Change in brain asymmetry reflects level of acute alcohol intoxication and impacts on inhibitory control

Abstract

Alcohol is one of the most commonly used substances and frequently abused, yet little is known about the neural underpinnings driving variability in inhibitory control performance after ingesting alcohol. This study was a single-blind, placebo-controlled, randomized design with participants (N = 48 healthy, social drinkers) completing three study visits. At each visit participants received one of three alcohol doses; namely, a placebo dose [equivalent Blood Alcohol Concentration (BAC) = 0.00%], a low dose of alcohol (target BAC = 0.04%), or a moderate dose of alcohol (target BAC = 0.08%). To measure inhibitory control, participants completed a Go/No-go task paradigm twice during each study visit, once immediately before dosing and once after, while their brain activity was measured with time-domain functional near-infrared spectroscopy (TD-fNIRS). BAC and subjective effects of alcohol were also assessed. We report decreased behavioral performance for the moderate dose of alcohol, but not the low or placebo doses. We observed right lateralized inhibitory prefrontal activity during go-no-go blocks, consistent with prior literature. Using standard and novel metrics of lateralization, we were able to significantly differentiate between all doses. Lastly, we demonstrate that these metrics are not only related to behavioral performance during inhibitory control, but also provide complementary information to the legal gold standard of intoxication (i.e. BAC).

Figure 1

Overview of study design. (a) The structure of each experimental session: participants performed an inhibitory control task (Go/No-go) while wearing the Kernel Flow1 headset. Next, depending on the predetermined randomized order, they were given a beverage with placebo (zero), low, or moderate doses of alcohol. Following alcohol consumption, there was a wait period during which BAC was measured every five minutes using a breathalyzer until the target BAC level was reached (10–30 min; gray vertical lines indicate BAC measurements). After the wait period, participants performed another round of the Go/No-go task. (b) Schematic of the Go/No-go block structure. A few representative trials at the start of a go-no-go block are shown. (c) BAC values for each participant (each dot) measured using a breathalyzer after the BAC rise phase and before the second Go/No-go task. BAC values after the moderate dose were significantly higher compared to those after the low dose of alcohol (paired t-test p = 2.19 × 10^–17; N = 46; shown are mean ± s.d.).

Figure 2

Main effect of alcohol level on task performance as measured by d-prime. (a) Pre Go/No-go performance is plotted against post Go/No-go performance for placebo (left, blue, N = 48), low (middle, purple, N = 47), and moderate (right, red, N = 47) levels of intoxication. Solid lines depict the line-of-best-fit and the shaded regions show the 95% confidence intervals. Dashed lines indicate the line of unity. Task performance is computed as the d-prime in the go-no-go blocks. (b, left) Colored bars show the average d-prime (task performance) across the pre tasks for placebo, low, and moderate sessions separately. (Right) The change in performance, measured as post d-prime–pre d-prime are shown for the different doses. Colored dots show individual subject measures (bar plots: mean ± s.e.m). Behavioral performance was significantly impaired in the post task compared to the pre task for the moderate dose (paired t-test p = 8.13 × 10^–4; N = 47).

Figure 3

Prefrontal brain activation during an inhibitory control task under the influence of different alcohol doses as quantified by generalized linear models. (a) HbO in pre-dose Go/No-go tasks showed increased right prefrontal activation during go-no-go (GNG) blocks compared to the go-only (GO) blocks (as indicated by warmer colors on the right side). This activation is accompanied by anterior deactivation in the left prefrontal cortex (indicated by cooler colors). Each line is an individual channel and only significant channels at p < 0.05 (uncorrected) are displayed. (b) Same as (a), but showing activation patterns for placebo (left), low (middle), and moderate (right) doses of alcohol during the post Go/No-go task. Different dosing sessions exhibited different strengths and patterns of lateralization. (c,d) Same as (a,b) except for HbR. In all subpanels, the color scale indicates the t-statistic. (e) Average right – average left brain activity, i.e. hemispheric difference (GLM effect size for go-no-go – go-only contrast) plotted against average right brain activity. Dose-separated distributions are displayed via box plots along the top and right margins. The distributions were compared against zero using one-sample t-test (**,***significance at p < 0.01 and p < 0.001, respectively; ^†indicate trends towards significance; N = 48 (placebo), 47 (low), and 46 (moderate)). (f) Right – left average GLM effect size (GLM-based lateralization) was significantly higher for low- versus moderate-dose (shown are paired t-test statistics; N = 45). Dashed line indicates the line of unity.

