Toward a further understanding of object feature binding: a cognitive neuroscience perspective.

Abstract

The aim of this thesis is to lead to a further understanding of the neural mechanisms underlying object feature binding in the human brain. The focus is on information processing and integration in the visual system and visual shortterm memory. From a review of the literature it is clear that there are three major competing binding theories, however, none of these individually solves the binding problem satisfactorily. Thus the aim of this research is to conduct behavioural experimentation into object feature binding, paying particular attention to visual short-term memory. The behavioural experiment was designed and conducted using a within-subjects delayed responset ask comprising a battery of sixty-four composite objects each with three features and four dimensions in each of three conditions (spatial, temporal and spatio-temporal).Findings from the experiment,which focus on spatial and temporal aspects of object feature binding and feature proximity on binding errors, support the spatial theories on object feature binding, in addition we propose that temporal theories and convergence, through hierarchical feature analysis, are also involved. Because spatial properties have a dedicated processing neural stream, and temporal properties rely on limited capacity memory systems, memories for sequential information would likely be more difficult to accuratelyr ecall. Our study supports other studies which suggest that both spatial and temporal coherence to differing degrees,may be involved in object feature binding. Traditionally, these theories have purported to provide individual solutions, but this thesis proposes a novel unified theory of object feature binding in which hierarchical feature analysis, spatial attention and temporal synchrony each plays a role. It is further proposed that binding takes place in visual short-term memory through concerted and integrated information processing in distributed cortical areas. A cognitive model detailing this integrated proposal is given. Next, the cognitive model is used to inform the design and suggested implementation of a computational model which would be able to test the theory put forward in this thesis. In order to verify the model, future work is needed to implement the computational model.Thus it is argued that this doctoral thesis provides valuable experimental evidence concerning spatio-temporal aspects of the binding problem and as such is an additional building block in the quest for a solution to the object feature binding problem.