UBC Faculty Research and Publications

Motor sequence learning occurs despite disrupted visual and proprioceptive feedback Vidoni, Eric D; Boyd, Lara A Jul 25, 2008

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

Item Metadata


52383-12993_2008_Article_166.pdf [ 525.75kB ]
JSON: 52383-1.0221342.json
JSON-LD: 52383-1.0221342-ld.json
RDF/XML (Pretty): 52383-1.0221342-rdf.xml
RDF/JSON: 52383-1.0221342-rdf.json
Turtle: 52383-1.0221342-turtle.txt
N-Triples: 52383-1.0221342-rdf-ntriples.txt
Original Record: 52383-1.0221342-source.json
Full Text

Full Text

ralssBioMed CentBehavioral and Brain FunctionsOpen AcceResearchMotor sequence learning occurs despite disrupted visual and proprioceptive feedbackEric D Vidoni1 and Lara A Boyd*2,1Address: 1Department of Physical Therapy & Rehabilitation Science, University of Kansas Medical Center Kansas City, KS, USA and 2School of Rehabilitation Sciences, University of British Columbia Vancouver, British Columbia, CanadaEmail: Eric D Vidoni - evidoni@kumc.edu; Lara A Boyd* - lara.boyd@ubc.ca* Corresponding author    AbstractBackground: Recent work has demonstrated the importance of proprioception for thedevelopment of internal representations of the forces encountered during a task. Evidence alsoexists for a significant role for proprioception in the execution of sequential movements. However,little work has explored the role of proprioceptive sensation during the learning of continuousmovement sequences. Here, we report that the repeated segment of a continuous tracking taskcan be learned despite peripherally altered arm proprioception and severely restricted visualfeedback regarding motor output.Methods: Healthy adults practiced a continuous tracking task over 2 days. Half of the participantsexperienced vibration that altered proprioception of shoulder flexion/extension of the activetracking arm (experimental condition) and half experienced vibration of the passive resting arm(control condition). Visual feedback was restricted for all participants. Retention testing wasconducted on a separate day to assess motor learning.Results: Regardless of vibration condition, participants learned the repeated segmentdemonstrated by significant improvements in accuracy for tracking repeated as compared torandom continuous movement sequences.Conclusion: These results suggest that with practice, participants were able to use residualafferent information to overcome initial interference of tracking ability related to alteredproprioception and restricted visual feedback to learn a continuous motor sequence. Motorlearning occurred despite an initial interference of tracking noted during acquisition practice.BackgroundMotor learning requires the ability to adjust future per-formance based on information regarding prior execu-tion. The feedback that is necessary for this process cancome from exogenous sources such as coaching, or endog-tion.[1] Recently, vision and proprioception have receivedconsiderable attention in the literature regarding their roleduring motor learning.[2] The findings suggest that thetwo modalities support learning of different environmen-tal characteristics. For example, it appears that only visionPublished: 25 July 2008Behavioral and Brain Functions 2008, 4:32 doi:10.1186/1744-9081-4-32Received: 29 April 2008Accepted: 25 July 2008This article is available from: http://www.behavioralandbrainfunctions.com/content/4/1/32© 2008 Vidoni and Boyd; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 11(page number not for citation purposes)enous sources such as self-evaluation of performance andafferent feedback, including vision and propriocep-is necessary for adapting to new kinematic environmentssuch as a visuomotor shift.[2] In contrast, learning to con-Behavioral and Brain Functions 2008, 4:32 http://www.behavioralandbrainfunctions.com/content/4/1/32trol novel dynamics such as manipulating a new objectcan be acquired through the proprioceptive systemalone.[3] In particular, previous work has demonstratedthe importance of proprioception in adapting movementcoordination to external forces encountered by the limbduring movement.[4,5]Our understanding of the importance of proprioceptionfor motor sequence learning is limited by two factors.First, past work investigating the interactions among pro-prioception, vision and motor learning has typicallyemployed discrete movements [5-7] that participants arealready familiar with and thus, have an established, well-learned motor plan.[6] Second, research has not carefullydissociated permanent changes in behavior that representmotor learning from short-term performance changes [8]by employing follow-up, retention testing session.[2,7,9]Commonly, participants have been allowed to familiarizethemselves with (i.e., practice) the task prior to data col-lection.[2,5,10] Though this creates a controlled environ-ment for investigating the role of proprioception in short-term motor performance, it confounds our understandingof motor learning by essentially pre-training participantson the task. This issue was inadvertently highlighted byBevan and colleagues [10] who anecdotally reported thatparticipants demonstrated the best performance for thetasks that they initially practiced. Rather than examiningadaptation of familiar movements in novel environ-ments, in the present research we sought to examinemotor sequence learning of an entirely new motor patternunder conditions that altered visual and proprioceptivefeedback.Little prior work has focused on the importance of propri-oception during continuous motor sequence learning.This omission is surprising considering that propriocep-tion has been suggested as an integral component of feed-back-based skill learning.[11,12] During motor learning,proprioceptive feedback may form a template for compar-ison to a motor plan; perhaps through the tuning of elec-tromyographic activity via feedback-basedadaptation.[11] Extending these findings, Hwang andShadmehr noted that computer simulations of musclespindle-based learning closely matched human learningof a reaching task in a force field.