Multi-Source Information Acquisition with Optional Stopping

Abstract

An optional stopping experiment was performed in which the subject had to decide not only how much information to gather prior to making a two-choice decision, but also which of three information sources to consult. The cost of consulting these sources was varied systematically, with each source having a known, fixed diagnosticity. Two loss structures were used: (1) penalty for miss = penalty for false alarm, and (2) penalty for miss = 3 × penalty for false alarm. Results indicated subjects bought too little information relative to an optimal Bayesian model. The subjects indicated a reluctance to expend the resources required to obtain the best information available. Although subjects responded to the loss variables in the appropriate direction, they did not respond with sufficient magnitude to be mathematically optimal. Future research areas are suggested.