Purchase this article with an account.
Joseph Schmidt, Gregory Zelinsky; Manipulating the availability of visual information in search. Journal of Vision 2007;7(9):715. doi: 10.1167/7.9.715.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Visual search studies typically assume that an exact picture of the target is available to guide the search process, but targets are often defined categorically, with varying degrees of visual specificity. Is search still guided to complex targets in the absence of pictorial target descriptions? We manipulated the availability of target visual information in three target preview conditions: (1) a picture of the target (e.g., an orange), (2) a visually precise target label (e.g., the word “orange”), and (3) a visually imprecise target label (e.g., the word “fruit”). The preview was followed by a search display depicting 5 pictures of photorealistic objects; no pictures or object categories were repeated. Consistent with the literature, we found strong evidence for guidance to pictorial targets. Compared to the non-pictorial conditions, pictorial search was faster (1961 ms vs. 2675 ms), required fewer fixations to reach the target (1.94 vs. 2.87), and subjects fixated on fewer distractor objects before the target (.57 vs. 1.17). Importantly, guidance differences were also found between the two non-pictorial target conditions. Compared to imprecise-label search, precise-label search was again faster (2510 ms vs. 2840 ms) and resulted in fewer fixations to the target (2.71 vs. 3.03) and fewer fixated distractors before the target (1.07 vs. 1.26). However, in all conditions, even with an imprecise target label, we found above chance levels of guidance as measured by the cumulative probability of target fixation. We interpret these data as broadly suggesting that guidance is related to the availability of target-defining visual information. Moreover, this visual information need not be obtained from a picture of the target. Although generally less precise, to the extent that visual information can be extracted from a target label and loaded into working memory, this information too can be used to guide search.
This PDF is available to Subscribers Only