Retrieval practice in motor learning

Human Movement Science xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

Human Movement Science

journal homepage: www.elsevier .com/locate/humov

Retrieval practice in motor learning

0167-9457/$ - see front matter � 2012 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.humov.2012.10.002

⇑ Corresponding author. Address: Leibniz Research Centre for Working Environment and Human Factors, Ardeystra44139 Dortmund. Tel.: +49 (0)231 1084 237; fax: +49 (0)231 1084 308.

E-mail address: [email protected] (A. Boutin).

Please cite this article in press as: Boutin, A., et al. Retrieval practice in motor learning. Human Movement(2013), http://dx.doi.org/10.1016/j.humov.2012.10.002

Arnaud Boutin a,⇑, Stefan Panzer b, Yannick Blandin c,d

a IfADo – Leibniz Research Centre for Working Environment and Human Factors, Dortmund, Germanyb Saarland University, Saarbrücken, Germanyc National Centre of Scientific Research (CNRS; CeRCA – UMR 7295), Poitiers, Franced University of Poitiers, Poitiers, France

a r t i c l e i n f o a b s t r a c t

Article history:Available online xxxx

PsycINFO classification:233023402343

Keywords:Retrieval practiceTestingEncodingConsolidationEffector transferMotor learning

In this study we sought to determine whether testing promotes thegeneralization of motor skills during the process of encoding and/or consolidation. We used a dynamic arm movement task thatrequired participants to reproduce a spatial-temporal pattern ofelbow extensions and flexions with their dominant right arm. Gen-eralization of motor learning was tested by the ability to transferthe original pattern (extrinsic transformation) or the mirrored pat-tern (intrinsic transformation) to the unpractised left arm. Toinvestigate the testing effects during both encoding and consolida-tion processing, participants were administered an initial testingsession during early practice before being evaluated on a post-practice testing session administered either 10 min (Testing-Encoding group) or 24 hr apart (Testing-Consolidation group),respectively. Control groups were required to perform a post-prac-tice testing session administered after either a 10-min (Control-Encoding group) or 24-hr delay (Control-Consolidation group).The findings revealed that testing produced rapid, within-practiceskill improvements, yielding better effector transfer at the 10-mintesting for the Testing-Encoding group on both extrinsic and intrin-sic transformation tests when compared with the Control-Encod-ing group. Furthermore, we found better performance for theTesting-Consolidation group at the 24-hr testing for extrinsic andintrinsic transformations of the movement pattern when com-pared with the Control-Consolidation group. However, our resultsdid not indicate any significant testing advantage on the latent,between-session development of the motor skill representation(i.e., from the 10-min to the 24-hr testing). The testing benefits

ße 67, D-

Science

http://dx.doi.org/10.1016/j.humov.2012.10.002

mailto:[email protected]


http://www.sciencedirect.com/science/journal/01679457

http://www.elsevier.com/locate/humov


2 A. Boutin et al. / Human Movement Science xxx (2013) xxx–xxx

Please cite this article in press as: Boutin, A.,(2013), http://dx.doi.org/10.1016/j.humov.20

expressed at the 10-min testing were stabilised but did not extendduring the period of consolidation. This indicates that testing con-tributes to the generalisation of motor skills during encoding butnot consolidation.

� 2012 Elsevier B.V. All rights reserved.

1. Introduction

Behavioral and neurophysiological studies suggest that motor skill learning is characterized by dis-tinct and successive phases. The acquisition of a new motor skill relies on a fast, within-practice phaseof performance improvement that is followed by a slow learning phase consisting of delayed, time-dependent performance improvements occurring off-line between practice sessions (e.g., Karni & Ber-tini, 1997; Karni et al., 1995; Korman, Raz, Flash, & Karni, 2003; Robertson, Pascual-Leone, & Press,2004; see also Krakauer & Shadmehr, 2006; Stickgold & Walker, 2007, for reviews). The off-line learn-ing process refers to a spontaneous improvement in performance without additional practice (Walker,2005), intent or awareness (Stickgold & Walker, 2007), where the new and initially labile task repre-sentation is strengthened and becomes integrated into the network of pre-existing and long-lastingmotor skill representations (>24 hours; see Krakauer & Shadmehr, 2006, for a review). These post-practice processes are essential to the formation and long-term storage of motor representations,and have been grouped under the term ‘‘consolidation’’ (e.g., Krakauer & Shadmehr, 2006; McGaugh,2000; Robertson, Pascual-Leone, & Miall, 2004; Stickgold & Walker, 2007).

