Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Does interlimb transfer of locomotor adaptations depend on limb dominance? Crombeen, Matthew 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_fall_crombeen_matthew.pdf [ 1.21MB ]
Metadata
JSON: 24-1.0074231.json
JSON-LD: 24-1.0074231-ld.json
RDF/XML (Pretty): 24-1.0074231-rdf.xml
RDF/JSON: 24-1.0074231-rdf.json
Turtle: 24-1.0074231-turtle.txt
N-Triples: 24-1.0074231-rdf-ntriples.txt
Original Record: 24-1.0074231-source.json
Full Text
24-1.0074231-fulltext.txt
Citation
24-1.0074231.ris

Full Text

  Does Interlimb Transfer of Locomotor Adaptations Depend on Limb Dominance?  by  Matthew Crombeen  B.Sc., Wilfrid Laurier University, 2011  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in The Faculty of Graduate and Postdoctoral Studies (Kinesiology)  THE UNIVERSITY OF BRITISH COLUMBIA (VANCOUVER)  August 2013  ? Matthew Crombeen, 2013  ii  Abstract  Generalization of these adaptations have been found to occur across task, workspace and between limbs. Interlimb adaptation transfer appears to depend on limb dominance. Transfer of adaptation from the non-dominant to the dominant limb involves faster rate of adaptation in movement trajectory patterns, while transfer from the dominant limb to the non-dominant limb involves a faster rate of adaptation positioning related parameters of movement. Although such observations are robust for upper limb adaptations, the extent of interlimb transfer during locomotor tasks is still unclear. Studies so far suggest that there is weak interlimb transfer of locomotor adaptation, but none have examined whether interlimb transfer during locomotor tasks depends on limb dominance. The objective of this study was to determine whether locomotor adaptations to a velocity-dependent resistance transfers asymmetrically depending on dominance associated with the legs. It was expected that transfer of adaptation will occur according to dominance, with the dominant limb showing faster adaptation in terms of foot trajectory following non-dominant limb learning; and the non-dominant limb showing faster adaptation in terms of heel strike position following dominant limb learning. Twenty able-bodied adults who were right hand and right leg dominant walked unipedally in the Lokomat robotic gait orthosis, which applied a velocity-dependent resistance against leg movements. The resistance was scaled to 10% of the individual?s maximum voluntary contraction of the hip and knee flexors. Subjects performed a heel targeting task that was scaled to their individual step length. Subjects were then randomly assigned to either the R?L training group, testing transfer to the non-dominant limb, or to the L?R training group, testing transfer to the dominant limb. Muscle activity (surface electromyography) and joint kinematics were recorded from the lower limbs. The adaptation rate in the initial foot trajectory slope and end point error were compared between the groups and iii  across trials using a 2 by 3 repeated measures ANOVA. There was no difference between the groups for either initial foot trajectory slope (p = 0.106) or end point error (p = 0.763). There was also no evidence for transfer of motor adaptations between the lower limbs in the other gait variables. These results suggest that interlimb transfer of locomotor adaptations is limited, but further studies are warranted to understand the neuromechanical mechanisms controlling locomotor adaptations.                 iv  Preface  This project was approved by the UBC Behavioural Research Ethics Board (certificate #H08-01300)                    v  Table of contents Abstract ......................................................................................................................................................... ii Preface ......................................................................................................................................................... iv Table of contents ........................................................................................................................................... v List of tables ................................................................................................................................................ vii List of figures ............................................................................................................................................. viii Acknowledgements ...................................................................................................................................... ix Introduction ................................................................................................................................................... 1 Overview ................................................................................................................................................... 1 Literature Review .......................................................................................................................................... 2 Adaptability of walking ............................................................................................................................ 2 Adaptation to sustained altered sensory information ................................................................................ 6 Cerebellar involvement in adaptation ..................................................................................................... 10 Generalization of motor adaptations ....................................................................................................... 12 Could transfer be dependent on limb dominance? .................................................................................. 16 Is it appropriate to apply hypotheses about interlimb transfer from upper limb reaching to walking tasks? ....................................................................................................................................................... 18 Purpose ........................................................................................................................................................ 20 Hypothesis .................................................................................................................................................. 21 Methods ...................................................................................................................................................... 23 Participants .............................................................................................................................................. 23 Lokomat .................................................................................................................................................. 23 Protocol ................................................................................................................................................... 24 Data analysis ........................................................................................................................................... 28 Statistics .................................................................................................................................................. 30 Results ......................................................................................................................................................... 31 Unipedal vs. bipedal walking .................................................................................................................. 31 Task performance ? end point error ........................................................................................................ 34 Task performance ? initial foot trajectory slope ..................................................................................... 35 Adaptation to resistance .......................................................................................................................... 38 Discussion ................................................................................................................................................... 44 Methodological considerations ............................................................................................................... 44 vi  Source estimation model of learning and the dynamic dominance hypothesis ....................................... 45 Consolidation of motor adaptations ........................................................................................................ 46 Where could transfer be occurring? ........................................................................................................ 48 Awareness and generalization ................................................................................................................. 52 Conclusions ................................................................................................................................................. 53 References ................................................................................................................................................... 54                    vii  List of tables  Table 1 - Subject demographics .................................................................................................................. 25 Table 2 ? Adaptation rates during baseline unipedal walking .................................................................... 33 Table 3 ? Adaptation rates for end point error and initial trajectory slope ................................................. 37 Table 4 - Average adaptation rates for descriptive measures ..................................................................... 43                                  viii  List of figures  Figure 1- Foot pedestal ............................................................................................................................... 26 Figure 2 - Visual feedback provided to subjects. ........................................................................................ 27 Figure 3 - Outline of the protocol ............................................................................................................... 28 Figure 4 - Bipedal vs. unipedal walking ..................................................................................................... 32 Figure 5 ? Step by step change in peak hip and knee flexion during unipedal walking ............................. 33 Figure 6 - End point error from target ......................................................................................................... 35 Figure 7 - Slope change from baseline bipedal ........................................................................................... 36 Figure 8 - EMG response to resistance ....................................................................................................... 38 Figure 9 - EMG and kinematic response to resistance ................................................................................ 39 Figure 10 - Step by step change in Rectus Femoris activity and peak knee angle ...................................... 41 Figure 11 - Group EMG adaptation rates .................................................................................................... 42                 ix  Acknowledgements  I would like to start by extending the utmost gratitude and appreciation to my supervisor Dr. Tania Lam. You have given me to change to explore the world of academia, learn from my mistakes, and guide me back in the appropriate direction if I have strayed. You have given me greater insight into the scientific process, a real understanding for the work of writing, and helped me to understand my own strengths and weaknesses. Without your guidance, support, and faith in my abilities I sincerely doubt I would be in the position I am today. I would also like to thank those on my committee, Dr. Nicola Hodges and Dr. Romeo Chua, for guidance and support through this process. I appreciate being able to bounce ideas around as a group, which has given me insight into the true collaborative process that is research.  I would like to thank those in the Human Locomotion Lab (past and present) and the members of the International Collaboration on Repair Discoveries (ICORD) on the third floor of the Blusson Spinal Cord Centre. Your assistance, friendship, and willingness to laugh at my terrible jokes have made graduate school more enjoyable then I ever thought possible. Coming into this environment brought a smile to my face every day. I would like to specifically thank Drs. Antoinette Domingo and Amanda Chisholm for you never ending help and willingness to talk through data analysis and collection with me.  Thank you to my family, fianc?e and friends for their willingness to listen as I shared joys, frustrations, and progress throughout this entire process. Thank you Sara for reminding me of the beautiful scenery around Vancouver and sharing some great adventures with me. Thank x  you Anthony for making sure I enjoyed life outside of the classroom. You have all made me feel confident in my abilities and given me strength to believe that is degree is achievable. Lastly, thank you to the School of Kinesiology Graduate Department and the University of British Columbia for you knowledge and help for all matters throughout this degree. I have left this degree with a better understanding of who I am, for which I am forever grateful.                                     xi         To Sara1  Introduction Overview Afferent feedback is essential for locomotor adaptations to various types of perturbations. There is a large amount of evidence showing that sensory information aids in the control and coordination of locomotion (Forssberg, 1975; Lam and Pearson, 2002; Lam et al., 2006; Houldin et al., 2011). Furthermore, with sustained exposure to a perturbation the nervous system develops predictions of the sensory consequences to the disturbance. Internal models, which are neural representations of limb dynamics during movement, are thought to be involved in adaptations to sustained perturbations (Shadmehr and Mussa-Ivaldi, 1994). The internal model is stored and updated to reflect the altered movement dynamics (Shadmehr and Mussa-Ivaldi, 1994).  Exposure to a perturbation for extended periods of time creates anticipatory motor adaptations to compensate for the new environment. For example, increases in flexor muscle activity are seen when resistance is applied to the leg during the swing phase of walking in cats, humans infants, and adults (Lam and Pearson, 2001; Lam et al., 2006; Lam et al., 2003). The response to the perturbation appears very quickly, especially when the perturbation poses a potential threat to stability and safety during walking. The involvement of internal models in the locomotor adaptation is revealed by the after effects in the stepping pattern that result when the perturbation is unexpectedly removed (Lam et al., 2006; Lam et al., 2003).   