Reliability of ON-OFF perimetry measurements. a–c, The means of total error (a), missed targets (b), and reaction times (c) decrease logarithmically with the number of stimulus trials. Decreasing the number of trials from 579 to 100 only causes a modest reduction in visual performance (<4% of total error, <3% of missed targets, <30 ms of reaction times). Notice that the maximum x value plotted is 500 trials. The y value is zero at x: 579 trials. d, Same as a for individual subjects. Each circle illustrates the average total error from each subject measured with the number of trials shown at the x-axis. e–g, The means of total error (e), missed targets (f), and reaction times (g) decrease logarithmically with the number of subjects. Decreasing the number of subjects from 30 to 1 causes a relatively modest reduction in average visual performance (<8% of total error and missed targets; <80 ms of reaction time). h, Average time between contrast measurements, from the highest contrast that was tested first (20%) to the lowest contrast that was tested last (5%). Error bars are SEM.