Cognitive Psychology Colloquium Thursday 13th November

When: Thursday 13th November, 12-1pm
Where: Keats Reading Room, Psychology Building
What: Nathan Evan's PhD Confirmation Seminar 

Evaluating the applicability of collapsing thresholds in computational models of perceptual decision making

Evidence accumulation models have successfully been fit to human decision data for decades, modeling decisions from simple perceptual choice to consumer choice. Traditionally, these models have assumed that decision thresholds (i.e., the amount of evidence required to trigger a decision) remain constant over the course of the decision (fixed thresholds). However, recent evidence has suggested that decision thresholds may in fact decrease as decision time unfolds (collapsing thresholds), though the experimental evidence associated with collapsing thresholds has been the target of criticism. This thesis aims to explore how applicable collapsing thresholds are to the decision process, and if they are applicable, under what conditions. The first experiment aims to further explore the difference in decision strategy found between humans and non-human primates (fixed thresholds and collapsing thresholds, respectively), and the idea that this is caused by a difference in experimental procedure. The second experiment aims to directly measure decision thresholds through an adjusted version of a standard perceptual choice paradigm, by allowing the stimulus evidence level to change over the course of a trial. The third experiment aims to examine a potential condition under which collapsing thresholds may be most applicable to decision making, with this condition being one that informs participants of reward rate optimality (a goal that collapsing thresholds better achieves than fixed thresholds). The fourth experiment aims to contrast fixed thresholds, and evidence accumulation models in general, to the urgency gating model; a variant of the collapsing thresholds model that suggests that there is no evidence accumulation involved in the decision making process. These experiments, and this thesis as a whole, will provide further insight into which decision thresholds best represent the decision process, and help to insure the accuracy of evidence accumulation models used in the past, present, and future.