Purchase this article with an account.
Jing Xing, Albert J. Ahumada; Estimation of human-observer templates in temporal-varying noise. Journal of Vision 2002;2(7):343. doi: https://doi.org/10.1167/2.7.343.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Estimation of human-observer templates in temporal-varying noise
The spatial templates that human-observers use for signal detection and discrimination can be estimated by constructing stimulus classification images (Ahumada and Beard, ARVO 1999). The purpose of this study was to elucidate whether the spatial-temporal templates could be estimated with the classification method. We performed signal detection experiments using a two interval forced choice (2IFC) method. The stimulus was a Gabor signal within additive white noise. Both the signal amplitude and the noise pattern varied during the 500ms display period. The 2 cycle/degree Gabor signal was presented in the middle of the noise image. The amplitude of the signal was varied according to a temporal function. Independent white noises were presented sequentially at a given rate during the display. The task was to determine whether the Gabor signal was presented or not. Each block included 100 trials. Within a block each noise sequence was repeated twice. Each of the two observers performed 40–60 blocks for every tested condition. We computed the linear classification images by averaging noises separately for each stimulus-response trial type. Results showed that at the high noise-changing rate (60Hz) the classification images contained no spatial templates or temporal templates. At low noise-changing rates (6Hz and 12Hz), the images had the templates similar to the shape of the Gabor signal. Moreover, the amplitudes of the cross-correlation between the classification images and the Gabor signal had temporal distributions similar to the temporal function of the signal, suggesting that the observers used temporal templates for signal detection. We further estimated the level of internal noise using the probabilities of agreed responses to the same noise sequence. At the high noise-changing rate, the internal noise overwhelmed the external noise. Thus the observers could not make a decision based on the contribution of each external noise pixel.
This PDF is available to Subscribers Only