Age-related reduction in motor adaptation: brain structural correlates and the role of explicit memory

Abstract

The adaption of movement to changes in the environment varies across life span. Recent evidence has linked motor adaptation and its reduction with age to differences in "explicit" learning processes. We examine differences in brain structure and cognition underlying motor adaptation in a population-based cohort (n = 322, aged 18-89 years) using a visuomotor learning task and structural magnetic resonance imaging. Reduced motor adaptation with age was associated with reduced volume in striatum, prefrontal, and sensorimotor cortical regions, but not cerebellum. Medial temporal lobe volume, including the hippocampus, became a stronger determinant of motor adaptation with age. Consistent with the role of the medial temporal lobes, declarative long-term memory showed a similar interaction, whereby memory was more positively correlated with motor adaptation with increasing age. By contrast, visual short-term memory was related to motor adaptation, independently of age. These results support the hypothesis that cerebellar learning is largely unaffected in old age, and the reduction in motor adaptation with age is driven by a decline in explicit memory systems.

Keywords: Ageing; Cerebellum; Explicit memory; Medial temporal lobe; Motor control; Sensorimotor adaptation.

Fig. 1

Visuomotor rotation learning task. (A) Illustration of the task in which participants moved a stylus-controlled cursor so as to hit a target. The target appeared pseudo-randomly in one of the 4 locations on the screen (once in each of the 4-trial cycles). Participants could not see their hand, and the visual feedback of the cursor was either veridical (pre-exposure and post-exposure phases) or rotated by 30° (exposure phase) relative to the stylus. (B) Participant movement adaptation was assessed by looking at the changes in their initial trajectory error θ_E, calculated 1 cm after starting the movement. (C) Mean trajectory error across the experimental cycles (±1 standard error shaded). Dashed vertical lines separate the phases: pre-exposure (left), exposure (middle), and post-exposure (right). For illustration purposes only, data were split into 3 age groups of a similar size (“young” = 18–45 years, N = 109; “middle” = 46–65 years, N = 102; “old” = 66–89 years, N = 108), although all analyses were performed with age as a continuous variable.

Fig. 2

Final adaptation across age. (A) Example of the model fit in a representative participant. The model consisted of 2 sequential exponential curves, fit with a robust bisquare weight function. The main parameter of interest was “final adaptation.” (B) Correlation between final adaptation and age (with marginal histograms). Solid line indicates the linear regression fit with 95% confidence interval (gray shade).

Fig. 3

Structural imaging results. (A) Axial sections (left; numbers indicating z coordinate) and 2 coronal sections (right; with y coordinate), showing significant clusters (red) where there was a significant (p < 0.05, FWE-corrected) conjunction between the positive association with adaptation and negative association with age. These clusters included the striatum, bilateral premotor cortex, superior parietal lobule, and lateral frontal cortex. No such effect was found in the cerebellum. (B) Sagittal sections (numbers indicating x coordinate), showing 3 significant clusters (blue) where there was a significant (p < 0.05, FWE-corrected) positive interaction between final adaptation and age in relation to gray matter volume. These clusters included 1 in the right middle and inferior temporal lobe, and 2 clusters in the medial temporal lobes, 1 on the left and 1 on the right, each encompassing the hippocampus and amygdala. (C) Illustration of the positive interaction from (B). The interaction in all clusters was driven by a more positive relationship between gray matter volume and final adaptation in older adults than in younger participants. Mean gray matter volume extracted for left medial temporal lobe cluster for illustration of interaction direction. Groups split by age as in Fig. 1 for illustration purposes only. Abbreviation: FWE, family-wise-error.

Fig. 4

Explicit memory performance and motor adaptation by age. (A) Illustration of the positive interaction between age and declarative LTM performance in the Story Recall task in relation to final adaptation. LTM scores were the total number of items recalled after a delay period, such that higher values indicate better LTM. Groups split by age as in Fig. 1 for illustration purposes only. Solid line indicates the linear regression fit with 95% confidence interval (gray shade). (B) As in (A), but for short-term memory score in the visual STM task. STM scores were the RMSE of the difference between target and reported color, such that higher values indicate worse STM performance. Abbreviations: LTM, long-term memory; RMSE, root-mean-square error; STM, short-term memory.