Structured Action Preparation during Visuomotor Decision-making

Decision-making and movement are often tightly intertwined, a phenomenon that has been primarily studied in the case of low-level perceptual decision-making. How higher-level decision processes interface with the motor system during action selection is less clear. Here, we used psychophysics and computational modeling to examine information flow between deciding and acting when actions had to be retrieved from a newly learned visuomotor mapping. Human subjects (N = 182) learned de novo visuomotor mappings with (or without) an imposed latent structure that linked visual stimulus features (e.g., color, shape) to intuitive motor distinctions, like hands and pairs of adjacent fingers. In subjects who learned structured visuomotor mappings, transitional response times between trials indicated that retrieving the correct action from memory invoked “traversal” of a tree-like mental graph of the learned mapping. Control experiments and analyses helped rule out alternative explanations related to intrinsic switch costs between finger responses and perceptual properties of the stimuli. We then used a forced response time paradigm to show that when people traverse these learned visuomotor mappings during action selection, they sequentially potentiate multiple actions along the way in accordance with the structure of the mapping. Our results point to a direct coupling between complex internal memory and decision-making processes and the systematic preparation of actions.