ADAPTATION TO VISUAL PERTURBATIONS WHILE LEARNING A NOVEL VIRTUAL REACHING TASK By Sachin Devnathan Narayanan A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Kinesiology – Master of Science 2019 ABSTRACT ADAPTATION TO VISUAL PERTURBATIONS WHILE LEARNING A NOVEL VIRTUAL REACHING TASK By Sachin Devnathan Narayanan The movements we do to perform our day-to-day activities have always been riddled with perturbations, to which we adapt and learn. The studies looking at this aspect of motor learning should consider, the biomechanical differences that exist between individuals and create a novel task that can test every individual without any bias. This was achieved in our study by using a virtual environment to perform a novel motor skill in order to investigate how people learn to adapt to perturbations. 13 college age participants (females = 7, Mean = 21.74 ± 2.55) performed upper body movements to control a computer cursor. Visual rotation of the cursor position was introduced as a perturbation for one half of the practice trials. Movement time and normalized path length were calculated. One way repeated measures ANOVA was performed to analyze significance between the performance at different times of the task. Significant learning seen while learning the initial baseline task (p<0.0001) and significant drop in performance upon immediate exposure to the perturbation (p =0.005). No significant adaptation over practice with the perturbation (p = 0.103) or significant after-effects on removal of the perturbation (p = 0.383). Results suggests differences in adaptation when the task is novel, when compared to other adaptation studies and such novel tasks trigger a different type of learning mechanism when compared to adaptation. ACKNOWLEDGEMENTS To my parents: Thank you for being the pillars of support throughout my life, especially during these two years, always telling me that “everything is fine”. To Mei: For the constant guidance and mentorship without which I would have not been able to complete this thesis. To Rajiv and Florian: For all their valuable feedback and effort in helping me shape my thesis proposal and defense, all in such short notice. To Priya: For being the reality check during my two years with the SD Lab and always ready to help me in the time of need. To Tzu, Rak and Aimee: Thank you for being the constant positive support and all the valuable discussions we have had during our time in the lab. To Tim and all the undergrads: Thank you for being such wonderful juniors, helping me in the data collections through your busy schedules. To the people at Smith Center: Thank you for understanding my increasingly complex daily schedule and making my job and the Smith center to be two things that I shall always cherish. To all my friends: Thank you for taking the stress off during the overwhelming times and always making feel like I am doing a great job. iii TABLE OF CONTENTS LIST OF FIGURES .........................................................................................................................vi CHAPTER 1................................................................................................................................. 1 INTRODUCTION & REVIEW OF LITERATURE .......................................................................... 1 Motor adaptation in research ........................................................................................... 1 Visuomotor adaptation ..................................................................................................... 2 Performance in visuomotor adaptation studies ............................................................... 3 Lack of task novelty in existing studies ............................................................................. 4 Studies using novel tasks ................................................................................................... 5 Need for this study ................................................................................................................. 5 Specific aims ...................................................................................................................... 6 Aim 1 ............................................................................................................................ 6 Aim 2 ............................................................................................................................ 6 Proposed hypothesis ......................................................................................................... 6 CHAPTER 2.................................................................................................................................7 METHODOLOGY .....................................................................................................................7 Participants ........................................................................................................................7 The body machine interface ..............................................................................................7 Structure of a body machine interface .............................................................................9 Signal acquisition ..........................................................................................................9 Transformation of body signals into control space ......................................................9 The control interface .................................................................................................. 10 Experimental setup ......................................................................................................... 10 Dimensionality reduction ................................................................................................ 12 The task interface ............................................................................................................ 13 The G.U.I .......................................................................................................................... 14 Experimental flow ........................................................................................................... 14 The experimental protocol .............................................................................................. 17 The visual perturbation blocks.................................................................................... 18 Data analysis .................................................................................................................... 19 Performance metrics .................................................................................................. 19 Statistical analysis............................................................................................................ 