Purchase this article with an account.
Louis Chan, William Hayward; How do people quit visual search? Justifications for a deadline model. Journal of Vision 2011;11(11):1305. doi: 10.1167/11.11.1305.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
How do people decide when to stop a search when there is no target? Most discussions on this question revolve around two models. In their simplest form, the drift diffusion model suggests that a decision is reached when sufficient evidence of target presence or absence is accumulated, and the deadline model suggests that people estimate a deadline for each search and stop when it is due. We conducted an experiment that tests against a deadline model, which assumes that search deadlines are calibrated for the observer's search experience in target-present trials. We “produced” some experimental blocks that were generally faster (or slower) by choosing more (or fewer) easy trials in these blocks, while the target-absent trials in these blocks were not censored. We expected a sooner search deadline for a “fast” block, rendering faster absent RTs. Results confirmed this expectation for large set sizes, but not for a small set size, justifying a deadline model. Analysis of data suggests that the faster absent RTs were not due to rhythmic responses. It is documented elsewhere that median absent RT generally overestimates miss rate with regard to the corresponding present RT distribution, violating the deadline model (Wolfe, Palmer, & Horowitz, 2010). However, we replicated this violation only for a small set size, but estimations based on median absent RT were very accurate for large set sizes. This mirrored the above results. In light of the present results, we suggest that observers generally adopt a deadline approach, unless a search is expedited by some evidence of an absence of target. In our case, evidence of absence may come about when a single glimpse is sufficient to confirm all items as distractors. In other cases, uniformity of stimulus or learned statistical signals of the scene may contribute.
This PDF is available to Subscribers Only