Proceedings of the Annual Meeting of the Cognitive Science Society

Simulations of many people’s decisions are used in public health and safety as well as to support policymaking.These simulations rely on creditable models of individual decision-making. An obvious approach is to develop a list of plausibleactions and to then evaluate the benefits of each in the current situation to make the decision. However, such evaluations canbe implausible, e.g., zero-intelligence traders in economics, or impracticable because the approach is computationally intensivefor large-scale simulations. As a result, a commonly used approach is to select randomly from the plausible actions. Withoutdata on how people would actually chose, a random number from a uniform distribution over the plausible options is often usedto represent the unknown cognition. However, we claim that substituting a uniform random distribution for how people makedecisions is making very strong claims about the process and we will present data demonstrating it is simply wrong.