Abstract
After a stroke, many patients underuse their paretic arm despite substantial recovery, which can lead to a secondary decline in motor function. The mainstream theory explaining this overuse of compensatory patterns (such as using the less affected arm or compensating with the trunk) is based on the hypothesis of a learned underestimation of the utility of the paretic arm through behavioural conditioning; however, an alternative explanation is possible. Nonuse of the paretic arm could result from an optimal control strategy aimed at minimising effort and maximising upper limb function in the presence of muscle weakness and impaired intermuscular coordination. To test this hypothesis, we conducted several studies in healthy young and elderly participants and stroke patients using upper limb motor tasks (seated reaching, circular steering).Our results showed that when healthy seated participants were asked to reach a target placed in front of them while holding a heavy dumbbell, they leaned forward to perform the task. When instructed to minimise trunk compensation, healthy participants were able to reduce compensation at the cost of increased shoulder muscle activation and greater perceived effort. This shows that it is possible to induce shoulder-elbow nonuse (proximal arm nonuse) in healthy subjects without requiring a learning phase. Conversely, by lightening the arm of post-stroke patients, they decreased their mandatory and non-mandatory trunk compensations, thereby spontaneously increasing arm use. A detailed analysis revealed that the nonuse strategy, by favouring compensations, avoids excessive muscle activations due to shoulder weakness and thus reduces the perceived effort.In addition, by reducing motor variability through reduced muscle activation and improving movement control through increased reliance on the less affected trunk, the proximal arm nonuse strategy also improves motor control of paretic arm movements and overall upper limb function in stroke patients. Consequently, both mandatory and non-mandatory compensations are highly adaptive, increasing when the target is smaller, or conversely decreasing when the target is higher (a higher target making trunk compensations less useful or even counterproductive). We also demonstrated a correlation between paretic limb impairment, motor and brain compensations, confirming that motor and brain compensations arise to compensate for motor deficits. Finally, we demonstrated that, within an optimal control approach, only a cost function that accounts for actual joint strength reserves can accurately model trunk compensations and predict proximal arm nonuse in the presence of shoulder strength deficit.These results suggest that the strategy of nonuse of the paretic limb in post-stroke patients is, at least in part, an optimal solution to reduce effort and improve motor function. We propose a method for identifying patients who would benefit most from interventions aimed at reducing compensations (CIMT, TRMT) and discuss the role of specific shoulder strength training in improving arm use in post-stroke patients.