Performance.

The modified Stroop task successfully evoked interference effect as indicated by lower accuracy (F(1,19) = 40.8, p<0.0001 ) and longer reaction times (F(1,19) = 130.5, p<0.0001) to INCONG as compared to all other conditions, Fig. 1. Subjects were more accurate under placebo than alcohol as indicated by the main effect of beverage (F(1,19) = 12.3, p<0.005). There was no main effect of beverage or beverage by condition interaction on RTs. However, there was a strong main effect of condition (F(3,57) = 82.5, p<0.0001), with the slowest responses to INCONG (820±126 ms). Responses to CONG were the fastest (639±89 ms), differing from both RTs to NEUT (693±98 ms, F(1,19) = 81.7, p<0.0001) and READ (683±100 ms, F(1,19) = 12.9, p<0.005). Participants made more corrective responses on INCONG trials compared to all other conditions (χ² = 4.8, p<0.05), and more under intoxication compared to placebo especially on INCONG trials (χ² = 4.5, p<0.05), Fig. 1. Reduced accuracy under alcohol correlated with three measures of impulsivity: psychoticism scale of Eysenck’s Personality Questionnaire (r = −0.59, p<0.001), impulsivity scale on Eysenck’s Impulsiveness and Venturesomeness Scale (r = −0.54, p<0.02), and thrill and adventure seeking scale of Zuckerman Sensation Seeking Scale (r = −0.58, p<0.02), suggesting that alcohol may impair the ability to inhibit the prepotent but erroneous responses.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Performance measures.

Accuracy, reaction times, and the percentage of trials on which participants corrected their responses immediately after an error (means ± standard errors) are shown for each task and beverage condition. The Stroop interference effect is indicated by lower accuracy and slower reaction times to conflict-inducing incongruous (INC) stimuli overall. Alcohol did not affect RTs, but participants responded less accurately and made more corrective responses on INC trials when intoxicated. Significant alcohol vs. placebo comparisons for each condition are marked, *p<0.05.

https://doi.org/10.1371/journal.pone.0043957.g001

Post-experimental questionnaire and mood ratings.

At the end of each experimental session participants were asked to provide ratings of various aspects of their subjective experience using Likert scales. On the scale from 1 (definitely contains no alcohol) to 5 (definitely contains alcohol) subjects rated the beverage contents as 4.7±0.5 under alcohol and significantly lower under placebo, 1.3±0.4, χ2 = 20.0, p<0.0001. Participants estimated that the alcoholic beverage contained 2.3±0.7 “alcoholic drinks” which was a slight underestimate of the actual average amount containing 2.8 standard drinks defined as 1.5 fl oz of vodka. In contrast, they estimated that the placebo beverage contained 0.0±0.1 “alcoholic drinks”. On the scale from 1 (not at all) to 5 (very much), participants reported feeling moderately intoxicated (2.7±0.7) under alcohol, but not at all intoxicated under placebo (1.0±0), χ² = 20.0, p<0.0001. They were only slightly more dizzy under alcohol than placebo (1.8±0.8 vs. 1.1±0.2), but this difference was significant, χ2 = 12.0, p<0.001. On the scale from 1 (easy) to 5 (difficult), subjects rated the task as being moderately difficult (2.8±0.9) but the perceived difficulty was not influenced by beverage.

Mood ratings were available for 19 participants. Overall, they felt less stimulated and more sedated at the end of the recording, as indicated by the main effects of BrAC phase on the BAES stimulation (F(2,36) = 6.1, p<0.005) and sedation subscales (F(2, 36) = 3.6, p<0.05). Baseline ratings that were obtained prior to beverage administration did not differ between sessions thus they were subtracted from the ratings obtained at the ascending and descending BrAC limbs. Relative to baseline, subjects reported feeling more stimulated on the ascending phase of BrAC under alcohol than under placebo (F(1,18) = 9.3, p<0.01). There were no main effects of interactions for the sedation BAES scale.

T1 (120–270 ms).

