Tilted trials: Assessing performance with varying angle perturbations in visuomotor adaptation task

INTRODUCTION

Visuomotor coordination enables humans to perform a wide range of activities, from daily tasks such as typing, driving, cooking, playing music, sports and computer games to complex procedures such as laparoscopic surgery.^[1] For instance, surgery requires ease of adaptation to visuomotor changes during minimally invasive procedures that require attention and precision.^[2] Previous studies have shown that learning a new visuomotor relationship relies on precise arm movements that require the conversion of sensory inputs into motor commands.^[3-5] In the process of acquiring a new visuomotor relationship between visual and proprioceptive feedback, the ability to distinguish between motor movements and somatosensory information is the key. This process, known as adaptation, is involved in acquiring new skills in response to perceived changes to achieve a target.^[5,6]

Adaptation to visuomotor rotations is one of the most extensively researched motor learning paradigms.^[4] Visuomotor adaptation can be used as a dynamic task to study the adaptation processes that include transformation, integration, modification and storage of visuospatial and kinaesthetic change.^[7] It is a screen-cursor transformation with a directional bias (perturbation) around the hand, giving rise to motor learning. The introduction of variable perturbation leads to recalibration that occurs during visuomotor transformation.^[8] The magnitude of the imposed perturbation determines the degree to which adaptation reduces errors in response to altered conditions.^[9]

Understanding motor adaptation and its strategies, informed by skill transfer evidence from virtual to real environments in diverse populations, can enhance robotic learning and adaptability.^[10] Visuomotor transformation, extending its applications beyond everyday tasks to rehabilitative practices, presents significant benefits for stroke victims through methods like error-based learning.^[11]

Visuomotor transformation involves an internal model to detect sensory error and provide motor instructions for hand movement. This is achieved by applying an external force, like an angular perturbation, to study how the internal model learns and adapts.^[12,13] Studies have shown that when the perceived angle change matches the perturbation angle, errors decrease, indicating sensory-motor adaptation.^[14] Motor adaptation aims to minimise the difference between predicted and actual sensory feedback, though measuring these predictions can be challenging.^[15] This concept can be used to evaluate the sensorimotor abilities of human participants.^[16]

It is crucial to study the sensory-motor components since implicit motor learning has become more effective and practical in neurological rehabilitation.^[17] This approach is especially relevant for Parkinson’s disease, which involves complex motor learning deficits, with a critical focus on balance and gait impairments in neurological research.^[18]

Research has also shown that ageing negatively impacts motor learning, affecting both explicit and implicit processes.^[19] This affects both explicit and implicit processes, such as conscious and unconscious learning. Explicit learning is deliberate and conscious, while implicit learning is gradual and occurs without conscious awareness. Explicit learning involves adjusting one’s aim to counter detected perturbations, while implicit learning relies on sensory prediction errors.^[20] For example, playing tennis, the learner is consciously aware of the movements they are making, and they are deliberately practising and refining their technique to improve their performance.^[21]

Explicit and implicit learning rely on overlapping neural networks but are weighted differently.^[22] An important issue to address is how implicit learning (adaptation) can offer some compensation for explicit learning deficits. Patients with motor impairments often require relearning and readjustment of motor skills during neurorehabilitation.^[23] In contrast to experimental paradigms utilising a single perturbation angle, our approach involved introducing various perturbation angles, both small and large, to gain insights into the strategies employed during movement execution, encompassing both implicit and explicit learning.

The angle of perturbation serves as an influential parameter that modulates implicit and explicit learning and their associated neural correlates.^[24] While the neural systems that govern these two learning processes are not entirely understood,^[25] the selected angles differentiate implicit (automatic) and explicit (conscious) learning processes.^[26,27] Implicit learning enables automatic adaptation to sensory deviations, while explicit learning supports strategic planning and execution based on conscious awareness.^[28,29]

These angles (30° Clockwise - 45° CW–60° CW) were used in the VMT task for their impact on mechanical stability, perceptual-motor performance and cognitive load, facilitating efficient learning with minimal training.^[30,31] Moderate degrees of rotations facilitate new motor mapping without overwhelming participants, thereby enhancing learning efficiency.^[32] The 30° angle, commonly used in experiments, balances movement facilitation with adaptation demands^[33] as the body is engaged in less complex trajectories, which can optimise conditions for implicit learning.^[34]

Conversely, research indicates that at 45°, participants often utilise both implicit learning strategies and explicit cognitive control, which involves the prefrontal cortex’s active engagement.^[35,36] The 60° angle introduces greater complexity, challenging both implicit and explicit learning.^[37] Larger rotations impose greater cognitive and motor demands, potentially exceeding adaptation limits and leading to task disengagement.^[38]

