Abstract
Visual searches for targets indicated by picture cues consistently result in better performance than those in which targets are indicated using word cues, with no definite answer as to why. (Wolfe, Horowitz, Kenner, Hyle & Vasan, 2004; Bravo & Farid, 2009; Schmidt & Zelinsky, 2009). One explanation for this phenomenon is that the two cue modalities provide different amounts of information. Picture cues provide specific information about the target features, while word cues evoke a vague distribution of possible target features. Our study tests this explanation by quantitatively varying the feature specificity of picture cues. If increased specificity explains the search advantage of picture cues, then reducing cue specificity should reduce performance. Participants performed a conjunction search for an oval or rectangle of a particular color in an array of similar elements after briefly observing a cue indicating the target, which was always present. The search cue itself was a picture, word, or nonspecific cue. Picture cues could exactly match the target, or be drawn from a distribution centered on the target's feature values. Picture cue specificity was manipulated by changing the feature variance of the picture cue color distributions, which were highly narrow, equivalent to those applied to word cues, or wider. We characterized the specificity of word cues using a hue selection task like that of Bae, Olkkonen, Allred, Wilson & Flombaum (2014). Search performance was measured using reaction time, which increased as picture cue specificity decreased. Overall these results are consistent with specificity being the source of picture cue search performance.
Meeting abstract presented at VSS 2018