Document Actions

Areas of research interest

My research is concerned with visual attention. Visual processing in the brain is massively parallel, yet our ability to identify more than one object at a time is severely limited. That is: vision is selective, and understanding the mechanisms of selection ('attention') is essential for understanding vision.

One line of my work has investigated the visuo-spatial orienting system, which guides the allocation of attention (and eye movements) to target locations. This work has shown that the orienting system consists of two separate, interacting, 'voluntary' and 'reflexive' orienting mechanisms. The voluntary mechanism is initiated by spatial expectancies (advance knowledge of the likely target location); the reflexive mechanism is triggered by certain stimuli in the environment (e.g. rapid onsets and offsets). The reflexive mechanism, once engaged, is difficult to interrupt by voluntary control; the voluntary mechanisms is more easily interrupted by a competing stimulus activating the reflexive mechanism. The mutual interruptibility is caused by mutually inhibitory connections between voluntary and reflexive orienting. Current work examines the nature of the mutual inhibition (e.g. is it 'continuous' or 'all-or-none'?).

A second line of work has been to build a connectionist network that simulates visual search and selection processes. This network, termed SERR (SEarch via Recursive Rejection), provides a computationally explicit account of how search operates within the system's functional ('attended') field. In essence, SERR operates via the grouping of display items on the basis of their similarity and via the recursive rejection ('selection') of groups of distractor stimuli. SERR produces linear search reaction time/display size functions if distractor stimuli are heterogeneous. SERR thus demonstrates that linear ('serial') search functions can be produced by a system that works in a spatially parallel manner. Further, SERR gave rise to several novel predictions that were tested successfully in experiments with normal and neurological human subjects. Future modelling work will extend SERR to deal with a richer world of stimuli.

Professor Hermann J Müller