Exploring Spaces to Make the Right Choice: The Cognitive Science of Search

One important question in cognitive science is how humans search for useful resources in the environment. Indeed, understanding the search process is often a critical step for studying how an agent learns, adapts, and behaves in an uncertain environment. More generally, search is required whenever an agent faces a problem and there are uncertainties involved during the process of solving the problem. Given its ubiquity, the search process is found to be central in many cognitive activities, ranging from vision, memory retrieval, problem solving, decision making, web navigation, to social selection. However, research on search has a tendency to fragment into multiple areas. The goal of this symposium is to lead an integrative discussion of the over-arching principles underlying the search process, and highlight how search plays a central role in cognition. To this end, participants in this symposium will present research results that show how humans search in different spaces such as information spaces (Wai-Tat Fu), decision spaces (Thomas Hills), motivational spaces (Art Markman), and social and non-social spaces (Peter Todd). Overview of Presentations Wai-Tat Fu Exploratory Information Foraging Information search has become an important part of human activities as people acquire new information about the world and adapt to the changes. The theory of information foraging analogizes information search with animal foraging, and assumes that the search process can be characterized as an optimization process that maximizes the intake of information while minimizing the costs of switching between information patches. Computational cognitive models based on the theory of information foraging provide good prediction on how people search for specific information on WWW. Recent research, however, shows that information foraging behavior is often exploratory, in which the information goal evolves as new information is discovered and integrated into their internal representations of the environment (Fu, Kannampallil, & Kang, 2009). This dynamic updating of internal representations and search policies is found to be highly adaptive to the characteristics of the environment. Modeling the exploratory information foraging behavior can provide a more complete understanding of emergent patterns of individual and aggregate search behavior in large-scale information spaces. Thomas Hills & Ralph Hertwig Individual Differences and Executive processing in Information Search Recent work on individual differences in information search reveals that people differ in how they mediate local versus global search policies. In other words, people differ in how long they search in a local region of the information space before making a global transition to another local region of the space. This may be a consequence of a general executive search process that mediates search across domains, including spatial search and problem solving (Hills, Todd, & Goldstone, 2008), memory search (Hills & Pachur, in review), and external information search among gambles (Hills & Hertwig, 2010). We investigated individual differences in local-versusglobal search when people were searching for information prior to making a decision between several options. In these studies, individuals were invited to explore several options that each produced different payoff distributions. Following a period of unconstrained information search (participants can explore options as they please for as long as they like), participants make a decision about which option they will choose for real, and then receive its associated payoff. Across multiple studies, we consistently find that some individuals search one option comprehensively before transitioning to a second option; other individuals frequently transition back and forth between options, using a piecewise search. These different search policies correspond to different final decisions, which are consistent with different cognitive models for how these individuals process the information when comparing the two samples. Moreover, these search and decision differences correlate with working memory span and are consistent with a broader model of a domain general cognitive control of search, which applies for both external and internal patterns of information search. Arthur B. Markman, Bradley C. Love & A. Ross Otto Modeling the motivational and environmental factors affecting exploration Intelligent agents need to resolve the tradeoff between trying new options (exploration) and relying on actions that