Purchase this article with an account.
Zhaoping Li, Robert J Snowden; A psychophysical test of the saliency map in V1. Journal of Vision 2003;3(9):208. doi: 10.1167/3.9.208.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
It has been observed (Snowden J. Exp. Psychol. 1998) that although random variations in colour of the stimulus bars does not disrupt visual search for a target bar defined by its unique orientation among distractor bars of a different orientation, it does impair texture segmentation between textures each defined by a uniform orientation of texture bars. These observations can be accounted for by the proposal that V1 produces a saliency map, as demonstrated by a V1 model (Zhaoping, Soc. Neurosci. abstract, 2002). The ease of the task is determined by the saliency of the target or the texture border, which in turn is determined by the evoked V1 responses subject to contextual influences from the contextual bars. Random colour variations affect performances by altering the contextual influences via intra-cortical interactions. The model predicts that the colour interference on the texture segmentation defined by orientation should become more pronounced as the stimulus bars become thicker and shorter. To test this prediction, subjects reported the orientation, either vertical or horizontal, of an elongated target texture region defined by an area of 2×8 bars that differed in orientation by 90 degrees from the background texture bars. The texture bars were either all fat or all thin. The two cases were equated such that the fat bars were twice as fat as the thin bars which were in turn twice as long. The reaction times of the subjects were measured, and as expected, they were longer with than without the random colour variations of the texture bars. This reaction time increment was used as a measure of the colour interference. The results showed that the fat texture bars resulted in at least a 65% increases in colour interference relative to the case with the thin bars. These results are consistent with the model prediction and thus support the proposal of a saliency map from V1.
This PDF is available to Subscribers Only