[12]We addressed previous experimental gaps by asking par-ticipants to practice a novel, continuous motor patternover the course of 2 days and return on a separate day forretention testing to assess motor learning. We severelyrestricted visual feedback and introduced vibration toalter proprioception. Vibration has previously been usedto alter proprioception during upper extremity move-who cataloged its illusory effects on sensory perception.This effect has been repeated and further qualified[4,5,14-18] making vibration a useful experimental toolfor the study of motor control and learning.[19] Theapplication of vibration to the arm has been used in pastexperiments to determine the role of proprioception inthe execution of previously learned movements [17,18]and the control of the limb against external forces.[4,5]Vibration produces the illusion of movement by stimulat-ing primary spindle fiber afferents.[14] Furthermore,vibration can mask the report of accurate afferent infor-mation regarding antagonist muscle stretch via stimulat-ing primary spindle fiber afferents [15] and influencingmuscle activation patterns.[16] In light of these findings,in Experiment 1, we characterized the impact of vibrationat the shoulder on proprioception using a limb positionmatching task. We confirmed that application of vibrationat the shoulder predictably altered proprioception andcaused participants to misjudge motion in the limb-matching task.After confirming that our method of vibration alters pro-prioception, we used the same method to apply vibrationduring practice of a continuous tracking task and exam-ined the effect of altered limb proprioceptive sensation onmotor sequence learning. We hypothesized that if propri-oception was essential for motor sequence learning, alter-ing this feedback during practice would be evident inaccuracy measures at retention testing. Alternately, it ispossible that the normal motor system is flexible androbust enough to facilitate sequence learning even whenproprioception is altered. The possibility that individualscan demonstrate motor sequence learning even when pro-prioception is altered is important to neuroscientists andclinicians alike. Knowledge of the significance of accurateproprioceptive information during motor sequence learn-ing may facilitate the genesis of novel models that predictthe capacity skill acquisition.Experiment 1MethodsTo confirm the putative effects of vibration in advance ofour investigation of motor learning (Experiment 2), wefirst verified the effect of vibration on both passive andactive continuous, whole-arm movement patterns. Previ-ous studies have applied shoulder muscle vibration dur-ing whole arm movements to examine motorperformance.[18] Because our experimental task requiredshoulder flexion and extension movements to push andpull a frictionless lever mounted at shoulder height, appli-cation of vibration to the proximal arm musculature wasideal. We expected vibration to predictably alter percep-tion of arm movement during limb matching in Experi-Page 2 of 11(page number not for citation purposes)ment. Vibration can predictably shift perception of move-ment as demonstrated by Goodwin and colleagues [13]ment 1.Behavioral and Brain Functions 2008, 4:32 http://www.behavioralandbrainfunctions.com/content/4/1/32Participants and ApparatusFifteen healthy adults (8 males, 7 females; mean age 33.1[range 26–46]) with no reported diabetes or upperextremity sensorimotor impairment participated. Eachprovided informed consent in the manner prescribed bythe University of Kansas Medical Center Humans SubjectsCommittee, in compliance with the Helsinki Declaration.Participants engaged in a previously reported limb posi-tion matching task modified for the upper extremity.[20]The protocol required the participant to continuously esti-mate the movement of a passively displaced extremitywhile experiencing vibration to one of their arms. Wemodified and extended the protocol to include vibrationof both active and passive upper extremities. Two nearlyfrictionless, horizontally mounted levers were positionedat shoulder level to allow participants to grasp one in eachhand (Figures 1a and 1b). The levers were attached topotentiometers that registered angular displacement.Three eccentrically loaded motors within a cuff providedthe vibratory stimulus. The cuff was secured to the domi-nant arm, as identified by the Edinburgh HandednessInventory,[21] with an elastic wrap at the level of the del-toid insertion. The vibrating motors were positioned onthe anterior, lateral and posterior aspects of the upperarm. In this manner aspects of the biceps brachii, tricepsbrachii and deltoids were vibrated.Vibration-induced movement illusion occurs over a widerange of stimulation frequencies, but most effectively at50–100 Hz.[15,22] To avoid the possibility that partici-pants would accommodate to vibration, thus potentiallylimiting its impact on limb proprioceptive sensation, wevaried the frequencies of vibration (50, 60, 70, 80 Hz)across trials; for Experiment 1 two trials were performed ateach frequency in a quasi-randomized order. The same16, 30s random movement patterns were tested for eachcondition. These patterns were similar in design to thoseseen in Figure 1c, without a repeated component. Follow-ing the waveform creation protocol of Wulf andSchmidt,[23] these waveforms were balanced across themidline with regard to amplitude, meaning that flexionand extension movements of equal magnitude wererequired for each trial.Experimental ConditionsTo confirm that our method of vibration altered proprio-ceptive sensation, we tested limb position matching dur-ing both passive and active movement, with and withoutvibration. Eight trials were performed under each condi-tion. The participant's eyes were closed throughout eachtrial. In all trials one passive arm was driven through acontinuous movement pattern by an experimenter whileFor example, a right-handed participant would be fittedwith the vibration cuff on the right upper arm. In the pas-sive, driven condition the examiner would guide the rightarm and the individual would match those movementswith the left arm for 8 trials without vibration and 8 trialswith vibration. Arm guidance was accomplished by theexperimenter supporting arm weight at the elbow andmoving the lever. Care was taken to minimize experi-menter-participant contact. In the active, matching condi-tion the examiner would guide the left arm and theindividual would attempt to match with the right (Table1). Thus we were able to compare typical limb positionmatching ability to that when vibration was applied eitherto the active, matching limb or to the passive, driven limb.To avoid potential vibration aftereffects, non-vibration tri-als were always performed prior to vibration trials.