Skill development during consolidation is characterised by either a quantitative increase in perfor-mance or a qualitative representational change (see Robertson, 2009). For instance, qualitativechanges might be expressed by a shift in the strategy used to solve a problem (e.g., Wagner, Gais, Haid-er, Verleger, & Born, 2004) or by a shift in reliance from one coding system to another (e.g., Boutinet al., 2012a; Kovacs, Muehlbauer, & Shea, 2009). In accordance with this notion, it has recently beenproposed that latent formation of motor representations is supported by the development of two dis-tinct skill components that operate together to mediate off-line learning (Boutin et al., 2012a). Theskill components that develop off-line can be distinguished as either intrinsic or extrinsic coding sys-tems (e.g., Criscimagna-Hemminger, Donchin, Gazzaniga, & Shadmehr, 2003; Hikosaka et al., 1999;Lange, Godde, & Braun, 2004). Each coding system contributes to movement production and can pro-duce specific learning and transfer capabilities (e.g., Boutin et al., 2012a; Boutin, Fries, Panzer, Shea, &Blandin, 2010; Panzer, Krueger, Muehlbauer, Kovacs, & Shea, 2009).

While the visually acquired information on movement and target positions is initially encoded inan eye-centred, extrinsic world-based reference frame, the muscular activation patterns are then en-coded in an intrinsic, body-centred reference frame (e.g., Colby & Goldberg, 1999; Soechting & Flan-ders, 1989). The intrinsic code is represented as an internal model of joint representations(Criscimagna-Hemminger et al., 2003), musculoskeletal forces, and dynamics (Krakauer, Ghilardi, &Ghez, 1999) that takes the relative orientation of body segments into account (Lange et al., 2004;Soechting & Flanders, 1989). The intrinsic coding system is thought to result in a representation ofthe motor skill that is effector-dependent and lacks transfer capability. Conversely, the extrinsic codereflects the Cartesian coordinates of the task space with respect to the visual display and results in aneffector-independent representation with effector transfer capability (Hikosaka et al., 1999).

Recently, Boutin and colleagues (2012a) investigated the effects of practice on effector transfer andthe associated skill representational changes that occur during the within-practice and between-practicephases. Interestingly, they found that practice induces rapid and non-transferable (effector-dependent)motor skill improvements, with the extrinsic component of the skill developing early and remaining thedominant coding system during practice. Conversely, their findings revealed latent and practice-depen-dent reorganisations of the motor skill during the off-line period: limited practice induced an off-linedevelopment of the extrinsic component, whereas prolonged practice subserved an off-line developmentof the intrinsic component. Practice has been proposed to trigger rapid and latent reorganisations of themotor skill representation, thus yielding distinct immediate and long-term transfer capabilities.

et al. Retrieval practice in motor learning. Human Movement Science12.10.002


A. Boutin et al. / Human Movement Science xxx (2013) xxx–xxx 3

The general conclusion drawn by Boutin et al. (2012a) was that practice induces long-lasting qual-itative representational shifts of the motor skill that are dependent on the organisation of the practicesession itself. They stated that the parsing of practice into multiple sessions (administered 24 hr apart)narrows effector transfer capabilities in comparison to a single practice session. Interestingly, how-ever, they found that the retrieval of information during a testing session (i.e., retention and transfertests) administered between the delayed practice sessions promoted skill transfer by inducing repre-sentational changes of the motor task. Testing might thus be considered as a critical learning factorthat potentially induces long-term skill enhancements through encoding and consolidation.

