[PS-3.14]Statistically defined chunks show similar within/ between-object processing to real objects

Lengyel, G. , Nagy, M. & Fiser, J.

Department of Cognitive Science, Central European University, Budapest, Hungary

Although many studies demonstrated visual statistical learning (VSL), only few focused on its consequences on perception. We investigated how acquiring visual regularities constrain subsequent visual processing.
Our method was based on the well-documented within-object processing advantage over across-object processing. We combined a VSL paradigm with a visual search task to assess whether participants detect a target better within statistical chunks than across chunks. In alternating blocks of observation and search trials, complex multi-shape visual scenes were presented, which unbeknownst to the participants, were built from pairs of abstract shapes without any segmentation cues. The visual chunks could only be extracted by tracking the statistical contingencies of shapes across scenes. During observation, participants passively observed the scenes, while during search, they performed a 3-AFC task deciding whether T letters appearing in the shapes formed horizontal or vertical pairs. Despite identical distance between positions, participants performed significantly better if the target pairs appeared within a statistical chunk than across two chunks.
Thus, statistical contingencies facilitate visual processing of elements that belong to the same statistical chunk. This similarity between the effects of true objects and statistical chunks support the notion that VSL has a central role in the emergence of object representations.