Abstract
Is it possible to take a picture of a visual perceptual event induced by stimulation of the visual cortex? While the question sounds philosophically impenetrable, we have developed a concept, perceptography, to approximate such pictures; perceptograms. We define a perceptogram as an altered image, which is indiscriminable for the observer from the state of being cortically stimulated. To test the concept, we chose the primary visual cortex (V1) as a testbed because the perceptual events resulting from stimulation of area V1, known as phosphenes, are well studied in both humans and monkeys, yet loosely characterized in shape, timing, and luminance. A macaque monkey was trained to behaviorally detect short electrical impulses (250 μA, 40Hz bipolar square wave, 200 ms) delivered to its V1 via a Utah array while fixating at a dynamic pattern of low-frequency enhanced white noise subtending 23 degrees. Each trial lasted 1s, included 5 refreshments of the noise pattern, randomly included cortical stimulation halfway through the trial and was followed by two response targets for the monkey to indicate if the trial was “stimulated” or not. We then marked the trials that led to a false alarm (12% of trials). We hypothesized that a false alarm is more likely if a pattern of noise happens to mimic the visual timing and features of the electrically induced phosphene. Thus, at first pass we averaged all the noise patterns that led to ~800 FAs across ~6000 behavioral trials and 3 cortical positions. For different cortical positions on V1, we found phosphenes significantly standing out of the noise (mean difference from average gray 4.6%, St.D 0.22, p<0.01, corrected for multiple comparisons). These phosphenes matched the location and timing of saccades induced by electrical stimulation of their corresponding cortical position with the fixation condition relaxed.