[24]Outcome MeasureThe displacement of each lever and thus the movement ofeach arm was sampled at 40 Hz, raw position data weresmoothed using a 100 ms moving average, and data fromeach arm was corrected for constant error. To demonstrateand quantify the effect of our vibratory manipulation, theperceptual shift stimulated by vibration was indexed bythe ratio of the active, matching and passive, driven armmovement amplitudes (Eq. 1).RMS ratio = RMS position of active tracking arm/RMS position of driven arm Root mean square (RMS) = SQRT(∑xi2/n) where xi = limb position (1)RMS ratios over 1.0 indicate movement amplitude of theactive, matching arm was greater than that of the drivenarm. RMS ratios during no-vibration trials were averagedfor each participant. Additionally, correlation coefficientswere calculated to assess spatiotemporal coordinationbetween arms. Based on previous work,[13,20] we pre-dicted that vibration delivered to the passive, driven armwould result in the perception of movement amplitude inthat arm to be greater than in reality. In that experimentalcondition, individuals would overestimate the magnitudeof active limb motion required to match the position ofthe passive, driven arm. The opposite was expected whenvibration was applied to the active, matching arm.One-way ANOVA was used to compare active, matchingarm vibration, passive, driven arm vibration and no-vibra-tion conditions. Post-hoc comparison of vibration andno-vibration conditions was planned using one-tailedStudent's t-test to further explore expected condition dif-ferences, α = .025.ResultsPage 3 of 11(page number not for citation purposes)the participant matched that movement with the oppo-site, active arm. Vibration was introduced in half the trials.We hypothesized that if our method of vibration resultedin altered proprioceptive sensation, then regardless ofBehavioral and Brain Functions 2008, 4:32 http://www.behavioralandbrainfunctions.com/content/4/1/32Page 4 of 11(page number not for citation purposes)Experimental SetupFigu e 1Experimental Setup. a) Participants were seated before a computer monitor and gripped one (Experiment 2) or both (Experiment 1) horizontally mounted levers. A vibrating cuff was secured to one arm. b) Draping was drawn over the shoul-ders to prevent visualization of arm movement, represented by a dashed line. c) In Experiment 2, participants followed a pat-tern of movement similar these two example trials. Following a 3s stable baseline, sine-cosine waveforms dictated target movement. Two trial waveform patterns, each assembled from 1 random and 1 repeating sequence, are overlaid. The random epoch comes first, followed by the repeated sequence epoch during both trials for ease of visualization.cbaBehavioral and Brain Functions 2008, 4:32 http://www.behavioralandbrainfunctions.com/content/4/1/32which arm was vibrated (active tracking or passive driven)participants would interpret the vibrated arm as experi-encing greater excursion than in reality. As can be seen inFigure 2, vibration to the shoulder musculature resulted inchanges in perceived upper limb movement consistentwith our expectations. Our condition ANOVA revealedstatistically significant differences between conditions,(F(2,42) = 6.997, p = 0.002). The average RMS ratio wassignificantly different on non-vibration trials than on thepassive driven arm vibration trials (p = 0.016) and activevibration (p = 0.003) conditions. Limb movementbetween sides was closely related across conditions (meanr = 0.93 ± 0.07). Anecdotally, some participants reportedfeeling vibration both proximally in the neck as well asdistally in the elbow and "tingling" into the wrist andhand.ConclusionThese data demonstrate that vibration to the upperextremity at frequencies between 50 and 80 Hz resulted inaltered proprioceptive sensation that did not accuratelyreflect the true state of the limb. Importantly, every 0.1difference in the RMS ratio (Figure 2) translates into anaverage 3 cm difference between hand positions at thetransition between flexion and extension movements. Asour participants demonstrated vibration-induced RMSratio differences of 0.19 in the active, matching arm vibra-tion condition and 0.13 in the passive, driven arm vibra-tion condition vibration substantially altered trackingaccuracy.Based on our findings, and other recent work by Bock etal. [19] outlining the problematic and invasive nature ofother methods of sensory disruption (i.e., ischemic cuffs,peripheral nerve blocks) we elected to use the samemethod of vibration in Experiment 2 to disrupt proprio-of peripherally altered proprioception on motor sequencelearning.Experiment 2In Experiment 1 we found that our method of applyingvibration to the upper arm alters proprioceptive signalingand results in a misperception of limb state. In Experi-ment 2 we sought to capitalize on this effect to examinethe impact of altered proprioceptive sensation duringmotor skill practice on learning of a continuous motorsequence.MethodsParticipantsTwenty-five healthy adults (9 males, 16 females; mean age27.0 [range 22–43]) with no reported diabetes, or upperextremity muscular or sensory impairments agreed to par-ticipate. Each provided informed consent in the mannerprescribed by the University of Kansas Medical CenterHumans Subjects Committee, in compliance with theHelsinki Declaration. Three of these individuals also par-ticipated in Experiment 1. The dominant arm, as deter-mined by the Edinburgh Handedness Inventory [21] wasused for the task.Tracking TaskSeated in front of a computer monitor, participants usedTable 1: Limb Position Matching ProtocolActive, MatchingVibraton ConditionPassive, Driven Vibraton ConditionLeft Right Left RightSide Vibrated X XActive Matching Arm X XPassive