The representation of contrast and other stimulus parameters by single neurons in area 17 of the cat

Publisher Website
Google Scholar
Abstract
The responses of neurons in area 17 were tested as a function of various stimulus parameters. The thresholds of individual cortical neurons were at contrasts between 0.01 and 0.1 (increment of 0.5×10−1 cd/m−2 on a background of 3 cd/m−2), the dynamic ranges were 1.0–2.0 log units of increment, and all cells showed a response decrease at increments above a certain maximum (supersaturation response). The averaged contrast/response curve for all neurons was S-shaped in the logarithmic plot, had a dynamic range of 2.5 log units, reached its maximum at a contrast of 0.75 and supersaturated above this level. The contrast/ sensitivity curves changed their slope under different stimulus conditions. They became flatter when the non-dominant eye was stimuated as compared to dominant eye stimulation or when the stimulation was done at a non-optimal orientation or direction, and they became steeper when both eyes were stimulated. But the maximum was reached at the same contrast and supersaturation was seen above maximum contrast no matter whether a cell was stronlgy (e.g. binocular stimulation at optimal orientation) or weakly excited (non-dominant or non-optimal orientation stimulation). After normalization, the averaged population contrast/response curves were virtually identical at all stimulus conditions. It was concluded, that range as well as maximum and supersaturation of cortical contrast/response curves are determined before the input reaches the cortex, and that the cortical cells summate, essentially, linearly. The findings furthermore demonstrate that the supersaturation of the cortical input must be due to subtractive inhibition, and that the same is true for the orientation sensitive inhibition in the cortex itself. Both, the peripheral contrast and the cortical orientation dependant inhibition cannot be explained by multiplicative inhibition. The fact, that the responses of neurons depend on many variables relativates their significance for feature representation.