Effect of locomotor demands on cognitive processing

Abstract

Understanding how brain dynamics change with dual cognitive and motor tasks can improve our knowledge of human neurophysiology. The primary goals of this study were to: (1) assess the feasibility of extracting electrocortical signals from scalp EEG while performing sustained, physically demanding dual-task walking and (2) test hypotheses about how the P300 event-related potential is affected by walking physical exertion. Participants walked on a treadmill for an hour either carrying an empty rucksack or one filled with 40% of their body weight. During the walking conditions and during a seated control condition, subjects periodically performed a visual oddball task. We recorded scalp EEG and examined electrocortical dynamics time-locked to the target stimulus. Channel-level event-related potential analysis demonstrated that it is feasible to extract reliable signals during long duration loaded walking. P300 amplitude was reduced during loaded walking versus seated, but there was no effect of time on task. Source level activity and frequency analysis revealed that sensorimotor, parietal, and cingulate brain areas all contributed to the reduced P300 amplitude during dual-task walking. We interpret the results as supporting a prioritization of cortical resources for walking, leading to fewer resources being directed toward the oddball task during dual-task locomotion.

Figure 1

Data processing pipeline. Data processing workflow. EEG data were first processed using fairly standard cleaning methods (grey boxes). After running AMICA, the data were processed in two different ways in order to perform both the channel-level analysis (orange boxes) and component-level analysis (green boxes).

Figure 2

Psychological and Behavioral Results. Psychological and Behavioral Results. Panel A shows subjects responses to the BORG Rating of Perceived Exertion over time. It was administered before and after every bout of the cognitive task. Panel B shows the reaction time to respond to the visual target during the seated (early and late) and walking (early and late) cognitive task bouts. Panel C depicts the Target Accuracy for the same cognitive bouts as Panel B.

Figure 3

Feasibility of measuring ERPs during loaded walking. Panel A shows topographical plots of the channel-level ERPs across time for non-target vs. target of the loaded walking condition. Data were time-locked to the presentation of the stimulus. Panel B displays the t-scores testing the significant differences between the non-target and target trials across all channels using cluster mass permutation. Channels are ordered based on the 256 Biosemi electrode cap (A1 – H32) and displayed by quadrant. Non-significant values were set to zero (green). The topographical plots below the figure represent the spatial location of the ERP t-test results for 400, 600 and 800 ms. The black symbols denote channel location and levels of significance (p < 0.01) for significantly different ERP responses.

Figure 4

Early and late task channel-level ERPs. Channel-level ERPs over time. The plots show ERP traces averaged across subjects in the central-parietal region of interest for the seated (early and late) and walking (early and late) cognitive bouts for the unloaded and loaded conditions (left and right panels, respectively).

Figure 5

Channel-level ERPs across conditions. Panel A displays the channel-level ERP in response to the target stimulus averaged across all subjects during seated, unloaded, and loaded conditions for the Central-Parietal ROI (see location shown in panel inset). Panel B shows the pair-wise differences between all three levels of locomotor demand (Seated, Unloaded, Loaded), α = 0.017 to account for multiple comparisons. Non-significant cluster t-values were set to zero (green). The topographical plots below each figure show the spatial location of the ERP responses for 375 and 445 ms. The black symbols denote channel location and levels of significance **p < 0.01;×p < 0.05) for significantly different ERP responses.

Figure 6

Component-level ERPs across conditions. Each plot depicts the average component-level ERPs for the Seated, Unloaded, and Loaded conditions in the approximate location of 12 ROIs across the scalp. Under each plot, the red and blue bars denote significant differences between Seated vs. Unloaded (blue bar) and Seated vs. Loaded (red bar) over time as determined by the cluster mass permutation test. No significant differences were found between the Unloaded and Loaded conditions.

Figure 7

Component-level ERSPs. Panel A displays the event related spectral perturbations (ERSPs) for the seated, unloaded, and loaded conditions (left, right and middle columns, respectively) for six relevant ROIs across the scalp. Panel B shows the significance-masked cluster mass permutation results for Seated compared to Unloaded and Seated compared to Loaded conditions (left and right columns, respectively) for the same ROIs as in Panel A. Non-significant cluster t-values were set to zero (green). No significant differences were found between the Unloaded and Loaded conditions. Note that these ROIs are the same as those depicted in Fig. 4, but only data from the right side of the brain is shown here due to the lateral similarities found across the scalp.