The ability to generalization motor adaptations between similar motor tasks, environments or across limbs implies a certain level of efficiency in motor control, decreasing the amount of exposure that is required to adapt when presented with a different perturbation. There is strong evidence that motor adaptations can be transferred between the upper limbs during reaching tasks (Criscimagna-Hemminger et al., 2003; Galae et al., 2007; Malfait and 2  Ostry, 2004), catching tasks (Morton et al., 2001) and drawing tasks (Vangheluwe et al., 2004), but the specific movement parameter that is transferred appears to depend on limb dominance (Sainburg, 2002; Sainburg and Wang, 2002; Sainburg and Kalakanis, 2000). During walking, the evidence for interlimb transfer between the lower limbs is less clear (van Hedel 2002; Stockel and Wang, 2011; Houldin et al., 2012). There is some evidence that interlimb transfer between the lower limbs also depends on limb dominance (Stockel and Wang, 2011), but whether this is a factor involved in interlimb transfer of locomotor adaptations has not been determined. Thus the objective of this project is to test the possibility that interlimb generalization of locomotor adaptations depends on lower limb dominance. Literature Review Adaptability of walking  During walking, we encounter numerous scenarios that require modification of existing patterns in order to successfully continue walking. To enable the adaptations that allow us to successfully walk in different environments, the gait cycle must be flexible enough to accommodate all situations encountered. Our ability to make all the adjustments occur automatically  and with seeming ease, but the neural networks underlying the control and adaptability of walking are complex.  The circuitry responsible for producing the basic motor output for walking can be found in the spinal cord. This circuitry is known as the ?central pattern generator? (CPG) and is responsible for activating the appropriate sequence of neurons and interneurons used for transmitting and interpreting descending and ascending information (Rossignol, 2002). Even when the spinal cord has been removed of all inputs, it is possible to trigger the basic alternating 3  pattern of flexor and extensor activity (Grillner and Zangger, 1979). The ability for the spinal cord to generate the basic locomotor pattern underlies some of the automaticity of walking.  In order for walking to be meaningful, sensory feedback during movement is critical for modulating the motor pattern produced by the CPG. Proprioceptive information from the legs is gathered from various sources, such as cutaneous receptors on the foot, muscle spindles, and Golgi tendon organs in the tendons of muscles. The information obtained from the sensory apparatus is sent to the spinal cord via different sensory pathways. For example, CPG output is modulated by signals indicating hip position and limb loading to regulate extensor muscle activity during the stance phase of locomotion. Even in the absence of any cortical input, as in the decerebrate cat preparation, treadmill locomotion is initiated by hip extension (Grillner and Rossignol, 1978; Andersson et al., 1981; McVea et al., 2005). When the hip is manually stretched into extension, initiation of swing phase occurs at a fairly consistent angle (as long as the contralateral limb is in stance phase) (Grillner and Rossignol, 1978). Conversely, directly stretching the hip flexors, imitating a greater hip angle, during the extension phase of the gait cycle results in the cessation of ongoing extensor muscle activity and immediate initiation of the swing phase of gait (Hiebert et al., 1996). Similarly, hip angle appears to provide a critical signal for the transition from swing to stance, as its positions varies the least among the hindlimb joints at the onset of extensor activity at the end of swing across a variety of locomotor tasks (McVea et al., 2005).  Proprioceptive signals about the load experienced by the extensor muscles also regulate the level and timing of extensor muscle activity during stance. If there is a load maintained on the limb, the swing phase is prevented and/or the stance phase is prolonged (Conway et al., 1987; Duysens and Pearson, 1980; Hiebert and Pearson, 1999). With excessive load placed on the 4  extensor muscles, the rhythmic activation of the flexors seen during locomotion is inhibited until the load on the extensor muscle is decreased (Duysens and Pearson, 1980). Conversely, unloading the limb during stance results in a reduction of extensor muscle activity. For example, Hiebert and Pearson (1999) used a foot-in-hole paradigm to determine how limb loading affected extensor muscle activation. While the animals walked on a motorized treadmill, the hind limb stepped in a hole to unload the limb. As soon as the limb was unloaded, extensor muscle activity decreased by an average of 68% to 74% in vastus lateralis, medial gastrocnemius, and lateral gastrocnemius (Hiebert and Pearson, 1999).  Proprioceptive feedback signaling limb load are also utilized to make ongoing corrections in flexor muscle activity during the swing phase of walking. For example, Lam and Pearson (2001) used a decerebrate cat preparation to investigate modulation of hip flexor activity during resisted and assisted locomotion. During manually assisted swing phase, there was a decreased activation of hip flexor muscles. This was accompanied by an earlier onset of extensor muscle activity (McVea et al., 2005). When flexion is resisted, there was an increased in amplitude and duration of hip flexor muscle activity (Lam and Pearson, 2001).  The increase seen in muscle activation shows the use of proprioceptive information relating to hip angle and limb resistance to alter the muscle activation and change the locomotor output. Sensory information also shapes locomotor output in human. Human infants have an automatic stepping response that can be elicited even before they are independently walking, thereby providing an opportunity for studying the neural control of human locomotion prior to the maturation of descending pathways. Research on infants help to understand the function of the spinal cord and CPG in modulating motor output because of their underdeveloped equilibrium system and immature cortical systems (Forssberg, 1985; Pang and Yang, 2000).  5  When perturbations are applied to the stance limb of infants during locomotion, unique responses are seen based on the afferent information regarding limb load and hip position (Pang and Yang, 2000). An example is the response during a mid-stance perturbation that involved placing a piece of cardboard under one foot to keep in stationary. In this situation, the contralateral limb continued stepping while locomotor activity in the stationary limb was replaced by tonic extensor activity (Pang and Yang, 2000). Another example is during a forward perturbation of the stance limb, where the limb was slid forward by the cardboard as soon as heel strike was made. Two separate responses were seen - 1) the disturbed limb did not re-initiate swing phase until it was fully extended and unloaded, prolonging stance phase and step length and 2) swing phase was re-initiated immediately with a flexed hip if the limb load was extremely low (Pang and Yang, 2000). The hip position and load placed on the limb were both very important in selecting the appropriate response to the perturbations experienced. These results corroborate data from animal studies and illustrate the common principles in sensory feedback regulation of locomotion across species. In adults, the alteration to the limb dynamics during movement has been shown to modify the muscle activation patterns used to maintain locomotor output. Corrective EMG responses to a brief resistance applied to one leg during walking can be observed throughout the entire gait cycle. However, the contralateral (unperturbed) leg only shows altered EMG activity depending on when the perturbation is applied in the gait cycle (Ghori and Luckwill, 1989; Ghori and Luckwill, 1990). These changes in EMG did not alter swing and stance phase durations, suggesting the body can alter locomotor output while maintaining balanced walking (Ghori and Luckwill, 1989). In another study, a velocity-dependent resistance was applied against leg movements during swing phase, resulting in an immediate increase in swing phase rectus femoris 6  activation during all steps against resistance (Lam et al., 2006). Once the resistance was removed this response in rectus femoris disappeared, suggesting the involvement of reflex mechanisms in dealing with the altered environmental conditions (Lam et al., 2006). Rapid responses in extensor muscle activity to changes in body loading are another example illustrating the corrective responses to alterations in the dynamics of the body during locomotion (Stephens and Yang, 1999). With both sudden and sustained changes in limb loading through the addition of 30% of body weight to center of mass, there was an increase in extensor muscle burst. For example, during sudden loading changes the quadriceps have a 134% increase in amplitude of muscle activity during early stance phase (Stephens and Yang, 1999).These studies help support the importance of afferent information for the correction and alteration of locomotion in different environments.  Adaptation to sustained altered sensory information  In addition to reactive movement corrections, sensory information is also used to make sustained, long-term adaptations to movement. While the feedback-mediated modifications of the gait cycle described above can be effective for online corrections during movement, predictive changes in the walking pattern allow for greater movement efficiency with sustained perturbations. The development of these anticipatory motor changes is typically discussed within the context of ?internal models? (Shadmehr and Mussa-Ivaldi, 1994; Kawato 1999). Internal models are a conceptual framework for understanding motor control and adaptation. The framework assumes that motor control involves two separate control loops to perform predictive and online corrections to feedback from a movement. The forward model creates a prediction of sensory feedback based on the movement command and the environment (Wolpert and Ghahramani, 2000). The inverse model uses sensory feedback to perform real time corrections to 7  the movement based on incoming sensory information (Wolpert and Ghahramani, 2000). If there is a discrepancy between the predicted sensory feedback and the actual sensory feedback, corrections to the internal model are implemented for subsequent movements. This update of an internal model leads to the adaptation of a movement. Martin et al., (1996) defined motor adaptation as: ?modification of movement in which three criteria are satisfied: (i) the movement retains its identity as being of some particular pattern of muscle activation or end result but changes with regard to some parameter or set of parameters; (ii) the change occurs only with repetition of the behaviour and is gradual and continuous; (iii) once adapted, subjects cannot retrieve the prior behaviour; instead they must change the adapted  behaviour with practice in the same gradual, continuous manner back to the prior state.? Internal models are therefore thought to be updated through adjustments to incongruent sensory information as the central nervous system (CNS) learns the new dynamics of the limb and/or the environmental changes.  Evidence for the formation of an updated internal model is observed with the appearance of ?after-effects? with the removal of the altered limb dynamics/environment. Movement trajectories during after effects mirror the initial movement errors that result from the first exposure to the perturbation (Shadmehr and Mussa-Ivaldi, 1994). For example, Shadmehr and Mussa-Ivaldi (1994) created a force field environment in which participants performed a reaching movement to 8 different targets while holding a robot manipulandum. Subjects reached through varied forces, according to the location of the target, for 250 trials on 4 separate exposures without visual feedback to guide their movements. During the initial exposure to the 8  environment, subjects reaching movements were destabilized in the direction of the force causing them to stop and perform a secondary corrective movement, resulting in a ?hook-like? trajectory to the final end point. During the training, subjects were unaware they were adapting to the force field, but did report that their sense of effort decreased as training progressed. When the force field was suddenly removed, after-effects were immediately observed. The direction of the corrective ?hook? movement had become the mirror to what was observed during initial exposure to the force field.  Updates of internal models can be accelerated by increasing the size of the error experienced during the movement (Emken and Reinkensmeyer, 2005). Emken and Reinkensmeyer (2005) used a robotic device to apply an assistive force that altered peak toe height during the swing phase of locomotion. The rate of adaptation could be manipulated by imposing a larger perturbation of the first step. Rate of adaptation was measured using observed peak toe height during swing phase of the gait cycle. When this was done, average adaptation rate of peak toe height to the perturbation was 5.2 steps, compared to 7.2 steps in the control exposure of an equal amplitude perturbation throughout the exposure. After-effects were seen when the perturbation was removed and were manifested as a decrease in peak toe height, suggesting that the internal model had incorporated the assistance provided by the robot for movement.  Analysis of changes in EMG activity has also revealed separate feedback and feedforward locomotor adaptations. Lam et al., (2006) investigated the effects of a velocity dependent resistance (using the Lokomat robotic gait orthosis) on locomotor output. Muscle activation in the rectus femoris during the swing phase was elevated immediately upon exposure to the perturbation, along with decreases in peak hip, knee and ankle flexion angle. With 9  prolonged exposure to the resistance, lower limb kinematics generally recovered to baseline values accompanied by appearance of increased pre-swing knee flexor muscle activity. During catch trials, subjects produced a high stepping pattern suggesting an altered internal model used for walking in the new environment. The high stepping response in the catch trials was characterized by the persistence of the pre-swing knee flexor muscle activity, suggesting that this response was part of a feedforward adaptation. On the other hand, the elevated rectus femoris activity that was seen during all steps against resistance was not evident during the catch trials, suggesting that this response was part of a feedback adaptation (Lam et al., 2006).   