Driven Arm X XTest conditions for a right-handed subject in the limb position matching task, Experiment 1. The vibrating cuff was applied to the dominant upper limb (in this case, the right shoulder). Then, each arm was moved, or "driven", by the experimenter, while the participant actively matched those movements with the opposite arm. In this manner, the right arm was vibrated both while it was actively matching and while it was passively driven.Effect of Vibration on Limb Excursion SenseFigure 2Effect of Vibration on Limb Excursion Sense. During the limb position matching task, participants interpreted the vibrated arm as having moved to a greater extent than in reality. This resulted in significantly reduced (active matching arm vibration) or increased (passive driven arm vibration) ratio of RMSE measures when vibration was applied as com-pared to when vibration was not applied.Page 5 of 11(page number not for citation purposes)ception throughout practice of a continuous motor track-ing task. This approach allowed us to ascertain the impacttheir dominant arm to track a continuously moving targetthat followed a sine-cosine wave pattern [23,25]: 23 right-Behavioral and Brain Functions 2008, 4:32 http://www.behavioralandbrainfunctions.com/content/4/1/32handed, 2 left-handed. The same lever set-up used inExperiment 1 (Figures 1a and 1b) was moved with shoul-der flexion and extension to track an on-screen cursor ver-tically up the screen (shoulder flexion) or down the screen(shoulder extension). Naturally, elbow extension fol-lowed shoulder flexion and elbow flexion accompaniedshoulder extension in a parasaggital plane. The targetappeared as a white box and participant's movementswere represented as a yellow circle cursor. The lever appa-ratus necessitated 31 cm of angular excursion over a max-imum of 60°, to accurately track the waveform; eachparticipant was easily able to move through this range ofmotion. As in Experiment 1, lever displacement samplingwas performed at 40 Hz. All stimuli were presented at 40Hz using custom software developed on the LabView plat-form (v. 7.1; National Instruments, Austin, TX).The pattern of target movement was predefined accordingto a method modified from Wulf and Schmidt.[23] Foreach 33s trial, a unique target wave was assembled fromone 3s baseline, presented at the middle of the screen andtracking range (to allow participants to orient their arm totask midline), and two 15s component sine-cosine wavesegments, or "epochs" (Figure 1c). In each tracking trial,participants were exposed to a novel random waveformepoch and an epoch that contained a repeated waveformsequence. To avoid order effects, presentation of therepeated sequence epoch randomly occurred as the first orsecond segment with a trial. The same presentation orderof trials was employed for every participant.Experimental DesignUpon enrollment, individuals were randomly assigned toeither have their passive, non-tracking arm vibrated as anexperimental control condition (CTL), or to have theirhand-dominant active, tracking arm vibrated (AV). Priorto starting, participants were instructed to track the targetas accurately as possible by controlling the position cursorwith shoulder flexion/extension movements of the lever.Individuals practiced the experimental tracking task 50 tri-als a day for two days (Table 2). During these trainingdays, vibration was applied according to group assign-ment. The possibility of accommodation to the vibratorystimuli was avoided by randomly varying the frequency ofstimulation after each trial; frequencies of 50, 60, 70 and80 Hz were randomly arranged and then delivered in thesame order for all participants.On a separate third day, participants returned for reten-tion test trials. We wished to examine the impact of prop-rioception on motor sequence learning over the twoprevious training days without the transient effects offitted according to group assignment but no vibration wasintroduced. To confirm the effect of vibration on motorperformance as compared to motor learning, after theretention test was completed without vibration, partici-pants performed an additional block with vibrationapplied according to their group assignment (CTL = vibra-tion to the passive, non-tracking arm; AV = vibration tothe active, tracking arm).The possibility that participants rely on vision to compen-sate for altered proprioception was accounted for by 1)preventing participants from seeing either of their arms,and 2) severely restricting visual feedback of the cursorposition. For both groups draping was used to preventvision of the arms throughout the entire study. The drapewas placed over, but did not come in contact with, the par-ticipant's upper body to avoid the possibility that brush-ing against it would provide cutaneous sensory cueing.Additionally, over the first 20 practice trials, visual feed-back regarding lever position was faded (i.e., linearlyreduced). We determined that initially (early practice)some visual feedback of cursor position was necessary forparticipants to understand the task; however, this feed-back was removed quickly (our schedule of fading wasbased on that reported by Winstein and colleagues).[26]Past work investigating continuous sequence productiondemonstrated that when visual feedback for cursor move-ments was delivered at 500 ms/1 sec or less it actually dis-rupted the use of visual feedback to guide movement.