Australasian Mathematical Psychology Conference 2019

The speed-accuracy tradeoff in probabilistic categorization: Selective influence?

David Sewell
School of Psychology, The University of Queensland
Alexander Stallman
The University of Queensland

For many choices, taking longer to deliberate improves the quality or accuracy of decision outcomes. However, there are practical limitations to how long one can consider a decision, resulting in a need to balance the speed of decision-making against accuracy. Traditionally, this speed-accuracy tradeoff has been understood in terms of differences in the thresholds people use when decisions are made under speed vs. accuracy emphasis. Recent work has shown that this explanation does not always hold, and that emphasizing response speed can also affect the quality of information available at the decision stage. We examine the speed-accuracy tradeoff in a simple probabilistic categorization task to determine the longer-term impact of speed emphasis on learning. We model learning data using both the diffusion decision model as well as an integrated model that uses an associative learning front-end to drive a diffusion decision process. At the group level, the data are highly consistent with the traditional interpretation of the speed-accuracy tradeoff. However, at the individual level, the majority of participants are better described by a model that allows evidence to vary for speed- and accuracy-emphasis conditions. Surprisingly, the quality of evidence is consistently higher under speed emphasis.