Perturbations can be applied during stance phase or swing phase to test internal model updates during locomotion. Noel et al., (2009) used an ankle robotic orthosis to apply perturbations during mid-stance, which increased dorsiflexion that altered the loading on the limb and position of the individual. The perturbation required active adaptation to return the ankle angle to baseline characteristics of movement. During the post-adaptation phase, subjects were required to actively adapt to return to normal movement in the null environment. Blanchette and Bouyer (2009) used elastic tubing cut to generate approximately 40% of the maximum force of the hamstring muscles for each participant and the tubing connected to the subjects foot. During swing phase there was a large increase in the peak foot velocity. EMG displayed a decreased in EMG activity during the exposure trial to return to baseline walking. During the post-adaptation phase, there was a decrease in peak foot velocity during swing phase because of the decreased EMG activity and required adaptation to the null environment. The duration of after effects appears to depend on the length of exposure to the perturbations (Fortin et al., 2009). Retention of adaptations appear evident up to 1 day following, suggesting there could have been a separate internal model created for movement in that specific environment 10  (Fortin et al., 2009). The adaptation and modification of motor output with different perturbations in these studies show modifications of internal models for locomotor movements. The updating of the feedforward portion of the internal model optimizes motor output for the demands of movement in the new environment. Cerebellar involvement in adaptation  Internal models are thought to be formed in the cerebellum. Anticipating movements in an environment have produced evidence of the internal model framework being found in the cerebellum. Investigations of adaptations of eye movements offer an opportunity to understand the cerebellar involvement in predicting movement outputs. The cerebellum receives large inputs from the visual system and helps to coordinate eye movements based on both visual feedback, and during saccade movement, the anticipated location of the stimulus (Kawato, 1999). For example, Alahyane et al., (2008) used two different tasks to understand cerebellar control of saccades. Subjects performed two different types of saccade movements. First, a reactive saccade task was performed where a sudden visual stimulus appeared causing a saccade movement. Second, a voluntary saccade task was performed with a moving complex environment requiring voluntary saccades to move through the environment. It was found that a medial cerebellar lesion shows deficits in anticipation of location during reactive saccade movements, whereas lateral cerebellar lesions create deficits in location anticipation during voluntary saccades. Similarly for gait, it is thought that the medial cerebellar structure (vermis, fastigial nuclei) and the flocculonodular lobe could be responsible for predictive control of gait and balance because inputs from the dorsal spinocerebellar tract with information about the state of the limbs, outputs via the ventral spinocerebellar tract influencing interneuron activity and primary vestibular afferents (Morton and Bastian, 2004). 11  The role of the cerebellum in motor adaptations can be illustrated by studies of individuals with cerebellar lesions. Cerebellar lesions create large deficiencies in postural control and coordination of locomotor output (Fuentes and Bastian, 2007; Bastian 2006; Bastian 2008; Morton and Bastian, 2006). Coordination of information from both the cortex and vestibular system allow the cerebellum to coordinate postural control and movement output. When there is damage to the cortico-cerebellar pathway, postural sway increases when compared to a normal subject (Morton and Bastian, 2004). There is an even larger increase in postural sway compared to healthy subjects when the vestibulocerebellar pathway is damaged (Morton and Bastian, 2004). Studies involving cerebellar lesion patients have shown that these patients have a decreased ability to update predictive movements, resulting in them relying on feedback mediated responses to stimuli (Bastian 2011; Shadmehr and Krakauer, 2008).  Morton and Bastian (2006) used a split-belt treadmill to determine how feedback is utilized to adapt to an altered environment. Subjects with cerebellar damage showed adaptation in stride length and stance time within a few strides in the altered environment, which are indicative of feedback controlled adaptations. Step length and double support time did not show adaptation, which are thought to be controlled through predictive updates to the internal model. Stride length and stance time are thought to be feedback controlled because they involve interlimb coordination, relying on spinal cord connections for adaptation (Reisman et al., 2005). Step length and double support time are thought to be feedforward controlled in the split belt treadmill paradigm because they involve intralimb coordination which would require CNS involvement for adaptation (Reisman et al., 2005). This evidence helps support that patients with cerebellar damage rely on feedback information to control movement and locomotion.  12  It is also possible to directly manipulate cerebellar activity in order to study the role of the cerebellum in motor adaptation. Yanagihara and Kondo (1996) decreased concentration of nitric oxide (NO) (an essential compound for motor learning to occur) in the cerebellum of cats and investigated changes in adaptation to split-belt treadmill locomotion. In control animals, adaptation to split-belt treadmill locomotion is accompanied by a gradual decrease in variability of step cycle and double-support phase durations. However, when NO concentration was decreased, there was large variability in step cycle and double support phase duration, throughout split-belt training trials. When levels of NO were returned to normal, the variability decreased and returned to the control values. In another study, direct stimulation of the cerebellum was used to demonstrate the involvement of cerebellar activity of locomotor adaptations. Jayaram et al., (2012) applied transcranial direct current stimulation (tDCS) to the cerebellum to show cerebellar control of locomotor adaptation. Subjects walked on a split-belt treadmill while current was applied to the ipsilateral cerebellar hemisphere of the fast leg. In control conditions, the rate of adaptation to split-belt locomotion was 12.6 strides. With the anode electrode placed on the ipsilateral hemisphere, the rate of adaptation was reduced to 8.7 strides. But when the cathode electrode was placed on ipsilateral hemisphere, the rate was slowed to 31.1 strides. These studies in the intact cerebellum support the idea that the cerebellum is important in regulating the rate of adaptation during a locomotor task.  Generalization of motor adaptations  Adaptations are even more useful if they can be generalized between environments and limbs. When similar movements are used in multiple tasks, generalization of adaptation could occur between tasks (Wang et al, 2011; Bhatt and Pai, 2009). For example, in the study by Bhatt and Pai, (2009) individuals performed two separate tasks: a sit-to-stand or walking task and 13  changes in the control of balance after training were measured. Training consisted of repeated exposure to slips created by forward moving platforms under each foot during the sit-to-stand task. Subjects were then exposed to an unannounced slip during the walking task that was similar to the slip in the sit-to-stand task. After training, subjects showed a significantly lower incidence of falls and balance loss when compared to control, which was associated with improvements in center of pressure control  Much like task or environmental generalization of adaptation, interlimb generalization would be useful for decreasing the time and number of exposures needed to learn a task once a limb has already learned it.  Interlimb generalization of motor adaptation have been found for many tasks of the upper limbs (Morton et al., 2001; Vangheluwe et al., 2004; Sainburg 2002; Sainburg and Wang 2002; Criscimagna-Hemminger et al., 2003; Malfait and Ostry, 2004; Galae et al., 2007). Studies of upper limb movements have the advantage of the ability to isolate training to one arm. Studies have shown that inter-limb transfer of adaptation can allow for the optimal performance of a task without ever practicing the task with that limb. For example, Morton et al., (2001) used a catching task where participants practiced catching balls of differing weights while not allowing their hands to drop outside the limits of a defined range. After training with one limb, participants switched to the opposite limb. Performance in the opposite limb was better than the initial performance of the trained limb, as shown by a decrease in the initial error during the catching task and also an increase the learning rate of the task (Morton et al., 2001). Similarly, Vangheluwe et al., (2004) used a bimanual drawing task to determine transfer between limbs performing differing tasks. After practice in one configuration of the drawing task, subjects showed faster learning rate while performing the same tasks with the opposite arm (Vangheluwe et al., 2004).  14   Given the bipedal nature of walking, one would expect that interlimb generalization should be very important for locomotor adaptations. The limbs move in coordination with one another and the sharing of information is important between limbs. The connection between the lower limbs is evident in people with complete spinal cord injury performing assisted unipedal walking on a treadmill. Coordinated locomotor-like muscle activation was seen in both limbs during assisted unipedal walking (Ferris et al., 2004). The activation was similar to that seen during assisted bipedal walking and appears that both legs are utilizing information to determine appropriate muscle activation (Ferris et al., 2004). van Hedel et al., (2002) investigated lower limb generalization using an obstacle avoidance task during walking. Subjects were asked to step over an obstacle while minimizing the distance of their foot over the obstacle. Visual information of the treadmill was removed and extraneous auditory information was suppressed to minimize reliance on other senses to perform the task. Feedback on performance was provided by a series of auditory tones corresponding to height over the obstacle. Subjects were required to step over the obstacle with a specific ?optimal? height. After training with one limb, there was similar performance found in the opposite leg (van Hedel et al., 2002). The transfer was equal whether training was done on the left or right leg, with significant decreases in error in the opposite leg after training (van Hedel et al., 2002). This transfer allows for the optimization of performance with smaller amounts of practice and is very similar to that found in the arms.  In contrast, recent findings in our lab did not show strong evidence for interlimb transfer of locomotor adaptations. Houldin et al. (2012), found limited evidence for transfer between lower limbs in healthy subjects. Subjects walked unipedally against a velocity dependent resistance during swing phase. Catch trials were interspersed throughout the exposure to the perturbation.  After opposite limb training, there was only transfer of after effects in the hip 15  measured as peak hip flexion (Houldin et al., 2012). A potential reason for the conflicting evidence for interlimb generalization during walking between studies (cf. Houldin et al., 2012 and van Hedel et al., 2002) could be that the task in the van Hedel experiment required greater cognitive attention to the task compared to that of Houldin et al. (2012). Cognitive awareness to the task has been shown to influence motor adaptations. Malfait and Ostry (2004) found that gradual application of a perturbation did not result in internal model updates, even though there was a large amount of practice. The perturbation had to be introduced suddenly, with the participant aware of the change in the environment for feedforward updates to occur (Malfait and Ostry, 2004).  Interlimb transfer of locomotor activity may be limited because information from afferent input may be limited the limb that supplied the information. Lavrov et al., (2008) performed a spinal cord transection and unilateral deafferentation in rats to determine if and what afferent information was crossed between limbs with epidural stimulation. They found limited recovery on the side that was deafferented, suggesting that even though there are large bilateral communication networks between the limbs afferent information from the ipsilateral limb is important for determining the appropriate movement patterns (Lavrov et al., 2008). Understanding that the spinal cord can only use information for the ipsilateral leg brings the importance of higher structures being involved in adaptation transfer. The cortex appears to play an essential role in allowing for interlimb transfer (McVea and Pearson, 2007). Houldin et al., (2012) did not find evidence of transfer, perhaps because the task used was implicitly motivated and did not engage cognitive structures. Since the spinal cord shows limited ability to transfer sensory information from the opposite leg despite large bilateral connections, the task may have been inappropriate for testing interlimb transfer during walking. We know from Malfait and 16  Ostry (2004) that cognitive awareness of the perturbation is required for adaptation to occur and indeed, van Hedel et al., (2002) who used an obstacle avoidance task requiring cognitive awareness and explicit feedback of task performance, did find transfer. Thus, interlimb transfer of motor adaptation is perhaps dependent on whether the task involves explicit goals and cognitive awareness. Could transfer be dependent on limb dominance?   