Orientation discrimination performance is predicted by GABA concentration and gamma oscillation frequency in human primary visual cortex

Abstract

Neuronal orientation selectivity has been shown in animal models to require corticocortical network cooperation and to be dependent on the presence of GABAergic inhibition. However, it is not known whether variability in these fundamental neurophysiological parameters leads to variability in behavioral performance. Here, using a combination of magnetic resonance spectroscopy, magnetoencephalography, and visual psychophysics, we show that individual performance on a visual orientation discrimination task is correlated with both the resting concentration of GABA and the frequency of stimulus-induced gamma oscillations in human visual cortex. Behaviorally, a strong oblique effect was found, with the mean angular threshold for oblique discrimination being five times higher than that for vertically oriented stimuli. Similarly, we found an oblique effect for the dependency of performance on neurophysiological parameters. Orientation detection thresholds were significantly negatively correlated with visual cortex GABA concentration for obliquely oriented patterns (r = -0.65, p < 0.015) but did not reach significance for vertically oriented stimuli (r = -0.39, p = 0.2). Similarly, thresholds for obliquely oriented stimuli were negatively correlated with gamma oscillation frequency (r = -0.65, p < 0.017), but thresholds for vertical orientations were not (r = -0.02, p = 0.9). Gamma oscillation frequency was positively correlated with GABA concentration in primary visual cortex (r = 0.67, p < 0.013). These results confirm the importance of GABAergic inhibition in orientation selectivity and demonstrate, for the first time, that interindividual performance on a simple visual task is linked to neurotransmitter concentration. The results also suggest a key role for GABAergic gamma oscillations in visual discrimination tasks.

Figure 1.

A, Schematic of the orientation discrimination task used in the experiment. B, Example data from the orientation discrimination task showing the two interleaved staircases—in this case for obliquely oriented stimuli. Filled dots represent turning points in the staircases.

Figure 2.

A, Cortical mesh model from a representative participant showing the location of the MRS voxel (blue box) and the source localization of induced gamma band (40–60 Hz) activity (orange) in the calcarine sulcus. B, Time–frequency representation of the peak location of gamma band activity in primary visual cortex showing a clear sustained oscillation that is present for the entire duration of the stimulus and is shown as an integrated spectrum in D. C, Edited MR spectra for the same participant (black) demonstrating a clearly resolved peak for GABA at 3 ppm, with the fitted GABA model indicated in red.

Figure 3.

Vertical and oblique angular orientation discrimination thresholds as a function of both GABA concentration (A, B) and gamma frequency (C, D). The crosshairs in the top right show estimates of uncertainty for each measure, averaged across participants. For gamma frequency this is the mean SEM obtained for the four participants with repeat data (see Materials and Methods) and was calculated to be 0.8 ± 0.2 Hz. For GABA the crosshairs depict the mean within-session difference between the two GABA concentration estimates, which was 0.043 ± 0.006 IU. For the behavioral thresholds, the crosshairs depict the SEM of the last 10 staircase reversals, averaged over both interleaved staircases. For oblique discrimination, this was 0.29 ± 0.03° and for vertical discrimination, it was 0.09 ± 0.01°.

Figure 4.

A, Gamma frequency is plotted as a function of GABA concentration. B, Stability of the observed significant correlations assessed using bootstrapping. The crosshairs depict the 95% confidence intervals.