Commentary on Hurford

Word Count:
Abstract: 63
Main Text: 750
References: 62
Total Text: 929

Grammar originates in action planning, not in cognitive and sensorimotor visual systems

Bruce Bridgeman

Department of Psychology

University of California

Social Sciences 2 office (831) 459 4005

Santa Cruz, Ca. 95064 fax (831) 459 3519

USA

bruceb@ucsc.edu

http://psych.ucsc.edu/Faculty/bBridge.shtml

Abstract: While the predicate(x) structure requires close coordination of subject and predicate, both represented in consciousness, the cognitive (ventral) and sensorimotor (dorsal) pathways operate in parallel. Sensorimotor information is unconscious, and can contradict cognitive spatial information. A more likely origin of linguistic grammar lies in the mammalian action planning process. Neurological machinery evolved for planning of action sequences becomes applied to planning communicatory sequences.

It is tempting to relate ideas in linguistics with ideas in neurophysiology, because at base much of linguistics is about the design and operation of a neurophysiological machine, in the language areas of the brain. In the spirit of consilience, such efforts are necessary. Hurford’s effort at tying together a formal logic developed within linguistics with the interactions between brain areas echoes another effort in BBS a decade ago (Landau & Jackendoff, 1993). Like that effort, though, this one founders by proposing a parallel that on closer examination turns out to be illusory.

The heart of Hurford’s article, identifying the dorsal and ventral streams of visual processing with a logical predicate(x) structure, misses the mark because there is a tight logical relationship between subject and predicate; but information in the two visual streams can be independent and even contradictory, running in parallel to subserve different functions (Bridgeman et al. 2000; Milner & Goodale 1995). The dorsal/ventral terminology is somewhat deceptive, for some cortical areas that are anatomically dorsal to the primary visual cortex are shared by both pathways, or even belong to the ‘ventral’ pathway. Terms that capture the contrast in functions of the two pathways are more useful. ‘What’ and ‘where’, as used by Hurford, are misleading because both pathways carry useful ‘where’ information — it’s just that the ‘where’ information in the two pathways is sometimes contradictory. The terms ‘cognitive’ and ‘sensorimotor’ are preferable, as they describe the distinct functions of the pathways, while Milner and Goodale (1995) suggest ‘what’ and ‘how’. Functional terms are preferable because they are less likely to lead to oversimplification or overinterpretation.

The idea of using logical grammar notations developed within linguistics for describing the information processing in brains is a productive one that promises to enrich neuroscience. The literal application of the logical structures to describe information processing within the two-visual-systems context, however, is wide of the mark because the linguistic structures and the logical structures of visual architecture are not parallel.

Subject & predicate are both conscious, in the minimal sense that one can talk about them. Their identities and relationship can be described, their application can be planned in language, and they define inseparable parts of a single linguistic act. The sensorimotor pathway, however, can function without cognitive participation and without conscious intervention all the way from stimulus to response, an example of ‘vertical modularity’ (Bridgeman, 1999).

A recent method of dissociating cognitive and sensorimotor information exploits the Roelofs effect (Bridgeman et al., 2000), without confounds from motion of the eye or of the visual stimulus. The Roelofs effect is a tendency to misperceive the position of a target presented along with an off-center background. A rectangular frame offset to an observer’s left, for instance, causes the position of a target presented within the frame to be mislocalized to the right. Despite this mislocalization, observers could jab the target accurately, without the frame affecting their behavior. The effects may be due to the frame biasing the observers’ subjective straight ahead, stored unconsciously in a sensorimotor system.

Anatomical connections between dorsal & ventral streams do not contradict the separability of their functions, any more than communication between two people contradicts their distinctness. Communication between the two streams is needed to initiate action (usually a cognitive-system function), to monitor progress in the execution of the action, and to modify goals of actions.

Rizzolatti & Arbib (1998) also use the language of language to describe neurophysiological relationships, but they explicitly specify a prelinguistic grammar, to distinguish it from spoken language. Thus their ‘grammar’ refers only to a set of rules by which the brain processes information. It is unrelated to language in the usual sense. Semantics in vision and in neurophysiology refers to a relation of images with meaning; its relation to language is more metaphorical than literal.

If the distinction between cognitive and sensorimotor pathways of the visual system does not offer a source for the evolution of the logical relations necessary for language, what does? A more likely alternative is the planning process that all mammals possess, and that becomes particularly important and well-developed in primates. Plans for action exist separate from the sensory or motor worlds, and their steps must be executed in a particular order to be effective. Grammar may have appropriated an existing capability for planning of action sequences to the planning of communicatory sequences (Bridgeman, 1992). Language, then, is a new capability built mostly of old parts, but the parts originate in motor planning, not in visual coding.

References

Bridgeman, B. (1992) On the origin of consciousness and language, Psycoloquy (refereed electronic journal) 92.3.17. http://psycprints.ecs.soton.ac.uk/archive/00000239/

Bridgeman, B. (1999) Vertical Modularity in the Visual System. In: Stratification in cognition and consciousness, ed. B.M. Velichkovsky & B. Challis. John Benjamins Publishing.

Bridgeman, B., Gemmer, A., Forsman, T. & Huemer, V. (2000) Properties of the sensorimotor branch of the visual system, Vision Research 40:3539-52.