Abstract
The rich literature on scene understanding has shown that human observers may quickly extract and represent several types of high-level information when viewing a scene. It has been suggested that the category of a scene is automatically accessed during viewing, but no previous studies have examined whether other types of knowledge concerning the scene are accessed in an automatic fashion as well. In this study, we examined whether such automaticity exists for actions and objects that are highly associated with the scene's schema. We used a lexical-decision task combined with a picture-word interference paradigm, in which participants had to decide whether a string of letters presented on a scene image was a word or a pseudoword. Words were scene, action and object labels, highly associated or weakly associated with the image. We manipulated task difficulty by varying the similarity between the pseudowords and the words from which they were derived. The stimuli were presented until response. We found slower response times in the weakly versus highly associated condition only for scene labels. Moreover, this interference effect was independent of label's length and of task difficulty. This shows that it was not modulated by cognitive load and available attentional resources. Our results suggest that automaticity of knowledge access may be restricted to the category of the scene when primed by the precise scene label, without embracing other components of the scene's schema, which might therefore be inferred as a second processing step during scene understanding.
Meeting abstract presented at VSS 2018