Figure 4

Time courses of prefrontal lateralization for different alcohol doses. Epoched time course of left (green) and right (orange) prefrontal activity for go-no-go blocks (top) and go-only blocks (bottom) separated by session. From left to right, pre-dose and placebo-, low-, and moderate-dose time courses are shown (mean ± s.e.m). Solid and dashed lines represent HbO and HbR, respectively. Note the (expected) opposite trends in the two chromophores.

Figure 5

Altered patterns of prefrontal lateralization with different alcohol doses. (a) Block-averaged time courses of left (green) and right (orange) prefrontal HbT for go-no-go blocks (top) and go-only blocks (bottom) separated by session. From left to right, pre-dose and placebo-, low-, and moderate-dose time courses are shown (mean ± s.e.m). (b) Population distribution of lateralization (direction metric) for go-no-go (top; black) and go-only (bottom; gray) blocks across all participants and sessions. Note the right lateralization for go-no-go blocks (shift towards positive values, t-test: p = 5.08 × 10^–7, N = 282) and left lateralization for go-only blocks (shift towards negative values, t-test: p = 1.73 × 10^–7, N = 282). (c) The distributions of the direction metric for go-no-go blocks across post-dose sessions (*significant difference between placebo and low dose: paired t-test p = 0.037, N = 47). (d) Population average parametric curves showing the trajectory of right prefrontal vs. left prefrontal activity through time for go-only (gray) and go-no-go (black) blocks, split by pre-dose and session type, as in (a). Stars indicate initial time points. (e) The distributions of the angle metric, split by pre-dose and session type, as in (a). *,** Indicate significant deviation from uniformity (at p < 0.05 (N = 46) and p < 0.01 (N = 47) respectively; Rayleigh test) when computing pairwise differences in the angle metric between dosing sessions. (f) Distributions of the angle metric, split by no placebo-effect/placebo-effect (left/right) and go-no-go blocks/ go-only blocks (top/bottom). ***A significant difference in the population mean in the go-no-go condition (p = 9.80 × 10^–5; Watson–Williams test) comparing no placebo-effect (N = 23, circular mean = 176.50 deg) and placebo effect (N = 23, circular mean = 34.45 deg).

Figure 6

Lateralization metrics improved the prediction of behavioral performance in addition to measured BAC. (a) The change in behavioral performance as measured by post d-prime – pre d-prime was significantly correlated with BAC (top; Pearson r = − 0.29, p = 5.24 × 10^–4, N = 142), magnitude × sin(angle) (middle; Pearson r = 0.19, p = 2.63 × 10^–2, N = 141) and direction metric (bottom; Pearson r = 0.22, p = 7.47 × 10^–3, N = 141. (b, top) Statistical significance of the different factors in the four OLS models (four blocks shown on the x-axis) in predicting the post behavioral performance are reported as − log₁₀ (p value). We build off of the three factors in the basic model (pre behavioral performance: blue, age: yellow, sex: green), and include additional variables in separate OLS models: (1) BAC model included the direct measure of intoxication (BAC values, red) in addition to the basic model; (2) direction model consisted of the direction metric (dark purple) in addition to the BAC model variables; (3) angle model included the weighted sine and cosine of the angle metric (pink and light purple respectively) in addition to the BAC model variables. Note that both BAC values and lateralization metrics were significant factors in predicting post performance (BAC model: N = 142, neural models: N = 141). Dashed lines indicate a significance level of 0.05 and 0.05/4 (after correcting for multiple comparisons using Bonferroni correction; “Methods”). (Bottom) Overall OLS model performances for the four models shown in terms of the change in R-squared values and the change in AIC (lower AIC values indicate better performance). (c) Same as (b) but when only considering the sessions with alcohol consumption (i.e. low and moderate doses; BAC model: N = 94, neural models: N = 93).