20 CHAPTER 3............................................................................................................................... 22 RESULTS ............................................................................................................................... 22 Movement time............................................................................................................... 22 Path length ...................................................................................................................... 24 Immediate after-effects .................................................................................................. 27 CHAPTER 4............................................................................................................................... 29 iv DISCUSSION .......................................................................................................................... 29 Summary ......................................................................................................................... 29 Baseline performance ..................................................................................................... 30 Perturbation block performance ..................................................................................... 30 Washout block performance and after-effects ............................................................... 31 Limitations ....................................................................................................................... 32 Future direction ............................................................................................................... 32 APPENDIX ............................................................................................................................. 34 REFERENCES ......................................................................................................................... 38 v LIST OF FIGURES Figure 1. Schematic diagram of a BoMI (adapted from Casadio et al., 2012) .............................8 Figure 2. Cursor coordinates mapping equation .......................................................................... 10 Figure 3. IMU position and experimental setup ........................................................................... 11 Figure 4. Matlab G.U.I to set up experimental parameters and control overall task protocol .... 14 Figure 5. Simulink layout of the perturbation block ..................................................................... 15 Figure 6. Customization window with the red dot task ................................................................ 16 Figure 7. The task interface (a) Cursor while reaching to a peripheral target. (b) Cursor upon reaching a target and holding there for 500 ms (c) Position of all eight targets with the effect of visual perturbation seen while reaching for the top target ............................................... 18 Figure 8. Overall protocol common for all participants ......................................................... 19 Figure 9. Performance graphs plotted for movement time against training number. The green part of the graph focuses on the baseline blocks, the blue part on the perturbation blocks and the grey part on the washout blocks. (a) Movement times (s) of individual participants plotted with respect to the training block number. (b) Movement time (s) averaged across all participants, represented with respect to the 12 blocks (c) Movement time comparison between males and females averaged across the participants plotted with respect to the training blocks ................................................................................................. 23 Figure 10. Path length trajectories at various stages of the protocol (a) Training 1 (first baseline block) (b) training 5 (last baseline block) (c) Training 6 (First block of perturbation) (d) training 10 (last block of perturbation) (e) (Training 11 (first block of washout). This clearly shows that gradual learning has happened to an extent along the perturbation block but there are not big differences on removal of the perturbation ........................................ 25 Figure 11. Performance graphs are plotted for normalized path length against training number. The green part of the graph focuses on the baseline blocks, the blue part on the perturbation blocks and the grey part on the washout blocks. (a) Normalized path lengths of individual participants plotted with respect to the training block number. (b) Normalized path length averaged across all participants, represented with respect to the 12 blocks (c) Normalized path length comparison between males and females averaged across the participants plotted with respect to the training blocks ........................................................ 26 vi Figure 12. Comparison between the average of reaches for the last cycle of training 10 and the first cycle of training 11. The performance parameters (movement time and normalized path length) have been plotted for each participant, averaged across each cycle ............... 28 Figure 13. One-way repeated measures analysis for movement time averaged for all participants with blocks as the repeated measures factor .................................................... 35 Figure 14. One-way repeated measures analysis for normalized path length averaged for all participants with blocks as the repeated measures factor .................................................... 36 Figure 15. One-way repeated measures analysis of movement time, comparing the last eight reaches of training 10 to the first eight reaches of training 11 .............................................. 37 Figure 16. One-way repeated measures analysis of normalized path length, comparing the last eight reaches of training 10 to the first eight reaches of training 11 .............................. 