Subject: Re: Temporary binding of descriptions in perception From: Neil Todd <todd(at)HERA.PSY.MAN.AC.UK> Date: Wed, 21 May 1997 12:00:39 +0100Dear Al, Alain, Pierre There is a fundamental difference between the collectivity of bound features which constitute a particular object, a concept of the collectivity of bound features which constitute a particular object and the word used to refer to the collectivity of bound features. The cortex does not have a problem binding the features of multiple copies of the same object, i.e. a green ball on a red table next to a red ball on a green table, because they are spatially segregated in the visual cortex and there are multiple copies of the individual feature detectors, i.e. colour, edges, corners, etc. distributed throughout the visual cortex. The problem arises when we consider the representation of the concept of the object constituted by the collectivity of bound features and a word which we humans might use to refer to the concept. The fact that pongids cannot speak but are able to achieve a vocabulary of up to 130 signs indicates that it is possible to have a concept of things and actions without words, or even a faculty for words. In the case of normal humans the evidence appears to be that visual and auditory representations of a word are processed independently (Peterson et al, 1988). This implies that there are at least two representations of the same word. Further insight into the representation of words is given by considering the case of patients with particular kinds of aphasia such as anomia. These are usually caused by lesions in the posterior speech zone in the left hemisphere. Patients with anomia have deficits in (a) synonym judgements (b) naming to definition (c) categorisation (d) property judgements. Common area of damage for these patients is posterior temporal-inferior parietal (T-IP) region => single word semantic processing in left T-IP region. The TPO junction includes posterior 22, 21 and 39, 37 (phylogentically recent areas). Possible function of earlier part (22, 21) is higher-level elaboration exclusively within domain of auditory and visual domains. Possible function of more recent parts (39, 37) are multimodal processing and integration with language. However, there appears to be even further detailed dissociations. For example, certain patients may have very specific categorical deficits, e.g. difficulty naming fruits or animals, certain lesions may erase very selective portions of semantic memory. Some patients have shown a distinct deficit in verbally accessing visual properties of semantic categories, such as animals, although can make judgements in other non-verbal tasks. This dissociation may be associated with relationship between inferior temporal region (visual processing) and left superior posterior T-IP region (lexical semantic processing). The above implies that in the normal state purely verbal knowledge is separate from higher-order visual and other perceptual information, but all these systems are highly linked. The verbal system contains tags or connections to visual and auditory systems. Representations then, reside in connections between the verbal and sensory systems. So long as the sensory/perceptual systems can segregate and represent multiple copies of the same object then this separation will be maintained in the connections to the verbal system. So, my guess is that, in response to Al and Dick, the independent habituation of the shared word at left and right is in the independent connections between the segregated voices and the verbal system. The habituation is not the node, but the connections to the node. Neil