To address this issue, the same experimental procedure as that used by Boutin et al. (2012a) wasapplied in the present study, but a shorter spacing interval was used between the practice sessions.Using a short 10-min interval rather than spacing the practice sessions across days (i.e., 24 hr) wouldallow us to determine whether the long-lasting testing benefits can be ascribed to encoding- or con-solidation-related processes; a 10-min interval is considered insufficient for the consolidation processto be completed (e.g., Shadmehr & Brashers-Krug, 1997; Shadmehr & Holcomb, 1997). To ascertain therapid and latent skill representational changes following the completion of a testing session adminis-tered during practice (i.e., between the two practice sessions), participants’ performance was testedshortly after the end of the two practice sessions (i.e., encoding processing) or after a 24-hr delay(i.e., consolidation processing).

Based on previous research investigating the learning and generalisation of motor skills (e.g., Bou-tin et al., 2012a; Boutin, Panzer, Salesse, & Blandin, 2012b; Witt, Margraf, Bieber, Born, & Deuschl,2010), we used retention and effector transfer tests, respectively. The traditional way is to train onelimb on a particular motor task and then to test the ability of the learner (1) to retrieve the informationacquired and stored with practice (i.e., retention test), and (2) to transfer the newly acquired task com-ponents to the unpracticed limb (i.e., transfer tests). More specifically, to test the generalisation ofboth extrinsic and intrinsic task components to the unpracticed limb, we used two effector transfertests. In the ‘‘intrinsic transformation’’ test, the original pattern was mirrored so that the sequentialmovements remain the same when transferred to the unpracticed limb (i.e., the same pattern of mus-cle activation and limb joint angles). This test assessed the contribution of the intrinsic component totransfer and learning. In the ‘‘extrinsic transformation’’ test, the original pattern was preserved butwas performed with the unpracticed limb (i.e., the same goal movement pattern). This test assessedthe contribution of the extrinsic component to transfer and learning because it implicated the samespatial positions but a different pattern of muscle activation compared with that used during practice.

Therefore, we investigated whether retrieval practice affects encoding and/or consolidation pro-cessing as evaluated by retention and transfer performance shortly after the end of practice (i.e.,10-min testing) or after a 24-hr break (24-hr testing), respectively. We predicted that performingan initial testing session during early practice should produce rapid and long-lasting skill enhance-ments, expressed by better performances on the retention and both transfer tests at the 10-minand 24-hr testing when compared with control groups.

2. Method

2.1. Participants

Forty-eight self-declared right-handed undergraduate students (17–26 years, 19.6 ± 1.2 years,mean age ± standard deviation, 20 females) volunteered to participate in this study. Participantshad no prior experience with the experimental task and were unaware of the specific purposes ofthe study. Informed consent was obtained before the experiment, and the protocol was approvedby a local ethics committee.

2.2. Apparatus

The apparatus consisted of a horizontal lever supported at one end by a vertical axle that turned ina ball-bearing support in a manner that was almost frictionless. The support was fixed on a table

Please cite this article in press as: Boutin, A., et al. Retrieval practice in motor learning. Human Movement Science(2013), http://dx.doi.org/10.1016/j.humov.2012.10.002



facing the participant, allowing the lever to move on a horizontal plane over the table surface. A ver-tical handle was fixed at the other end of the lever. The position of the handle could be adjusted so thatwhen the participant grasped the handle, his or her elbow was aligned with the axis of rotation (Fig. 1,top left). A potentiometer with an output sampled at 1000 Hz was attached to the lower end of theaxis to record the position. The potentiometer data were used to provide lever position informationto the participant and were stored for later analysis. A wooden cover was placed over the table to pre-vent participants from seeing the lever and their arm. A video projector was used to display the goalmovement pattern, the on-line position of the lever and the knowledge of results (KR) on the wall fac-ing the participant. Participants were seated approximately 2 m from the wall where a 1.64 � 1.23 mimage was projected. All aspects of the experiment were programmed with Matlab R2008b softwarefrom MathWorks� (The MathWorks, Inc., Natick, MA) using the Psychophysics Toolbox extensions(Brainard, 1997; Kleiner, Brainard, & Pelli, 2007; Pelli, 1997).

