Purchase this article with an account.
Yoon Bai, Yuzhi Chen, Wilson Geisler, Eyal Seidemann; Human and monkey detection performance in natural images compared with V1 population responses. Journal of Vision 2015;15(12):577. doi: https://doi.org/10.1167/15.12.577.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Detection is a fundamental task that is critical to visual behavior. Our aim here was to measure and model behavioral and neurophysiological performance for detection of targets under naturalistic conditions. We first measured behavioral detection performance in three humans and two macaque monkeys. Localized oriented stimuli were presented for 250 ms and contrast psychometric functions were measured on uniform backgrounds and for several contrasts of the natural image backgrounds. We found that (i) threshold contrast power is a linear function of background contrast power for both humans and macaques, and (ii) the relative threshold functions (the ratio of thresholds under natural and uniform background as a function of background contrast) for humans and macaques are in good agreement, although (iii) the macaques are less sensitive overall. Subsequently, we used voltage-sensitive dye imaging (VSDI) to measure the neural population activity in V1 for the same stimuli, while the monkeys held fixation. The spatial scale of VSDI measurements was sufficient to resolve orientation columns over the whole region activated by the target. The average VSDI response to high-contrast targets on a uniform background was used to define a “matched template”. Then, for each condition we applied the template to the responses on target + background trials and background alone trials and computed the signal-to-noise ratio (d’), which specifies the neurometric functions. We found that (i) the threshold contrast power for the VSDI responses is also linear with background contrast power and (ii) the relative thresholds for the VSDI responses are in fairly good agreement with the human and macaque behavioral thresholds. Conclusions: the macaque is a good model of human detection behavior in natural images, and population responses at the columnar scale in V1 predict behavioral detection performance in natural images.
Meeting abstract presented at VSS 2015
This PDF is available to Subscribers Only