While we might think that the quality of our decisions depends on the content of the decision and how knowledgeable we are about it, in fact, many other factors can affect our ability to make timely, accurate decisions. The proposed research investigates how the difficulty of some decisions affects the speed and accuracy of subsequent ones. We examine individual performance in series' of simple decisions, manipulating the difficulty of each decision. For one condition, groups (or blocks) of decisions (trials) with relatively similar degree of difficulty are presented. For another, blocks of trials with more dissimilar degrees of difficulty are used. The latter condition is referred to as a "high variability" condition and the former as a "low variability" condition.
The main focus of the study is to test the whether decisions are both slower and less accurate under higher variability conditions than low variability conditions. This assumption has been employed by various models applied to binary-decision tasks, but has never been tested empirically. Two preliminary experiments generated data unfavorable to this assumption and its consequences in the models. A new round of experimental sessions will gather data to enable the comparison of response time distributions and accuracy for each of two conditions. These new data sets will be used to test some decision-making models' predictions for individual subject data. Specifically, we will examine whether the slower error responses are indeed associated with changes in decision variability, and we will investigate how the models explain this phenomenon. In case the models fail, the results of this project should point at theoretical improvements for the set of decision models that need this assumption.
Regardless of outcome, this project will be valuable because it explicitly tests an assumption already widely used within the RT modeling community for several years. In the future, studies such as this may help aid the betterment of displays based on which a judgment will be made -- like contrast-based displays in aircraft cockpits, or aid decision makers such as managers by helping identifying the amount, at what pace, and with what amount of diversity the information relayed to them should come in order for decisions to be made at the desired level of accuracy and speed.