[27]Thus, in instances where visual feedback is less than orequal to 500 ms/1 sec, Kao showed that its brevity ren-dered it virtually useless for guiding hand-controlled cur-sor movements. Applying this finding, we linearlyreduced the amount of time the position cursor appearedbeyond the threshold reported by Kao. This ensured thatvisual error-feedback could not be used to continuouslyguide movement. In the present task, arm position infor-mation was faded from continuous delivery on trial 1Table 2: Testing Schedule and ConditionsDay Blocks Trials Vibration Visual FeedbackTraining Day 1 1–2 20 Yes Faded3–5 30 200 ms/2sTraining Day 2 6–10 50 Yes 200 ms/2sRetention Day 3 11 10 No 200 ms/2s12 10 Yes 200ms/2sEach subject completed three (3) days of training and retention testing. Visual and proprioceptive conditions followed this schedule. Participant groups (CTL, AV) differed only according to the side on which vibration was applied.Page 6 of 11(page number not for citation purposes)altered proprioception influencing performance. There-fore, during the first block of retention testing the cuff was(block 1), to a 200 ms position cursor presentation every2s by trial 19 (block 2) and kept at this level for theBehavioral and Brain Functions 2008, 4:32 http://www.behavioralandbrainfunctions.com/content/4/1/32remainder of the study. To maintain motivation for thisdifficult task and encourage accurate tracking, participantswere provided summary feedback after each trial duringacquisition as a percentage of time the position cursorspent within a 10° bandwith of the target. Because sum-mary feedback regarding overall tracking accuracy wasprovided only for motivational purposes, did not containsufficient information to alter performance, and was notexplicitly manipulated across the groups we also providedthis information at retention.Outcome MeasuresThe primary outcome measure was root-mean-squarederror of velocity changes (RMSE) (Eq. 2) separately calcu-lated for the random and repeated sequence epochs as thearea difference between target and participant movementvelocity.RMSE = SQRT(∑(xi - Xi)2/n) where xi = probe velocity, Xi = target velocity (2)RMSE from each 10 consecutive trials was averaged to rep-resent 1 block of sequence performance during the initialtwo days of training, and learning at retention. We consid-ered the pattern of continuous velocity changes ratherthan absolute position [23,28,29] as recent work hasemphasized the encoding of velocity-based informationby the proprioceptive system. [7,30]Statistical AnalysesFirst, to ensure that no baseline motor control or epoch-related differences biased performance, two-way ANOVAof Group (AV, CTL) and Epoch (random, sequence) atBlock 1 with repeated measures correction for Epoch wasperformed on RMSE. Next, change in tracking RMSE overthe 2 practice days was assessed via three-way ANOVA ofGroup × Epoch × Block (1–10) with repeated measurescorrection for Epoch and Block. Sequence-specific learn-ing was assessed using the retention test with two-wayANOVA of Group (AV, CTL) × Epoch with repeated meas-ures correction of Epoch using both position and velocityerror data. During practice and at retention, our analysisof random versus sequence epochs allowed us to parse outimproved motor control or non-specific learning frommore permanent changes in behavior as a result ofsequence-specific learning for the repeatedepoch.[23,25,28] Finally, we assessed the cost of vibrationto sequence-specific performance by calculating the differ-ence in RMSE between random and sequence epochswhen vibration was again introduced on day 3. One-wayANOVA of Group was used to test for change in perform-ance. All analyses were tested at α = .01 to protect againstType I error. A Greenhouse-Geisser correction was usedBox-plot analysis revealed that one individual in the AVgroup performed poorly enough to be statistically consid-ered an outlier when compared to the group during everyblock of practice. This participant was excluded from fur-ther analysis.ResultsContinuous Tracking During Acquisition PerformanceAt the beginning of training in block 1, before visual feed-back of the position cursor was faded, performance for theAV and CTL groups was similar. No main effects of Groupor Epoch or interaction of the two were noted (p > 0.05).Visual inspection of the data shows that over the course ofthe two training days all participants improved(decreased) sequence tracking error as compared to ran-dom tracking performance (Figure 3). This was confirmedvia a three-way ANOVA; the acquisition of sequence-spe-cific knowledge across practice is evident in the significantEpoch × Block interaction (F(9,198) = 24.211, p < .001).However, no Group effect or interaction was evident (p >0.1).Continuous Tracking at RetentionAt the day 3 retention test without vibration both groupsdemonstrated sequence-specific learning of task regulari-ties that allowed them to maintain improved trackingability for the sequence epoch when compared to randomperformance (Figure 4). This was confirmed with a twofactor ANOVA Group × Epoch where only the main effectof Epoch reached significance (F(1,22) = 37.407, p <.001).When vibration was reapplied according to group assign-ment on the retention test day, we noted a cost to expres-sion of sequence-specific learning for the AV group butnot the CTL group. That is, vibration resulted in a mean-ingful (Effect Size = 0.74) decline in tracking accuracy forthe repeated sequence. The interfering effect of vibrationwas evident in the loss of learning-related differencebetween tracking error for repeated as compared to ran-dom epochs for the AV but not the CTL group (Figure 5).This effect trended toward significance (F(1,22) = 3.667, p= 0.069).ConclusionThe purpose of this study was to examine the impact ofaltered proprioception and reduced visual feedback oncontinuous motor sequence learning. A large body of lit-erature demonstrates that information regarding bodystate is crucial for motor control.[4,6,31-33] In this study,we sought to determine whether this was also the case formotor sequence learning. In the past, Rothwell and col-Page 7 of 11(page number not for citation purposes)where appropriate. leagues [34] suggested that motor learning might be dele-teriously impacted by absent proprioception via their caseBehavioral and Brain Functions 2008, 4:32 http://www.behavioralandbrainfunctions.com/content/4/1/32report of a deafferented individual. In this case, theauthors reported that learning new, complex sequences ofhand movements was difficult when deafferentation waspresent. We wondered if similar negative effects would bepresent for continuous motor sequence learning whenproprioception was shifted by vibration. We discoveredthat in the short-term, peripherally altered proprioceptionand reduced visual feedback impacted motor perform-ance; however, given two days of practice, the extraordi-narily robust human motor learning system was able toovercome the challenge presented by shifted propriocep-To our knowledge the present study represents the firstexperimental investigation of the impact of altered prop-rioception on continuous motor sequence learning. Theexperimental design employed for the current work differsfrom previous studies in several important ways. First, weused a continuous tracking task that required participantsto use their entire upper extremity to produce movement.It has been suggested that investigation of complex move-ments (i.e. those that involve more degrees of freedom orgreater muscle activation) are critical for understandingmotor learning and behavior.