Behavioral detection of electrical microstimulation in different cortical visual areas

Abstract

The extent to which areas in the visual cerebral cortex differ in their ability to support perceptions has been the subject of considerable speculation. Experiments examining the activity of individual neurons have suggested that activity in later stages of the visual cortex is more closely linked to perception than that in earlier stages [1-9]. In contrast, results from functional imaging, transcranial magnetic stimulation, and lesion studies have been interpreted as showing that earlier stages are more closely coupled to perception [10-15]. We examined whether neuronal activity in early and later stages differs in its ability to support detectable signals by measuring behavioral thresholds for detecting electrical microstimulation in different cortical areas in two monkeys. By training the animals to perform a two-alternative temporal forced-choice task, we obtained criterion-free thresholds from five visual areas--V1, V2, V3A, MT, and the inferotemporal cortex. Every site tested yielded a reliable threshold. Thresholds varied little within and between visual areas, rising gradually from early to later stages. We similarly found no systematic differences in the slopes of the psychometric detection functions from different areas. These results suggest that neuronal signals of similar magnitude evoked in any part of visual cortex can generate percepts.

Figure 1

Two-alternative forced choice task. During fixation, a stimulus was delivered during one of two 250 ms time intervals that were marked by auditory tones and separated by 500 ms. Two-hundred and fifty milliseconds after the end of the second interval, two response targets appeared, and the animals indicated which interval contained the stimulus by making a direct saccade to the appropriate target (target 1 for period 1). The electrical stimuli were 250 ms trains of constant current pulses at 200 Hz, with the current randomly selected on each trial.

Figure 2

Representative psychometric function. Behavioral performance at a V1 site. Fifty repetitions of each of 10 currents were delivered in random order. The points show the average performance (±1 SE), and the curve is the best fitting psychometric function (see Methods). Threshold at each site was taken as the current corresponding to 82% correct performance on this curve (dashed lines).

Figure 3

Distributions of detection thresholds. Median thresholds are marked by a triangle. Medians and interquartile ranges for the two animals were (in microamps): Animal 1: V1, 5.2 (4.4-6.4), 6.6 (5.2-8.6); V2, 6.3 (4.9-7.6), 8.3 (5.8-11.5); V3A, 8.9 (7.7-12.2); MT, 10.1 (8.1-11.9); IT, 10.3 (8.2-15.5), 11.3 (7.5-15.2).

Figure 4

Distributions of psychometric slopes. Summary slope histograms for all sites with reliable slope estimates (see Methods). Medians and interquartile ranges for the two animals were: V1, 4.5 (3.0-5.2), 6.0 (4.4-8.1); V2, 5.6 (4.9-7.6), 8.4 (5.2-11.5); V3A, 5.0 (3.6-6.6); MT, 5.2 (3.7-6.5); IT, 4.8 (3.3-5.7), 5.2 (4.0-7.0).