The main effect of beverage was observed in most regions during the early time window, with stronger event-related theta power under placebo than alcohol overall. The only conflict-related effect was increased theta power in the right ACC under placebo (F(1,19) = 8.6, p<0.01) but not under alcohol (F(1,19) = 1.5, ns).

T2 (320–470 ms).

Conflict-related increase in theta power was observed under placebo but not alcohol in the right ACC as well as lateral frontal and parietal regions (Table 1, Fig. 2). Direct comparison of the beverage effects on the conflict-related theta power (Alc - Plac x INCONG - CONG interaction) showed significant attenuation by alcohol in the right ACC (F(1,19) = 9.9, p<0.01) and right SFG (F(1,19) = 11.3, p<0.005).

T3 (480–670 ms).

Bilateral ACC and right lateral frontal regions showed conflict-related increase in theta power under placebo, but not under alcohol (Table 1). The effect of beverage on conflict-related theta power was significant in the late time window in the left parietal region (F(1,19) = 5.7, p<0.05). Under alcohol, event-related theta power in the right ACC was negatively correlated with RTs to INCONG – the longer the RTs, the weaker theta (r = −0.65, p<0.005).

Not all frontal regions showed sensitivity to conflict. In the middle time window, left IFG showed the largest response to neutral stimuli as compared to all others under both placebo (F(1,19) = 8.0, p<0.02), and alcohol (F(1,19) = 10.2, p<0.005), in agreement with its role in semantic processing [35]. A similar effect was found in the late time window under placebo (F(1,19) = 12.5, p<0.005), but not under alcohol (F(1,19) = 2.2, ns).

Baseline theta.

Total theta power in the prestimulus period was analyzed in order to examine the possibility that the observed effects of alcohol on event-related theta power are due to baseline changes. While there were no beverage or condition effects in any of the left hemisphere regions, the baseline theta was marginally reduced in the right ACC (F(1,19) = 3.7, p<0.1), IFJ (F(1,19) = 4.0, p<0.1), and SFG (F(1,19) = 4.6, p<0.05) under alcohol. Greater sensitivity of the right hemisphere to the effects of alcohol has been reported previously [36], [37], [38]. However, since event-related theta power is calculated by subtracting the baseline power from the raw power, a decrease in the baseline theta power under alcohol would result in increased event-related theta power. Therefore, it is unlikely that alcohol-induced reduction in event-related theta power is caused by changes in the baseline power, but it reflects task related changes in theta power independent from baseline effects.

Pre-response theta.

Pre-response theta power was calculated for the −200 to −50 ms time window immediately preceding the motor response, time-locked to button presses (Fig. 4). It showed sensitivity to conflict in the ACC bilaterally only under placebo (left: F(1,19) = 6.6, p<0.02 right: F(1,19) = 6.7, p<0.02), but not alcohol (F(1,19) <0.6, ns). Alcohol reduced the response preparation theta power bilaterally in the ACC (left: F(1,19) = 8.8, p<0.01; right: F(1,19) = 4.4, p<0.05) and the motor cortex (left: F(1,19) = 8.0, p<0.02; right: F(1,19) = 7.0, p<0.02), indicating alcohol-induced impairment of the regions essential for motor preparation and execution. The effect of Beverage X Condition interaction was not significant in any of the ROIs. Theta to INCONG was negatively associated with RTs, r = −0.53, p<0.02, so that lower theta was associated with longer RTs to high-conflict stimuli.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Group-average timecourses of response-locked theta source power estimates in the left ACC and MOT areas.

Time 0 ms corresponds to the button press. The time window immediately preceding motor responses is indicated by a horizontal bar. During this time, only ACC showed sensitivity to conflict with stronger theta on INCONG than CONG trials during placebo (* p<0.02), suggesting the ACC engagement in response selection and execution. Alcohol attenuated pre-response theta overall. The y-axis depicts baseline-corrected noise-normalized source power expressed in arbitrary units.

https://doi.org/10.1371/journal.pone.0043957.g004

MRI.