37 vii CHAPTER 1 INTRODUCTION & REVIEW OF LITERATURE While learning and performing motor skills daily, we see that there exist various disturbances to our movements, to which we constantly try to adjust and adapt with ease. This process of adjusting or modifying our movements to new demands by gradual trial-to-trial basis using the feedback from the errors, in order to reduce the same is called motor adaptation and this gives rise to motor learning (Martin et al., 1996; Bastian, 2008). These demands or disturbances may be external, such as environmental conditions/forces, use of certain devices like prisms, or internal like muscle fatigue or injuries (Krakauer, 2009). Common examples of adaptation are walking on slippery or snowy surfaces, where our walking patterns would have to adjust to the change in the nature of the surface or trying to use computers which have different mouse sensitivities, to name a few. Motor adaptation in research In order to help us understand the effects of adaptation in our real life movements, studies have extensively used tasks based on error-based paradigms, such as visual or mechanical perturbations (Shadmehr et al., 2010; Mazzoni & Krakauer, 2011). Study-specific examples of such paradigms being used include perturbations being introduced to the upper body such as, reaching and pointing movements. Studies with visual perturbations include visuomotor transformations using prism goggles (Helmholtz, 1962; Newport & Schenk, 2012; Redding & Wallace, 1996; Rossetti et al., 1998) or virtual rotation of the hand or finger displacement using 1 a mouse cursor on the computer screen (Wigmore et al., 2002; Scheidt et al., 2011). Alternatively, mechanical perturbations can be used in the form of lateral external forces being brought about using a robot arm manipulandum. In these studies, there are constant modifications to the movements in the task from trial-to-trial, where the movement retains the identity of action, but over practice, changes in terms of other parameters such as movement direction or intensity of force (Martin et al, 1996). Visuomotor adaptation Between the two types of perturbations, a lot of studies have used visuomotor rotations and visual perturbations to study adaptation and the subsequent learning associated with it. There are many reasons and findings that support this. First, many studies using this type of perturbation have provided results that provide important results regarding procedural learning and memory. Second, rotational learning is implicit, which means that the participants tend to adapt to the perturbation even without awareness and without any explicit strategies regarding the perturbation being provided. Visuomotor adaptation studies also have tasks that are predominantly reaching based, which has a significance that the human nervous system plans reaching movements as a vector consisting of separate details for extent and direction (Gordon et al., 1994; Vindras and Viviani, 1998; Ghez et al., 2000). Reaching tasks primarily involve arm movements and visuomotor rotation works by introducing a direction based change in the reach target position, which in turn is related to the arm movement. Therefore, this can be used to study the adaptation in important movements such as reaching (Krakauer, 2009). 2 In such adaptation studies using visual perturbations, we see certain common trends in the structure of the experimental protocol used. There is a primary, task familiarization phase, where the participant learns to perform a certain movement for the task in hand. This phase usually consists of a certain number of trails, after which the participant is exposed to the perturbation of the task. Then there is a series of trials where the participant performs the task with the perturbation influencing the movements and it is over this period of performance where people study the effects and rate of adaptation taking place. After a considerable period of practice under the perturbation, it is taken off, and the task becomes similar to the initial familiarization task. This “washout” block is to see how the effects of the adaptation get carried over to the baseline movements. Performance in visuomotor adaptation studies Looking at the performance in such adaptation studies, we see that initially in the familiarization block, the participants perform well with minimum errors in the task parameters. This was because most of the tasks involve movements that participants already have some idea about. Then, upon introduction of the perturbation, the participant has a significant drop in performance and a rise in the errors, as they are not aware of the perturbation and do not yet know how to develop counter-measures to adapt. With practice over several trials, the participants gradually get better in performing the task with the perturbation, showing the presence of adaptation. This is seen in many studies by the trial-by-trial reduction in the errors made during the perturbed phase (Buch et al., 2003, Shadmehr et al., 2011). On removing the perturbation in the washout block, it is seen that there are significant after-effects, which are 3 seen in the form of a sudden increase in the occurrence of errors. The presence of after-effects indicates that the participant does not merely react to the perturbation that has been introduced but also gradually adapts and anticipates the expected dynamics of the visually rotated environment. This therefore shows that motor adaptation appears to work by updating the internal model of an existing baseline environment, when changes are made (Huang and Krakauer, 2009). Lack of task novelty in existing studies However, in the tasks performed in most of the visuomotor adaptation studies done earlier, there is always the possibility of some participants having prior knowledge about the task. For example, if it is a dart-throwing task, some participants who have been prior experience throwing darts, or if they are trained in dart-throwing before. This would make these participants use strategies that they already know to result in better performance that all the other participants who have not experienced dart-throwing before. Another potential issue is that, there might exist certain biomechanical body differences between various participants, which might cause the them to get an edge over others. This for example can be understood in reaching tasks, where people with longer arms might be able to reach further when compared to people with shorter arms. In short, most of the previous studies involve tasks which try to look at adaptation through re-parameterization of an already existing, well-learn movement coordination pattern (Lee et al., 2018). These two aspects of prior task experience and the biomechanical differences play a crucial role in every task because these might prevent studies from looking at true learning from scratch. Therefore, tasks that are novel enough in the 4 movements executed by the participants, so that they have no prior experience advantage, nor do they have any edge in the performance due to biomechanical differences, should be used. Studies using novel tasks To overcome these issues, there have been newer few studies that have taken task novelty into consideration and have structured a task which involves the learning of a novel movement coordination patterns such as using a split-belt treadmill to perform walking based adaptation tasks (Reisman et al., 2007) or the use of a data glove (Liu et al., 2011). But even in these studies, for example in Liu et al., 2011, the movements performed by the data glove were not completely novel and exploratory. The participants were made to wear a data glove and control a cursor on the screen, but when