Object-Knowledge Sharpens Vision
It is widely held that the brain derives object information from features of images. The predictive coding approach holds that object information should influence feature processing. We tested this idea using ambiguous stimuli such as the one on the right. Initially such image appears as an incoherent set of blobs but after seeing the full image it appears coherent (ie containing objects). We report that peoples’ perception of such images changed how sensitive they were to the introduction of new features to the image, even though sensory stimulation was identical. This shows that object representations in the brain sharpen feature-detection, optimising performance for the current context.
Author: Prof Steven Dakin
Collaborators: Christop Teufel (Cardiff, UK) Paul Fletcher (Cambridge, UK)
Published: Teufel, Dakin & Fletcher (2018) Prior object-knowledge sharpens properties of early visual feature-detectors, Scientific Reports #10853.