There is evidence that limb dominance could be an important factor in interlimb transfer (Criscimagna-Hemminger et al., 2003; Malfait and Ostry, 2004; Galae et al., 2007; Sainburg 2002; Sainburg and Wang 2002; Stockel and Wang, 2011; Pryzbyla et al., 2012). Transfer in both directions, dominant (D) to non-dominant (ND) and ND to D have been assessed to understand transfer. Criscimagna-Hemminger et al. (2003) determined that transfer of learning to the new arm dynamics between the arms occurred in only the D to ND direction. Transfer appeared to rely on an extrinsic coordinate system regarding the velocity of the hand during movement (Criscimagna-Hemminger, 2003). However, Sainburg (2002) found that transfer did occur in both D to ND and ND to D directions during dynamic perturbations, depending on what movement parameter was used to index transfer. Contrary to Criscimagna-Hemminger et al., (2003) transfer in the direction of D ? ND occurred mainly for end-point accuracy, whereas transfer in the ND ? D direction occurred for movement trajectory. Transfer in both directions has also been found during visuo-motor perturbations (Sainburg and Wang, 2002).  From the evidence in the literature, it appears that we have to have the correct movement parameters to quantify interlimb transfer. Sainburg (2002) proposed a Dynamic Dominance Hypothesis to explain the differences in D?ND and ND?D interlimb transfer of adaptation. 17  This hypothesis states that transfer of adaptation is dependent on the dominance associated with the upper limb. The dominant arm is considered to be more aptly suited to code adaptations related to the dynamic portions of movement, such as joint torque interactions. On the other hand, the non-dominant arm is considered to be more aptly suited to code adaptations related to position or end-point accuracy. Thus, transfer between the limbs will be asymmetrical depending on the limb being trained and the movement parameter being tested (Sainburg, 2002; Sainburg and Kalakanis, 2000).  The hypothesis is supported by results of reaching adaptation studies involving visuo-motor rotations or alterations to the dynamic properties of the reaching movement (Sainburg, 2002; Sainburg and Kalakanis, 2000; Sainburg and Wang, 2002; Wang and Sainburg, 2006). When the non-dominant limb is trained to perform movement task in these environments, subsequent testing of the dominant arm shows that adaptation of the dominant side occurs at a faster rate for the dynamic portions of movement. Conversely, when the dominant arm is exposed to a perturbation for a training period, subsequent testing of the non-dominant arm reveals faster adaptation of the end point accuracy of the movement (Sainburg, 2002; Sainburg and Kalakanis, 2000; Sainburg and Wang, 2002). Similar results are observed even during tasks where movement trajectories are not constrained by a manipulandum. The dynamic dominance is seen in unsupported reaching tasks also (Tomlinson and Sainburg, 2011). There is a greater degree of flexibility seen in movements that are not restrained to an experimental manipulandum. Subjects performed reaching movements with the shoulder and elbow, while the wrist was splinted, to three different targets placed at different angles from the start point. It was found that the non-dominant arm produced movements with increased precision and end point accuracy of movement and the dominant arm showed efficient movement patterns with well-coordinated 18  muscle activation patterns (Tomlinson and Sainburg, 2011). Unsupported reaching offers numerous different solutions to make a movement. To observe the principles of the Dynamic Dominance Hypothesis still present during unsupported reaching shows the need to measure the appropriate variables to understand adaptation transfer.   There is also some evidence that there is asymmetrical transfer of adaptation for lower limb movements. Stockel and Wang (2011) showed transfer dependent on practice order and task completion. Subjects performed a spatial targeting task with their leg movements while they were placed on a swinging sled apparatus. There was a force plate perpendicular to the sled for the foot to push on and a light to provide an end point goal. The task involved a knee extension movement to propel the swing to either reach the lighted target area with their head or to maintain a force output. For the force training paradigm, the dominant leg showed faster rate of adapatation when the non-dominant leg was trained first. For training of end point accuracy, the non-dominant leg showed faster adaptation rates when the dominant was trained first (Stockel and Wang, 2011). As with the upper arms, these results are consistent with the Dynamic Dominance Hypothesis since the transfer of adaptation was dependent on the lower limb dominance.  Is it appropriate to apply hypotheses about interlimb transfer from upper limb reaching to walking tasks?  Unlike the reaching or pointing tasks typically used in the aforementioned studies, walking is not a discrete movement but uses a continuous pattern for the gait cycle to take place. However, locomotion and reaching movements could involve similar neural structures and control, with the cortex involved with visual coordination of the movement (Georgopoulos and Grillner, 1989). Although locomotion is a continuous motion, there are definite portions of the 19  gait cycle that have defined start and end positions, cycling repeatedly with each step. For instance, during walking, foot trajectory during swing is closely regulated (Winter, 1992), as evidenced by our ability to precisely control foot position over an obstacle (e.g., van Hedel, 2002). Similarly, foot placement following the end of swing (heel strike) must be carefully controlled in order to ensure a safe landing site with each step. This motion mimics a reaching movement whereby the end point position must be precisely regulated.  The precise control of foot trajectory during walking can be appreciated when locomotion is performed with cortical damage (Adkins et al., 1971; Dubrovsky et al., 1974; Eidelberg and Yu, 1981). Eidelberg and Yu (1981) lesioned the corticospinal system in multiple areas to understand the role of the different areas on locomotor kinematics. There were no differences in kinematics found between the different lesioned areas but the locomotor pattern on the contralateral side of the ablation showed changes, with large increases in the amount of extension at the hip, knee and ankle joints during walking (Eidelberg and Yu, 1981). These differences became attenuated with increased walking practice, with joint angle returning to baseline two weeks post-surgery (Eidelberg and Yu, 1981).  Adkins et al., (1971) described deficits in paw placement and jumping in cats after the portions of the motor or sensory cortices had been damaged. The altered descending input from the cortex alters the locomotor behaviour seen in the animals. Although, baseline gait cycle returns, movement during the paw contact placement response becomes irregular and inaccurate in sensory cortex damage and disappears completely in motor cortex damage. The deficits seen in paw placement make movement in complex environments nearly impossible.  Other investigators have been able to show specific changes in firing pattern in cortical areas during locomotor accuracy tasks (Beloozerova and Sirota, 1993a; Beloozerova and Sirota, 20  1993b; Beloozerova et al., 2010; Drew, 1993; Drew et al., 1996). Beloozerova et al., (2010) recorded from neurons within the motor cortex of a cat trained to walk on a ladder with different cross beams widths. The cats were tested on 4 different tasks: simple locomotion over flat ground and stepping on ladder rungs that were 18cm, 12 cm in width or 5 cm in width. There was an increase in the precision of discharge timing in neurons of the motor cortex that was very evident in the transition from the simple walking task over flat ground to walking across the 18cm beam width. Modulation of firing was also evident, such that firing was shorter but more intense as ladder widths decreased. There was also a large decrease in the variability of stepping location for the animal as it progressed through the decreasing width seen in the ladder (Beloozerova et al., 2010). This data shows that the motor cortex has involvement in controlling accuracy of movement during difficult locomotor tasks. Purpose  It is clear that key features of the walking pattern must be precisely controlled in order for locomotion to proceed successfully in a variety of conditions. However, it remains unclear the extent to which interlimb transfer is involved in locomotor adaptations. Previous studies of interlimb transfer of locomotor adaptations have generally been limited to select indices of performances as the measure of transfer. However, the determination of whether transfer occurred could critically depend on what specific movement parameter is being tracked. Also, studies to-date have not considered the possibility that interlimb transfer is dependent on limb dominance. Therefore, the overall objective of this project is to understand whether interlimb transfer of locomotor adaptation is asymmetrical and dependent on limb dominance. A precision locomotor task will be used whereby subjects will be instructed to aim each heel strike to a 21  virtual target (corresponding to a point on the treadmill). Locomotor adaptations will be induced by imposing a velocity-dependent resistance against the hip and knee movements. Subjects will walk unipedally, allowing us to test interlimb transfer of adaptations without interference from contralateral limb responses to the perturbation. The specific objectives of this research are to determine whether: 1. Transfer from the D to the ND limb will result in a faster rate of adaptation in end point placement during locomotion. 2. Transfer from the ND to the D limb will results in a faster rate of adaptation in the dynamic performance of the movement during locomotion. Hypothesis   The main hypothesis of this study is that interlimb transfer of locomotor adaptations to a velocity-dependent resistive force will be asymmetrical and follow the predictions of the dynamic dominance hypothesis. Specifically, it is expected that: 1. Training of the dominant leg to a velocity-dependent resistance during unipedal walking will result in faster adaptation rate of end-point position when the non-dominant leg is subsequently tested. 2. Training of the non-dominant leg to a velocity-dependent resistance during unipedal walking will result in a faster adaptation rate of movement trajectory when the dominant leg is subsequently tested. 22  3. During subsequent exposures to a velocity-dependent resistance following an initial training trial, the adaptation rate will be faster for both end-point position and movement trajectory within the same leg during unipedal walking. 4. There will be no difference in the adaptation rates between the dominant and non-dominant leg group for either end-point position or movement trajectory during the initial training trials to a velocity-dependent resistance force field   23  Methods Participants  Twenty healthy subjects (male and female, ages 18-40) were recruited for this study. Recruitment took place at the University of British Columbia. All participants were free of any known neurological or motor disabilities. All subjects were both right-handed and right leg dominant.  Hand dominance was determined using the 10-item version of the Edinburgh inventory (Oldfield, 1971), while leg dominance was determined using the Waterloo Footedness questionnaire (Elias et al., 1998). Participants were excluded if they had previously participated in activities where asymmetrical limb training was involved (i.e., soccer). All participants signed an informed, written consent in accordance with the University of British Columbia (UBC) ethics committee. Lokomat  The lower limbs were measured to ensure proper fit within the Lokomat robotic gait orthosis (Hocoma AG, Volketswil Switzerland). The Lokomat system incorporates a body weight support system that suspends the participant over a treadmill and a pair of robotic limbs attached to the legs. The subject is attached to the robotic limbs with thick Velcro cuffs around the upper thigh, upper shank and lower shank. The Lokomat allows for movement hip and knee flexion and extension, while the ankle is free to move unimpeded.  The Lokomat was programmed to apply the velocity dependent resistance to movement defined by:  24   Where M is the instantaneous amount of torque applied, B is the viscous (or resistive force) coefficient, and ? is the instantaneous angular velocity of the hip (H) and knee (K) joints (Lam et al, 2006). When B is set to zero, no force is applied to the limbs (null field). B values used to apply velocity dependent force was based on 10% of the subject?s hip and knee flexor maximum voluntary contraction (MVC) and average joint angular velocity during baseline bipedal walking. Hip and knee flexor MVC were measured by the force sensors embedded in the Lokomat (Bolliger et al., 2008). Three successive contractions were averaged to determine a value in Nm. Average hip and knee joint angular velocity during swing were measured from a period of baseline bipedal walking in the null field  Protocol  Subjects visited the lab on two separate occasions. The first visit was a familiarization visit during which subjects practiced bipedal walking with the Lokomat in the null field. In addition, MVC were measured during this visit. Participants were asked to walk at 3 km/h (0.83 m/s). If they were unable to walk at this speed with the Lokomat, they were excluded from the study. Once the subject was comfortable with moving in the robot, baseline recording of hip and knee joint kinematics during walking was recorded for subsequent B value calculations for the velocity dependent resistance.    25    Subject Gender Age Height (cm) Weight (kg) Footedness Score Handedness Score R to L transfer group 1 F 29 168 54 10 84.62 2 F 24 178 71 6 85.71 3 M 23 172 79 5 100 4 M 26 164 92 9 70 5 F 18 150 65 15 100 6 F 19 160 45 14 81.82 7 M 24 180 65 4 100 8 M 20 180 74 15 80 9 M 25 178 78 12 87.5 10 F 19 140 55 12 87.5  Mean (SD) 22.7 (3.59) 167 (13.67) 67.8 (13.96) 10.2 (4.1) 87.71 (9.86) L to R transfer group 1 M 29 167 95 8 100 2 M 26 181 64 15 100 3 M 31 183 70 18 76.47 4 M 18 174 65 8 100 5 M 18 173 65 12 68.42 6 F 23 178 68 9 88.24 7 F 23 168 68 8 80 8 F 19 155 51 17 87.5 9 F 20 164 57 9 86.67 10 F 18 160 50 14 76.47  Mean (SD) 22.5 (4.79) 170.3 (9.14) 65.3 (12.6) 11.8 (3.94) 86.36 (11.16) Table 1 - Subject demographics During the second laboratory visit, subjects underwent the experimental protocol. Unipedal walking was used to isolate training as much as possible to one limb. A pedestal was built to hold the foot a minimal distance above the moving treadmill belt so that subjects could stand on one leg and step with the other. The pedestal attached to the side of the Lokomat and used a counter weight to ensure the platform did not touch the treadmill. During all unipedal walking trials, subjects were asked to perform a targeted walking task and received feedback 26  about foot position at heel strike. The task was used to simulate walking in a target oriented situation that requires accurate foot placement, such as when crossing a river by stepping on stones.  