Racing Against The Clock: Evidence-Based Vs. Time-Based Decisions

Classical dynamic theories of decision making assume that responses are triggered by accumulating a threshold amount of information. Recently, there has been a growing appreciation that the passage of time also plays a role in triggering responses. We propose that decision processes are composed of two diffusive accumulation mechanisms---one evidence-based and one time-based---that compete in an independent race architecture. We show that this Timed Racing Diffusion Model (TRDM) provides a unified, comprehensive and quantitatively accurate explanation of key decision phenomena---including the effects of implicit and explicit deadlines and the relative speed of correct and error responses under speed-accuracy tradeoffs---without requiring additional mechanisms that have been criticized as being ad-hoc in theoretical motivation and difficult to estimate, such as trial-to-trial variability parameters, collapsing thresholds, or urgency signals. In contrast, our addition is grounded in a widely validated account of time-estimation performance, enabling the same mechanism to simultaneously account for interval estimation and decision making with an explicit deadline.