Deconstructing the vigilance decrement: Changes in bias, lapse rate, and guess rate, but not sensitivity

Authors:

Abstract
The vigilance decrement-a gradual decline in performance over the course of a sustained attention task-has been studied extensively, but its underlying mechanisms remain the subject of debate. We analyzed psychometric curves from a vigilant monitoring task to isolate the contributions of response bias, sensitivity, mental lapses, and guesses to the effect. Stimuli were pairs of small visual probes that varied in their spatial separation from trial to trial. Participants (N = 366) performed a psychophysical task that required them to report occasional trials on which the separation between probes exceeded a criterion value. Stimuli were embedded in dynamic noise, and trials occurred at a forced pace of 40 per minute, conditions designed to elevate processing demands and encourage sensitivity losses. Model comparisons revealed decisive evidence for conservative bias shifts, increases in mental lapse rates, and decreases in positive guess rate over time, but gave evidence against a loss of sensitivity. A reanalysis of data from an earlier study indicated that apparent evidence for a sensitivity loss in that report was the result of model misspecification. Results in total suggest that the vigilance decrement most generally reflects changes of response bias and increases in the frequency of mindless responses, not a fundamental loss of signal-noise discriminability. Open Practices Statement: Data and code are available at https://osf.io/3np2v/ . Experiment was preregistered at https://osf.io/2b58e and https://osf.io/zynx9 .