Motor learning is driven by error and reward feedback which are considered to support two different mechanisms: error-based and reward-based learning. While in the real-world we normally experience them together, in lab-based tasks we typically isolate them to study the different learning mechanisms. Yet, dissociating between them in the real-world is not trivial.
In previous studies, we established a paradigm to study real-world motor learning using the game of pool billiards, capturing full-body movement using motion tracking (Haar, van Assel, & Faisal, 2020) and brain activity through wireless electroencephalography (EEG) system (Haar & Faisal, 2020). In this paradigm, we showed the use of varying contributions of error-based and reward-based learning mechanisms in different individuals during real-world learning in the pool task (Haar & Faisal, 2020). We then incorporated it in an embodied Virtual Reality (EVR) environment (Haar, Sundar, & Faisal, 2021) to allow perturbations and visual manipulations to be applied. In the EVR setup, while the participants perform the shot in the real world on a physical pool table, we provide visual feedback through the VR system, which allows us to introduce feedback manipulations (isolating error and reward feedback) and visuomotor perturbations.
In this study, we used the EVR setup to study the effects of the forced use of a single learning mechanism on learning and related brain activity. Each of the 32 participants attended the lab for two sessions to learn visuomotor rotation (clockwise or counterclockwise) with each feedback (error or reward) separately, counterbalancing among participants by the order of the feedback and rotation directions. The error feedback provided in the EVR showed the cue ball trajectory only up to the point of ball collision, whereas the reward feedback consisted of the ball pocketing with a fixed trajectory for all rewarded trials while for all unrewarded trials, the ball disappeared after the shot. Rewarded trials were defined based on a reward zone based on improvement from previous trials, following previous works in lab-based tasks (Therrien, Wolpert, & Bastian, 2016).
Our behavioural results showed significant differences between the learning with the different feedback, but in both cases the participants learned to correct for the rotation, at least partially.
In their brain activity, we were looking at the post-movement beta rebound (PMBR) response, which was linked to motor learning performance in many previous studies. Specifically, previous studies showed different trends of PMBR over reward-based and error-based learning.
Our results show the presence of different trends over reward and error-based learning, in line with the suggested role of the PMBR as a neural marker for learning mechanisms. Specifically, in the reward condition, participants’ PMBR displayed a steadily decreasing behaviour, in line with the initial hypothesis in the pool study held in the real-world. However, unlike previous results in multiple studies, in the error feedback, participants’ PMBR displayed no significant trend.
Our study provides preliminary evidence for the potential to manipulate brain activity through the manipulation of visual feedback in a real-world motor learning task.