Human judgment is often based on integrating information accrued from experience and from description (e.g., the Internet). My research aims to specify cognitive principles of integrating experience and description of probabilistic information, and to compare these principles to rational (statistical) principles.
I designed an experiment in which I asked people to integrate information about two samples of red and blue chips drawn from a virtual bag. For each bag, research participants experienced one sample and received a description of a second sample: experience consisted of observing a sequence of binary events (i.e., a sequence of red and blue chips in a sample drawn from the bag); description consisted of a summary (total number of chips and proportion of red chips in a sample drawn from the bag). Participants were asked to integrate this information and to predict the proportion of red chips in the bag. The trials in some conditions of the experiment consisted of receiving experience followed by description; in other conditions, the presentation order was reversed.
The data (the participants estimates of red chips in the bags) were analyzed using a computational model that mimics how humans allocate attention between description and experience. The results show that cognitive principles of information integration deviate from rational principles in two ways. One, the impact of information depends on its source: experience-based information had more impact on cognitive integration than description-based information. Two, cognitive integration was affected by the presentation order of the two sources: the impact of experience was more pronounced when it was obtained after description rather than before it.
The findings of my research can be used to guide the development of decision aids (e.g., product recommendation systems). The findings also contribute to theories of cognition that address the role of attention in information integration.