The strategic control of gaze direction during problem solving

T.L. Hodgson1, A. Bajwa1, A.M. Owen2, C. Kennard1

1Division of Neuroscience & Psychological Medicine, Imperial College School of Medicine, Charing Cross Hospital, 10th Floor East Wing, Fulham Palace Rd, London W6 8RF, United Kingdom (e-mail:t.hodgson@ic.ac.uk);
2MRC Cognitive and Brain Sciences Unit, University of Cambridge, Chaucer Road, Cambridge, United Kingdom

Eye movements were recorded while subjects solved the "Tower of London" task, a test used in the neuropsychological assessment of problem solving and planning ability. Subjects viewed a series of pictures showing two arrangements of coloured balls in billiard pockets. They were asked to plan the shortest sequence of ball movements to rearrange the balls in one half of the display (Workspace) to match the arrangement of balls in the opposite half of the display (Goalspace). Subjects had to plan, rather than execute the ball movements required to solve the problem, and then pressed a response button as soon as they had worked out the correct solution. Two groups of subjects were presented with an identical set of Tower of London pictures. However, one group was instructed to solve the problems with the Workspace arrangement in the upper half of the display, whilst the other group were instructed to solve the problems by rearranging the balls in the lower half of the display.

For difficult problems, gaze was more likely to be directed towards the Goalspace in the initial period after presentation of each problem picture. Later on in the trial, subjects directed gaze towards locations in the Workspace. Finally, prior to pressing the response key, fixations were once again more likely to land on the Goalspace. This pattern was found regardless of whether the subjects solved the problems by rearranging the balls in the lower or upper visual fields, demonstrating that the biases in fixation correlated with discrete phases in problem solving. A second experiment demonstrated that the pattern of fixations was dependent upon the problem moves being internally rehearsed by the subject, independent of the influence of salient stimulus features. Marked differences were also found in the distribution of fixations made by subjects who solved problems accurately and those who made errors.

It is concluded that efficient visual problem solving involves the co-ordination of appropriate gaze strategies. The detailed analysis of eye movement patterns can provide insights into problem solving and higher cognitive processes.