calibrating the movements, they were not given complete freedom to explore and define their movements, but rather made to perform repetitions of a random sequence of 24 hand postures corresponding to the static finger spelling characters from the American Manual Alphabet (AMA). Further, subjects were provided with photographic images of an expert in AMA performing these postures. This therefore restricted participants from completely exploring and defining their own set of movements for calibration. Need for this study Therefore, it was critical for us to investigate the effects of visual perturbations when learning a novel, exploratory motor task to understand the true effects of adaptation and give us a better understanding in real life. To achieve these task goals of novelty and lack of biomechanical differences, we used a Body Machine Interface (BoMI). There are various kinds of 5 Body Machine Interfaces which basically help to connect body movements and transform them to perform new functional tasks such as controlling a computer cursor, a prosthetic arm or even a wheelchair. A body-machine interface does not take account of previous biases: subjects perform novel virtual tasks that are independent from their prior experience and their biomechanical characteristics (Casadio et al., 2012). We used one such interface, which uses upper body movements to control a cursor on a computer screen (Casadio, Ranganathan, & Mussa-Ivaldi, 2012). With this, we focused on answering our following research questions. Specific aims Using a novel and exploratory virtual reaching task, we investigated Aim 1: Adaptation on introduction to visual perturbation with practice. Aim 2: After-effects of the adaptation to perturbation on its removal. Proposed hypothesis For Aim 1, we expected to see an initial drop in the performance upon introduction of the visual perturbation, but also expect to see a gradual improvement in the performance over practice over a period of reaching trials. For our second aim, we expected to see significant after- effects when the visual perturbation is removed, which would see a sudden drop in the performance similar to the drop seen on the onset of perturbation (Helmholtz, 1867; Liu and Scheidt, 2011; Pierella et al., 2015). 6 CHAPTER 2 METHODOLOGY Participants In order to answer our research questions, we had 13 healthy college-age adult participants (Mean = 21.74, SD = 2.25) with no recent history of upper body disabilities, out of which 7 were females. All the participants were undergraduate and graduate student from the Department of Kinesiology in Michigan State University (East Lansing, Michigan) and received an extra credit in one of their courses to perform the experiment as compensation. The procedures were approved by Michigan State University Human Research Protection Program (MSU HRPP) and all the participants were given consent forms to sign and a copy of the signed consent form was given to them. In this study, all the participants performed the same task with the same experimental protocol and were all exposed to the visual perturbation when they were performing a novel and exploratory virtual reaching task using a Body Machine Interface (BoMI). The body machine interface The Body Machine Interfaces are a means for the human body to interact with an external device. These interfaces play an essential role in assisting the people with reduced motor skills on a daily basis. The main aspect of this Body Machine Interface that makes it significant in our study is that it does not take account of previous biases: subjects perform novel virtual tasks that are independent from their prior experience and their biomechanical characteristics (Casadio et al., 2012). General scheme for a body machine interface is shown in the following figure: 7 Human Body Neural Firing EEG Muscle activity EMG Movements (limbs, head, tongue, gaze...) Forces - Pressure Body Machine Interface (BoMI) Device to be controlled Figure 1. Schematic Diagram of a BoMI (adapted from Casadio et al., 2012) A typical Body Machine Interface consists of three main components. The body being the first, refers to the human body from which signals are obtained. The second component is the machine, which refers to the device or machine that is to be controlled. The final component is the one that links these two, which is the interface. The interface plays an important role in the functioning of the BoMI by receiving and transforming the body signals into commands for controlling the device. The general aim of a body-machine interface is to enable the user to retain a complete or shared control over the device (e.g a prostethic limb, a wheelchair or the cursor position on a computer monitor) through signals derived from the user’s body. These signals may be extracted directly from body motions, using goniometers, magnetic or infrared sensors, accelerometers, cameras, force sensors and pressure switches. 8 Structure of a body machine interface Signal acquisition The initial part of a BoMI involves acquiring the signals from the body, which is done using various measuring devices and, in our case, we use four wireless Inertial Measurement Units (IMUs). These IMU sensors are placed on the shoulders of the participants (two on each shoulder) to record the movement. Transformation of body signals into control space The acquired body signals are mapped onto the control space, i.e. the space defined by the commands of the external device. This mapping process is typically obtained by executing a dimensionality reduction which allows mapping the n-dimensional body signal into the m- dimensional control signal (with m1) or contract (gain <1) the workspace, whereas the offset and rotation values will allow respectively the translation and rotation of the workspace. For our study purposes, we use a Gain of +1 and an Offset value of 0. Both parameters are selected in accordance to participant’s preferences, in order to let them move the cursor in the entire workspace more intuitively. After selecting the “Save customization parameters” button, the user must select a group (Group 1, Group 2). Each group is characterized by a different protocol and the selection must be done according to the protocol that the participant must follow. The experimental protocol The participants were told to move their upper body (shoulders and torso) to control a cursor on the monitor and perform a center-out reaching task, designed in the custom made Matlab/Simulink program. The overall goal of the participant was to move the cursor into a specific target circle as fast and as close to the center as possible. On reaching the target circle, they would have to stay in it for at least 500ms. All participants performed 12 training blocks each consisting of 24 trials where the target circles appeared in random order in both the cardinal and diagonal directions at 11.5 cm from the center of the screen. They went back and forth between these peripheral targets and a central base target. In these blocks, the participant received a score that reflected their individual performance for each trial. Among the 12 blocks, the first five blocks were without any kind of visual perturbation, but rather a baseline or familiarization block. 