if population-wide frequencies of it are low, and vice versa. I argue that at least one alternative hypothesis exists that explains her data equally well. Alternative hypotheses must be formulated and tested before any theory can be validated.

Selection theory requires multiple, simultaneously-actualized states. In cognition, each thought changes the “selection pressure” against which the next is evaluated; they are not simultaneously selected amongst. Cognitive change occurs not through selection among discrete “neural configurations,” but through interaction between conceptual web and context. This introduces a non-Kolmogorovian probability distribution, hence a classical formalism (e.g., selection theory) cannot be used.

If selection is interpreted as involving repeated cycles of replication, variation, and environmental interaction so structured that environmental interaction causes replication to be differential, then selection in gene-based biological evolution and the reaction of the immune system to antigens are relatively unproblematic examples of selection processes. Operant learning and cultural evolution pose more serious problems. In this response we deal with operant learning as a selection process.

Mazur & Booth (1998) (M&B) suggested that high testosterone (T) relates to status, dominance, and (anti-) social behaviour. However, low T also relates to status and to formal dominance. The General Trait Covariance (GTC) model predicts both relations under the assumption that high and low T modulates the genotype in ways that enforce the development of almost polar covariant patterns of body, brain, intellectual, and personality traits, irrespective of race. The precise modelling of these dose-dependent molecular body-intelligence-personality-behaviour relations requires that causes, mechanisms, and effects enjoy equal operational standing.

Mazur & Booth's (1998) target article concerns basal and reciprocal relations between testosterone and dominance, and has its roots in Mazur's (1985; 1994) model of primate dominance-submissiveness interactions. Threats are exchanged in these interactions and a psychological stress-manipulation mechanism is suggested to operate, making sure that face-to-face dominance contests are usually resolved without aggression. In this commentary, a recent line of evidence from human research on the relation between testosterone, cortisol, and vigilant (dominant) and avoidant (submissive) responses to threatening “angry” faces is discussed. Findings, to a certain extent, converge with Mazur & Booth's theorizing. However, the strongest relations have been found in subliminal exposure conditions, suggesting that biological instead of psychological mechanisms are involved.

Although the sensorimotor account is a significant step forward, it cannot explain experiences of entoptic phenomena that violate normal sensorimotor contingencies but nonetheless are perceived as visual. Nervous system structure limits how they can be interpreted. Neurophysiology, combined with a sensorimotor theory, can account for space constancy by denying the existence of permanent representations of states that must be corrected or updated.

Although Barsalou is right in identifying the importance of perceptual symbols as a means of carrying certain kinds of content, he is wrong in playing down the inferential resources available to amodal symbols. I argue that the case for perceptual symbol systems amounts to a false dichotomy and that it is feasible to help oneself to both kinds of content as extreme ends on a content continuum. The continuum thesis I advance argues for the inferential content at one end and perceptual content at the other. In between the extremes, symbols might have aspects that are either perceptual or propositional-linguistic in character. I argue that this way of characterising the issue preserves the good sense of Barsalou's recognition of perceptual representations and yet avoids the tendency to minimise the gains won with symbolic representations vital to contemporary cognitive science.