eSSENCE@LU 7:4 - Method development: Analysing real life eye tracking data using realistic computational models of cognition



Decision making is a core human competence, which has been the subject of study in approaches ranging from the computational and conceptual to the purely empirical. In the past decades, the area has begun to focus on eye tracking as a means to access “the mind’s eye” and to provide insight into the processes of decision making. Eye tracking methodology is, however, a research area in its own right and the methods and tools used there are not straightforwardly relevant for decision making research. In particular, there is a focus on the visually available options and how attention is directed to these, instead of internal factors, such as experience and memory, which provide values to the options and (partially) steer the decision making and visual attention.

To study the latter, the intention is to use a computational model of cognition when analysing eye tracking data. Unlike many psychological models, this will explicitly model the processes that give rise to the observed behaviour, rather than the structure of the data itself. Furthermore, eye tracking data has been obtained in the natural setting of individuals buying food in a normal shop supplemented with detailed knowledge of each shopper’s thoughts on what matters when buying the relevant products (e.g.: Is it important that food is organic? How much does the price matter?). The data set is highly unusual, not only for eye tracking research, but also for computational models of cognition. The cognition model has been developed, but methods need to be devised for efficient fitting of the large data set.

Thus, the project aims to merge two lines of research, eye tracking and cognition modelling, which will provide something novel in both areas; it will challenge the modelling line of research with eye tracking data that is perfectly representative of everyday decision making and will make a cognitively precise and realistic model available to the eye tracking line of research. The preciseness of such a model is an improvement over previous pure sampling or aggregation-based methods.
Kort titeleSSENCE@LU 7:4
Gällande start-/slutdatum2021/01/012022/12/31