17 a b c Figure 7. The task interface (a) Cursor while reaching to a peripheral target. (b) Cursor upon reaching a target and holding there for 500 ms (c) Position of all eight targets with the effect of visual perturbation seen while reaching for the top target The visual perturbation blocks From the sixth block, the visual perturbation was introduced. The perturbation was in the form of a constant clockwise rotation of 45⁰ of the cursor position (Figure 6.c). Previous literature on cursor rotation tasks show that the response time. error reduction and adaptation rate are 18 maximum at 45⁰ (Krakauer, 2009; Newell, 2012). The cursor rotation was brought about by applying a rotation matrix on the final 2-D task space vector. The participants were not given any information about the perturbation and were completely oblivious to its onset. Such a perturbation was applied to the next five blocks (training blocks 6-10). At the end of these five perturbed training blocks, the participants performed two more blocks of the original unperturbed condition, which was a washout block to investigate the presence of after-effects of the perturbation. The overall layout of the study is shown in Figure 5. Baseline Block 120 Trials Perturbation Block 120 Trials Washout Block 48 Trials Figure 8: Overall protocol common for all the participants Data analysis The data that we received was in the form of .mat files, with sensor information and cursor position and time information in the files. Based on this, we will be looking at two different parameters to evaluate the participant’s performance. Performance metrics The two different performance metrics we will be using are movement time and normalized path length, where movement time was our primary variable because all the targets 19 were equidistant from one center target and we also looked at path length to check for the straightness in the path. We used such these measures because previous literature has shown that there is a gradual straightening of the path in reaching tasks over practice (Shadmehr and Mussa-Ivaldi, 1994; Mosier et al., 2005). Movement time is defined as the time at which the cursor leaves the center target to the time at which the cursor reaches the destination target and stays inside the target for the next 500 ms. Normalized path length between two targets is defined as the ratio of the actual distance travelled by the cursor from the center target to the destination target, to the straight line distance between the two targets. We use normalized path length as our secondary measure to analyze how the cursor has travelled, especially when there have been high movement times. If the path length has also been correspondingly large, it means that the participant did not have a very good control of the cursor during that trial. If the path length was comparatively smaller, it would mean that the participant moved slowly, but still tried to maintain a straight line. This therefore gives us a better idea as to how participants reach and learn the task. Statistical analysis In order to answer our research questions, we analyzed the performances of all the twelve training blocks and investigated the effects of learning and adaptation across these. First, we wanted to confirm if learning of the baseline task had taken place, therefore we compared the performance between training 1 and training 5, i.e. the first and last training blocks of the baseline version of the task. Next, in order to see if the introduction of the perturbation 20 has affected the participant’s performance, we look at the performance between training 5 and training 6, i.e. the last baseline training block and the first block of perturbation. Then, to study the rate of adaptation, we would compare training 6 and 10, which are the first and last blocks of perturbation. This would be followed by the washout blocks. Here, we would compare training 10 and 11 to check for the presence of after-effects upon the removal of the perturbation. Also, in order to check for any immediate after-effects upon the removal of the perturbation, we closely investigated the last cycle of target reaches of training 10 (last perturbation block) with the first cycle of target reaches of training 11 (first washout block). For all these comparisons, we ran individual one-way repeated measures ANOVA with the training block (baseline, perturbation and washout) being our independent variable and taking the performance measures (movement time and normalized path length) as our dependent variables. Since all participants performed both the baseline and the perturbation blocks, there were no separate groups or group-effects. Post-hoc Tukey’s tests were done to analyze the significance of the comparisons. The significance levels were set at p < .05. All the statistical analyses were performed using Jamovi version 0.9.6.9. 21 CHAPTER 3 RESULTS Movement time Baseline (Training 1 to training 5) : As we see from Figure 8, there was significant improvement in the performance initially between training blocks 1 and 5 (df = 48.0, t = 5.943, Ptukey <.001), which shows that considerable learning of the baseline task has happened. This was supported by a significant drop in the movement time. Perturbation (Training 6 to training 10) : Upon introduction of perturbation, we saw that there is a significant drop in performance, which was shown by the sudden increase in movement time between trainings 5 and 6 (df = 48.0, t = -3.699, Ptukey = 0.005). Over the period of five training blocks under the visual perturbation, we noticed that there was a gradual reduction in the movement time, although not statistically significant (df = 48.0, t = 2.518, Ptukey = 0.103). Washout (Training 11 and training 12) : Upon removal of the perturbation, at the end of training 10, we saw that there was no significant change in the movement time curve, which in fact reduces a little bit. We saw that the performance from training 10 to 11 (last block of perturbation and first block of washout) was not significant (df = 48.0, t = 1.804, Ptukey = 0.383). We also looked to see if there were any gender-related differences with movement time and found that there were no significant differences at any point through the 12 training blocks (df = 11.0, t = 0.320, Ptukey = 0.755). 