Abstract
Targets in most search tasks can be defined either visually or categorically. This potential for a dual-code begs the question of how to weight these two factors when interpreting search behavior. Two experiments are reported to address this question. Exp 1 had Chinese and non-Chinese subjects search for a pre-designated Chinese character among other Chinese character distractors. A target character was presented for 1 sec, followed by a search display. Manipulations included target-display SOA (1.1, 1.5, or 6 sec), set size (1, 9 or 17 items), target presence or absence, and the number of strokes comprising the characters (5 or 15). Search slopes for non-Chinese subjects were dramatically affected by both SOA and stroke whereas Chinese subjects showed no SOA x Stroke interaction. Exp 2 used an eye movement methodology to quantify visual and categorical search patterns in terms of oculomotor variables. The stimuli were a subset of the same 5 and 15-stroke Chinese characters used in Exp 1. Displays consisted of 9 characters arranged in a circle around a central character. The task was to indicate whether the central character (the target) was present in the array. Eye movements of Chinese and non-Chinese subjects were analyzed in terms of target refixations (i.e., how often gaze was directed back to the central target during search). Chinese subjects rarely refixated the target regardless of stroke condition whereas the refixation rate for non-Chinese subjects was higher overall and interacted with visual complexity. Based on these refixation rates, we estimate that non-Chinese subjects were able to check about 3 distractors before visual interference forced them to refresh their memory for the target. Taken together, these findings highlight the need to consider categorical target definitions when interpreting search behavior, but also argue for a purely visual component to search, one that is highly susceptible to target pattern degradation.