Purchase this article with an account.
Albert Ahumada, Keith Billington, Jerry Kaiwi; Searching the horizon for small targets. Journal of Vision 2011;11(11):500. doi: 10.1167/11.11.500.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We have simulated the search for small targets on the horizon using a display that has only a solid dark blue ocean below a solid light blue sky. The target is a single pixel the same color as the ocean in the sky adjacent to the ocean. Our goal was to model this one-dimensional search process. The first model we tried was random search constrained by an initial a priori distribution and no memory. This model predicts that the standard deviation of the number of fixations is about the same as the mean number of fixations, while the actual standard deviation was closer to the prediction of perfect memory. A simple model with a saccade-distance penalty and inhibition-of-return with a temporal decay was able to predict the first order distribution of saccades and the fixations standard deviation. However, when we looked at the distribution of saccades conditional on the location of the previous saccade we found that although most saccades continue in the same direction, there were a large number that went backwards and that the spatial distribution of these saccades showed no tendency to avoid the location of the previous fixation. Search models usually contain bottom-up processes, such as a saliency map, and top-down processes, such as a priori distributions over the possible locations to be searched. A situation that needs neither of these features is the search for a very small target near the horizon when the sky and the ocean are clear. How can search performance be lacking in memory (Horowitz & Wolfe, 1998, Nature), yet still have overall statistics that are so nearly optimal (Najemnik & Geisler, 2008, J Vision). We are now trying to find heuristic search strategies for this situation that have these properties.
This PDF is available to Subscribers Only