Purchase this article with an account.
Rachit Dubey, Chun Siong Soon, Po-Jang (Brown) Hsieh; A blurring based model of peripheral vision predicts visual search performances. Journal of Vision 2014;14(10):935. doi: 10.1167/14.10.935.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The role of peripheral vision has been under-specified in most studies of visual search. Since attention is expensive, the peripheral visual field may provide information to help identify potential search targets. A recent work argued that information available in peripheral vision is a key determinant of search difficulty (Rosenholtz et al, 2012). Here, we build upon Rosenholtz's patch-based discrimination strategy within the periphery, but instead of summary statistic computation, we consider how the larger receptive field sizes of peripheral neurons affect potential target identification. We suggest that larger receptive field sizes lead to a loss of spatial specificity, resulting in "blurring" of target and distracter items. Difficult search might arise when the blurred representations in the periphery are indistinguishable between target present and target absent patches. We first carried out seven different search tasks of varying difficulty in which participants had to find a target amongst an array of distracter items. For each search task, we then measured the discrimination performance on target present vs. target absent patch representations generated by our model. If a search condition is easy, then discriminating between a target present patch representation and target absent representation should also be easy, and vice versa. Results demonstrate a strong correlation between search performance and discriminability of patch representations generated by our model (R2 = 0.94, p <0.05). Our results support the idea that visual search involves discriminability of peripheral patches, and is affected by the receptive field sizes of peripheral processing stream neurons, as proposed in our model.
Meeting abstract presented at VSS 2014
This PDF is available to Subscribers Only