Temporal Adaptation with a variable delay

The consequences of an action almost always occur immediately. Delaying the consequences of an action (eg by delaying visual feedback) drastically impairs performance on a wide range of tasks. A few minutes of exposure to a delay can, however, induce sensorimotor temporal adaptation. Here we ask whether a stable delay is necessary for temporal adaptation. Specifically, we examined performance in a driving simulator (where subjects could control the direction but not the speed of travel). The delay was on average 250 ms, but fluctuated rapidly (36 Hz) and randomly between 50 and 450 ms. Overall, subjects were able to learn to drive a virtual car with a variable delay. In one experiment, we found that the adapted state also improved performance on untrained streets (generalisation). In a second experiment, performance with immediate feedback was measured both before and after delay training. We found a strong negative aftereffect (approximately 50 drop in performance from pre- to post-test). While some behavioural strategies (eg slow gradual changes in steering wheel angle) might mitigate the impact of a variable delay, these strategies do not totally eliminate the variability, particular for fast speeds and sharp corners