[35,36] Our task met thosecriteria by including multiple joints, greater movementexcursion and longer movement patterns. Also, previousstudies of proprioception have often employed discrete,reaching-type tasks for which participants likely alreadyhave at least a rudimentary motor plan.[2,3,6,7] Our useof an entirely novel continuous tracking task allowed us tomore fully examine novel motor sequence learning.Finally, we engaged individuals in two days of practiceand a separate, delayed, retention test. In this mannerlearning versus performance improvements were clearlydifferentiated.[8] Because no prior studies of the role ofproprioception have employed a retention test design, ithas not been clear whether altered proprioception woulddeleteriously impact motor learning.[2,7]We hypothesized that if veridical proprioceptive sensationSequence-specific LearningFig re 4Sequence-specific Learning. At retention, when vibration was removed but visual feedback continued to be disrupted, improvements on sequence epoch tracking persisted regard-less of group, showing that altered proprioception during acquisiton performance did not impair continuous motor sequence learning. Decreased RMSE, towards graph bottom, denotes performance improvements.Tracking ErrorFigure 3Tracking Error. Average RMSE over skill practice (days 1 and 2). Open shapes represent performance on the sequence epoch, closed shapes represent the random epoch. Panel (a) displays the performance of the AV group that experienced vibration to the active tracking arm. Panel (b) displays the performance of the CTL group that experienced vibration to the passive, unused arm. Block 1 represents initial perform-ance with attenuating visual feedback but also vibration. The remaining blocks with vibration and minimal visual feedback, show an interaction of epoch and block suggesting improve-ment on sequence epoch tracking over time. Decreased RMSE, towards graph bottom, denotes performance improvements.Page 8 of 11(page number not for citation purposes)tive sensation and motor learning of a repeated continu-ous sequence occurred.was essential for sequence learning, peripherally alteredproprioceptive information that did not reflect the trueBehavioral and Brain Functions 2008, 4:32 http://www.behavioralandbrainfunctions.com/content/4/1/32state of the limb would diminish both acquisition andretention of the repeated motor sequence. We discoveredthat the opposite was true; all participants were able tolearn sequence-specific regularities as compared to ran-dom epoch performance. The finding that individuals canlearn to accurately and continuously track a repeatingsequence even when vibration was applied to the armbeing used suggests that accurate and intact propriocep-tion are not absolute prerequisites for encoding and con-solidating movement regularities. We found that thegroup that practiced with altered proprioception (AVgroup) and minimal visual feedback was able to improvein the same manner as the group who experience onlycontrol vibration to the non-tracking limb. Additionally,we found that when vibration was re-introduced at reten-tion, motor learning in the AV group was masked byaltered performance; this effect was not observed for theCTL group. These findings were facilitated by our experi-mental design; had we stopped data collection after 1 dayas past work has done we would not have noted the posi-tive effect of task practice in overcoming altered proprio-ceptive feedback.proprioceptors may be important for the execution ofmovement sequences. Based on this, we posited that pro-prioception would also be critical for learning the spatio-temporal regularities of a repeated continuous sequence.Rather, we found that accurate proprioceptive informa-tion was not essential for learning our experimental task.Nor was continuous visual feedback. Recent work hasreported that motor sequence learning can occur in arange of other environmental experiences. Overduin andcolleagues [38] demonstrated that sequence learningoccurred independently of learning predictable shifts inthe dynamic environmental state. Our work supports andextends these findings to show that motor sequence learn-ing can occur despite changes in visual and peripheralproprioceptive information.Simply becoming aware of the repeating sequence is onepossible reason that the AV group was able to learn thecontinuous tracking task. It is certainly possible thatobservation of target movement was sufficient to stimu-late learning. Indeed, sequence learning has been demon-strated following stimulus observation alone [39]especially when individuals attend to the task.[40,41] Inaccordance with these findings we cannot rule out thepossibility that untrustworthy proprioception was com-pensated for by paying greater attention to target motion.Another plausible explanation for our finding that alteredproprioception did not diminish learning may be thataccurate afferent sensation from more distal segments ofthe arm might have been preserved and exploited. Wecannot totally rule out this possibility with the presentexperimental setup. Single joint elbow muscles as well aswrist and finger musculature, joint and cutaneous affer-ents were possibly spared from vibratory disruption(though several subjects reported "numbness and tin-gling" into the forearm and wrist). Furthermore, second-ary spindle afferents appear to be relatively insensitive tovibration.