Structural MRI images were acquired in order to define the model for the volume conductor and the solution space for MEG source localization analysis. Images were acquired with a 3 T Siemens Trio whole-body scanner (Siemens, Erlangen). For each subject, two high-resolution 3D MP-RAGE T1-weighted sequences that optimize contrast for a range of tissue properties were obtained with the following parameters: TR = 2.53 sec, TE = 3.25 ms, flip angle = 7 degrees, FOV = 256, 128 sagittal slices, 1.33 mm thickness, in-plane resolution 1×1 mm. Structural images were used to reconstruct each person’s cortical surface [75], [76]. Inner skull surface was derived from the segmented MRI data and used for a boundary element model of the volume conductor in the forward calculations. The solution space was approximated by ∼5000 free-rotating dipoles along the gray-white matter surface in the cortex, with spacing between dipole locations ∼7 mm.

MEG.

High-density MEG signals were recorded from 204 channels (102 pairs of planar gradiometers) with a whole-head Neuromag Vectorview system (Elekta) in a magnetically and electrically shielded room. The signals were recorded continuously with 600 Hz sampling rate and minimal filtering (0.1 to 200 Hz). The position of magnetic coils attached to the skull, the main fiduciary points such as the nose, nasion and preauricular points, as well as a large array of random points spread across the scalp were digitized with 3Space Isotrak II system for subsequent precise co-registration with structural MRI images. Trials with incorrect responses were excluded from all further analysis. In addition, the number of included trials was equated across beverage and task conditions for each subject in order to eliminate potential bias due to unequal number of trials. The MEG analysis stream primarily uses our custom Matlab functions and it relies partially on publicly available packages including FieldTrip [77], EEGLAB [78], and OpenMEEG [79]. Single trial MEG data were low-pass filtered at 100 Hz and epoched in two different ways in order to gain insight into both stimulus-related processing and response preparation and execution stages. Data were epoched from −500 ms to 1000 ms relative to each stimulus onset for the stimulus-locked analysis and from −500 to 500 ms relative to each button press for the response-locked analysis. For each epoch, the data were downsampled by a factor of 2, linear trend was removed, and mean activity across the entire epoch was subtracted from each time point. Epoched data were then passed through automatic threshold rejection to remove trials that were contaminated with eye-blinks or other artifacts. Independent component analysis [78] was used to remove heart-artifact components.

Estimated source power constrained to cortical surface was calculated based on the spectral dynamic statistical parametric mapping approach [80], by applying anatomically constrained MEG (aMEG) method based on cortically constrained minimum norm estimate [81] to the complex power spectrum. Complex power spectrum was calculated for each trial using convolution with complex Morlet wavelets [82] in 1 Hz increments from 4 to 7 Hz, with a constant time resolution of 80 ms and frequency resolution of 2 Hz. Wavelet width varied from 2 to 4 cycles in the specified frequency range to ensure constant wavelet length of 500 ms for all frequencies in the theta band. The first and last 250 ms of each epoch were discarded to remove edge artifacts potentially resulting from wavelet analysis. Theta band power was plotted for each individual epoch to further visually inspect the wavelet results and reject additional artifact contaminated trials that had not been detected via the automatic threshold rejection procedure. Complex power spectra were downsampled by a factor of 4 and entered into inverse calculations resulting in 13 ms temporal sampling rate for the spatio-temporal power estimates. To estimate the noise covariance for calculation of the inverse and to prevent biasing the inverse solution against spontaneous brain oscillations, we used empty room data that were detrended and band-pass filtered between 3 and 50 Hz. The signal-to-noise ratio (SNR) equaling 5 [80] was used for scaling of the noise covariance matrix in calculation of the inverse operator. The identity matrix was used for noise-sensitivity normalization of the source-space solution. The noise-sensitivity normalized estimates of total source power were obtained at each location on the cortical surface at each frequency. For each subject, a map of total source power was calculated by averaging across theta band frequencies (4–7 Hz) and across trials. Finally, total event-related theta power was baseline-corrected by subtracting the mean theta source power estimate in the 250 ms prestimulus period. Intersubject averages were created by morphing each subject’s reconstructed surface onto an average representation after aligning their cortical sulcal-gyral patterns [83] and averaging individual source power estimates.

