Individual differences in subconscious motor control predicted by GABA concentration in SMA

Abstract

Subliminal visual stimuli affect motor planning, but the size of such effects differs greatly between individuals. Here, we investigated whether such variation may be related to neurochemical differences between people. Cortical responsiveness is expected to be lower under the influence of more of the main inhibitory neurotransmitter, GABA. Thus, we hypothesized that, if an individual has more GABA in the supplementary motor area (SMA)--a region previously associated with automatic motor control--this would result in smaller subliminal effects. We measured the reversed masked prime--or negative compatibility--effect, and found that it correlated strongly with GABA concentration, measured with magnetic resonance spectroscopy. This occurred specifically in the SMA region, and not in other regions from which spectroscopy measurements were taken. We replicated these results in an independent cohort: more GABA in the SMA region is reliably associated with smaller effect size. These findings suggest that, across individuals, the responsiveness of subconscious motor mechanisms is related to GABA concentration in the SMA.

Figure 1. Methodology for masked priming and GABA spectroscopy

a, target arrows were preceded by masked primes presented below the threshold for conscious discrimination. For the stimulus timing illustrated, responses tend to be slower when prime and target are the same (compatible) than when they are not (right hand illustration). This is the measure of subliminal suppression, and the magnitude differs between individuals. b, the MRS voxel (yellow, (3 cm)³ voxel) was placed over the anatomical location of SMA. As a check on voxel placement, for two participants we acquired a functional localiser for the SMA using fMRI (see Methods and bottom sagital view). Edited MR spectra (c) allow the quantification of GABA concentration by extracting the area under the GABA peak [6, 8, 9, 48] (glutamine/ glutamate, Glx, and N-acetyl-aspartate, NAA, peaks are also marked). The peak will also contain co-edited macromolecules. See Methods and supplementary fig S2 for more details and individual spectra.

Figure 2. Subliminal suppression correlates with GABA in the SMA region

Higher GABA concentration in the region around human SMA predicts smaller negative compatibility effect (NCE) across individuals (a). This result was replicated in a second cohort. b, There was no correlation between the NCE and GABA concentration in dorsolateral prefrontal cortex (DLPFC), parietal cortex, anterior cingulated cortex (ACC) or inferior frontal gyrus (IFG). Positioning of these MRS voxels is shown for one participant (yellow rectangles on inset brains). Note that although the word cortex is included in some labels (to follow standard abbreviations for these regions) all voxels necessarily included both grey and white matter. Filled symbols and bold R-values reflect measurements from the second cohort. GABA concentration measurements are stated in institutional units (i.u.). All p-values are given 2-tailed but uncorrected for multiple comparisons; the main relationship of interest between the NCE and the SMA was specified a priori [5], but even if it had not been (and the first p-value is corrected), the replication demonstrates that the relationship is robust.

Fig. 3. Correlations of the PCE with the NCE and GABA in the SMA

a,b, higher positive compatibility effect (PCE, subliminal activation) predicts higher (more negative) NCE (subliminal suppression). Further analyses of previously published data [3] also revealed strong correlations between NCE and PCE (experiment 1: r = −0.69, p< .013; experiment 2: r = −0.72, p< 0.008 & experiment 3: r = −0.63, p< .03). Thus it seems a general and robust phenomenon that the magnitudes of the PCE and NCE are correlated across individuals. However, although there is weak correlation in both cohorts between the PCE and GABA concentration in the SMA region (c, d), this was not significant (even across cohorts) and is presumably just mediated by the correlations between NCE and GABA and between NCE and PCE. Thus it appears that the common factor between NCE and PCE does not lie with GABA in the SMA.

Fig. 4. GABA in SMA region does not correlate with other potential mediating factors

(a: age, b: prime identification, c: mean reaction time, d: error rate, e: error compatibility effect, and f: fraction of grey matter in SMA MRS voxel) R-values are the correlation coefficient obtained putting both cohorts together; R1- and R2-values are the coefficient obtained for cohort 1 and 2 separately. There was also no significant correlation of any of these factors with the NCE (all |R|, |R₁| or |R₂| < 0.44, p> .16). Most importantly, when these factors were controlled for, the (partial) correlation (R_p) between the NCE and GABA in the SMA region remained. When controlling for the amount of grey matter, R_1p = 0.8, p< .003, R_2p = .53, p< .04, one-tailed. Similarly when controlling for age (R_1p = 0.77, p< .005, R_2p = .62, p< .035), average speed (R_1p = 0.84, p< .001, R_2p = .61, p< .03), prime visibility (R_1p = 0.8, p< .003, R_2p = .55, p< .03, one-tailed) and error rate (R_1p = 0.75, p< .008, R_2p = .51, p< .045, one-tailed). Note that as a neurotransmitter, the concentration of GABA is expected to higher in grey matter (GM) than in white matter, so one might predict a correlation between GM volume and GABA. However, the GM proportion in the voxel was very similar across participants (i.e. it was well controlled for), so there was little opportunity for a correlation to be revealed. GM proportion ranged from 49% to 54% in cohort 1 and from 46% to 55% in cohort 2. The essential point is not whether GM correlates with GABA, but that this relationship does not account for the correlation of GABA with the NCE.