Vision (Roger Watt )

Subject: Vision
From:    Roger Watt  <r.j.watt(at)STIR.AC.UK>
Date:    Wed, 28 May 1997 17:54:55 +0100

Maybe I could help tidy things up a bit here as someone whose main research lies in vision. The argument that coupled oscillations in the visual system are caused by high frequency eye movements does not seem very viable. The eye shows a tremor at between 30-70Hz, but of very low amplitude (typically around 1/4 of the half-height width of the optical point spread function). These will not normally lead to changes in the retinal stimulation of a sufficiently high contrast to cause measurable temporal effects on the visual system. More interestingly, perhaps for this discussion, in the visual system the various different attributes, such as colour, contrast, motion, depth and so on are all subject to very different speeds of neuronal conduction and different processing rates. Hence at the point where it is appropriate to ask how the information is combined into a single coherent representation, that information is not synchronous in arrival time. If it turns out to be synchronous nevertheless, then the implication must be that it has been pushed into synchrony at that stage. In vision, there is exquisite spatial precision as well as resolution but much less precision in the time domain, despite reasonable temporal resolution. Presumably this reflects the constraint that on the one hand photon noise is a problem at lower light levels, and on the other hand that most visual events to be treated as a single object are spatially delimited. I wonder whether anyone has any views on the situation in the auditory system: does it invest in high temporal precision (as well as high frequency resolution)? do events need to be temporally contiguous and correlated to become a single object? do events need to be spatially contiguous and correlated to become a single object? The drift of the discussion so far suggests that the auditory system maps temporal frequency onto spatial layout. How about temporal information (ie envelope information)? Roger Watt Professor of Psychology University of Stirling Scotland

This message came from the mail archive
maintained by:
DAn Ellis <>
Electrical Engineering Dept., Columbia University