Prefixational Object Perception in Scenes: Objects Popping Out of Schemas

Semantic influences of context on the ease of object identification in real-world scenes are commonly accepted, but when eye movements are taken into account the unanimity dwindles. The question is whether object-in-scene semantics only come into play when the object is foveated or already have an impact during extrafoveal, prefixational object processing, and if so, whether semantic consistency or inconsistency would enhance extrafoveal processing. A theoretical framework (mismatch theory) is borrowed from reading and word recognition to support the hypothesis that both consistency and inconsistency may facilitate extrafoveal processing. Two earlier studies of context-sensitive object identification in scenes are reanalyzed to provide an initial test of the validity of the theoretical framework. Analysis of context effects on gaze shift frequency, gaze shift destination and gaze shift latencies suggests that in the earliest stages of scene exploration scene-inconsistent objects are more salient saccade targets. However, this does not appear to be a popout phenomenon based on attentional capture by schema-inconsistent objects, but rather reflects a smaller useful field of view for such objects. If attention is selectively captured at the onset of scene exploration, it appears to be by schema-consistent rather than inconsistent objects.