Abstract
Many models of selection in visual search predict saccades to be biased towards conspicuous objects (often referred to as salient). In accordance, it has been demonstrated that initial saccades are indeed biased towards the most conspicuous element. Yet, only few studies have explicitly investigated how elements of different conspicuity affect scanning order beyond the initial saccade. One recent study (Siebold, Van Zoest & Donk, 2011) does investigate whether the saccade following the first is biased towards the most conspicuous of the potential candidates for the second saccade. Despite conspicuity differences between these candidates, no bias towards the most conspicuous candidate was found for the second saccade. Upon this finding they conclude that selection beyond the initial saccade is primarily under top-down control. However, this conclusion, which contrasts with many influential models of selection, is based on only a single conspicuity manipulation (three deviating orientations on a grid of vertical non-targets). Here we investigate whether their finding can indeed be generalized by introducing luminance variations to create conspicuous locations. In our experiment, three annuli of varying luminance were placed around a vertical target and two horizontal distractors to introduce differences in conspicuity. Observers were asked to find the vertical target as quickly as possible and were informed that luminance differences between the annuli were not predictive of the target’s location. In contrast to the findings of Siebold et al. we do find a large bias towards the most conspicuous candidate for the second saccade. This bias is present even for the longer inter-saccade intervals. Thus, bottom-up factors can still play a role in eye movements beyond the initial saccade. Whether second saccades are biased towards more conspicuous elements appears to depend on the type of contrast underlying the conspicuity.
Meeting abstract presented at VSS 2012