examples

Examples of empirical single-subject distributions observed in choice RT tasks (see Figure 1 below). Sample-based hazard (A, C, E) and conditional accuracy functions (B, D, F) are plotted for three tasks. (A, B). Visual search data from Panis et al. (2020). The participant searches for a target defined by a conjunction of “red” and “vertical”. (C, D). Spatial cueing data from Panis and Schmidt (2022). The participant indicates whether a peripheral target stimulus is presented to the left or right of fixation. The target is preceded by a central cue, and the central cue is preceded by a peripheral cue at the same (valid) or opposite (invalid) side as the target. (E, F) Unpublished data. The participant discriminates the direction of a horizontal arrow (imperative stimulus, presented at time zero), and when a stop-signal is presented with a certain stop-signal delay (ssd; see vertical lines), the participant is trying to suppress the overt response.

The hazard functions indicate, for each bin, the conditional probability that a button-press occurs in bin t, given that no button-press occurred in earlier bins. The conditional accuracy functions indicate, for each bin, the conditional probability of a correct response in bin t, given that the response is emitted in bin t.

In the search task, the effect of target presence is present in the left tail of the hazard functions (and in the right tail for the conditional accuracy functions). In the stop-signal task, the effect of stop-signal delay is present in the right tail of the hazard functions (compared to no stop signal). Regarding the spatial cueing task, we infer that the participant transitions through at least three motor states during a trial: a first state running from -160 to -40 ms, reflecting the “automatic” response to the peripheral cue; a second state running from 0 to 160 ms, where conditional accuracies reverse (likely due to selective suppression of the premature response tendency); and a third state where conditional accuracy increases to 1 and the hazard functions display the famous inhibition-of-return effect (i.e., a lower mean RT in invalid compared to valid cueing, which translates to a higher hazard for invalid compared to valid cueing).

Figure 1