Figure 1- Foot pedestal Foot pedestal to facilitate unipedal walking. a) top view, b) side view, c) front view to show height off of treadmill  An Optotrak smart marker was placed on the side of the treadmill for use as a reference point. During the baseline bipedal recording, an offset value was found in both the X and Y coordinates to allow for the virtual movement of the target during testing. To determine the offset values, subjects were asked to walk normally and the position of their foot at heel strike in reference to the physical placement of the reference marker was used as offset values.   Participants monitored their performance during the task via a 46-inch TV monitor at eye level, 3.35 meters in front of them. The output seen by the participants was similar to a ladder, with the middle line representing the target and subsequent lines above and below the target line appearing in 5cm intervals. The custom written Labview program used the data from force sensitive resistors (FSR) placed in the shoe to determine the instant that heel strike occurs during walking. The moment threshold was passed the placement of the foot was displayed.  The a) c) b) 27  subject was instructed that the goal of the task was to ensure that the heel lands on the target line in the anterior/posterior direction during each step. The heel placement was displayed for 750 ms after each step. The system was customizable to allow for differing stride lengths of subjects and created a task specific to each participant. Participants were verbally reminded throughout the testing that their goal was to place their heel onto the target line.   Figure 2 - Visual feedback provided to subjects.  The target was represented by the white line. The red lines correspond to 5 cm increments behind (below the white line) or beyond (above the white line) the target. Actual heel position was represented by a green Subjects were randomly allocated to two separate training groups, right to left transfer group (R?L group) or left to right transfer group (L?R group). The limb that received the training was defined as the trained leg, and the opposite leg was defined as the test leg. A total of 400 steps were performed against resistance for the trained limb and 200 steps against resistance were performed for the test limb. Baseline recordings were made for bipedal walking, and unipedal walking for both the trained leg and test leg. Subjects then performed 200 steps against resistance to the trained leg (Resist1). This was followed by a washout period where the trained leg performed unipedal walking without resistance (Washout). A second period of training then took place consisting of 200 steps against resistance (Resist2). This was immediately followed by the test trial, where the opposite leg walked unipedally against resistance (Transfer). A catch 28  trial, where resistance is removed for a single step, was used to help determine adaptation and was used after the first step against resistance (second step of the trial) and again at the 180th step of the trial.      Data analysis   The primary parameters used to determine transfer between the limbs was heel end point error and initial foot trajectory slope during swing phase. End point error was measured in the anterior-posterior direction and calculated as the distance between the target position and the actual heel position at foot contact during each step. Positive differences represented landing past the target, while a negative difference represented landing short of the target. The difference found was then converted to a percentage which corresponds to the change in stride length from baseline bipedal walking. A difference score of 0% meant that stride length during unipedal walking was the same length as the average stride length during bipedal walking. Foot trajectory slope was calculated over the period of the gait cycle starting at toe off and lasting in duration equivalent to 5% of the gait cycle. Slope values were normalized to the average foot trajectory Baseline Bipedal 50steps Baseline Unipedal 50steps (Test leg) Baseline Unipedal 50steps (Train leg) Resist1 200steps Washout 200steps Resist 2 200steps Transfer 200steps Washout 200steps Pre-Test Train Leg Test Leg Feedback task, with resistance Feedback task, no resistanceNo feedback, no resistance Figure 3 - Outline of the protocol Outline of the protocol used during the data collection session. Catch trials were spaced during the Resist 1, Resist 2 and Transfer trials at the 2nd step and the 180th step during the trial. 29  slope during baseline bipedal walking and multiplied by 100. A score of 100% meant the slope found was the same as the one during bipedal walking.  The time course of adaptation was calculated for each subject for all measured variables. Adaptation was described by fitting the data with the exponential function, y = a + b * exp-x/?, where a represents the offset, b is the gain, and ? is the time constant of adaptation, representing the number of steps it would take to obtain 63.2% of total adaptation (Lam et al., 2006). Adaptation was defined as being complete at 95% steady state, which was calculated by rounding up the value defined by ?? * ln (0.05) (Lam et al., 2006).  In instances where the data was best fit with a linear equation (suggesting the rate of adaptation was beyond the trial length), we assigned a value of 200 to represent the number of steps needed for adaptation (the maximum number of steps in the trial).   Secondary measures included changes in temporal gait parameters, hip and knee joint angles, and EMG activity at key points in the step cycle. We used this data to understand how changes in task performance are associated with changes in kinematic and EMG activation patterns during adaptation, and the time course of these adaptations. EMG activity was normalized to peak amplitude during baseline bipedal walking. Time periods of interest based on expected periods of activity change were chosen for each muscle and the average activation was calculated for a step by step comparison. Period of interest for the tibialis anterior (TA) was 50 to 90% of the gait cycle, 50 to 60% of the gait cycle was chosen for both soleus (SOL) and medial gastrocnemius (MG), 60 to 100% of the gait cycle for rectus femoris (RF) and finally 50 to 70% of the gait cycle for bicep femoris (BF). The hip and knee flexion angles were normalized to peak flexion angle during baseline bipedal walking. The change in peak angle was normalized for each participant to be used for step by step comparison. Stride length, defined by the length 30  between toe off and the subsequent heel strike of the same leg, was also used to understand the pattern of adaptation. This data was used to understand what strategies are used to enable the adaptation in the trained leg and the transfer of adaptation to the test leg. Statistics   All statistical analyses were performed with a commercially available software package (SPSS 12.0; SPSS Inc., Chicago, IL). Significance was assessed at 0.05 for all statistical evaluations. Descriptive statistics were calculated for age, weight, height, Waterloo Footedness Questionnaire score and Edinburgh Handedness Questionnaire score.   A 2 (group) x 3 (trial) repeated measures ANOVA was used to compare adaptation rates among the Resist1, Resist2, and Transfer trials between the two groups for end point error, initial foot trajectory slope, EMG amplitude and the kinematic.            31  Results Unipedal vs. bipedal walking   We used unipedal locomotion in order to isolate adaptation to a single leg. A comparison of the EMG activity and joint kinematics between bipedal and unipedal walking is shown in  Fig 4. Muscle activation patterns were largely similar, except in the BF where there was greater activation during the stance phase of unipedal walking compared to bipedal walking. However, the activation patterns in all the other muscles were largely comparable between bipedal and unipedal walking although there tended to be greater variability during unipedal walking. Hip and knee joint trajectories were also similar between bipedal and unipedal walking.  32   Figure 4 - Bipedal vs. unipedal walking Average group EMG and kinematic patterns for all subjects in the L?R group. The Bipedal walking is diagrammed in the grey lines (average, +/- standard deviation). The unipedal walkng is diagrammed in the black lines (average, +/- standard deviation).  All EMG plots are in uV and kinematic plots are in degrees. Steps were triggered off of hip angle, with peak hip flexion corresponding with heel strike and peak hip extension corresponding with toe off. Data has been normalized to 100% of the gait cycle. TibialisAnterior(uV)Soleus(uV)MedialGastrocnemius(uV)RectusFemoris(uV)BicepsFemoris(uV)HipAngle (Degrees)KneeAngle(Degrees)% S te p C y c leBipedalUnipedal02 04 06 0- 2 004 08 0- 2 02 06 01 0 0- 2 002 04 06 0- 2 002 04 06 0- 2 002 04 002 04 06 08 02 0 4 0 6 0 8 0 1 0 033   Figure 5 illustrates the average step by step change in peak hip and knee flexion during the baseline unipedal trials. Adaptation rates in both legs in both groups ranged from 12 to 30 steps (Table 2), which meant that subjects would have required 36 to 90 steps to reach 95% of steady state during baseline unipedal walking.  Figure 5 ? Step by step change in peak hip and knee flexion during unipedal walking Average step by step change in peak hip and knee flexion angles when normalized to baseline bipedal movement. The grey line represents peak joint angle that would be the same as that seen during bipedal walking.    Left Leg Right Leg HIP   R ?L 30.32 (19.77)  23.28 (23.54)  L ? R 12.47 (16.10)  29.10 (21.73)  KNEE   R ?L 19.34 (17.09)  15.35 (18.99)  L ? R 21.61 (20.60) 20.26 (21.17)  STRIDE LENGTH   R ? L  22.37 (23.96)  25.18 (22.08)  L ? R  22.70 (21.12)  23.56 (19.71)  Table 2 ? Adaptation rates during baseline unipedal walking Adaptation in the hip and knee flexion and stride length during baseline unipedal walking for both legsThe average adaptations rates (representing the number of steps required to reach 63.2% of steady state) across all subjects in each group are presented. Values in parentheses represent the standard deviation. 0 . 911 . 10 . 911 . 12 0 4 0 6 0#  o f  S t e p sNormalized to%ofpeak joint angle34  Task performance ? end point error  Figure 6 illustrates the average step by step change in end point error during Resist 1, Resist 2, and Transfer trials.  Resistance resulted in an overall end point error magnitude of about 2% of baseline step length in all trials except the Transfer trial in the R?L group, where the overall error magnitude was -1% of baseline step length. In the R?L group, the 63.2% steady state adaptation rate was 18 steps in Resist1, 35 steps in Resist2 and 18 steps in the Transfer trial. In the L?R group, the 63.2% steady state adaptation rate was 30 steps in Resist1, 19 steps in Resist2 and 31 steps in the Transfer trial. There was no main effect for group (F(1,18) = 0.094, p = 0.763) when comparing the R?L group to the L?R group. There was no significant main effect across the trials for end point placement (F(2,36) = 0.042, p = 0.959), nor any interaction effects between group and trial (F(2,36) = 0.920, p = 0.408).   35   Figure 6 - End point error from target Average group step by step displacement from the target position, expressed as a percentage of baseline bipedal stride length. The difference from heel contact placement from the target position in the anterior/posterior plane was calculated, and then normalized to the percent change in stride length. Average percent change and +/- standard deviation are represented as solid lines on plots.  Task performance ? initial foot trajectory slope  Figure 7 illustrates the step by step change in initial foot trajectory slope during Resist1, Resist2, and Transfer trials. The magnitude of the change in initial trajectory slope was variable between groups. The R?L group average initial trajectory slope was over 100% compared to baseline bipedal walking, or nearly equal to that slope. In the L?R group, the average slope during the Resist1 and Transfer trials was 130%, meaning the initial foot trajectory slope was 30% steeper during unipedal walking compared during bipedal walking. In the R?L group, the 63.2% steady state adaptation rate was 65 steps in Resist1, 12 steps in Resist2 and 53 steps in the R --> Lgroup-8.00-4.004.008.0040 80 120 160 200-8.00-4.004.008.0040 80 120 160200# of stepsEnd Point Error%ofstride length-8.00-4.004.008.0040 80 120 160 200L --> Rgroup-8.00-4.004.008.0040 80 120 160 200-8.00-4.004.008.0040 80 120 160 200-8.00-4.004.008.0040 80 120 160 200#of stepsResist1Resist2TransferEnd Point Error%ofstride lengthEnd Point Error%ofstride length36  Transfer trial. In the L?R group, the 63.2% steady state adaptation rate was 80 steps in Resist1, 87 steps in Resist2 and 62 steps in the Transfer trial. There was no main effect for group (F(1,18) = 2.903, p =0.106) when comparing the R?L group to the L?R group. There was no significant main effect across the trials for slope (F(2,36) = 0.481, p = 0.622), nor any interaction effects between group and trial (F(2,36) = 1.1817, p = 0.317).   Figure 7 - Slope change from baseline bipedal Average group step by step change in initial foot trajectory slope, expressed as a percentage of baseline bipedal slope. The change in slope during the initial swing phase, corresponding with 5% of the gait cycle, was normalized to the percent change when compared to baseline bipedal. Average percent change and +/- standard deviation are represented by solid lines on each plot.          