22 Figure 9. Performance graphs plotted for movement time against training number. The green part of the graph focuses on the baseline blocks, the blue part on the perturbation blocks and the grey part on the washout blocks. (a) Movement times (s) of individual participants plotted with respect to the training block number. (b) Movement time (s) averaged across all participants, represented with respect to the 12 blocks (c) Movement time comparison between males and females averaged across the participants plotted with respect to the training blocks 23 Path length Baseline (Training 1 to training 5) : From Figure 9, we saw that path length followed a trend that was similar to movement time. There was significant improvement in the performance initially between training blocks 1 and 5 (df = 48.0, t = 4.79269, Ptukey <.001), which shows that considerable learning of the baseline task had happened. This was supported by a significant drop in the normalized path length. Perturbation (Training 6 to training 10) : On introducing of perturbation, we saw that there was not such a significant increase in the normalized movement time, which indicated that there was no significant drop in performance between training blocks 5 and 6 (df = 48.0, t = -1.64893, Ptukey = 0.475). Over the period of five training blocks under the visual perturbation, we saw that there was a gradual reduction in the normalized path length almost to near baseline performance, which was not statistically significant (df = 48.0, t = 1.8334, Ptukey = 0.367). Washout (Training 11 and training 12) : When the perturbation is taken off at the end of training 10, we saw that there was almost no change in the normalized path length curve, which stayed almost exactly at the same level. We saw that the performance from training 10 to 11 (last block of perturbation and first block of washout) was not significant (df = 48.0, t = -0.19362, Ptukey = 1.000). Even for normalized path length, we did not see any significant gender-related differences at any point through the 12 training blocks, although we saw that the females had slightly lower normalized path length than the males (df = 11.0, t = -2.00, Ptukey = 0.070). 24 a c b d e Figure 10. Path length trajectories at various stages of the protocol (a) Training 1 (first baseline block) (b) Training 5 (last baseline block) (c) Training 6 (First block of perturbation) (d) Training 10 (last block of perturbation) (e) Training 11 (first block of washout). This clearly shows that gradual learning has happened to an extent along the perturbation block but there are no big differences on removal of the perturbation. 25 Figure 11. Performance graphs are plotted for normalized path length against training number. The green part of the graph focuses on the baseline blocks, the blue part on the perturbation blocks and the grey part on the washout blocks. (a) Normalized path lengths of individual participants plotted with respect to the training block number. (b) Normalized path length averaged across all participants, represented with respect to the 12 blocks (c) Normalized path length comparison between males and females averaged across the participants plotted with respect to the training blocks 26 Immediate after-effects Although we saw that there was no significant results for both movement time and normalized path length between training blocks 10 (last perturbation block) and training block 11 (first block of washout), which showed the absence of after-effects on removal of the perturbation, we wanted to see if there were any immediate after-effects on the first few trials of the washout block. For this, we compared the performance of participants in the last 8 of the 24 trials of the perturbation block (training 10) with the first 8 trials of the washout block (training 11). Eight reaches were chosen because they corresponded to the one cycle of reaches to all the eight targets. The results showed that the last cycle of target reaches for training 10 (perturbation block) were significantly higher than the first cycle of target reaches for training 11 (washout block) for movement time (df = 7.0, t = 4.76, Ptukey = 0.002). For normalized path length, although the reaches in the washout block were more than the ones in the perturbation block for certain targets, they were not statistically significant for us to make a defining conclusion (df = 7.0, t = - 0.287, Ptukey = 0.435). 27 Normalized Path Length Movement time (s) last cycle of 10 first cycle of 11 last cycle of 10 first cycle of 11 10 8 6 4 2 0 ) s ( e m i t t n e m e v o M 1 2 3 4 5 6 7 8 9 10 11 12 13 1 2 3 4 5 6 7 8 9 10 11 12 13 Participants Participants 3 2.5 2 1.5 1 0.5 0 h t g n e l h t a p d e z i l a m r o N Figure 12. Comparison between the average of reaches for the last cycle of training 10 and the first cycle of training 11. The performance parameters (movement time and normalized path length) have been plotted for each participant, averaged across each cycle. 28 CHAPTER 4 DISCUSSION Summary Learning a motor pattern is one of the most primitive and important processes in human development. This process of motor learning also leads to adjustment of movements, when encountered by forces or changes in the environment, which results in motor adaptation. During adaptation, eventually new motor patterns are learnt, and adaptation always tends to work in the direction of bringing back the performance to near baseline (unperturbed) conditions (Izawa et al., 2008). This is done through a feedback process using the errors we commit in the previous trials, which is used to correct movements in future trials (Wei and Kording, 2008). We initially aimed to investigate motor adaptation in a novel, exploratory virtual reaching task using a visual perturbation (visuomotor rotation). We identified that the lack of task novelty would cause some participants to have undue advantage due to prior task experience or biomechanical differences giving Previous literature has shown that the performance worsens on the onset of the perturbation but over practice, it does get better, which suggests that motor adaptation has taken place. Also, it was seen that once the perturbation is removed and the participant performs the baseline task, the performance goes bad once more, which suggests the presence of after-effects in the washout trials. This is also what we expected to see in our study. 