It demands increased cognitive resources to interpret sensorimotor feedback and engages additional neural substrates, including the hippocampus, which is involved in memory processes during procedural learning.^[39] Studies show that obtuse angles increase activation in areas like the precuneus, highlighting distinct neural pathways for explicit and implicit learning.^[28]

Therefore, based on existing literature, 30°, 45° and 60° angles were chosen to examine the shift from implicit to explicit learning in visuomotor adaptation.^[40,41] Smaller perturbations (30°) engage cerebellum-driven implicit recalibration, and larger ones (60°) require explicit strategy use through the prefrontal cortex,^[42] while the intermediate 45° condition captures their interaction, involving prefrontal and parietal cortices.^[43]

The primary objective of this study is to investigate the impact of varying angular perturbations on visuomotor adaptation. We acknowledge that excessive focus on the end goal can be counterproductive. To achieve consistent and successful performance in the face of variations, it is advisable to shift one’s attention away from the primary goal. We hypothesised that various angles of perturbation could impact the nature of adaptation across trials if the mapping between mouse and cursor movement changes frequently and unpredictably. The task structure was modified by introducing a visuomotor conflict, where participants were expected to reverse the angular lag. This was done to characterise the nature of adaptation to various angles of rotation, through an error-based learning experience.

For the administration of visuomotor tasks, utilising distinct colours for feedback can effectively differentiate between anticipated movements and unexpected perturbations, which is crucial for successful adaptation.^[44,45] Evidence indicates that when colour cues were used, participants could easily interpret feedback and associate it with various types of adaptation or necessary responses, thereby enhancing performance outcomes.^[46] Experimental reports suggest that enhanced colour perception refines visuomotor adaptation by providing contextual cues that aid task performance.^[47]

Evidence suggests that colour-based feedback influences the interplay between frontal and posterior brain regions, potentially enhancing top-down control and information integration during learning tasks.^[48] Studies also indicate that colour processing originates in distinct neural pathways contributing to cognitive load associated with movement adaptation.^[49] Although it could overwhelm the cognitive processing, it effectively guides the responses and reduces uncertainty about the task requirements.^[39] In this study, rather than acting as a distraction, colour feedback provided significant context that enhances cognitive efficiency during a visuomotor adaptation task (VMAT).^[50] Therefore, it was sensible to use colour feedback to distinguish between target and successful trials.

The present study was conducted to understand the variation of performance through directional errors and adaptation under the influence of imposed rotation between perturbation blocks to learn about implicit and explicit forms of learning separately. In addition, to indicate whether there is an effect of random sequence allocation for participants at various angles of perturbation, we compared the performance parameters, including mean directional errors (MDEs), adaptation, reaction time (RT) and movement time (MT) between angle sequences provided to participants in a randomised order.

MATERIALS AND METHODS

Task paradigm

The VMAT was administered and scripted using The MathWorks, Inc., MATLAB R2012b software in Natick, USA. A 17-inch flat panel liquid-crystal display (LCD) monitor with a refresh rate of 60 Hz (Dell Professional P170S) was used for the task, with participants seated at a distance of 70 cm from the monitor.

The participants were asked to perform a reaching task in which a target appeared on the computer screen in one out of eight radial directions, as shown in Figure 1. The participants were required to reach the target by moving the cursor from the fixation point (distance: 4 cm) using a mouse. The completion of a trial was detected by the algorithm, signalling the initiation of the next trial. In each trial, one target appeared randomly at one of the 8 locations on the screen, and 35 such trials were presented to each participant for familiarisation. The following instructions were given to the participants before the commencement of the experiment.

Close

Figure 1:

Visuomotor adaptation task: Types of Blocks (Baseline, Washout - No Perturbation) and Perturbation angles (at 30°-45°-60°), Mouse movement and the cursor movement shown by red and blue arrow.

Export to PPT

1. Before each trial, click on the red target (starting point)

2. You must aim at the blue target at a comfortable speed and stop at/within the target

3. You are not allowed to stop in between, change trajectory or correct their initial movement plan during their movement

4. Once you start, maintain a straight line and only stop after reaching the target

5. At the end of the movement, the target turns green

6. After a few trials, an angle change will be introduced in the cursor movement. Make out the angle and move in a straight line

7. Towards the end of the block, the angle change in the cursor movement will not be present.

Following the familiarisation session, the participants were given a 1-min rest period before the task was administered. The experimental design [Figure 1] for VMAT consisted of 7 blocks, consisting of baseline (one block of 20 trials), angle perturbation (three blocks of angle perturbation, 60 trials each) and subsequent washout (one block of 20 trials).

