To examine the feasibility of implementing a cardiorespiratory exercise stimulus during functional Magnetic Resonance Imaging (fMRI).

12 young adults (age: 18-22 years) completed progressive maximal exercise testing and a brain MRI scan. During scanning, participants completed three runs of functional MRI (volumes = 619; TR = 800 ms; multiband = 4; voxel size = 3 mm3). During each 8 minute fMRI run, participants completed an exercise challenge consisting of alternating blocks of exercise and rest. Exercise was implemented with a cardiostepper, an MRI-compatible device (similar to a Stairmaster) capable of generating a cardiorespiratory exercise stimulus. During exercise blocks, participants stepped at a rate of 60 Hz with pedal resistance determined by participants' fitness level. Heart rate and respiration data were collected during MRI. fMRI data were processed and analyzed using FMRIB Software Library (FSL). The ARtifact Detection Toolbox (ART) software was also used to identify volumes with significant artifact, and ICA-AROMA was used to remove motion-related BOLD signal components.

During exercise blocks, heart rate increased (mean = 131 beats per minute) compared to rest (mean = 87 beats per minute; t(34) = 4.3; p < .001). The mean heart rate during exercise blocks corresponds to an exercise intensity in the light to moderate intensity range for this age group. Motion (median framewise displacement) was significantly higher during exercise (mean = .53 mm) than rest (mean = .36 mm). Across all blocks, ART classified 19.8% of brain volumes as artifact-containing outliers, with 69% of the outliers occurring during exercise blocks. Although greater head motion was observed during exercise, the use of ICA-AROMA reduced the impact of motion considerably, recovering an additional 25% of the task-related signal, relative to noise. Comparison of fMRI activity during exercise versus rest revealed significant associations with primary and supplementary motor cortices, hippocampus, and the insula, among other regions.

The current study demonstrates the feasibility of eliciting light to moderate intensity cardiorespiratory exercise (using a lower body stepping exercise) during functional MRI. Although increased head motion was observed during exercise compared to rest, the degree of head motion was roughly approximate to the values published in previous fMRI studies and post image acquisition processing improved task-related signal. During exercise, increased brain activation was observed in regions associated with the central command network, which regulates autonomic nervous system and musculoskeletal function during exercise.

Timing, or the decision of when to act, is essential to mammalian behaviors from escaping predators to driving a car. It requires cognitive functions such as working memory for time-based rules and attention to the passing of time. Thus, it can be used as a proxy for higher order executive functions that are difficult to measure but are impaired in many neurological disorders. Therefore, insights from studies of interval timing, tasks which require estimating time intervals of several seconds, have great value for our understanding of human disease. Crucial to timing is the basal ganglia, which integrates cortical activity with midbrain dopamine signals and sends out signals to the spinal cord that regulate movement, motivation, and other behaviors. We have previously found that within the basal ganglia, medium spiny neurons of the striatum exhibit ramping activity in time-related tasks. In other words, they gradually increase or decrease firing frequency across a timed interval, and this is thought to encode time. Yet it is still unknown how the encoding of time is translated into time-based motor responses. To answer this question, we turned to the external globus pallidus (GPe) because it is a regulatory hub within the basal ganglia and is thus well positioned to regulate timing behavior. We sought to examine how the GPe functions in response to time-based demands.

We recorded from neuronal ensembles using 16 channel electrode arrays implanted in the GPe of five mice while they performed an interval timing task called the switch interval timing task. Spike sorting was then used to identify signal from individual neurons.

Data were compiled from 43 neurons over several trials. Principal component analysis of neural firing activity was then conducted and revealed a downward ramping pattern in GPe neurons during interval timing trials. Data were then separated based on trials in which mice made correct decisions and those in which mice made a mistake. We found that when mice make correct timing decisions, there is downward ramping activity in the GPe, yet when mice make timing mistakes, this ramping pattern is lost.

Our findings suggest that the GPe processes timing signals through ramping activity, before projecting to the output nuclei of the basal ganglia. This is a novel finding and contributes to a growing understanding of the temporal code of the basal ganglia. The full extent of this code is still unknown, but this insight contributes to a better understanding of how the globus pallidus represents cognition. If we can better explain the neural correlates of timing, we can use this knowledge to inform therapeutic interventions for basal ganglia dysfunction, which could have profound implications for diseases like Parkinson’s disease, which affects millions around the world.

Gait speed is associated with poorer executive functioning performance in older adults such that individuals with slower gait speed have shown declines in cognitive flexibility and set-shifting. Body mass index (BMI) is associated with sedentary lifestyles and slower gait speed, and has demonstrated negative effects on executive set-shifting in this population. However, the interaction between gait speed, BMI, and executive functioning has yet to be examined. The purpose of this study is to investigate the potential mediating effect of BMI on the negative relationship between gait speed and executive functioning in older adults.

