Impact of prior errors on visuomotor adaptation and savings: experimental considerations and clinical implications

Abstract

The motor system retains learning from visuomotor adaptation tasks in the form of “savings” to enable faster readaptation to similar perturbations in the future. Leow et al. (J Neurophysiol 116: 1603–1614, 2016) suggest that the experience of prior errors during relearning is necessary for savings while repetition of prior actions may not be sufficient. These findings provide novel insight into factors that contribute to visuomotor adaptation and can be applied to future experimental and clinical research.

to efficiently engage in everyday motor movements and successfully adapt to perturbations from our bodies and the environment, the motor system must quickly respond to and compensate for the discrepancies between predicted and observed outcomes. The motor system retains this learning in the form of “savings” to enable faster adaptation to similar perturbations in the future (Landi et al. 2011). One theoretical model that describes the adaptation process is the internal forward model, which suggests that motor control results from the prediction of sensory consequences of motor movements (Miall and Wolpert 1996). These predictions are gradually updated as a result of perturbations from internal and external sensory input (Miall and Wolpert 1996). For example, if someone picks up an empty soda can with the expectation that it will be full, they might pick up the can with greater force than is necessary. The next time they approach the can, they will likely adapt to this error by using less force.

Several studies have suggested that internal forward models may not solely explain successful adaptation and suggest that other theories such as use-dependent plasticity (Huang et al. 2011) impact savings. A debate exists in the literature as to which of these theories contributes to savings. If an error-based model in fact leads to savings, the underlying mechanisms of this process remain unclear. Additional processes, such as explicit strategy use (Morehead et al. 2015) and operant conditioning (Galea et al. 2015), may explain the underlying mechanisms of error reduction and its impact on savings.

Leow and colleagues (2016) compared these competing theories to determine the necessary conditions for savings. The authors hypothesized that the prior experience of errors would result in greater savings than prior repetition of the action. In other words, they asserted that the experience of errors has greater impact on future movements than does the repeated engagement of successful movements (Leow et al. 2016). This assertion seems to be consistent with other aspects of human behavior. For example, a person might be more likely to learn a math problem after getting it wrong than if they simply repeated the correct operations over and over. An error-based theory in motor adaptation suggests that individuals access memories of prior movement errors, leading to faster error reduction and greater savings when exposed to similar movement perturbations in the future. This theory lacks a comprehensive explanation of motor learning, however, because the mechanisms underlying memories of errors have not been identified.

Some authors have argued that because error-based theories rely solely on adaptation they may fail to account for additional learning processes that contribute to successful adaptation. Huang and colleagues (2011) assert that an error-based model alone is insufficient and suggest that other mechanisms, including use-dependent plasticity, may play a role in motor learning and savings. The use-dependent plasticity theory suggests that savings occurs as a result of repeating previous successful actions. This theory contends that repetition of a newly adapted movement induces directional biases toward the repeated movement (Huang et al. 2011). In other words, engaging in the successful movement leads to biases toward the movement that has been repeated.

Experimentally, visuomotor adaptation tasks consist of the following phases: 1) baseline with normal visual feedback of the movements (e.g., on a computer, a participant moves a cursor to a target on the screen); 2) learning trials where the visual feedback of the on-screen movement is distorted (e.g., the on-screen visual feedback is rotated); and 3) postlearning trials where visual feedback of the hand movement is back to normal. Successful learning can be measured by the presence of an aftereffect during postlearning trials (e.g., continuing to make errors without the distorted visual feedback). Savings are evaluated by measuring the reduction in adaptation speed during future exposures to the same visual perturbation.

In the present study conducted by Leow and colleagues (2016), participants were instructed to move a cursor from a starting point to a target circle as accurately and as quickly as possible, in one uncorrected movement. Online visual feedback of the cursor position was provided on the screen. Participants first completed 60 baseline trials, then they completed the test phase that included an adaptation block (A1), a null block (null), a savings block (B), and another adaptation block (A2). A1 was 119 trials and consisted of a 30° clockwise rotation of visual feedback on the movement path.

The authors used a 2×2 design to manipulate the prior error and prior repetition of the movement sequence conditions in A1. The prior-error conditions (Err⁺Rep⁺ and Err⁺Rep⁻) included abruptly imposed visual feedback rotations that evoked counterclockwise errors. The rotation was imposed and removed gradually for the no-prior-error conditions (Err⁻Rep⁺ and Err⁻Rep⁻) so that errors would be avoided. Prior repetition was manipulated in conditions Err⁺Rep⁺ and Err⁻Rep⁺ by rotating the target relative to the movement solution in the same direction and magnitude as the visual feedback rotation, therefore encouraging repetition of the movement sequence. Repetition of the movement solution was limited in conditions Err⁺Rep⁻ and Err⁻Rep⁻ as participants adapted to the rotation. In Err⁻Rep⁻, repetition of the solution was limited as participants adapted incrementally to the rotation. In this condition, the target stayed in the same position throughout A1 as the clockwise rotation was gradually imposed and removed. Repetition was limited in In Err⁺Rep⁻ by rotating the target in the same direction and magnitude as the feedback rotation.

Block A1 was followed by the 60-trial null block in which no rotation was imposed. Savings block B consisted of 60 trials in which a 30° counterclockwise rotation of visual feedback was abruptly imposed to observe the time in which participants would take to adapt to a new perturbation after having adapted to the original perturbation in A1. The final block A2 (60 trials) contained an abruptly imposed 30° clockwise rotation and was used to observe savings as well as anterograde interference from block B. Anterograde interference is a phenomenon in which subsequent learning occurs more slowly due to prior tasks having opposite demands (Leow et al. 2016).

