ABSTRACT
Gestalt theory drew upon facts of magnetic and electrical fields in physics and speculated about similar field influences in the brain. That is, it considered that both perception and its underlying physiology were modelled on organising principles that could not be reduced to the operations of simplified elements. The Gestalt psychologists emphasised the holistic aspects of perception rather than the analytic: They did not believe that perception could be explained by reducing either stimuli or responses to their simplest levels. This contrast between holistic and analytic approaches to perception is one that remains in theories of perception. Ironically, the subsequent advances in the neurophysiology of vision have been based on precisely the reductionist ideas that the Gestalt psychologists rejected. The detailed study of the structure and function of nerve cells has suggested the ways in which complex patterns of light striking the retina are analysed in terms of simplified features that they contain. Thus, cortical cells could be excited by chromatic or spatio-temporal changes in the pattern of light falling over particular regions of the retina. Thus features of the stimulus (like retinal location, contour orientation, direction of motion, and colour) are thought to be extracted by single cells in the visual cortex of higher mammals, and they are preferentially processed at subsequent cortical sites. The presence of such feature detectors and cortical maps has been related to certain simple phenomena like visual aftereffects and illusions. However, the range of phenomena that has been successfully interpreted in neurophysiological terms remains limited. The more recent developments in modelling the behaviour of neural nets, derived from computer science, may provide a basis for returning to a more holistic approach to perception. Such networks have capacities for learning that are not localised in particular units, but that are a property of the net as a whole.
