Identifying motor learning deficits in neurological conditions: a critical analysis of a perennial problem

Abstract

Identifying deficits in motor learning has the potential to serve as an indicator of brain function in several neurological conditions. However, evidence for motor learning deficits can be confounded by factors unrelated to learning. In this review, we critically examine the evidence for these deficits in neurological conditions, focusing both on conceptual and methodological issues. Across a wide range of neurological conditions, we found that a majority of evidence for motor learning deficits is difficult to clearly interpret as learning-related and is potentially confounded by the presence of baseline differences. In addition, the use of low sample sizes, short time durations of practice, and the narrow focus on sequence learning paradigms also raise questions about the validity and generality of this evidence. Given that deficits in motor learning have implications for the early detection of neurological status and the design of rehabilitation strategies, we highlight the need for greater rigor when addressing this important, but perennially challenging, question.

Keywords: Baseline; Learning; Parkinson’s disease; Performance; Skill; Stroke.

Fig. 1

Assessing learning deficits in the presence of baseline differences. Hypothetical learning curves are presented for two groups (indicated in blue and red), with practice on the x-axis and movement time/error on the y-axis in arbitrary units (such that decreasing values indicate “better” performance). Learning deficits are assessed using four different learning measures (final level, change, relative change, and rate). When there are no baseline differences, A all measures of learning tend to agree. However, in the presence of baseline differences (B)–(E), conclusions can be ambiguous because they are highly dependent on the choice of the learning measure

Fig. 2

PRISMA graph of study selection. A total of 58 papers (covering 72 studies) on motor learning deficits were included for this review

Fig. 3

A Neurological conditions where motor learning deficits were assessed. A majority of the studies (57%) reported deficits in motor learning. B Types of task paradigms used for assessing deficits and the nature of the deficit investigated. Sequence learning was the most popular paradigm, followed by adaptation and coordination tasks. Learning rate deficits were investigated more often, although sequence learning paradigms did tend to have a high number of studies that investigated learning mechanisms

Fig. 4

A Type of learning measure used for assessing deficits – change scores were the most common, which can be problematic when there are baseline differences. B Presence of baseline differences, i.e., difference in performance at the onset of learning between the clinical and the control groups. Over half of the studies reported differences in baseline performance. C Accounting for baseline differences was dependent on the task paradigm. In sequence learning tasks, the majority of paradigms used a control condition with a random sequence acting as a ‘within-subject’ comparison to offset the presence of any baseline differences. However, in other paradigms, there were statistical corrections (such as adding covariates), task adjustments, or no corrections

Fig. 5

A Sample size distribution across studies. Sample sizes were low, but typical of motor learning studies in the literature (~ 10–20/group). B Time duration of motor learning across studies. A majority of the studies focused only on a single day of practice, again typical of most motor learning studies