Each participant was randomly assigned to one of the six possible sequences of induced angle perturbation during the three perturbation blocks, without being told of the specific sequence that would be administered. The possible angle sequences were 30° 45° 60°, 30° 60° 45°, 45° 30° 60°, 45° 60° 30°, 60° 30° 45° and 60° 45° 30°.

During each perturbation block, the movement of the cursor was perturbed randomly by 30°, 45° or 60° angles compared to the position of the target. The path of the movement of the cursor on the screen acted as an indicator of the angle of the perturbation induced with respect to the target in each block. The participant attempted to reach the target while making corrections for the direction of movement of the mouse as soon as the target was missed in the initial perturbation trials, as shown in Figure 2. As instructed to the participants, the objective of the task was to reach the target with a minimum path length. The participants completed 60 trials in each perturbation block.

Close

Figure 2:

Order of events across trials (baseline, 60° perturbation, and washout trials) of the visuomotor adaptation task. Each trial began with a red central fixation point, followed by the appearance of a blue peripheral target at a random location. Participants clicked on the red dot and moved the mouse in a straight trajectory to reach the blue target. In the baseline and washout phases, cursor motion matched hand movement, while in the perturbation phase, a 60° angular deviation was introduced. The target turned green upon successful reach, and the next trial began within 1 second. Within 1 second of achieving the target, the second trial started with appearance of the target at another location randomly.

Export to PPT

The outcome variables included Mean directional errors (MDE), adaptation, reaction time (RT) and MT, calculated for each trial as shown in Figure 3. MDEs assess the accuracy of movements, where a decrease over time suggests a learning process in motor control. Adaptation percentage, calculated by normalising MDE against the perturbation angle, indicates how well participants adjust to perturbations, with values near 1 showing strong adaptation. As observed, the data were found to be partially normally distributed. Therefore, transforming non-normally distributed variables to approximate normality before applying multivariate analysis of variance (MANOVA).

Close

Figure 3:

Calculation of outcome variables (Reaction time, Movement time, Mean directional errors, Adaptation) during each trial of visuomotor adaptation task.

Export to PPT

General linear model MANOVA was used to compare performance parameters between the perturbation blocks and the three angles of rotation provided during the perturbation blocks for multiple dependent variables, to report statistical significance using the Statistical Package for the Social Sciences (SPSS) version 20.0. The dependent variables included MDEs, adaptation, reaction time and MT that were obtained during the task, while the independent variable included the three blocks and angles of perturbation presented across blocks. Whenever a significant difference between the blocks and angles was found, effect sizes were computed using Cohen’s d by taking the difference in mean scores divided by the pooled standard deviation.

RESULTS

The MDEs, reaction time (RT) and Movement time (MT) changed significantly with repeated trials along a block during the baseline or washout blocks in which the mouse and cursor moved simultaneously (i.e. no angle perturbation) [blue line in Figure 4]. Angle perturbations led to significantly higher MDE, RT and MT as compared to no angle perturbation.

Close

Figure 4:

Graph representing change in outcome variables. (a) Mean directional errors, (b) Reaction time and (c) Movement time for participants who performed visuomotor adaptation task (n = 72) during baseline/washout (no rotation): Blue, and for perturbation blocks (rotation). Asterisks*** represents P < 0.001 between baseline/washout and block 1. Hash### represents P < 0.001, Hash## represents P < 0.01 between baseline/washout and block 2. Asterisks* represents P < 0.05 between baseline/washout and block 3. Asterisks** represents P < 0.01 between baseline/washout and block 1.

Export to PPT

Further comparison of each angle perturbation with another angle was done to decipher if a specific angle perturbation caused more errors/RT/MT, or if adaptation is better for a specific angle of perturbation for each participant. Regardless of block, perturbation at all three angles resulted in significant differences in MDE, adaptation, RT and MT in most comparisons, suggesting that each angle of perturbation caused a change in explicit and implicit learning for each participant [Table 1].

Next, we analysed if the performance changed for each subject across blocks (e.g., 1^st block compared to 2^nd block and so on). With few exceptions (for 30° perturbation), performance improved significantly in block 2 compared to block 1 for the other angles of perturbation, suggesting both explicit and implicit learning in response to perturbations. The performance in blocks 2 and 3 was comparable for MDE and adaptation, though RT and MT varied [Table 2].