29 Baseline performance We initially saw that there was significant learning of the baseline task, which corresponded well with the results of previous studies done with the same baseline task (Lee et al., 2018). The performance curves were similar for both movement time and normalized path length. This suggested that our initial number of blocks given for practice was enough for the participants to learn the task, although it was novel. Perturbation block performance Our first aim was to see how participants would react to the perturbation and how the corresponding motor adaptation to the visuomotor rotation of the cursor position would be. We saw that the performance on both parameters, movement time and normalized path length, reduced considerably upon introduction of the visual perturbation. This suggested that all the participants had a significant reaction to the perturbation on its exposure, although none of them were given any feedback or knowledge about the perturbation or it’s onset. Over practice, where the number of trials were the same as that of the baseline block (5 training blocks, 24 trials each), we saw that there was a significant improvement in performance which was seen with the reduction in movement time. The normalized path length also gradually reduced although not statistically significant. This suggested that although the task was novel, adults could adapt their movements to visual perturbations after learning a novel task from scratch. 30 Washout block performance and after-effects On removing the perturbation, we expected to see considerable after-effects in the movements. The presence of after-effects has been found in most of the previous motor adaptation studies, even the few that have been done in novel tasks (Liu and Scheidt, 2011). It demonstrates that the participant doe not just react to the changes in the environment or the perturbation to movement, but rather also works according to a prediction-based model where he/she tries to anticipate the dynamics of the changed environment and perform movements accordingly. Therefore, there is an update that happens to the existing internal model of the external environment. In this study, we expected to see significant after-effects, but we found that participants exhibited no after-effects at all when the perturbation was removed. The performance with respect to both movement time and normalized path length remained almost the same after the perturbation was removed like it was before removal. We looked closely to check for any immediate after-effects in the few reaches, by comparing the last cycle of reaches of the final perturbation block with the first cycle of reaches in the first washout block. Even in that case, we found that there was not drop in performance which could indicate any after-effects. We saw that the movement time for the reaches of the last perturbation block was higher than the movement time for the washout block. This shows the lack of after-effects. Also, the performance at the beginning of the washout block was almost at the same level as that of the last block of baseline (both unperturbed conditions). 31 The lack of after-effects can be attributed to two possible reasons. One, the practice that occurred during the perturbation phase might have helped strengthen the original motor plan of the baseline block. In that case, it is not just adaptation to the visual perturbation but also reinforcement of the existing map. The second possibility could be that the novelty of the task might have made it too difficult for the participant to adapt to the perturbation and this might have caused the them to learn two separate motor plans one for the unperturbed condition and one for the baseline condition. Limitations A major limitation of our study was the small sample size. The statistical insignificance of the results can be attributed to the low power of the study, which in turn is related to the sample size. For now, although we see significance for adaptation over the period of visual perturbation, the results are not so conclusive for the after-effects phase, which could be understood better if we had a bigger population. Future direction Few suggestions for the future would include trying to look at similar adaptation in different populations, as previous literature has suggested that there are significant age-related differences when the task performed is novel. For children, we expect the populations that are slightly older (around 12 years) to not show any after-effects and perform similar to the adults, whereas expect the younger populations of children to have lower performances with significant 32 after-effects. That being said, this study can be potentially used as an effective tool to investigate the effects of aging on the ability to adapt to visuomotor perturbations an also in understanding motor adaptation, motor planning and execution. Another suggestion would be to look at the data from a different perspective. Although in this study we have considered movement time to be the primary variable to measure learning, literature has looked at other parameters such as error in path deviation and directional error (Kagerer et al., 1997, Buch et al., 2003). These are measures of the angle and distance and with our perturbation being a visuomotor rotation, this might give us a better understanding of how it affects movement. This is because, a participant can reach to a target very fast, but can have randomness or deviation in their movement. It is this angular deviation that might give us a better reading about how the visual rotation has affected the movement. 33 APPENDIX 34 Data Analysis Tables Movement Time Figure 13. One-way repeated measures analysis for movement time averaged for all participants with blocks as the repeated measures factor. 35 Normalized path length Figure 14. One-way repeated measures analysis for normalized path length averaged for all participants with blocks as the repeated measures factor 36 Immediate after-effects Movement time Figure 15. One-way repeated measures analysis of movement time, comparing the last eight reaches of the training 10 to the first eight reaches of training 11. Normalized path length Figure 16. One-way repeated measures analysis of normalized path length, comparing the last eight reaches of the training 10 to the first eight reaches of training 11. 37 REFERENCES 38 REFERENCES Bastian, A. J. (2008). Understanding sensorimotor adaptation and learning for rehabilitation. Current opinion in neurology, 21(6), 628. Bo, J., Contreras-Vidal, J. L., Kagerer, F. A., & Clark, J. E. (2006). Effects of increased complexity of visuo-motor transformations on children’s arm movements. Human movement science, 25(4-5), 553-567. Buch, E. R., Young, S., & Contreras-Vidal, J. L. (2003). Visuomotor adaptation in normal aging. Learning & memory, 10(1), 55-63.] Casadio, M., Ranganathan, R., & Mussa-Ivaldi, F. A. (2012). The body-machine interface: a new perspective on an old theme. Journal of Motor behavior, 44(6), 419-433. Cunningham, H. A. (1989). Aiming error under transformed spatial mappings suggests a structure for visual-motor maps. Journal of experimental psychology: Human perception and performance, 15(3), 493. Daffertshofer, A., Lamoth, C. J., Meijer, O. G., & Beek, P. J. (2004). PCA in studying coordination and variability: a tutorial. Clinical biomechanics, 19(4), 415-428. Farshchiansadegh, A., Abdollahi, F., Chen, D., Lee, M. H., Pedersen, J., Pierella, C., Roth, E.J., Gonzalez, I.S., Thorp, E.B. and Mussa-Ivaldi, F. A. (2014, August). A body machine interface based on inertial sensors. In Engineering in Medicine and Biology Society (EMBC), 2014 36th Annual International Conference of the IEEE (pp. 6120-6124). IEEE. Ferrel-Chapus, C., Hay, L., Olivier, I., Bard, C., & Fleury, M. (2002). Visuomanual coordination in childhood: adaptation to visual distortion. Experimental Brain Research, 144(4), 506- 517. Hansen, S., Elliott, D., & Tremblay, L. (2007). Online control of discrete action following visual perturbation. Perception, 36(2), 268-287. Hayashi, T., Yokoi, A., Hirashima, M., & Nozaki, D. (2016). Visuomotor map determines how visually guided reaching movements are corrected within and across trials. eNeuro, 3(3). Helmholtz, H. von. (1962). Helmholtz's treatise on physiological optics (Vol. 1). Dover Publications. 