Use-dependent biases primarily originate from a contaminated motor plan

Abstract

Repetition of a specific movement biases subsequent actions towards the recently practiced movement, a phenomenon referred to as use-dependent learning (UDL). UDL has been attributed to shifts in the tuning of neurons in the motor cortex. However, recent studies employing a forced reaction time task, including the eLife article by Marinovic et al (2017), indicate that these biases may also arise from a contaminated motor plan, one that is biased towards the practiced direction. We advanced this line of inquiry, seeking to establish the relative contribution of execution and planning processes to UDL in a center-out reaching task in which participants were able to initiate movements of their own volition. On most trials, the target appeared at a designated “frequent” location; on other trials, the target appeared at one of six “rare” locations. In Experiment 1, participants exhibited a robust movement bias towards the frequent target when movements were self-initiated quickly, but a small movement bias when movements were self-initiated slowly – the signature of a contaminated motor plan. Strikingly, the heading angles were bimodally distributed, with one peak at the frequent target location and the other at the rare target location – a finding reinforced by a re-analysis of two widely cited studies on UDL. Notably, the latter peak was shifted in the frequently practiced direction, a signature of a motor execution bias. To eliminate the contribution of planning-related UDL, we imposed a delay between target onset and movement initiation in Experiment 2. As predicted, the heading angles became unimodally distributed around the rare target. The peak of this distribution was again shifted towards the location of the frequent target, indicative of a persistent bias in motor execution. Taken together, these results highlight two distinct components of UDL even when movements are self-initiated: First, the temporal dynamics underlying movement planning, in which a default plan is progressively overridden by a new plan, produces a pronounced motor planning bias. Second, there is a small, temporally stable bias that may reflect shifts in motor unit tuning.

Understanding implicit sensorimotor adaptation as a process of proprioceptive re-alignment

“…Skill learning requires a much more flexible system, one that can exploit multiple sources of information and heuristics to create novel movement patterns (Yang et al, 2021). These capabilities draw on multiple learning processes that use a broad range of error and reinforcement signals (Galea et al, 2011;Shmuelof et al, 2012;Tsay, Kim, Saxena, et al, 2021) that may be attuned to different contexts (Avraham, Taylor, et al, 2020;Heald et al, 2021). While we anticipate that these processes are sensitive to multimodal inputs, it will be useful to revisit these models with an eye on the relevance of proprioception.…”