We address two kinds of criticisms of our Bayesian framework for generalization: those that question the correctness or the coverage of our analysis, and those that question its intrinsic value. Speaking to the first set, we clarify the origins and scope of our size principle for weighting hypotheses or features, focusing on its potential status as a cognitive universal; outline several variants of our framework to address additional phenomena of generalization raised in the commentaries; and discuss the subtleties of our claims about the relationship between similarity and generalization. Speaking to the second set, we identify the unique contributions that a rational statistical approach to generalization offers over traditional models that focus on mental representation and cognitive processes.

The “somatic marker mechanism” (SMM; Damasio 1994) is proposed as the cognitive and neural basis of the theory of mind mechanism. The SMM evolved for evaluating the intentions, dispositions, and relationships of conspecifics; hence, it is adaptive in the social domain. It is predicted that chimpanzees will indeed have theory of mind (ToM) ability, but that this will be socially domain-specific. Domain-general ToM will be found only in primates with abstract, symbolic language (adult humans). Putative ToM tests require revision in the light of these distinctions.

Rendell and Whitehead note the necessary, complementary relationship between field and laboratory studies in other species, but conclude their article by de-emphasizing the role of laboratory findings in cetacean research. The ambiguity in field studies of cetaceans should argue for greater reliance on the laboratory, which has provided much of the available research supporting the hypothesis of cetacean culture.

Cowan makes an intriguing case for a fundamental limit in the number of chunks that can be stored in short term memory (STM). Chunks are collections of concepts that have strong associations to one another and much weaker associations to other chunks. A translation of this definition for the visual domain would be that a visual chunk is a collection of features that belong to the same perceptual group (see also Mahoney & Ullman 1988). Here, we will first address the neuronal mechanisms that may demarcate visual chunks. Then we critically evaluate to what extent these mechanisms might be responsible for the limit on the number of chunks that can be held in STM. We conclude that the clarity with which the psychophysical data point to the number four is not matched by a similarly clear limit imposed by physiological mechanisms.

The authors note that theoretical revision is likely to be required to refine and clarify a general theory of selection so that it is capable of accounting for selection processes across different subject matters. Some aspects of conceptual language from the biological realm work across all three domains. However, terminological revision may be required where concepts of “replication” and “information” pose particular problems for the general theory as that is applied to operant research.

To eliminate the leap of faith required to explain how visual consciousness arises from visual representation, O'Regan & Noë focus on the sensorimotor interaction with the outside world and ban internal representations from their account of vision. We argue that evidence for transsaccadic representations necessitates a central position for an internal, on-line stimulus rendition in any adequate theory of vision.

Social learning in the strict sense is learning by observation or instruction. Learning by observation appears to be an analogue process while learning by instruction is digital. In evolutionary biology this distinction is currently thought to have implications for the extent to which mechanisms can function successfully as an inheritance system in an evolutionary process.

Cowan argues that the true short-term memory (STM) capacity limit is about 4 items. Functional neuroimaging data converge with this conclusion, indicating distinct neural activity patterns depending on whether or not memory task-demands exceed this limit. STM for verbal information within that capacity invokes focal prefrontal cortical activation that increases with memory load. STM for verbal information exceeding that capacity invokes widespread prefrontal activation in regions associated with executive and attentional processes that may mediate chunking processes to accommodate STM capacity limits.

Heyes's (1998) triangulation approach to distinguishing a “theory” of mind (ToM) from a “theory” of behavior (ToB) in chimpanzees fails. The ToB theorist can appeal to the explicit training sessions and analogical reasoning to explain/predict the chimpanzees' behaviors. An alternative triangulation experiment is sketched, demonstrating how the removal of such training sessions paves the way toward solving the distinguishability problem.

The author does not wish to respond to the Commentators.

The commentators do not contest the target article's claim that there is no compelling evidence of theory of mind in primates, and recent empirical studies further support this view. If primates lack theory of mind, they may still have other behavior control mechanisms that are adaptive in complex social environments. The Somatic Marker Mechanism (SMM) is a candidate, but the SMM hypothesis postulates a much weaker effect of natural selection on social cognition than the theory of mind hypothesis (on inputs to cognitive mechanisms, not on the mechanisms themselves), and there is currently no evidence that it is specific to social stimuli or to primates. “Two Guesser” training would make the goggles test too chauvinistic, and in its current form the goggles problem could not be solved by physical matching because, while wearing goggles, an individual cannot see itself seeing.

The authors show that some long-standing confusions and problems can be avoided by thinking of perception in terms of sensorimotor contingencies, a close kin to my heterophenomenological approach (Dennett 1991). However, their claim that subjects do not have any commitments about the resolution of their visual fields is belied by the surprise routinely expressed by subjects when this is demonstrated to them.

The sacrilegious proposition of the existence of cultures in whales and dolphins should open the discussion of cultures in other animals, allowing us to find what is unique in human cultures. The ethnographic approach used by all anthropologists is the key in this investigation and revealed that cultural differences are present in animals and could result from different learning mechanisms.

Hull et al. see responses and properties of responses as units of selection in behavioral change. However, this perspective cannot account for goal-directed behavior in which organisms employ variable means to reliably attain intended consequences. An alternative perspective is offered in which the intended consequences (goals) of behavior serve as the units of selection in behavior change.

Cowan defines a chunk as “a collection of concepts that have strong associations to one another and much weaker associations to other chunks currently in use.” This definition does not impose any constraints on the nature and number of elements that can be bound into a chunk. We present an experiment to demonstrate that such limitations exist for visual short-term memory, and that their analysis may lead to important insights into properties of visual memory.

Schyns et al.'s (1998) target article raises a conflict between the need for a fixed functional architecture in an explanatory cognitive science and the need for a system to learn to detect new features. This conflict can be resolved by avoiding the scope slip in which properties of objects are erroneously viewed as being properties of their representations.

Selection in behavior analysis fits the criteria of replication, variation and interaction proposed by the authors except for information under replication. If information requires physical structure, behavior analysis does not fit that model because functional analysis may provide parallels between behavior, neurology, and biochemistry but not sequencing. The three sciences are not unified by the model but another is available.

Genomic conflicts are potentially involved in the evolution and maintenance of culture. Maternal genes contributing to neocortical development could influence biases in the acquisition of information. Specifically, relatedness asymmetries due to multiple paternity are expected to lead to an increased reliability and receptivity of matrilineally-transmitted information. This view complements the gene-culture coevolutionary model adopted by Rendell and Whitehead.

Some of the evidence for a “magical number 4” has come from the study of visual cognition, and Cowan reinterprets such evidence in terms of a single general limit on memory and attention. We evaluate this evidence, including some studies not mentioned by Cowan, and argue that limitations in visual processing are distinct from those involved in other memory phenomena.

Two visual phenomena are described in which oculomotor activity (saccades) changes our conscious perception: (1) some geometrical visual illusions disappear when saccades are suppressed, and (2) misperceptions occur in an antisaccade task with attentional precues. The first phenomenon shows that what we consciously perceive depends on how we look. The second phenomena indicates that a saccade itself may remain unconscious together with the accompanying changes of the retinal image.