Finger-specific representations are sharpened during a fatiguing motor task

2054 

Abstract Submission 

Caroline Heimhofer^1,2, Susanne Koblitz¹, Marc Bächinger¹, Nicole Wenderoth^1,3

¹ETH Zurich, Zurich, Switzerland, ²Neuroscience Center Zurich (ZNZ), University of Zurich, Federal Institute of Technology Zurich, University and Balgrist Hospital Zurich, Zurich, Switzerland, ³Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence and Technological Enterprise (CREATE), Singapore, Singapore

Caroline Heimhofer  
ETH Zurich|Neuroscience Center Zurich (ZNZ), University of Zurich, Federal Institute of Technology Zurich, University and Balgrist Hospital Zurich
Zurich, Switzerland|Zurich, Switzerland

Susanne Koblitz  
ETH Zurich
Zurich, Switzerland

Marc Bächinger  
ETH Zurich
Zurich, Switzerland

Nicole Wenderoth  
ETH Zurich|Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence and Technological Enterprise (CREATE)
Zurich, Switzerland|Singapore, Singapore

Motor fatigability is a frequent symptom in neurological disorders. It can be quantified through the decrease in movement speed, when low-force movements are performed repeatedly with maximal speed. In this study, we measure motor fatigability in healthy with fast finger tapping. Previous research has shown that the decrease in movement speed, or motor slowing, is associated with a rise in BOLD activity, a reduction in surround inhibition in the primary sensorimotor cortex (SM1), and an increase in co-activation of antagonistic muscle groups involved in the movement (Bächinger et al. 2019). However, it remains an open question of whether motor slowing and the associated release of inhibition causes a reduction of signal-to-noise ratio for movement-specific information. Here, we aim to answer this question by assessing finger representations using representational similarity analysis (RSA) when participants perform fatiguing tapping with the index or middle finger (Fig 1A). We hypothesized that a reduction of movement-specific information would be associated with the index and middle finger representations in SM1 getting "blurred" over time due to a gradual break-down of surround inhibition. Thus, if the signal-to-noise ratio of movement-specific information decreases in parallel with motor slowing, we would expect finger representations to become more overlapping than is predicted by BOLD activity changes alone. Vice-versa, if the signal-to-noise ratio of movement-specific information increases despite motor slowing, we would expect sharper finger representation than purely predicted by the BOLD activity increase (Fig 1B).

26 healthy young participants performed a motor slowing finger tapping task during functional MRI. The participants performed 30s of maximal speed finger tapping with the index and the middle finger, alternating between trials. For the first-level general linear model, the fingers were regressed separately and the 30s of tapping were further split into 3 x 10s regressors (time bin 1, bin 2, bin 3). We performed RSA separately on each 10s regressor for the anatomically defined regions of interest (ROI) M1 and S1 hand area (Diedrichsen, et al. 2013; Walther et al. 2016) and therefore obtained a dissimilarity measure for each time bin (Fig 2B&E, actual dissimilarity, purple). A mixed effects model with the factor time was used to test whether dissimilarity changed across time bins. Since the change in dissimilarity over time bins might purely be explained by changes in BOLD activity, we extracted an estimate for BOLD activity for each time bin and used a mixed effects model with the factor time to test whether activity increased. We then predicted expected dissimilarities for bin 2 and bin 3 based on the changes in activity (Berlot et al. 2021; Fig 2B&E, predicted dissimilarity, pink). Using another mixed effects model with the factor dissimilarity type (predicted, actual) and time (bin2, bin3), we tested whether the predicted and actual dissimilarities differed. Post-hoc comparisons were then performed on each time bin. These analyses were done separately for each ROI.

Behaviourally, motor slowing was observed, as tapping speed significantly decreased in each finger over time (F(1,85)<=10.67, p<=.001, Fig1A). BOLD activity increased over time for S1 and M1 (F(2,50)<=20.09, p<.001, Fig 2A&D). Comparison of the predicted versus the actual dissimilarity revealed a significant difference in both ROI (F(1,75)<=37.62, p<.001). Post-hoc comparison showed that time bin 2 and bin 3 both have significant differences between predicted and actual dissimilarity in both ROI (p<.005, Fig 2B&D).

We conclude that the finger representations in the sensorimotor cortex become more distinct with motor slowing, even when correcting for the increase in activity. This suggests that the signal-to-noise ratio of movement-specific information is increased, potentially to compensate for supraspinal changes caused by fatigability.

Multivariate Approaches ²

Other Methods

Motor Behavior Other ¹

Cortex

FUNCTIONAL MRI

Motor

Multivariate

NORMAL HUMAN

Other - motor fatigability