Edelman's target article purports to be about veridical representations. I argue that it would be a mistake to think it has much to do with veridicality as normally understood.

Edelman's memory-based approach to visual representation is preferable to parts-based alternatives. However, the existing algorithms for learning the shape prototypes are biologically implausible because they are nonlocal and nonconstructive. There is an alternative learning algorithm that constructs a mixture model of prototypes on-line, using only local information, and is more biologically plausible and may perform sufficiently well.

The attempt to provide a faithful mapping from distal shape space to proximal state space in terms of a higher order relationship defined over proximal similarity space stumbles on the context sensitivity of higher order relationships. Proportional analogy problems using quadruples of figures illustrate that for a number of interesting perceptual problems, the number of relevant dimensions cannot be reduced.

Proximal mirroring of distal similarities is, at present, the only solution to the problem of representation that is both theoretically sound (for reasons discussed in the target article) and practically feasible (as attested by the performance of the Chorus model). Augmenting the latter by a capability to refer selectively to retinotopically defined object fragments should lead to a comprehensive theory of shape processing.

The species-specific organizational property of speech is a continual mouth open-close alternation, the two phases of which are subject to continual articulatory modulation. The cycle constitutes the syllable, and the open and closed phases are segments – vowels and consonants, respectively. The fact that segmental serial ordering errors in normal adults obey syllable structure constraints suggests that syllabic “frames” and segmental “content” elements are separately controlled in the speech production process. The frames may derive from cycles of mandibular oscillation present in humans from babbling onset, which are responsible for the open-close alternation. These communication- related frames perhaps first evolved when the ingestion-related cyclicities of mandibular oscillation (associated with mastication [chewing] sucking and licking) took on communicative significance as lipsmacks, tonguesmacks, and teeth chatters – displays that are prominent in many nonhuman primates. The new role of Broca's area and its surround in human vocal communication may have derived from its evolutionary history as the main cortical center for the control of ingestive processes. The frame and content components of speech may have subsequently evolved separate realizations within two general purpose primate motor control systems: (1) a motivation-related medial “intrinsic” system, including anterior cingulate cortex and the supplementary motor area, for self-generated behavior, formerly responsible for ancestral vocalization control and now also responsible for frames, and (2) a lateral “extrinsic” system, including Broca's area and surround, and Wernicke's area, specialized for response to external input (and therefore the emergent vocal learning capacity) and more responsible for content.

We applaud the spirit of MacNeilage's attempts to better explain the evolution and cortical control of speech by drawing on the vast literature in nonhuman primate neurobiology. However, he oversimplifies motor cortical fields and their known individual functions to such an extent that he undermines the value of his effort. In particular, MacNeilage has lumped together the functional characteristics across multiple mesial and lateral motor cortex fields, inadvertantly creating two hypothetical centers that simply may not exist.

We agree with MacNeilage's claim that speech stems from a volitional vocalization pathway between the cingulate and the supplementary motor area (SMA). We add the vocal self- monitoring system as the first recruitment of the Broca-Wernicke circuit. SMA control for “frames” is supported by wrong consonant-vowel recurring utterance aphasia and an imaging study of quasi-reiterant speech. The role of Broca's area is questioned in the emergence of “content,” because a primary motor mapping, embodying peripheral constraints, seems sufficient. Finally, we reject a uniquely peripheral account of speech emergence.

The jaw movements of speech are most probably derived from jaw movements associated with vocalisation. Cyclicity does not argue strongly for derivation from a cyclic pattern, because it arises readily in any system with feedback control. The appearance of regular repetition as a part of ritualisation of a display may have been important.

MacNeilage's target article develops a theory for the evolution of human speech articulation along the lines of “slot- filler” structure. His content/frame schema commits him to the tenets of embodiment, muscle sense, and a memory for speech. My commentary ties these aspects together in their historical and current perspective.

If syllables are the link between nonhuman calls and human speech, as MacNeilage suggests, then that link is actually revealed in the “syllabic” sounds of the 3-month-old infant, well before the reduplicative babbling of the 8-month-old. Anatomical, acoustic, cognitive, and social perceptual evidence supports this earlier landmark.

The frame/content theory suggests that chewing was tinkered into speaking. A simple extrapolation of this approach suggests that syllable structure may have been tinkered into syntax. That would explain the widely noted parallels between sentence structure and syllable structure, and also the otherwise mysterious pervasiveness of the grammatical distinction between sentences and noun phrases.

There is evidence of reciprocal connectivity, similarity of oscillatory responses to stimulation of multiple motor and somatosensory cortices, whole system oscillation, and short- latency responses to behavioral perturbation. These suggest that frame/content may be instantiated by overlapping neural populations, and that the genesis of frame oscillations may be profitably thought of as an emergent property of a distributed neural system.

The cyclic nature of speech production, as manifested in the syllabic organization of spoken language, is likely to reflect general properties of sensori-motor integration rather than merely a phylogenetic progression from mastication, teeth chattering, and lipsmacks. The temporal properties of spontaneous speech reflect the entropy of its underlying constituents and are optimized for rapid transmission and decoding of linguistic information conveyed by a complex constellation of acoustic and visual cues, suggesting that the dawn of human language may have occurred when the articulatory cycle was efficiently yoked to the temporal dynamics of sensory coding and rapid retrieval from referential memory.

The frame/content theory of speech production is restricted to output mechanisms in the target article; we suggest that these ideas might best be viewed in the context of language production proceeding as a coordinated dynamical whole. The role of the medial premotor system in generating frames matches the important role it may play in the internally dependent timing of motor acts. The proposed coevolution of cortical architectonics and language production mechanisms suggests a significant divergence between primate and cetacean species corresponding to major differences in areal differentiation trends in cerebral cortex.

Constraints on the types of speech errors observed can be accounted for by a frame/content distinction, but connectionist modeling shows that they do not require this distinction. The constraints may arise instead from the statistical properties of our language, in particular, the sequential biases observed in the vocabulary. Nevertheless, there might still be a role for the frame/content distinction in syntactic planning.

This commentary discusses whether abstract metrical frames are stored. For stress-assigning languages (e.g., Dutch and English), which have a dominant stress pattern, metrical frames are stored only for words that deviate from the default stress pattern. The majority of the words in these languages are produced without retrieving any independent syllabic or metrical frame.

The segmentation of phonation by articulation is a characteristic feature of speech that distinguishes it from most nonhuman vocalizations. However, apart from the trivial fact that speech uses some of the same muscles and, hence the same motoneurons and motorcortical areas used in chewing, there is no convincing evidence that syllable segmentation relies on the same pattern generator as mastication. Evidence for a differential cortical representation of syllable segmentation (“frame”) and syllable “content” is also meager.

Many animals, including dogs, produce vocal signals in which their mouths open and close producing “syllables.” In contrast, the vocal signals of species other than humans are tied to emotional states. The Broca's-Wernicke's area model of the brain bases of language is wrong.

It is likely that the system controlling speech has evolved from the one that controls feeding. However, the idea that frames and content are programmed independently by two different cortical areas is not plausible. Models of the speech control system must also take into account the need to coordinate the respiratory, laryngeal, and articulatory musculature.

The frame/content theory justifiably makes tinkering an important explanatory principle. However, tinkering is linked to the accidental and, if completely decoupled from functional constraints, it could potentially play the role of an “idiosyncracy generator,” thus offering a sort of “evolutionary” alibi for the Chomskyan paradigm – the approach to language that MacNeilage most emphatically rejects. To block that line of reasoning, it should be made clear that evolutionary opportunism always operates within the constraints of selection.