Abstract
When observers see several instances of a scene category, the objects and relations and category should be primed, according to mainstream cognitive theories. This priming should then facilitate perception of a new instance of the scene. We tested this idea with the powerful full-report method, which provides rich data about what observers encode from a scene. Observers saw a series of the prime-scenes (1 sec each), followed by a new target scene. The primes were either from the same category as the target (congruent), from a different category (incongruent), or were a neutral sequence. The cognitive prediction is that congruent primes should prime a network for that scene, and should produce more detailed and accurate full reports than the other two prime-types. We tested this prediction in two experiments with differing sets of stimuli; one was standard scenes and one was less typical, creative scenes. Initial analyses indicate that observers reports were detailed and highly accurate in general. However, there was no evidence that prime type influenced the richness of the full reports; word-counts were similar for targets in each of the three priming conditions, in both experiments (23 words/picture). We suggest that category priming effects may be limited to when primes induce observers to prepare a response or interpretation in advance of the target. When the task is more open-ended, as in the present case, there may be no effects of prior semantic and visual information on the efficiency of scene perception.
Meeting abstract presented at VSS 2018