[Evaluation algorithm for eye movement patterns during a problem solving task]

Cognitive processes are accompanied by eye movements, which are interrupted by fixations of an almost infinite number of points in the visual field. To investigate these processes, a systematic analysis of those movements is required to structure the long sequences of fixations. This task is possible if there are interrelationships between the ongoing thought processes and the fixations. Such interrelationships may manifest themselves through certain patterns of eye movements. It often happens that subsets of visual points are being looked at repeatedly. This is because a cognitive connection among the subset's elements is being discovered or created by such repetition. The method presented here attempts to achieve automatic recognition of fixation point patterns. Its usefulness is demonstrated using data from an experiment where eye movements were recorded while subjects were solving Raven Test items. The method may be characterized as a variant of time series analysis for nominal data. It is suitable for non-numerical data in general.