[ARTICLE] Learning and transfer of complex motor skills in virtual reality: a perspective review – Full Text

Abstract

The development of more effective rehabilitative interventions requires a better understanding of how humans learn and transfer motor skills in real-world contexts. Presently, clinicians design interventions to promote skill learning by relying on evidence from experimental paradigms involving simple tasks, such as reaching for a target. While these tasks facilitate stringent hypothesis testing in laboratory settings, the results may not shed light on performance of more complex real-world skills. In this perspective, we argue that virtual environments (VEs) are flexible, novel platforms to evaluate learning and transfer of complex skills without sacrificing experimental control. Specifically, VEs use models of real-life tasks that afford controlled experimental manipulations to measure and guide behavior with a precision that exceeds the capabilities of physical environments. This paper reviews recent insights from VE paradigms on motor learning into two pressing challenges in rehabilitation research: 1) Which training strategies in VEs promote complex skill learning? and 2) How can transfer of learning from virtual to real environments be enhanced? Defining complex skills by having nested redundancies, we outline findings on the role of movement variability in complex skill acquisition and discuss how VEs can provide novel forms of guidance to enhance learning. We review the evidence for skill transfer from virtual to real environments in typically developing and neurologically-impaired populations with a view to understanding how differences in sensory-motor information may influence learning strategies. We provide actionable suggestions for practicing clinicians and outline broad areas where more research is required. Finally, we conclude that VEs present distinctive experimental platforms to understand complex skill learning that should enable transfer from therapeutic practice to the real world.

Introduction

The goal of rehabilitation interventions for clients with neurological impairments is to (re)learn motor skills during therapeutic practice and transfer those improvements to functional activities in daily life. Researchers and clinicians seek to understand the content and structure of practice that facilitates such learning and transfer for different tasks, environmental contexts and clinical populations [1]. Although (re)learning activities of daily living is the focus of neurological rehabilitation, much of the evidence base for therapeutic interventions stems from basic or clinical research on simple experimentally-controlled tasks, such as reaching to a target in the horizontal plane or learning a finger tapping sequence. While these simplified tasks are very different from the tasks of daily life, they facilitate precise quantification of performance variables and stringent hypothesis testing, providing insights into basic principles of motor control and learning. However, their deliberately reduced testbeds lack a feature that is pervasive in real-world tasks: the affordance of multiple options to achieve a movement goal [2]. Hence, principles of learning derived from these simple movement paradigms may not translate into useful transfer-oriented principles for rehabilitation [3].

With some exceptions, e.g., Constraint-Induced Movement Therapy [4], few rehabilitation interventions can consistently demonstrate evidence for transfer from practiced tasks to non-treatment contexts. This is also true for the rehabilitation-based use of virtual environments (VEs): computer hardware and software systems that generate simulations of real or imagined environments with which participants interact using their own movements [5]. VEs differ according to viewing medium, level of immersion, and type of interaction [6]. While practice in a variety of VEs offers promising evidence for skill acquisition as compared to conventional interventions in many rehabilitation populations, [e.g. 7, 8, 9, 10] the focus has been predominantly on training simplified movements. This may be one reason why successful transfer of skill learning to non-practiced tasks and real-life contexts often remains a challenge [11, 12, 13, 14, 15, 16]. As such, the design of both virtual and conventional interventions requires greater understanding of how humans acquire, retain and transfer real-world skills. We propose that VEs themselves can serve as useful experimental platforms to gain this knowledge as they allow the study of these complex skills with sufficient experimental control to draw scientifically tractable conclusions [2].

Complex real-world tasks have nested redundancy

In the motor learning literature, the adjective “complex” is often treated synonymously with “difficult” [17, 18]. For example, a task can be labelled as difficult or complex when reaction time or movement time are relatively long, when skill improvement requires long hours of practice, or when the task poses high demands on the learner’s attention and memory [3]. To sharpen the discussion, we reserve the term ‘complex’ for tasks with nested redundancy. Redundancy is present when there is a greater number of execution variables than variables that define the result of the task. The well-known example for motor redundancy is pointing to a target with one’s fingertip, which can be achieved with many different joint configurations, because the arm (without the hand) has 7 degrees of freedom, while the target is defined in 3 degrees of freedom.

However, real-world tasks have another level of redundancy that lies in the task itself. Imagine you are asked to point to a line, where each location on the line is equally correct. Here, the task itself allows an infinite number of “solutions”. And of course, each of those solutions can be achieved with an infinite number of joint configurations. Further, each of the points on the target line can be reached with an infinite number of trajectories from the starting point towards the target line. It is these nested redundancies that characterize the challenge and the richness of real-world tasks. Figure 1 illustrates these nested redundancies with the example of hammering a target on an anvil. The traces are the original recordings of Bernstein from the 1930s, showing the tip of a hammer in the sagittal plane [19]. The added simplified arm with three joints can take on infinite configurations for any position of the hammer endpoint in the 2D plane (intrinsic redundancy). Next, the trajectories of the repeated endpoint actions take on many different shapes, in fact infinitely many shapes, while all of them hit the anvil (extrinsic redundancy). Finally, the anvil or target itself is not a point but a line, where any contact is regarded as a successful hit (task redundancy) [20]. Examples for these nested redundancies are ubiquitous in real life, from combing one’s hair to cutting a steak with a knife and fork. Performers must choose (implicitly or explicitly) from an infinite range of possible solutions, each leading to successful task accomplishment [2]. We define such actions as ‘complex’ skills. To gain insight into these ever-present control challenges and opportunities, scientific inquiry must move beyond simple tasks where redundancy has been purposefully removed and begin to examine more complex tasks.

Open image in new window

Fig. 1Nested redundancies in a hammering task

[…]

Continue —->  Learning and transfer of complex motor skills in virtual reality: a perspective review | SpringerLink