[15] The central nervous system could have pref-erentially attended to these signals for information regard-ing performance.However, we suggest that the hypothesis outlined abovecannot completely explain our results because this same"unaltered" afferent information did not overcome vibra-tion-induced changes as shown by the limb positionmatching task in Experiment 1 or during reintroduction ofvibration at Experiment 2 retention testing. These findingssupply convergent evidence that vibration was disruptiveto motor control. Based on these findings, it appears thatvibration induced at least some shift in the afferent feed-back from the shoulder and elbow spanning musculatureto the central nervous system that altered motor output.Reintroduction of VibrationFigu e 5Reintroduction of Vibration. On day 3, during the last block of the study, sequence-specific learning improvements in tracking that were demonstrated by the AV group at the retention test when vibration was removed were masked by the re-introduction of vibration. That is, performance on ran-dom and sequence epochs were similar as shown by the lack of difference (low change score) between random and repeated sequence tracking error for the AV group. In con-trast, CTL participants maintained sequence-specific improvement that was seen in no-vibration retention testing. Larger change in RMSE denotes greater performance differ-ence between the sequence than the random epoch.Page 9 of 11(page number not for citation purposes)Cordo and colleagues [33,37] have suggested that thedynamic position and velocity information supplied byMotor sequence learning appears to have occurred despitethis shift in the veracity of limb proprioceptive sensation.Behavioral and Brain Functions 2008, 4:32 http://www.behavioralandbrainfunctions.com/content/4/1/32It has been previously noted that vision is critical whenproprioceptive sensation is diminished or absent[34,42]Ghez et al. [31] reported that individuals with large fibersensory neuropathy improved their aim on discrete reach-ing tasks when able to visualize arm position beforemovement. To explore the contribution of proprioceptionwithout the confound of visual feedback, we reduced vis-ual information available to the participant via severalcontrols. First, we occluded vision of the arm via draping.Next, we quickly faded feedback regarding cursor positionover the first 20 trials to an intermittency exceeding thatwhich Kao [27] cited as being disruptive to continuoustracking. However, we chose to preserve some visual feed-back to reduce cumulative error which might haveobscured improved motor control associated with learn-ing [43] by displaying the arm position cursor for 200 msat 1800 ms intervals. It is possible that even this minimalvisual information may have allowed participants to eval-uate their performance and adjust accordingly in theabsence of trustworthy proprioceptive feedback. How-ever, based on the past work of Kao [27] we find thisexplanation of our conclusions highly improbable.Our finding of preserved continuous sequence learningdespite restricted visual feedback and altered propriocep-tion reflects the dynamic and robust nature of a motorlearning system that is able to compensate for inaccurateafferent information through redundant physiologicaland cognitive systems. One or some combination of all ofthe mechanisms proposed above may have facilitatedlearning for participants in this research. Though thesefindings do not directly support our original hypothesesthat altered proprioception would disrupt motorsequence learning, they are not without precedent. Skilllearning has been reported in dorsal rhizotomized mon-keys.[44,45] The juxtaposition between our findings andTaub et al.'s are in contrast to reports by others,[46,47]who have reported disruption of skill learning followingsensoricortical damage. These seemingly contradictoryresults may be a function of the difference between centraland peripheral neural damage/disruptions. It remains tobe seen if those with chronic sensory impairment result-ing from damage to central sensory cortical or thalamicregions have difficulty learning new motor skills. Futurework should consider this possibility in persons withmedical conditions characterized by reduced propriocep-tion.Competing interestsNeither author has any personal or financial relationshipto declare regarding this manuscript.Authors' contributionsing the manuscript. Both authors have read and approvedthe final version of this manuscript.AcknowledgementsThis work was supported in part by the Kansas Partners in Progress, Inc., the Vancouver Coastal Health Research Institute and Foundation, the Heart and Stroke Foundation of Canada, and the North Growth Founda-tion. We appreciate the assistance of Rebecca Maletsky for programming assistance. We also thank Janice Eng, Barbara Quaney, Brenda Wessel and Nicole Acerra for graciously making time to review early drafts.References1. Schmidt RA, Lee TD: Motor Control and Learning: A Behavioral Approach3rd edition. Champaign, IL: Human Kinetics; 1999. 2. Bernier PM, Chua R, Bard C, Franks IM: Updating of an internalmodel without proprioception: a deafferentation study.  Neu-roreport 2006, 17:1421-1425.3. Krakauer JW, Ghilardi MF, Ghez C: Independent learning ofinternal models for kinematic and dynamic control of reach-ing.  Nat neurosci 1999, 2:1026-1031.4. Vercher JL, Sares F, Blouin J, Bourdin C, Gauthier G: Role of sensoryinformation in updating internal models of the effector dur-ing arm tracking.  Prog Brain Res 2003, 142:203-222.5. Pipereit K, Bock O, Vercher JL: The contribution of propriocep-tive feedback to sensorimotor adaptation.  Exp Brain Res 2006,174:45-52.6. Gordon J, Ghilardi MF, Ghez C: Impairments of Reaching Move-ments in Patients Without Proprioception. I. Spatial Errors.J Neurophysiol 1995, 73:347-360.7. Hwang EJ, Smith MA, Shadmehr R: Dissociable effects of theimplicit and explicit memory systems on learning control ofreaching.  Exp Brain Res 2006, 173:425-437.8. Salmoni AW, Schmidt RA, Walter CB: Knowledge of results andmotor learning: a review and critical reappraisal.  Psychol Bull1984, 95:355-386.9. Bernier PM, Chua R, Inglis JT, Franks IM: Sensorimotor adaptationin response to proprioceptive bias.  Exp Brain Res 2007,177:147-156.10. Bevan L, Cordo P, Carlton L, Carlton M: Proprioceptive coordi-nation of movement sequences: discrimination of joint angleversus angular distance.  