Poster E137, Monday, March 27, 2:30 – 4:30 pm, Pacific Concourse
Mechanisms of Information Accumulation across Speed-Accuracy Tradeoff
Christina M Merrick1, Kate T Duberg1, Anne GE Collins1, Richard B Ivry1; 1University of California Berkeley
Decisions that involve uncertainty (e.g. identifying someone in a rainstorm) require information to be accumulated over time. Sequential sampling models such as the diffusion decision model (DDM; Ratcliff 1978) assume accumulation of information to be a noisy process that results in a response after a decision threshold is reached. The decision threshold is modulated according to the level of caution employed. In the current study we used the random-dot motion task, manipulating the quality of information (i.e., dot coherence) and task instructions (i.e., emphasis on speed or accuracy) in a 2 x 2 design. Two versions were compared, one in which conditions varied in a block design and the other in which conditions varied on a trial-by-trial basis. We fit the behavioral results with a hierarchical Bayesian drift-diffusion model (Wiecki, Sofer & Frank, 2013) including three parameters: drift-rate, threshold and non-decision time. The parameter estimates were similar for the blocked and mixed designs, indicating that participants were able to flexibly switch between states associated with speed or accuracy. Model comparison (DIC) indicated that the best fits were obtained when all three parameters were allowed to vary. Drift-rate increased with dot coherence and the threshold was lower on speeded trials, replicating previous work (Ratcliff & McKoon, 2002). There was a significant interaction in drift-rate across the instruction manipulation, such that the increase in drift-rate with increasing coherence was greater in the accuracy condition compared to the speed condition. Current experiments are examining EEG correlates of these behavioral effects and model-based parameters.
Topic Area: THINKING: Decision making