Resist1Resist2T ansfer4 08 01 2 01 6 02 0 04 0 8 0 1 2 0 1 6 0 20 0%changefromBipedal4 08 01 2 01 6 02 0 04 0 8 0 1 2 0 1 6 0 2 0 04 08 01 2 01 6 02 0 04 0 8 0 1 2 0 1 6 0 2 0 04 08 01 2 01 6 02 0 04 0 8 0 1 2 0 1 6 0 2 0 04 08 01 2 01 6 02 0 04 0 8 0 1 2 0 1 6 0 2 0 04 08 01 2 01 6 02 0 04 0 8 0 1 2 0 1 6 0 2 0 0%changefromBipedal%changefromBipedal# of steps # of stepsR --> Lgroup L --> Rgroup37   RESIST 1 RESIST 2 TRANSFER END POINT ERROR    R ?L 17.59 (18.84)  34.22 (54.90)  17.22 (23.49)  L ? R  29.18 (28.33)  19.00 (21.71)  30.68 (61.89)  INITIAL SLOPE    R ?L 64.25 (93.86)  11.70 (18.37)  52.63 (73.07)  L ? R  79.90 (83.67)  86.74 (86.70)  61.09 (69.24)  Table 3 ? Adaptation rates for end point error and initial trajectory slope The average adaptations rates in end point error and initial trajectory slope (representing the number of steps required to reach 63.2% of steady state) across all subjects in each group are presented. Values in parentheses represent the standard deviation.                                38  Adaptation to resistance  Figure 8 - EMG response to resistance Average group EMG response seen in TA, SOL, MG and RF for both R?L and L?R groups. The black trace represents the average EMG trace for the last 20 steps of the corresponding trial, while the light grey trace shows the average EMG trace during the baseline unipedal trial for the same leg seen used during the transfer trial.  In the R?L group, the Resist1 and Resist2 trials are performed with the right leg and the Transfer trial is performed with the left leg. The opposite is true for the L?R group. All EMG data is plotted in normalized units. RESIST 1 RESIST 2 TRANSFERTASOL% o f G a it C y c le % o f G a it C y c le % o f G a it C y c leBaseline UnipedalAverage Last 20 stepsMGRFL --> RR --> LL --> RR --> LL --> RR --> LL --> RR --> L00 . 5100 . 5100 . 5101201201200 . 61 . 200 . 61 . 200 . 61 . 200 . 61 . 200 . 61 . 200 . 61 . 201200 . 61 . 200 . 61 . 200 . 61 . 200 . 61 . 200 . 61 . 20120120120122 0 4 0 6 0 8 0 1 0 00122 0 4 0 6 0 8 0 1 0 00122 0 4 0 6 0 8 0 1 0 039   Figure 9 - EMG and kinematic response to resistance Average group EMG trace for BF and kinematic response seen in the hip and knee for both R?L and L?R groups. The black trace represents the average joint angle trace for the last 20 steps of the corresponding trial, while the light grey trace shows the average joint angle trace during the baseline unipedal trial.  In the R?L group, the Resist1 and Resist2 trials are performed with the right leg and the Transfer trial is performed with the left leg. The opposite is true for the L?R group. All EMG data is plotted in normalized units and all kinematic data is plotted in degrees.  The response to resistance during unipedal walking was primarily characterized by a change in hip and knee joint kinematics and EMG activity patterns during swing phase. Figures 8A and 8B show the response to resistance during the unipedal walking trials. RF activity showed an increase throughout the trials against resistance. There was an increase in the activity seen when comparing the last 20 steps against resistance to the baseline activity seen in the RF. RESIST 1 RESIST 2 TRANSFERHIPBaseline UnipedalAverage Last 20 stepsR --> LL --> R- 2 002 04 0- 2 002 04 0- 2 002 04 002 04 06 002 04 06 0- 2 002 04 0- 2 002 04 0- 2 002 04 002 04 06 02 0 4 0 6 0 8 0 1 0 0% o f G a i t C y c l e02 04 06 02 0 4 0 6 0 8 0 1 0 0% o f G a i t C y c l e02 04 06 02 0 4 0 6 0 8 0 1 0 0% o f G a i t C y c l e02 04 06 0KNEER --> LL --> RBFR --> LL --> R01201201201201201240  There was a decrease in MG activity when comparing the baseline activity to the last 20 steps against resistance. There was an increase in BF activity when comparing baseline activity to the last 20 steps against resistance. There was a change in TA activity timing, which can be seen when comparing the last 20 steps against resistance to baseline activity. There was no visible change in SOL activity between baseline and the last 20 steps against resistance.  Peak knee joint angle decreases in response to the resistance applied during the perturbation trials. But there is a large decrease in the knee joint trajectory by the last 20 steps against resistance characterized by less extension during stance phase and a decrease in peak joint angle. The range of motion observed by the hip changes between baseline movement and the last 20 steps against resistance, characterized by an increase in hip extension with movement against resistance. Figure 9 illustrates examples of the step by step change in kinematics (peak knee flexion) and EMG activity (RF EMG amplitude) during the steps against resistance. The rate of adaptation to resistance was calculated for hip and knee flexion as well as EMG amplitude during swing for Resist1, Resist2 and the Transfer trial for comparison across trials (Table 4).  41   Figure 10 - Step by step change in Rectus Femoris activity and peak knee angle The plots above are step by step adaptation changes for averaged group data showing the change in EMG activity for RF and peak knee angle. Grey line represents normalized average for RF activity and peak knee angle during baseline unipedal walking. Data has been normalized to baseline bipedal activity and angle. Data is presented and a percent change from baseline, with a score 1 equal to activity/angle during baseline bipedal walking.   There were no significant differences for the main effects of group and trial and no interaction effect between group and trial in any of EMG amplitude or kinematic parameters, except for TA and SOL. For TA EMG amplitude, the adaptation rate in the R?L group was significantly faster than that in the L?R group (F(1,18) = 6.727, p = 0.018, Fig. 10A) but there were no differences across trials (F(2,36) = 0.964, p = 0.391) and no interaction effects (F(2,36) = 1.936, p = 0.159). For the SOL, the adaptation rate in the R?L group was similarly faster than that in the L?R group (F(1,18) = 4.787, p = 0.042, Fig. 10B), but there were no differences across trials (F(2,36) = 0.521, p = 0.598) and no interaction effects (F(2,36) = 0.307, p = 0.737). RESIST 1 RESIST 2 TRANSFERR --> LL --> RRFR --> LL --> RKNEE#  o f  S t e p s #  o f  S t e p s #  o f  S t e p s00 . 5100 . 5100 . 5100 . 5100 . 5100 . 5100 . 40 . 81 . 200 . 40 . 81 . 200 . 40 . 81 . 24 0 8 0 1 2 0 1 6 0 2 0 0 4 0 8 0 1 2 0 1 6 0 2 0 0 4 0 8 0 1 2 0 1 6 0 2 0 000 . 40 . 81 . 200 . 40 . 81 . 200 . 40 . 81 . 2Normalized EMG amplitudePeak knee flexion angle(normalized)42   Figure 11 - Group EMG adaptation rates The number of steps to reach 63.2% of steady state for EMG activity in the TA and SOL for R?L group (grey diamond) and L?R group (black diamond).  Statistical differences between group adaptation rates were found, with the R?L group showing an increased rate of adaptation when compared to the L?R group for both muscles.               A. TAB. SOL-40-20020406080100120140160Resist 1 Resist 2 Transfer#ofsteps toreachsteadystate-60-40-20020406080100120140160Resist 1 Resist 2 Transfer#ofsteps toreachsteadystateR-->LgroupL-->Rgroup43    RESIST 1 RESIST 2 TRANSFER TA    R ?L  20.08 (34.22) 11.14 (11.60) 11.66 (15.17) L ? R  22.12 (19.58) 40.09 (50.37) 66.11 (79.09) SOL    R ?L  8.95 (8.51) 12.54 (16.59) 8.41 (7.67) L ? R  45.07 (73.72) 54.2 (80.74) 26.69 (61.98) MG    R ?L  22.20 (34.10) 23.53 (62.12) 43.22 (82.80) L ? R  51.54 (79.65) 11.98 (18.41) 31.01 (61.52) RF    R ?L  16.00 (14.54) 6.80 (4.96) 28.78 (60.88) L ? R  9.91 (7.974) 11.98 (18.41) 31.01 (61.52) BF    R ?L  29.39 (61.11) 16.33 (13.12) 32.41 (52.04) L ? R  60.14 (85.02) 28.86 (60.47) 41.78 (65.93) HIP    R ?L  21.71 (52.94) 7.31 (10.59) 26.86 (58.79) L ? R  13.89 (23.69) 31.35 (60.91) 65.32 (93.61) KNEE    R ?L  29.48 (35.90) 10.79 (11.34) 15.44 (13.18) L ? R  39.77 (57.91) 30.86 (32.47) 34.04 (61.11) STRIDE LENGTH    R ? L  78.63 (86.04) 40.38 (58.39) 81.72 (87.47) L ? R 49.38 (60.60) 51.41 (57.66) 108.76 (82.97)  Table 4 - Average adaptation rates for descriptive measures Average number of steps required to reach 63.2% of steady state in each of the EMG and kinematic variables. Values in parentheses represent standard deviation.     44  Discussion In this study, we examined whether adaptations to Lokomat-applied resistance in one leg transferred to the opposite leg. Interlimb transfer was evaluated in accordance with the concepts put forth by the dynamic dominance hypothesis, which postulates that transfer of learning occurs asymmetrically depending on limb dominance. Information about end point position is preferentially coded by the non-dominant limb while information about movement trajectory preferentially transfers to the dominant limb. Here we tested the predictions of the dynamic dominance hypothesis during a locomotor task. Subjects performed a heel targeting task, which required them to heel strike on a target location with each step. The distance from the target at heel strike and the initial foot trajectory slope at the beginning of swing were used to track adaptation and transfer. Adaptation to resistance was evidenced by changes in EMG activity and kinematic patterns in the lower limbs. However, there were no significant differences in the adaptation rates of end point error and foot trajectory slope between the R?L and L?R group and there was also no evidence of transfer of EMG or kinematic adaptations to resistance. These results suggest that transfer did not occur from the trained leg to the untrained leg in either the R?L or L?R directions.  Methodological considerations    We used a unipedal walking paradigm in order to better isolate and test adaptations in one leg at a time. There was no evidence of transfer in our experiment, which raises the question as to whether a unipedal walking paradigm was appropriate to assess interlimb transfer during locomotion. However, previous research has shown that when a resistance is applied unilaterally during bipedal walking, the opposite leg also shows altered kinematic patterns and EMG activity that persisted even after the resistance was removed (Savin et al., 2010). We therefore chose to 45  use a unipedal walking paradigm because this evidence of adaptation in the non-resisted leg could have been a potential confounding factor when evaluating interlimb transfer during locomotion. We did observe some differences in EMG activity patterns between unipedal and bipedal walking. Most notably, BF activity during unipedal walking was much larger than that during bipedal walking, especially during stance. This is likely due to an increased demand to extend the hip during unipedal walking. The platform used to create the unipedal paradigm raised the stationary leg slightly higher off the treadmill compared to the moving leg. This could require the BF to recruit a greater number of muscle fibers in order to aid in hip extension to successfully propel the body forward during unipedal walking (Silverman et al., 2008). There was also decreased activation in MG during stance phase, which would have reduced the propulsion from the plantarflexors and added to the demand on BF to initiate swing. However, even though there were changes in activity, there was no discernible difference in hip or knee kinematic patterns between unipedal and bipedal walking. Source estimation model of learning and the dynamic dominance hypothesis  Understanding how movements can generalize across environments, tasks, and limbs is often considered within the framework of internal models (Bhatt and Pai, 2009; Morton et al., 2001; Wang and Sainburg, 2002; Stockel and Wang, 2011; van Hedel et al., 2002; Alexander et al., 2012; Lam and Dietz, 2004).  This framework uses the assumption that updates to a perturbation occur in mutually exclusive intrinsic or extrinsic frames of reference (Berniker and Kording, 2008). This idea of adaptation may not be entirely correct, however, because there are both internal and external forces placed on the body during all movements. The source estimation model uses the same general principles as the internal model framework, except that it 46  postulates that the central nervous system not only estimates the required corrections, but also takes into account the source of errors and weights the amount of error accordingly (Berniker and Kording, 2008; Berniker and Kording, 2011). This model helps to explain generalization through the idea that errors attributed to the world would generalize better to other contexts whereas error attributed to limb properties are independent of the specific movement apparatus (Berniker and Kording, 2008). If the source of the error is estimated to originate from the environment, then generalization should occur across limbs as the motor commands are updated to account for environmental factors (Berniker and Kording., 2008). One possibility in our experimental paradigm is that movement errors created by walking while placed in the Lokomat were associated with an internal source of error. Since the legs of the robot fit to each individual, the potential for the error to have a greater weighting to the internal portion of movement are maximized. The custom fit could have biased the CNS to associate the resistance occurring from an internal source, and not the environment which movement is occurring in. Considering that the nervous system weights the sources of error, the source-estimation model may help explain why there was no evidence of transfer like that postulated in the dynamic dominance hypothesis. Consolidation of motor adaptations  The CNS can be considered in a hierarchy, with learning and control at the top, adaptation and modification of movement in the middle, and the lowest level involving spinal control and reflexive modifications to movement (Huang and Krakauer, 2009; Kahle, 2005). Thus, there are multiple pathways and limb configurations that can be used to complete a task. Adaptation can become a hindrance to learning when there is an inability to reach the highest levels of the CNS hierarchy, therefore limiting the learning of the new movement and generalization to new environments (Huang and Krakauer, 2009). For learning (and therefore 47  generalization) to occur, modification of the motor output must occur at the highest level of that skill, which requires extensive practice (Kahle, 2005; Huang and Krakauer, 2009). Consolidation of a motor memory requires sufficient practice on an initial exposure to be resistant to interference. The consolidation of anticipatory models creates the presence of savings, or a faster rate of adaptation found in a subsequent exposure or after transfer (Krakauer and Shadmehr, 2006). Malone et al., (2011) found that constant practice in adaptation/washout split belt treadmill practice paradigm resulted in retention of the adapted walking that was present even the next day. Subjects were also perturbed with opposite belt speeds on the second day and were found to adapt at a faster rate to the previous day walking pattern (Malone et al., 2011).  