Abstract
Humans maintain internal models to generate expectations and predictions about the external world. Critical for the reliance on such models is that they be updated to reflect salient environmental changes. One possible signal of environmental change is that incoming sensory information and mental model predictions mismatch. Deficits in updating spatial representations have been well demonstrated following brain injury, especially right parietal injury. Here we explore the ability of healthy individuals to generate and update temporal representations. An image of a red square was briefly displayed to experimental participants (n = 20). After each presentation, participants were asked to predict if the next presentation would be longer or shorter. Presentation durations were randomly chosen from 400 to 700 msec in duration (step size 50 ms). Following a period of guessing behaviour (approx. 50 trials), accuracies improved (range 56% to 71%) confirming that participants can form a mental model for these short time intervals. To evaluate if participants can update this representation when presented with a covert salient environmental shift, we shifted the range of temporal durations to 600–900 msec (50 msec steps) without informing participants. Confirming the ability to detect this slight shift, after a period of transition, participants shifted their behaviour for the stimulus presentation durations that were common to both conditions (600, 650, and 700 msec) from “next trial shorter” to “next trial longer.” However, performance was not as accurate, and did not change as quickly, for the second set of durations as it had for the first set of durations. In summary, we describe a task for the assessment of updating temporal representations of subsecond durations and show the effect of history on mismatch detection.