Optimal bounds, bounded optimality

Decision-Making is an integral part of everyday life. Most decisions share two features: uncertainty and impatience. Whilst much research has been done on the role of uncertainty, surprisingly little is known about the role of impatience. This dissertation tries to fill this gap.

One of the most important types of decision processes in psychological research is perceptual decision-making, where participants need to choose among several possible interpretations of a stream of noisy sensory information. The mathematical standard model for this type of decision process assumes that impatience plays no role in perceptual decisions. That is, participants should require a fixed amount of information before committing to a decision, irrespective of the amount of time they have already expended on the decision. However, a number of recent studies from neuroscience claim that impatience leads to more optimal decisions. Participants should, therefore, require less information for a decision the more time they have already expended.

Although this impatience hypothesis seems intuitive, the theoretical and empirical support for it is weaker than claimed by its proponents. A literature review shows a number of shortcomings in the setup and execution of experimental studies, as well as in the use of quantitative models supporting the impatience hypothesis. A theoretical analysis further shows that impatience only leads to markedly more optimal behaviour if task difficulty is high. Finally, an experimental study described here could not confirm the prediction that participants flexibly adjust their behaviour to the decision environment. In summary, the standard model of forty years still provides the best account of perceptual decision-making in humans.

Dissertation: http://hdl.handle.net/11370/f8564ba8-5d1d-49dc-badd-0d670685c34a