Consolidation of the motor output from the previous day led to a robust motor program that was resilient when perturbed and gives evidence to the use of sufficient practice. Our results show no evidence for savings when testing the trained leg for two separate bouts against resistance. The lack of consolidation could have resulted from the lack of practice in a null environment for unipedal walking. In our experiment, subjects had 50 steps of practice in unipedal walking for each leg, which may not have been enough practice to enable longer lasting changes during the baseline unipedal walking. Indeed, the kinematic adaptation rates calculated during unipedal walking in the null environment show that subjects did not reach a 95% steady state of adaptation in the amount of time provided for practice. Thus, it is possible that there was not enough practice in the unipedal walking environment to reach the highest levels of the CNS, and that unipedal walking in the Lokomat was not consolidated. There is data supporting the idea that adaptation could be found within the cerebellum, which could be thought of as the middle of the CNS hierarchy (Bastian, 2008; Reisman et al., 2005; Jayaram et al., 2011; Jayaram et al., 2012) while retention of adapted programs could be 48  found in the primary motor cortex (M1), which could be thought of as the highest level of the CNS hierarchy (Jayaram et al., 2011; Galea et al., 2011; Riek et al., 2012; Krakauer and Shadmehr, 2006). When there is sufficient practice, the updated motor program is transferred, where it can be retained and retrieved for future use (Shadmehr and Brashers-krug, 1997; Galea et al., 2011). In our experiment, we may have perturbed the movement (added resistance) prior to the underlying motor command (unipedal walking with the Lokomat) becoming consolidated within the CNS, as evidenced by the adaptation rates calculated during the baseline unipedal walking trials. Although we may consider that unipedal walking is an inherent component of locomotor circuitry (Choi and Bastian, 2007), recent evidence suggests that walking in the Lokomat ?null? environment still requires an update to the motor program to adjust to the inertial properties of the robotic legs (Zabukovec et al., 2013). The lack of transfer we observed here could have been a result of insufficient amount of practice to allow for savings of the updated motor output for both bipedal and unipedal walking. This potentially resulted in a motor output that was constantly changing to accommodate the environmental changes without a prior model (baseline unipedal walking) that had been stabilized for walking with the Lokomat. This could result in abnormally long adaptation periods to reach a steady state or never truly updating a motor command to compensate for the perturbation. Where could transfer be occurring?  The control and adaptation of locomotor patterns involves multiple levels of the CNS and there is evidence that different levels of the CNS have differing capacity for mediating interlimb transfer. Since transfer was not evident in our experiment, understanding how the different levels of the CNS interact is an important question. 49  The first level of the hierarchy is considered the CPG and spinal cord. Studies utilizing split-belt treadmill paradigm provide evidence that adaptive strategies likely mediated at the level of the spinal cord circuitry are stored independently and do not transfer. Choi and Bastian (2007) had subjects walk with one leg walking on a forward moving belt and the opposite leg walking on a backwards belt at different step rates between the legs. After effects showed that walking adaptations were stored independently and did not transfer across directions. There is also evidence for different control networks for slow and fast walking (Vasudevan and Bastian, 2010). There has also been evidence showing that there is limited transfer of proprioceptive information at the level of the spinal cord between the hindlimbs in animal models (Lavrov et al., 2008). This evidence support the idea that there would be limited transfer of adaptations between the lower limbs at this level of the CNS.  The cerebellum and other mid-brain structures used in the adaptation of movement are considered to lie in the second level of the CNS hierarchy (Nowak et al., 2005; Nowak et al., 2009). There is evidence that suggests that sensory information entering the cerebellum appears to generalize between upper limbs during reaching tasks (Nowak et al., 2005; Novak et al., 2009). Since it is thought there are separate CPGs for control of each lower limb, it suggests that the cerebellum has separate projections from each hemisphere to the corresponding CPG (Reisman et al., 2005; Morton and Bastian, 2006). The evidence for separate projections from the cerebellum come from the inability of cerebellar patients to adapt interlimb coordinated portions of locomotion, which include step length, percent of double support time and limb orientation during weight transfer, while walking on a split belt treadmill (Reisman et al., 2006; Morton and Bastian, 2006). This evidence points to the importance of the cerebellum in the transfer of adaptation between the limbs. 50  The motor cortex and other cortical structures are considered the highest level of the CNS hierarchy. There is evidence of transfer between the lower limbs when the experimental task directly involved conscious control for the completion of the movement (and therefore involvement of cortical structures) (van Hedel et al., 2002; Stockel and Wang, 2011). There is evidence for interlimb transfer during both continuous locomotor patterns (van Hedel et al., 2002) and during discrete tasks involving the lower limbs (Stockel and Wang, 2011). The motor cortex has bilateral projections to the lower limbs through the corticospinal tract, which could potentially mediate the transfer of adaptation between the lower limbs. This evidence supports the importance of cortical areas in the interlimb transfer of adapted movements.  In our experiment, it appears that there was adaptation to the resistance placed on the lower limbs as evidenced by the progressive changes in RF activity and peak knee flexion angle during the resistance trials. We also assume access to the highest level of the CNS through the use of the targeting task. However, we found no evidence of interlimb transfer here. One possibility is that there was a weighting emphasizing the production of cyclic flexion/extension pattern through the CPG during locomotion. Priority on the maintenance of this cyclic pattern rather than the sharing of resources from higher centers in CNS hierarchy could have precluded interlimb transfer in this situation. Future studies should be designed to understand this possibility. Perhaps if the lower limbs are placed in a protocol involving a discrete reaching or pointing task in a perturbing environment (similar to those seen in the upper limbs) there would be a different result compared to the ones found during locomotion. The difference of transfer between a discrete and continuous motion in a perturbing environment could lead to a better understanding of the function of the CPG during lower leg movements. 51  Interlimb transfer during locomotion has only been shown in one study, where subjects had to learn to perform obstacle crossing with a specific foot height over the obstacle (van Hedel et al., 2002). A feature of this task was that it would have engaged higher levels of locomotor control. In other studies where no transfer was evident (Houldin et al., 2012; Choi and Bastian, 2007), the adaptations may have been limited to the lower levels of the CNS hierarchy that could potentially not allow for transfer. For example, adaptations to split belt treadmills mainly require alterations in the phasing of interlimb coordination, which could mainly be mediated by spinal circuitry. In our experiment, the adaptations to the Lokomat?s inertial properties could also have been mediated mainly at the lower levels of the CNS, and therefore may have precluded interlimb transfer.  Another consideration when comparing the results to the obstacle crossing study (van Hedel et al., 2002) vs. our studies and the split belt paradigm (Choi and Bastain, 2007) is that the obstacle crossing task did not require an adaptation to a new sensory environment, but rather focused on the production of an explicit motor control task. This could potentially shift emphasis away from the lower two levels of the motor control hierarchy and allow for greater focus on explicit optimization of the locomotor task (control of foot height), requiring cortical areas. In other experimental paradigms requiring adaptation to a sensory perturbation (i.e., forces applied to the limb, interlimb coordination), the emphasis may have been more heavily weighted towards processing at the lower levels of the motor control hierarchy (e.g., locomotor CPG circuitry). If the adaptive mechanisms to split belt treadmill walking and velocity-dependent forces are restricted to the lower levels of the CNS, then we may expect limited interlimb generalization of new sensory environments.   52  Awareness and generalization  Awareness of the perturbation was thought to be an important factor when determining the presence of adaptation transfer between limbs. Malfait and Ostry (2004) found limited transfer between upper limbs in a reaching task when the perturbation was introduced gradually. Transfer was found when the perturbation was introduced abruptly, with the thought that conscious awareness of the perturbation allowed for the transfer (Malfait and Ostry, 2004). There is evidence in the literature suggesting that conscious thought is not needed for transfer, and there is no difference between gradual and abrupt introductions of a perturbation when assessing transfer (Wang et al., 2011, Saijo and Gomi, 2010; Taylor et al., 2011). Wang et al., (2011) used a visuo-motor rotation to perturb a reaching movement to a target. It was found there was no difference in the adaptation rates of the second limb when the exposure to the rotation was different between the abrupt, gradual and informed rotation groups. There is also evidence suggesting that declarative knowledge of the perturbation does not predict an increase in transfer and potentially predicts a strategic control to counteract the perturbation rather than a recalibration of the model for movement (Werner and Bock, 2007). This evidence suggests that the conscious knowledge of the performance may not be required to the extent used during our experiment. Subjects were given knowledge of their performance after each step, which may have interfered with the robust consolidation of an updated motor command for movement during the perturbation. With less frequent knowledge of performance subjects may have produced a more robust adapted pattern that would allow for access during transfer.  53  Conclusions  The results of this study indicate that there is no interlimb transfer of locomotor adaptations to a velocity-dependent force field. Although other studies have shown generalization between tasks and environment in the lower limbs, there appears to be limited transfer of motor adaptations between the legs during walking. It is possible that we did not observe transfer because the baseline motor pattern for unipedal walking with the Lokomat was not completely consolidated. Another possibility for the lack of transfer could be from the rhythmic activity produced by the CPG, which could have precluded interlimb transfer mediated by higher levels of the CNS. It is possible that interlimb transfer is limited to adaptations emphasizing the highest levels of the motor control hierarchy, as exemplified by studies that have used discrete, goal-directed movements. Future studies should be designed to address these issues in order to understand the features driving interlimb transfer of adaptations.  54  References Adkins, R.J., Cegnar, M.R., and Rafuse, D.D. (1971). Differential effects of lesions of the anterior and posterior sigmoid gyri in cats. Brain Research. 30: 411-414. Alahyane, N., Fonteille, V., Urquizer, C., Salemme, R., Nighoghossian, N., Pelisson, D., and Tilikete, C. (2008). Separate neural substrates in the human cerebellum for sensory-motor adaptation of reactive and of scanning voluntary saccades. Cerebellum. 7: 595-601. Alexander, M.S., Flodin, B.W.G., and Marigold, D.S. (2013). Changes is task parameters during walking prism adaptation influence the subsequent generalization pattern. Journal of Neurophysiology. 109: 2495-2504. Andersson, O., Forssberg, H., Grillner, S., and Wallen, P. (1981). Peripheral feedback mechanisms acting on the central pattern generators for locomotion in fish and cats. Canadian Journal of Physiology and Pharmacology. 59: 713-727. Bastian, A.J. (2006). Learning to predict the future: The cerebellum adapts feedforward movement control. Current Opinion in Neurobiology. 16: 645-649. Bastian, A.J. (2008). Understanding sensorimotor adaptation and learning for rehabilitation. Current Opinion in Neurology. 21(6): 628-633. Bastian, A.J. (2011). Moving, sensing and learning with cerebellar damage. Current Opinion in Neurobiology. 21: 596-601. Beloozerova, I.N., Farrell, B.J., Sirota, M.G., and Prilutsky, B.I. (2010). Differences in movement mechanics, electromyographic, and motor  cortex activity between accurate and nonaccurate stepping. Journal of Neurophysiology. 103: 2285-2300. Beloozerova, I.N., and Sirota, M.G. (1993). The role of the motor cortex in the control of accuracy of locomotor movements in the cat. Journal of Physiology. 461: 1-25. 55  Beloozerova, I.N., and Sirota, M.G. (1993). The role of the motor cortex in the control of accuracy of locomotor movements in the cat. Journal of Physiology. 461: 27-46. Berniker, M., and Kording, K. (2008). Estimating the sources of motor errors for adaptation and generalization. Nature Neuroscience. 11(12): 1454-1461. Berniker, M., and Kording, K. (2011). Estimating the relevance of world disturbances to explain savings, interference and long-term motor adaptation effects. PLoS Computational Biology. 7(10): 1-12. Bhatt, T., and Pai, Y.-C. (2009). Generalization of gait adaptation for fall prevention: From moveable platform to slippery floor. Journal of Neurophysiology. 101: 948-957. Blanchette, A., and Bouyer, L.J. (2009). Timing-specific transfer of adapted muscle active after walking in an elastic force field. Journal of Neurophysiology. 102: 568-577. Bolliger, M., Banz, R., Dietz, V., and Lunenburger, L. (2008). Standardized voluntary force measurement in a lower extremity rehabilitation robot. Journal of Neuroengineering and Rehabilitation. 5: 23. Chapman, J.P., Chapman, L.J., and Allen, J.J. (1987). The measurement of foot preference. Neuropsychologia. 