The sample included 154 community-dwelling older adults drawn from two clinical trials. Participants were recruited from the VA Palo Alto Health Care System and Stanford/VA Alzheimer’s Disease Research Center. Gait speed was measured using the six-minute walk test, with longer distances representing faster gait speeds. Weight and height were used to calculate BMI. Each participant completed the Trail Making Test Part B (TMTB), which was used to measure executive functioning. A simple mediation analysis was performed using SPSS PROCESS Macro version 3.5. The outcome variable for the analysis was TMTB completion times. The predictor variable was gait speed, and the mediator variable was BMI. Age was entered as a covariate. We hypothesized that gait speed will negatively predict time to completion on the executive functioning task. We also hypothesized that BMI will mediate this relationship.

The analysis found that gait speed negatively predicted executive functioning scores (b = -.12, p = .02). The overall mediation model was statistically significant (F(3,150) =9.17, p < .001). Gait speed and age negatively predicted BMI (p < .001). BMI was a significant predictor of executive functioning (p = .001). The direct effect of gait speed on executive functioning remained significant after including BMI in the model (p < .001), which suggests that BMI partially mediated the relationship between gait speed and executive functioning. The indirect effect of the model when including BMI was tested using the bootstrap estimation approach with 5,000 samples, and was found to be significant (95% CI [.03,.11]), indicating that mediation did occur in the analysis.

BMI partially mediated the relationship between gait speed and executive set-shifting. Thus, the path by which gait speed predicted executive functioning abilities was partially attributable to BMI, one measure of obesity. This finding suggests that older adults with slower gait speeds may have poorer executive function partially due to greater BMI. Given the importance of executive functions on independence and well-being in older adulthood, management of BMI could lead to improved functioning and quality of life. Interventions to decrease weight in older adults is likely to result in several positive health outcomes, and these results suggest that they may also promote important cognitive processes.

Functional near-infrared spectroscopy (fNIRS) is a non-invasive functional neuroimaging method that takes advantage of the optical properties of hemoglobin to provide an indirect measure of brain activation via task-related relative changes in oxygenated hemoglobin (HbO). Its advantage over fMRI is that fNIRS is portable and can be used while walking and talking. In this study, we used fNIRS to measure brain activity in prefrontal and motor region of interests (ROIs) during single- and dual-task walking, with the goal of identifying neural correlates.

Nineteen healthy young adults [mean age=25.4 (SD=4.6) years; 14 female] engaged in five tasks: standing single-task cognition (serial-3 subtraction); single-task walking at a self-selected comfortable speed on a 24.5m oval-shaped course (overground walking) and on a treadmill; and dual-task cognition+walking on the same overground course and treadmill (8 trials/condition: 20 seconds standing rest, 30 seconds task). Performance on the cognitive task was quantified as the number of correct subtractions, number of incorrect subtractions, number of self-corrected errors, and percent accuracy over the 8 trials. Walking speed (m/sec) was recorded for all walking conditions. fNIRS data were collected on a system consisting of 16 sources, 15 detectors, and 8 short-separation detectors in the following ROIs: right and left lateral frontal (RLF, LLF), right and left medial frontal (RMF, LMF), right and left medial superior frontal (RMSF, LMSF), and right and left motor (RM, LM). Lateral and medial refer to ROIs’ relative positions on lateral prefrontal cortex. fNIRS data were analyzed in Homer3 using a spline motion correction and the iterative weighted least squares method in the general linear model. Correlations between the cognitive/speed variables and ROI HbO data were applied using a Bonferroni adjustment for multiple comparisons.

Subjects with missing cognitive data were excluded from analyses, resulting in sample sizes of 18 for the single-task cognition, dual-task overground walking, and dual-task treadmill walking conditions. During dual-task overground walking, there was a significant positive correlation between walking speed and relative change in HbO in RMSF [r(18)=.51, p<.05] and RM [r(18)=.53, p<.05)]. There was a significant negative correlation between total number of correct subtractions and relative change in HbO in LMSF ([r(18)=-.75, p<.05] and LM [r(18)=-.52, p<.05] during dual-task overground walking. No other significant correlations were identified.

These results indicate that there is lateralization of the cognitive and motor components of overground dual-task walking. The right hemisphere appears to be more active the faster people walk during the dual-task. By contrast, the left hemisphere appears to be less active when people are working faster on the cognitive task (i.e., serial-3 subtraction). The latter results suggest that automaticity of the cognitive task (i.e., more total correct subtractions) is related to decreased brain activity in the left hemisphere. Future research will investigate whether there is a change in cognitive automaticity over trials and if there are changes in lateralization patterns in neurodegenerative disorders that are known to differentially affect the hemispheres (e.g., Parkinson’s disease).