The role of asymmetric processing latencies in multiattribute choices

Considerable behavioral and neural evidence indicates that decisions are dynamic processes that unfold over time. Here, we examine these temporal dynamics directly to better understand cognitive and neural decision mechanisms. Using multiattribute food choices from four independent experiments, we demonstrate that dynamic and asynchronous evidence accumulation models lead to new insights into multiattribute decision processes and how they will respond to changes in the environment or individual goals. We find that both how quickly and how strongly taste and healthiness attributes weigh into the decision process are key determinants of choice outcomes, and consequently, individual differences in self-control. By combining computational modeling with experimental manipulations of attention and neural activity, we show that the processes determining the relative weighting and speed of consideration for healthiness and taste attributes during food choices are dissociable at the cognitive and neural levels. At the cognitive level, we find that directing attention toward healthiness led to significant, but independent (i.e. uncorrelated), changes in both how quickly and how strongly healthiness and taste attributes affected choices. At the neural level, we show that cathodal tDCS over the the left dorsolateral prefrontal cortext (dlPFC) decreased self-control by causing an increase in the relative decision weight for taste but did not change the speed with which taste or healthiness attributes entered into the decision process. Lastly, in addition to generating a more complete understanding of multiattribute choice mechanisms, we find that asynchronous evidence accumulation models make better out-of-sample predictions about choice behavior than traditional logistic regression models.