25(3): 579-584. Conway, B.A., Hultborn, H., and Kiehn, O. (1987). Proprioceptive input resets central locomotor rhythm in the spinal cat. Experimental Brain Research. 68: 643-656. Criscimagna-Hemminger, S.E., Donchin, O., Gassangia, M.S., and Shadmehr, R. (2003). Learned dynamics of reaching movements generalize from dominant to nondominant arm. Journal of Neurophysiology. 89: 168-176. 56  Donovan, J.J., and Radosevich, D.J. (1999). A meta-analytic review of the distribution of practice effect: now you see it, now you don?t. Journal of Applied Psychology. 84(5): 795-805. Drew, T. (1993). Motor cortical activity during voluntary gait modifications in the cat I: cells related to the forelimbs. Journal of Neurophysiology. 70(1): 179-199. Drew, T., Jiang, W., Kably, B., and Lavoie, S. (1996). Role of the motor cortex in the control of visually triggered gait modifications. Canadian Journal of Physiology and Pharmacology. 74: 426-442. Dubrovsky, B., Garcia-rill, E., and Surkes, M.A. (1974). Effects of descrete precruciate cortex lesions on motor behavior. Brain Research. 82: 328-333. Duysens, J., and Pearson, K.G. (1980). Inhibition of flexor burst generation by loading ankle extensor muscles in walking cats. Brain Research. 187: 321-332. Eidelberg, E., and Yu, J. (1981). Effects of corticospinal lesions upon treadmill locomotion in cats. Experimental Brain Research. 43: 101-103. Emken, J.L., and Reinkensmeyer, D.J. (2005). Robot-enhanced motor learning: Accelerating internal model formation during locomotion by transient dynamic amplification. IEEE Transactions on Neural systems and Rehabilitation Engineering. 13(1): 33-39. Fahle, M. (2005). Perceptual learning: specificity versus generalization. Current Opinion in Neurobiology. 15: 154-160. Ferris, D.P., Gordon, K.E., Beres-Jones, J.A., and Harkema, S.J. (2004). Muscle activation during unilateral stepping occurs in the nonstepping limb of humans with clinically complete spinal cord injury. Spinal Cord. 42: 14-23. 57  Forssberg, H., Grillner, S., and Rossignol, S. (1975). Phase dependent reflex reversal during walking in chronic spinal cats. Brain Research. 85: 103-107. Fortin, K., Blanchette, A., McFadyen, B.J., and Bouyer, L.J. (2009). Effects of walking in a force field for varying durations on aftereffects and on next day performance. Experimental Brain Research. 199: 145-155. Fuentes, C.T., and Bastian, A.J. (2007). Motor cognition ? what is it and is the cerebellum involved. The Cerebellum. 6: 232-236. Galea, J.M., Miall, R.C., and Woolley, D.G. (2007). Asymmetric interlimb transfer of concurrent adaptation to opposing dynamic forces. Experimental Brain Research. 182: 267-273. Galea, J.M., Vazquez, A., Pasricha, N., Orban de Xivry, J-J., and Celnik, P. (2011). Dissociating roles of the cerebellum and motor cortex during adaptive learning: the motor cortex retains what the cerebellum learns. Cerebral Cortex. 21: 1761-1770. Georgopoulos, A.P., and Grillner, S. (1989). Visuomotor coordination in reaching and locomotion. Science. 245(4923): 1209-1210 Grillner, S., and Rossignol, S. (1978). On the initiation of the swing phase of locomotion in chronic spinal cats. Brain Research. 146: 269-277. Grillner, S., and Zangger, P. (1979). On the central pattern generation of locomotion in the low spinal cat. Experimental Brain Research. 34: 241-261. Hiebert, G.W., Whelan, P.J., Prochazka, A., and Pearson, K.G. (1996). Contribution of hind limb flexor muscle afferents to the timing of phase transitions in the cat step cycle. Journal of Neurophysiology. 75: 1126-1137. 58  Hiebert, G.W., and Pearson, K.G. (1999). Contribution of sensory feedback to the generation of extensor activity during walking in the decerebrate cat. Journal of Neurophysiology. 81: 758-770. Houldin, A., Luttin, K., and Lam, T. (2011). Locomotor adaptations and aftereffects to resistance during walking in individuals with spinal cord injury. Journal of Neurophysiology. 106: 247-258. Houldin, A., Chua, R., Carpenter, M.G., and Lam, T. (2012). Limited interlimb transfer of locomotor adaptations to a velocity-dependent force field during unipedal walking. Journal of Neurophysiology. 108: 943-952. Huang, V.S., and Krakauer, J.W. (2009). Robotic neurorehabilitation: a computational motor learning perspective. Journal of NeuroEngineering and Rehabilitation. 6(5): 1-13. Jayaram, G., Galea, J.M., Bastian, A.J., and Celnik, P. (2011). Human locomotor adaptive learning is proportional to depression of cerebellar excitability. Cerebral Cortex. 21: 1901-1909. Jayaram, G., Tang, B., Pallegadda, R., Vasudevan, E.V.L., Celnik, P., and Bastian, A.J. (2012). Modulating locomotor adaptation with cerebellar stimulation. Journal of Neurophysiology. 107: 2950-2957. Kawato, M. (1999). Internal models for motor control and trajectory planning. Current Opinion in Neurobiology. 9: 718-727. Kording, K.P., Tenenbaum, J.B., and Shadmehr, S. (2007). The dynamics of memory as a consequence of optimal adaptation to a changing body. Nature Neuroscience. 10(6): 779-786. 59  Krakauer, J.W., Ghilardi, M-F, and Ghez, C. (1999). Independent learning of internal models for kinematic and dynamic control of reaching. Nature Neuroscience. 2(11): 1026-1031. Krakauer, J.W., Ghez, C., and Ghilardi, M-F. (2005). Adaptation to visuomotor transformations: consolidation, interference, and forgetting. Journal of Neuroscience. 25(2): 473-478. Krakauer, J.W., and Shadmehr, R. (2006). Consolidation of motor memory. Trends in Neurosciences. 29(1): 58-64. Krakauer, J.W., Mazzoni, P., Ghazizadeh, A., Ravindran, R., and Shadmehr, R. (2006). Generalization of motor learning depends on the history of prior action. PLoS Biology. 4(10): 1798-1808. Lam, T., Anderschitz, M., and Dietz, V. (2006). Contribution of feedback and feedforward strategies to locomotor adaptations. Journal of Neurophysiology. 95: 766-773. Lam, T., and Dietz, V. (2004). Transfer of motor performance in an obstacle avoidance task to different walking conditions. Journal of Neurophysiology. 92: 2010-2016. Lam, T., and Pearson, K. (2001). Proprioceptive modulation of hip flexor activity during the swing phase of locomotion in decerebrate cats. Journal of Neurophysiology. 86: 1321-1332. Lam, T., and Pearson, K. (2002). The role of proprioceptive feedback in the regulation and adaptation of locomotor activity. Sensorimotor Control of Movement and Posture. 40: 343-355. Lam, T., Wolstenholme, C., van der Linden, M., Pang, M.Y.C., and Yang, J.F. (2003). Stumbling corrective responses during treadmill-elicited stepping in human infants. Journal of Physiology. 533.1: 319-331. 60  Lavrov, I., Courtine, G., Dy, C.J., van de Brand, R., Fong, A.J., Gerasimenko, Y., Zhong, H., Roy, R.R., and Edgerton, V.R. (2008). Facilitation of stepping with epidural stimulation in spinal rats: Role of sensory input. Journal of Neuroscience. 28(31): 7774-7780. Malfait, N., and Ostry, D.J. (2004). Is interlimb transfer of force-field adaptation a cognitive response to the sudden introduction of load? Journal of Neuroscience. 24(37): 8084-8089. Malone, L.A., Vasudevan, E.V.L., and Bastian, A.J. (2011). Motor adaptation training for faster relearning. Journal of Neuroscience. 31(42): 15136-15143. Mattar, A.A.G., and Ostry, D.J. (2007). Modifiability of generalization in dynamics learning. Journal of Neurophysiology. 98: 3321-3329. Martin, T.A., Keating, J.G., Goodkin, H.P., Bastian, A.J., and Thach, W.T. (1996). Throwing while looking through prisms II: Specificity and storage of multiple gaze-throw calibrations. Brain. 119: 1199-1211. McVea, D.A., Donelan, J.M., Tachibana, A., and Pearson, K.G. (2005). A role for hip position in initiating the swing-to-stance transition in walking cats. Journal of Neurophysiology. 94: 3497-3508. McVea, D.A., and Pearson, K.G. (2007). Long-lasting, context-dependent modification of stepping in the cat after repeated stumbling-corrective responses. Journal of Neurophysiology. 97: 659-669. Morton, S.M., and Bastian, A.J. (2004). Cerebellar control of balance and locomotion. Neuroscientist. 10: 247-259. Morton, S.M., and Bastian, A.J. (2006). Cerebellar contributions to locomotor adaptations during splitbelt treadmill walking. Journal of Neuroscience. 26(36): 9107-9116. 61  Morton, S.M., Lang, C.E., and Bastian, A.J. (2001). Inter- and intra-limb generalization of adaptation during catching. Experimental Brain Research. 141: 438-445. Noel, M., Fortin, K., and Bouyer, L.J. (2009). Using an electrohydraulic ankle foot orthosis to study modifications in feedforward control during locomotor adaptation to force fields applied in stance. Journal of Neuroengineering and Rehabilitation. 6: 16. Nowak, D.A., Hermsdorfer, J., Timmann, D., Rost, K., and Topka, H. (2005). Impaired generalization of weight-related information during grasping in cerebellar degeneration. Neuropsychologia. 43: 20-27. Nowak, D.A., Hufnagel, A., Ameli, M., Timmann, D., and Hermsdorfer, J. (2009). Interhemispheric transfer of predictive force control during grasping in cerebellar disorders. Cerebellum. 8: 108-115. Oldfield, R.C. (1971). The assessment and analysis of handedness: The edinburgh inventory. Neuropsychologia. 9: 97-113. Orban de Xivry, J-J., Marko, M.K., Pekny, S.E., Pastor, D., Izawa, J., Celnik, P., and Shadmehr, R. (2011). Stimulation of the human motor cortex alters generalization patterns of motor learning. Journal of Neuroscience. 31(19): 7102-7110. Przybyla, A., Good., D.C., and Sainburg, R.L. (2012). Dynamic dominance varies with handedness: Reduced interlimb asymmetries in left handers. Experimental Brain Research. 216: 419-431.  Riek, S., Hinder, M.R., and Carson, R.G. (2012). Primary motor cortex involvement in initial learning during visuomotor adaptation. Neuropsychologia. 50: 2515-2523. Reisman, D.S., Block, H.J., and Bastian, A.J.(2005). Interlimb coordination during locomotion: What can be adapted and stored? Journal of Neurophysiology. 94: 2403-2415. 62  Rochet-Capellan, A., Richer, L., and Ostry, D.J. (2012). Nonhomogeneous transfer reveals specificity in speech motor learning. Journal of Neurophysiology. 107: 1711-1717. Rossignol, S., Bouyer, L., Barthelemy, D., Langlet, C., and Leblond, H. (2002). Recovery of locomotion in the cat following spinal cord lesions. Brain Research Review. 40: 257-266. Saijo, N., and Gomi, H. (2010). Multiple motor strategies in visuomotor rotation. PLoS one.  5(2): 1-11.  Sainburg, R.L. (2002). Evidence for a dynamic-dominance hypothesis of handedness. Experimental Brain Research. 142: 241-258.  Sainburg, R.L., and Kalakanis, D. (2000). Difference in control of limb dynamics during dominant and nondominant arm reaching. Journal of Neurophysiology. 83: 2661-2675. Sainburg, R.L., and Wang, J. (2002). Interlimb transfer of visuomotor rotations: Independence of direction and final position information. Experimental Brain Research. 145: 437-447. Savin, D.N., Tseng, S-C., and Morton, S.M. (2010). Bilateral adaptation during locomotion following a unilaterally applied resistance to swing in nondisabled adults. Journal of Neurophysiology. 104: 3600-3611. Seidler, R.D. (2004). Multiple motor learning experiences enhance motor adaptability. Journal of Cognitive Neuroscience. 16(1): 65-73. Shadmehr, R., and Brashers-krug, T. (1997). Functional stages in the formation of human long-term motor memory. Journal of Neuroscience. 17(1): 409-419. Shadmehr, R., and Krakauer, J.W. (2008). A computational neuroanatomy for motor control. Experimental Brain Research. 185(3): 359-381. Shadmehr, R., and Mussa-Ivaldi, F.A. (1994). Adaptative representation of dynamics during learning of a motor task. Journal of Neuroscience. 14(5): 3208-3224.  63  Silverman, A.K., Fey, N.P., Portillo, A., Walden, J.G., Bosker, G., and Neptune, R. R. (2008). Compensatory mechanisms in below-knee amputee gait in response to increasing steady-state walking speed. Gait & Posture. 28: 602-609. Stephens, M.J., and Yang, J.F. (1999). Loading during the stance phase of walking in humans increases the extensor EMG amplitude but does not change the duration of the step cycle. Experimental Brain Research. 124: 363-370. Stockel, T., and Wang, J. (2011). Transfer of short-term motor learning across the lower limbs as a function of task conception and practice order. Brain and Cognition. 77: 271-279. Taylor, J.A., Wojaczynski, G.J., and Ivry, R.B. (2011). Trial-by-trial analysis of intermanual transfer during visuomotor adaptation. Journal of Neurophysiology. 106: 3157-3172. Tomlinson, T., and Sainburg, R.L. (2012). Dynamic dominance persists during unsupported reaching. Journal of Motor Behaviour. 44(1): 13-25. Torres-Oviedo, G., and Bastian, A.J. (2012). Natural error patterns enable transfer of motor learning to novel contexts. Journal of Neurophysiology. 107: 346-356. Vangheluwe, S., Puttemans, V., Wenderoth, N., Van Baelen, M., and Swinnen, S.P. (2004). Inter- and Intralimb transfer of a bimanual task: Generalisability of limb dissociation. Behavioural Brain Research. 154: 535-547.  van Hedel, H.J.A., Biedermann, M., Erni, T., and Dietz, V. (2002). Obstacle avoidance during human walking: Transfer of motor skill from one leg to the other. Journal of Physiology. 543.2: 709-717. Vasudevan, E.V.L., and Bastian, A.J. (2009). Split-belt treadmill adaptation shows different functional networks for fast and slow human walking. Journal of Neurophysiology. 103: 183-191. 64  Wang, J., Joshi, M., and Lei, Y. (2011). The extent of interlimb transfer following adaptation to a novel visuomotor condition does not depend on awareness of the condition. Journal of Neurophysiolog. 106: 259-264. Wang, T.-Y., Bhatt, T., Yang, F., and Pai, Y.-C. (2011). Generalization of motor adaptation to repeated-slip perturbation across tasks. Neuroscience. 180: 85-95. Werner, S., and Bock, O. (2007) Effects of variable practice and declarative knowledge on sensorimotor adaptation to rotated visual feedback. Experiment Brain Research. 178: 554-559. Wilke, C., Synofzik, M., and Lindner, A. (2013). Sensorimotor recalibration depends on attribution of sensory prediction error to internal causes. PLoS One. 8(1): 1-13. Winter, D.A. (1992). Foot trajectory in human gait: A precise and multifactorial motor control task. Physical Therapy. 72(1): 45-56. Wolpert, D.M., and Ghahramani, Z. (2000). Computational principles for movement neuroscience. Nature Neuroscience Supplement. 3: 1212-1217. Yanagihara, D., and Kondo, I. (1996). Nitric oxide plays a key role in adaptive control of locomotion in cat. Proceedings of the National Science Academy: Neurobiology. 93: 13292-13297. Zabukovec, J.R., Boyd, L.A., Linsdell, M.A., and Lam, T. (2013). Changes in corticospinal excitability following adaptive modification to human walking. Experimental Brain Research. 226: 557-564.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0074231/manifest

Comment

Related Items