Abstract
Under conditions in which response speed is stressed, response accuracy often suffers. Using a motion coherence task and functional magnetic resonance imaging (fMRI) chronometry, we sought to identify neural regions that are differentially sensitive to the mental set for speed versus accuracy. Regions of interest (ROIs) involved in the detection of motion coherence were first identified in a localizer run by comparing high coherence displays with static displays. The other fMRI runs stressed either speed (SPD) or accuracy (ACC) of the task in separate blocks of event-related trials (‘mixed-trial’ design). We used an experimental manipulation that allowed reaction time differences between ACC and SPD conditions to be resolved with fMRI: for each 20s-long trial, the proportion of dots with coherent motion was either nil (0% coherence) throughout the trial, started at 0% and increased 3% every 2s, or remained constant at a low coherence (< 10%). The behavioral data revealed the expected speed-accuracy tradeoff: the ability to detect and identify the direction of motion coherence (leftward vs rightward) was better under ACC instructions and, conversely, reaction times were shorter under SPD conditions. Time course analyses revealed that the temporal dynamics of the peak BOLD signal corresponded with reaction time differences between SPD and ACC conditions in a large network of sensory, motor and premotor regions. Furthermore, the mental set for speed versus accuracy had a more dramatic impact on frontal/prefrontal cortex activity than in sensory/perceptual cortex (MT+ and extra-striate cortex). These results suggest that the establishment of a mental set for speed versus accuracy is not equally implemented across the neural network mediating the detection and identification of motion coherence.