Abstract
Although Categorical Perception (CP) can be acquired as a result of learning, the mechanisms involved remain unknown and there is debate about whether it is a perceptual phenomenon. In a previous experiment (Notman, Sowden & Ozgen, in press, Cognition), we showed that acquired CP for spatial phase defined categories was specific to orientation (tuned with a bandwidth of 6.5 deg.) supporting an early locus of the effect. Here, using the same stimulus set, we examined the perceptual nature further. In experiment 1, we examined retinal location specificity of learned CP. Observers were trained to categorise stimuli briefly presented to one retinal location. Following training, CP was apparent at the trained location but not at locations 3.3 deg. away. This supports the idea that CP is dependent on perceptual processing at relatively early stages of visual analysis where receptive fields (RFs) are smaller. In experiment 2, we examined a possible explanation for the narrow 6.5 deg. bandwidth found previously. The original stimuli (at 8 deg.) spanned the RFs of multiple interconnected units in V1 and it may be that perceptual learning modified the strengths of intra-cortical connections between cells tuned to the same (or similar) orientations. Here, stimuli were viewed through a narrow Gaussian aperture (20% of original) to reduce the impact of lateral interactions. Over the course of 3 days, observers were trained to categorise stimuli at a single orientation, before and after completing a same different judgement task at each of nine orientations. Whilst training led to improved categorisation performance, there was no measurable improvement in discrimination performance and therefore no learned CP effect. In combination, these experiments are consistent with the possibility that CP may be mediated by intra-cortical connections at early, retinotopically organised, stages of visual analysis.
This research was funded by the ESRC