J Neurophysiol 1994, 71:1862-1872.11. Thoroughman KA, Shadmehr R: Electromyographic correlatesof learning an internal model of reaching movements.  J Neu-rosci 1999, 19:8573-8588.12. Hwang EJ, Shadmehr R: Internal models of limb dynamics andthe encoding of limb state.  J Neural Eng 2005, 2:S266-278.13. Goodwin GM, McCloskey DI, Matthews PBC: The contribution ofmuscle afferents to Kinaesthesia shown by vibration inducedillusions of movement and by the effects of paralysing jointafferents.  Brain 1972, 95:705-748.14. Roll JP, Vedel JP: Kinaesthetic role of muscle afferents in man,studied by tendon vibration and microneurography.  Exp BrainRes 1982, 47:177-190.15. Vedel JP, Roll JP: Muscle spindle contribution to the coding ofmotor activities in man.  Exp Brain Res 1983:253-265.16. Capaday C, Cooke JD: Vibration-induced changes in move-ment-related EMG activity in humans.  Exp Brain Res 1983,52:139-146.17. Cordo P, Gurfinkel VS, Bevan L, Kerr GK: Proprioceptive conse-quences of tendon vibration during movement.  J Neurophysiol1995, 74:1675-1688.18. Forwell LA, Carnahan H: Proprioception during manual aimingin individuals with shoulder instability and controls.  J OrthopSports Phys Ther 1996, 23:111-119.19. Bock O, Pipereit K, Mierau A: A method to reversibly degradeproprioceptive feedback in research on human motor con-trol.  J Neurosci Methods 2007, 160:246-250.20. van Deursen RW, Sanchez MM, Ulbrecht JS, Cavanagh PR: The roleof muscle spindles in ankle movement perception in humansubjects with diabetic neuropathy.  Exp Brain Res 1998, 120:1-8.Page 10 of 11(page number not for citation purposes)EDV participated in all aspects of the study. LAB partici-pated in study design, interpretation of findings and draft-21. Oldfield RC: The assessment and analysis of handedness: theEdinburgh inventory.  Neuropsychologia 1971, 9:97-113.Publish with BioMed Central   and  every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central Behavioral and Brain Functions 2008, 4:32 http://www.behavioralandbrainfunctions.com/content/4/1/3222. Sittig AC, Denier van der Gon JJ, Gielen CC: The contribution ofafferent information on position and velocity to the controlof slow and fast human forearm movements.  Exp Brain Res1987, 67:33-40.23. Wulf G, Schmidt RA: Variability of practice and implicit motorlearning.  J Exp Psychol Learn Mem Cogn 1997, 23:987-1006.24. Ishihara Y, Izumizaki M, Atsumi T, Homma I: Aftereffects ofmechanical vibration and muscle contraction on limb posi-tion-sense.  Muscle Nerve 2004, 30:486-492.25. Pew RW: Levels of analysis in motor control.  Brain Res 1974,71:393-400.26. Winstein CJ, Pohl PS, Lewthwaite R: Effects of physical guidanceand knowledge of results on motor learning: support for theguidance hypothesis.  Res Q Exerc Sport 1994, 65:316-323.27. Kao HS: Effects of intermittency of feedback on a compensa-tory tracking task.  Percept Mot Skills 1976, 43:1339-1345.28. Shea CH, Wulf G, Whitacre CA, Park JH: Surfing the implicitwave.  Q J Exp Psychol A 2001, 54:841-862.29. Boyd L, Winstein C: Explicit information interferes withimplicit motor learning of both continuous and discretemovement tasks after stroke.  J Neurol Phys Ther 2006, 30:46-57.discussion 58–59.30. Prochazka A, Gorassini M: Models of ensemble firing of musclespindle afferents recorded during normal locomotion in cats.J Physiol 1998, 507(Pt 1):277-291.31. Ghez C, Gordon J, Ghilardi MF, Christakos CN, Cooper SE: Roles ofproprioceptive input in the programming of arm trajecto-ries.  Cold Spring Harb Symp Quant Biol 1990, 55:837-847.32. Ghez C, Gordon J, Ghilardi MF: Impairments of reaching move-ments in patients without proproception. II. effects of visualinformation on accuracy.  J Neurophysiol 1995, 73:361-372.33. Cordo P, Carlton L, Bevan L, Carlton M, Kerr GK: Proprioceptivecoordination of movement sequences: role of velocity andposition information.  J Neurophysiol 1994, 71:1848-1861.34. Rothwell JC, Traub MM, Day BL, Obeso JA, Thomas PK, Marsden CD:Manual motor performance in a deafferented man.  Brain1982, 105(Pt 3):515-542.35. Cordo PJ, Gurfinkel VS: Motor coordination can be fully under-stood only by studying complex movements.  Prog Brain Res2004, 143:29-38.36. Wulf G, Shea CH: Principles derived from the study of simpleskills do not generalize to complex skill learning.  Psychon BullRev 2002, 9:185-211.37. Cordo P, Bevan L, Gurfinkel V, Carlton L, Carlton M, Kerr G: Prop-rioceptive coordination of discrete movement sequences:mechanism and generality.  Can J Physiol Pharmacol 1995,73:305-315.38. Overduin SA, Richardson AG, Bizzi E, Press DZ: Simultaneous sen-sorimotor adaptation and sequence learning.  Exp Brain Res2008, 184:451-456.39. Bird G, Osman M, Saggerson A, Heyes C: Sequence learning byaction, observation and action observation.  Br J Psychol 2005,96:371-388.40. Howard JH Jr, Mutter SA, Howard DV: Serial pattern learning byevent observation.  J Exp Psychol Learn Mem Cogn 1992,18:1029-1039.41. Willingham DB, Goedert-Eschmann K: The relation betweenimplicit and explicit learning: evidence for parallel develop-ment.  Psychol Sci 1999, 10:531-534.42. Cole JD, Sedgwick EM: The perceptions of force and of move-ment in a man without large myelinated sensory afferentsbelow the neck.  J Physiol 1992, 449:503-515.43. Hocherman S, Levy H: The role of feedback in manual trackingof visual targets.  Percept Mot Skills 2000, 90:1235-1248.44. Taub E, Ellman SJ, Berman AJ: Deafferentation in monkeys: effecton conditioned grasp response.  Science 1966, 151:593-594.45. Taub E, Williams M, Barro G, Steiner SS: Comparison of the per-formance of deafferented and intact monkeys on continuousand fixed ration schedules of reinforcement.  Exp Neurol 1978,58:1-13.46. Pavlides C, Miyashita E, Asanuma H: Projection from the Sensoryto the Motor Cortex is Important in Learning Motor Skills inthe Monkey.  J Neurophysiol 1993, 70:733-741.47. Sakamoto K, Arissian K, Asanuma H: Functional role of the sen- yours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 11 of 11(page number not for citation purposes)sory cortex in learning motor skills in cats.  Brain Res 1989,503:258-264.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items