39 Hinder MR, et al. The interference effects of non-rotated versus counter- rotated trials in visuomotor adaptation. Exp Brain Res 2007;80(4):629–40 Izawa, J., Rane, T., Donchin, O., & Shadmehr, R. (2008). Motor adaptation as a process of reoptimization. Journal of Neuroscience, 28(11), 2883-2891. Kagerer, F. A., Contreras-Vidal, J. L., & Stelmach, G. E. (1997). Adaptation to gradual as compared with sudden visuo-motor distortions. Experimental brain research, 115(3), 557-561. Krakauer, J. W. (2009). Motor learning and consolidation: the case of visuomotor rotation. In Progress in motor control (pp. 405-421). Springer, Boston, MA. Krakauer, J. W., & Mazzoni, P. (2011). Human sensorimotor learning: adaptation, skill, and beyond. Current opinion in neurobiology, 21(4), 636-644. Krakauer, J. W., Pine, Z. M., Ghilardi, M. F., & Ghez, C. (2000). Learning of visuomotor transformations for vectorial planning of reaching trajectories. Journal of Neuroscience, 20(23), 8916-8924. Lee, M. H., Farshchiansadegh, A., & Ranganathan, R. (2018). Children show limited movement repertoire when learning a novel motor skill. Developmental science, 21(4), e12614. Lee, M. H., Ranganathan, R., Kagerer, F. A., & Mukherjee, R. (2016). Body-machine interface for control of a screen cursor for a child with congenital absence of upper and lower limbs: a case report. Journal of neuroengineering and rehabilitation, 13(1), 34. Liu, X., & Scheidt, R. A. (2008). Contributions of online visual feedback to the learning and generalization of novel finger coordination patterns. Journal of Neurophysiology, 99(5), 2546-2557. Liu, X., Mosier, K. M., Mussa-Ivaldi, F. A., Casadio, M., & Scheidt, R. A. (2010). Reorganization of finger coordination patterns during adaptation to rotation and scaling of a newly learned sensorimotor transformation. Journal of neurophysiology, 105(1), 454-473. Martin TA, Keating JG, Goodkin HP, et al. Throwing while looking through prisms. II: Specificity and storage of multiple gaze-throw calibrations. Brain 1996;119:1199– 1211. [PubMed: 8813283] Mazzoni, P., & Krakauer, J. W. (2006). An implicit plan overrides an explicit strategy during visuomotor adaptation. Journal of neuroscience, 26(14), 3642-3645. Miall, R. C., Jenkinson, N., & Kulkarni, K. (2004). Adaptation to rotated visual feedback: a re- examination of motor interference. Experimental brain research, 154(2), 201-210. 40 Mussa-Ivaldi, F. A., Casadio, M., & Ranganathan, R. (2013). The body–machine interface: a pathway for rehabilitation and assistance in people with movement disorders. Expert review of medical devices, 10(2), 145-147. Mussa-Ivaldi, F. A., & Danziger, Z. (2009). The remapping of space in motor learning and human–machine interfaces. Journal of Physiology-Paris, 103(3-5), 263-275. Newport, R., & Schenk, T. (2012). Prisms and neglect: what have we learned? Neuropsychologia, 50(6), 1080-1091 Pierella, C., Abdollahi, F., Farshchiansadegh, A., Pedersen, J., Thorp, E. B., Mussa-Ivaldi, F. A., & Casadio, M. (2015). Remapping residual coordination for controlling assistive devices and recovering motor functions. Neuropsychologia, 79, 364-376. Redding, G. M., & Wallace, B. (1996). Adaptive spatial alignment and strategic perceptual- motor control. Journal of Experimental Psychology: Human Perception and Performance, 22(2), 379. Reisman, D. S., Wityk, R., Silver, K., & Bastian, A. J. (2007). Locomotor adaptation on a split belt treadmill can improve walking symmetry post-stroke. Brain, 130(7), 1861-1872. Rosetti Y, Rode G, Pisella L, Farne A, Li L, Boisson D, Perenin MT. Prism adaptation to a rightward optical deviation rehabilitates left hemispatial neglect. Nature 395: 166– 169, 1998 Rustighi, E., Dohnal, F., & Mace, B. R. (2010). Influence of disturbances on the control of PC- mouse, goal-directed arm movements. Medical engineering & physics, 32(9), 974- 984. Schmidt, R. A., & Lee, T. D. (2011). Motor control and learning: a behavioral emphasis 5th ed- Champaign, IL: Human Kinetics. Shadmehr, R., Smith, M. A., & Krakauer, J. W. (2010). Error correction, sensory prediction, and adaptation in motor control. Annual review of neuroscience, 33, 89-108. Tseng, Y. W., Diedrichsen, J., Krakauer, J. W., Shadmehr, R., & Bastian, A. J. (2007). Sensory prediction errors drive cerebellum-dependent adaptation of reaching. Journal of neurophysiology, 98(1), 54-62. Ugrinowitsch, H., Santos-Naves, S. P., Carbinatto, M. V., Benda, R. N., & Tani, G. (2011). Motor skill adaptation depends on the level of learning. International Journal of Human and Social Sciences, 6(3), 177-181. 41 Vaswani, P. A., Shmuelof, L., Haith, A. M., Delnicki, R. J., Huang, V. S., Mazzoni, P., ... & Krakauer, J. W. (2015). Persistent residual errors in motor adaptation tasks: reversion to baseline and exploratory escape. Journal of Neuroscience, 35(17), 6969-6977. Wang, J., & Sainburg, R. L. (2003). Mechanisms underlying interlimb transfer of visuomotor rotations. Experimental Brain Research, 149(4), 520-526. Wei, K., & Kording, K. (2009). Relevance of error: what drives motor adaptation? Journal of neurophysiology, 101(2), 655-664. White, O., & Diedrichsen, J. (2010). Responsibility assignment in redundant systems. Current Biology, 20(14), 1290-1295. Wigmore V, Tong C, Flanagan J. Visuomotor rotations of varying size and direction compete for a single internal model in motor working memory J Exp Psychol Hum Percept Perform 28: 447–457, 2002 Wigmore, V., Tong, C., & Flanagan, J. R. (2002). Visuomotor rotations of varying size and direction compete for a single internal model in a motor working memory. Journal of Experimental Psychology: Human Perception and Performance, 28(2), 447. Wolpert, D. M., & Flanagan, J. R. (2010). Motor learning. Current biology, 20(11), R467-R472. Wolpert, D. M., Diedrichsen, J., & Flanagan, J. R. (2011). Principles of sensorimotor learning. Nature Reviews Neuroscience, 12(12). 42