Binding of visual features in human perception and memory

Abstract

The leit motif of this thesis is that binding of visual features is a process that begins with input of stimulation and ends with the emergence of an object in working memory so that it can be further manipulated for higher cognitive processes. The primary focus was on the binding process from 0 to 2500 ms, with stimuli defined by location, colour, and shape. The initial experiments explored the relative role of topdown and bottom-up factors. Task relevance was compared by asking participants to detect swaps in bindings of two features whilst the third was either unchanged, or made irrelevant by randomization from study to test, in a change detection task. The experiments also studied the differences among the three defining features across experiments where each feature was randomized, whilst the binding between the other two was tested. Results showed that though features were processed to different time scales, they were treated in the same way by Visual Working Memory processes. Relevant features were consolidated and irrelevant features were inhibited. Later experiments confirmed that consolidation was aided by iconic memory and the inhibitory process was primarily a post-perceptual active inhibition.