To be able to explain the differential effect of 30° angle perturbation, the effect of specific sequence of perturbations was analysed [Table 3]. The significant findings are as follows:

1. In sequences 3, 4, 5 and 6, RT and MT improved consistently

2. When the first perturbation was at 30° angle (in sequences 1 and 2), the RT and MT failed to improve across block

3. Adaptation improved significantly from 1^st to 2^nd block in 4 out of five sequences (Sequences 1, 3, 5 and 6) and in 1^st block compared to 3^rd block in all except one sequence (sequence 4) where it remained unaltered (45°,60°, 30° perturbation).

DISCUSSION

For explicit learning, MDEs were assessed and averaged for the successive five trials. Target error (performance) is the sum of the explicit aiming direction and implicit learning of a forwards model minus the perturbation. This difference was confirmed by binning the aiming directions over the 260 trials of the rotation block into bins of five trials each.^[51] For implicit learning, the percentage change in performance at which participants adapted varied across the three different angles of perturbation. This variation in adaptation rates highlights the inherent variability in how individuals respond to and overcome challenges presented by varying degrees of task modification.

The directional errors and adaptation were examined in the participants as they experienced visuomotor learning based on angle deviation. Our findings indicate that changes in explicit and implicit learning occurred in most participants when perturbations were introduced at three angles, as evidenced by significant differences in MDE, adaptation, RT and MT across all blocks. It is inferred that each angle of perturbation had an impact on learning outcomes for each individual presented with a specific sequence, as shown in Table 1.

These observations suggest that exposure to angle perturbation led to more errors since participants were not familiar with the orientation. Moreover, the rate of adaptation varied for all three angles of perturbation, indicating the encountered variability. Since the goal of the present study was to provide a more comprehensive understanding of motor adaptation, we conducted an analysis to investigate the impact of perturbation on explicit and implicit learning by comparing the first block with the second and subsequent blocks for all three angles allocated [Table 2].

In subsequent blocks, motor learning was found to be associated with reduced directional errors (explicit) and improved adaptation (implicit). Similar findings have been reported by Bond and Taylor^[52] who showed that implicit and explicit learning processes were separable. Learning in our study was implicit, meaning that participants were unaware of the induced perturbation but lacked awareness of the specific perturbation angles.^[53] Both components were assessed, explicit learning to detect initial errors and implicit learning to adapt or integrate information about the difference between expected and actual outcomes.^[54] Adaptation reflected a decline in performance, probably due to saturation in learning after block 2. However, it was evident that participants failed to show improvements for smaller angles such as 30° and 45°, implicating the role of magnitude in achieving precision and fine tuning the movement. These flexible processes may allow them to derive maximum benefit from both types of learning.^[55]

We further compared one perturbation angle with the others to determine whether a specific angle caused more errors/RT/MT or whether an adjustment for specific angle sequence was better for each participant. Depending upon the visual feedback, we hypothesised that directional errors were expected to be proportional to the angle of perturbation, with the least adaptation at 60°. However, the most striking finding was that participants faced difficulty adapting to 30° perturbations, taking more time to perform. This could be an indication that larger angles are easier to detect as compared to smaller ones in a sequence, leading to better adaptation. Similar conclusions were drawn by Morehead et al.^[56] who suggested that the performance may depend on the change in aiming strategy, in addition to the absolute magnitude of angular rotation.

Upon further examination of this pattern across sequences, including reaction time and movement time showed improvement during blocks 1 and 2 for perturbation blocks for most sequences, but there were no significant differences in performance measures related to adaptation in further blocks. Our findings confirm those of earlier studies,^[57,58] where adaptation significantly improved when the sequences were presented in sequentially increasing or decreasing order for subsequent blocks. We could observe partial adaptation till block 2 in which the smaller angles were presented together (30° followed by 45° or vice-versa). Adaptation failed to occur across blocks for only one sequence, where a larger angle was presented between two smaller angles. This partial adaptation could be attributed to failure in proprioreceptive recalibration when the endpoint is unknown.

CONCLUSION

The study examined the impact of perturbation angles presented in a random order on performance parameters. The results revealed that participants had a greater degree of explicit learning for larger angles but struggled to adapt to a 30° angle perturbation, suggesting a lack of improvement in performance with minimum visual feedback. The study also found faster responses in subsequent blocks at the end of the task, regardless of the sequence, indicating that prior exposure to one rotation angle affects the adaptation strategy for subsequent angles presented during repeated exposure. These results suggest that visual representations of hand position may play a more significant role than proprioceptive representations in the process of motor learning. However, there is limited evidence on whether the degree of performance depends on the magnitude of perturbation or the order of presentation in subsequent blocks. Overall, the study highlights the importance of developing a counterintuitive strategy to achieve precision in motor learning.