The primary goal of this study was to determine whether the previous experience of errors or repetition of the movement solution was more likely to contribute to savings during a visuomotor task. Overall, the authors’ findings were consistent with their hypothesis that participants experiencing prior errors would display greater savings than those in the repetition condition. Participants in the prior repetition condition of the movement solution failed to display faster error reduction and savings. Similar to earlier research (e.g., Herzfeld et al. 2014), participants in the prior experience of similar errors condition exhibited savings even when the movement solution had not been previously executed. In sum, their findings suggest that prior experience of sensory prediction errors leads to faster reduction in errors and greater savings in comparison to repetition of previous successful actions. This study is unique in that it is the first of its kind to compare two major theories related to the conditions necessary for savings. Based on these findings, the error-based model is more likely to contribute to savings than the use-dependent plasticity model. Unfortunately, the mechanisms contributing to this error-based learning remain unclear.

Leow et al. (2016) concluded that explicit strategy use may have impacted the savings observed in the present study. Other research has indicated that explicit strategy use may result in faster error reduction than relying on implicit adaptation (e.g., error sensitivity, retrieval of previously successful actions) alone (Morehead et al. 2015). The experience of errors may lead participants to use explicit strategies to reduce future errors and increase savings. In other words, the experience and recognition of errors may lead to strategic planning of future actions. A third condition within this study that controlled strategic aiming strategies by employing explicit instructions, similar to the work of Morehead et al. (2015), would have been valuable for more effectively examining error reduction and savings outcomes. While measuring the effects of such a mechanism on savings may have been outside the scope of the present study, future work should aim to obtain measures of mechanisms such as explicit aiming strategies. A potential method for measuring the effects of explicit strategy use would simply be to include an aiming report task to determine whether participants have engaged in an explicit reaiming strategy (Taylor et al. 2014). It is unclear to what extent explicit processes may impact the present findings without such measurement.

Another theory that should be considered in relation to the mechanisms impacting the effects of prior errors on savings is operant conditioning. Specifically, the way in which various types of errors are reinforced within an experimental paradigm may impact savings. Perhaps savings resulting from the experience of prior errors are related to the feedback received when these errors are made. For example, the provision of online visual feedback with cursor location allows participants to be aware of sensory prediction errors and reward prediction errors as the task is being completed. Since reward-based feedback has been associated with greater savings (e.g., Galea et al. 2015), future experience of similar errors may result in avoidance of the consequences associated with errors (i.e., lack of reinforcement or punishment). In other words, participants may be more motivated to reduce errors to avoid the consequences of errors than to receive positive reinforcement associated with successful actions. The application of this hypothesis to the present findings may be consistent with the results of Galea et al. (2015). They found that negative feedback increased the speed of learning while positive feedback led to greater savings (Galea et al. 2015). Based on their findings, one could argue that negative feedback may influence the rate of error reduction in response to the experience of consequences associated with prior errors.

The manipulation of feedback in response to actions may allow for the effects of errors on savings to be more closely examined. One method for this manipulation would be to modify the online feedback provided to participants in one condition and to modify the target success (e.g., reward) feedback in another condition. The recent work of Brudner et al. (2016) aimed to separate two modes of reinforcement by providing variable feedback to participants. Error-based feedback was provided by displaying the terminal position of the cursor and outcome-based (i.e., reward) feedback was provided using points (Brudner et al. 2016). Perhaps a similar manipulation could be used to determine which type of error and subsequent feedback leads to a faster reduction of errors and results in greater savings during adaptation tasks.

The work by Leow et al. (2016) provides new insight into how the experience of prior similar errors might contribute to savings in adaptation tasks. Their finding that the experience of prior errors contributes to savings is important for the development of strategies to improve motor control in a variety of clinical populations. Increasing savings is a key aspect of intervention development for populations experiencing motor deficits such as individuals with autism spectrum disorder (ASD). ASD is a neurodevelopmental disorder characterized by deficits in social communication and by restricted, repetitive patterns of behavior. While most interventions target the social impairments, prevalence of motor deficits in individuals with ASD has been estimated to range from 20 to 100% (Gowen and Hamilton 2013). Inconsistent findings have been documented in the motor learning literature related to ASD regarding whether implicit learning processes are impaired and whether such deficits can be improved using explicit instruction. Some studies suggest that implicit learning may be impaired in individuals with ASD (Gordon and Stark 2007) while others suggest that both implicit and explicit processes aid in adaptation (Izadi-Najafabadi et al. 2015). The findings presented by Leow et al. (2016) may be useful when considering the motor learning deficits in individuals with ASD. In visuomotor adaptation tasks, individuals with ASD often fail to attend to visual feedback and adequately modify their motor movements, therefore they may not display sensitivity to prior errors. If explicit processes enable individuals with ASD to compensate for deficits in implicit learning one might assert that explicit reaiming strategies may be useful for increasing savings this population. It is possible, however, that the experience of prior similar errors and repetition of successful actions may have different effects on savings in this population.

In all, the work by Leow et al. (2016) significantly contributes to the debate regarding the factors that influence savings in visuomotor adaptation tasks. They highlight the importance of the experience of prior errors on relearning and savings. These findings play an important role in the motor adaptation literature as well as in the clinical application of motor learning interventions. Although their findings significantly add to current research in motor adaptation, further work is necessary to elucidate the mechanisms that may impact the way in which prior errors lead to greater savings and faster reduction in errors during relearning. Future work may also provide insight as to how these findings apply to clinical populations such as ASD. This work provides several viable avenues for further exploration in the field of motor adaptation both experimentally and clinically.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author.

AUTHOR CONTRIBUTIONS

M.W. drafted manuscript; M.W. edited and revised manuscript; M.W. approved final version of manuscript.