Region-of-interest (ROI) analysis was conducted to further examine possible interactions of the factors of beverage, task condition, and gender on event-related changes in theta power. Unbiased ROIs were selected based on the overall group average across all subjects, task and beverage conditions and comprised dipole locations along cortical surface with most notable source power. The same set of group-based ROIs was used for all subjects in a manner blind to their individual activations by applying an automatic spherical morphing procedure [83]. The ROIs primarily encompassed the cognitive control network in the fronto-parietal regions, as shown in Figs. 2 and 3. More specifically, bilateral ROIs included dorsal ACC on the medial surface, inferior frontal junction (IFJ) [84] touching on the posterior inferior frontal sulcus and precentral sulcus, hand motor region (MOT) in the central sulcus, and intraparietal sulcus (PAR). Additionally included were the left inferior frontal gyrus (IFG) and the right superior frontal gyrus (SFG).

Repeated measures ANOVAs (SPSS for Windows, SPSS Inc) were performed with the factors of Beverage (alcohol, placebo) and Condition (CONG, INCONG, NEUT, READ) for each ROI stimulus-related theta power averaged over time points in three time windows, T1 (120–270 ms), T2 (320–470 ms), and T3 (480–670 ms), chosen to include prominent peaks in theta timecourses across ROIs, Fig. 3, Table 1. Similarly, within the response-locked analysis stream, ANOVAs were carried out on baseline-corrected response-related theta power estimated during the motor preparation time window (−200 to −50 ms), immediately preceding response execution (Fig. 4). Beverage effects on conflict-related theta power changes were assessed by comparing the difference between INCONG and CONG under alcohol and placebo for each ROI (Table 1). In addition, uncorrected baseline power estimates were submitted to the same analysis in order to examine potential effects of beverage and task condition for the −250 to 0 ms time window. Mixed design ANOVAs including Gender as a between-group factor were initially performed separately for each ROI and time window. Since no significant effects of Gender were observed in any of the analyses, reported are the results of repeated measures ANOVA with factors Condition and Beverage. Non-parametric related-samples Friedman’s ANOVA by ranks was performed when the normality assumption was violated, such as percentage of self-corrections and post-experimental Likert scale ratings. Normality was assesed using Shapiro-Wilk Test and by visual inspection using Normal Q-Q plot.

EEG.

In order to provide a complementary measure of the effects of alcohol on event-related theta power and to relate our findings to previous EEG studies, EEG data were measured at Fz and Cz sites simultaneously with the MEG signal. An electrode placed on the tip of the nose served as the reference and the one on the right earlobe as ground. In addition, the electrooculogram (EOG) was recorded with bipolarly referred electrodes placed at the outer canthus of the left eye and just above the nasion. The electrode impedance was kept below 5 kOhms. Good quality complete EEG data sets were obtained from 17 participants and were submitted to the same signal-space analysis as described in the MEG section of the Methods. Total power was calculated at 4 to 12 Hz in 1 Hz increments using 500 ms wavelet length, 14 to 24 Hz in 2 Hz increments using 333 ms wavelet length, and 25 to 55 Hz in 5 Hz increments using 200 ms wavelet length. Grand averages of event-related power expressed as relative change to the prestimulus period were calculated (i.e. normalized power N(t,f) = (P(t,f) – B(f))/B(f), where P(t,f) is raw total power at timepoint t and frequency f and B(f) is mean power for frequency f in the prestimulus time window). Total event-related theta power was averaged for three time windows T1 (120–270 ms), T2 (320–470 ms), and T3 (480–670 ms) for each subject and entered into a repeated measures ANOVA analysis as described above.