2.3. Task, experimental groups and procedures

Participants were randomly assigned to one of four practice conditions (Fig. 1, bottom). Two exper-imental groups performed an initial testing session during early practice. The initial testing session wascomposed of retention and transfer tests (Fig. 1, top) and was scheduled between two practice ses-sions (Acquisition 1 and Acquisition 2, which were composed of 2 and 15 practice blocks of 9 trials,respectively). Participants were then administered a post-practice testing session either 10 min (Test-ing-Encoding group; N = 12) or 24 hr (Testing-Consolidation group; N = 12) after the completion of thetwo practice sessions. To evaluate the rapid and latent developments of the motor skill during encod-ing and consolidation, we included two control groups that were provided the same procedures as the

Fig. 1. (Top) Schematic illustrating the arm and criterion movement pattern used during acquisition, retention and transfertests. The goal movement pattern was projected onto the wall in front of the participant. Participants used their dominant rightarm during acquisition and retention. The non-dominant left arm was used for the extrinsic and intrinsic transformation tests.(Bottom) All participants completed the first (Acquisition 1) and second (Acquisition 2) practice sessions, composed of 2 and 15blocks of 9 trials, respectively. The retention (R), extrinsic transformation (Et) and intrinsic transformation (It) tests wereperformed between the practice sessions (initial testing) and/or 10 min after the completion of the practice sessions (10-mintesting) and/or 24 hr after the end of practice (24-hr testing), depending on group assignment.




experimental groups, except that they did not perform the initial testing session during practice. Par-ticipants in the control groups were merely administered a delayed post-practice testing sessioneither 10 min (Control-Encoding group; N = 12) or 24 hr (Control-Consolidation group; N = 12) afterthe end of practice.

Participants were tested individually in a silent and dimly lit room. Before beginning the experi-ment, each participant received written instructions and additional verbal information about the taskand procedures. They were asked to sit on a chair with an adjustable height, so that their lower armwas positioned at an approximately 80-degree angle to their upper arm in the starting position. Theywere asked to perform a sequence of extension-flexion movements with their dominant right arm thatreproduced the spatial and temporal aspects of the goal pattern projected in front of them (Fig. 1, top).The spatial-temporal pattern was created by summing two sine waves with similar periods (1 s) butdifferent amplitudes (30� for the first and 45� for the second sine wave, which refers to the lever an-gles). The horizontal visual angle of the display was about 12� to both sides from the vertical midlineof the screen. The duration of the goal pattern was 1500 ms, and the extension-flexion movements re-quired three reversals. The potentiometer output was sampled for 2000 ms, but only the first 1500 mswas retained for later analysis.

At the beginning of each trial, the goal movement pattern was displayed on the wall, and the par-ticipant was asked to move the lever to the starting position (1 degree area at the beginning of thegoal movement pattern). One second after achieving the starting position, a tone (50 ms in duration)indicated that they should perform the task. The task required moving the lever with the dominantright arm through a pattern of extension-flexion cycles (three reversals) to produce the requiredspatial-temporal pattern. As soon as the participant began moving, the goal movement pattern dis-appeared from the screen and a cursor representing the position of the lever was displayed. After a2-s interval following response completion, KR was provided by superimposing the goal movementpattern (white) over the actual pattern (green). In addition, the root-mean-square error (RMSE) ofthe actual movement from the goal movement was calculated and displayed on the screen. TheRMSE is the deviation of the actual pattern from the goal pattern calculated from the onset of move-ment until the end of the first 1500 ms. RMSE is sensitive to both response bias and within-partic-ipant variability. When presented, KR was displayed for 5 s. Participants were instructed toreproduce the goal movement pattern as accurately as possible and were requested to reduce theRMSE on subsequent trials.

All participants completed the first practice session (Acquisition 1). During this phase, KR wasprovided following each trial. Participants assigned to the Testing-Encoding (TE) and Testing-Consol-idation (TC) practice conditions were administered the initial testing session (i.e., retention and effec-tor transfer tests), whereas their Control-Encoding (CE) and Control-Consolidation (CC) practicecondition counterparts were not. The retention test consisted of 1 block of 9 trials without KRand required the participants to produce the movement pattern with the same limb that was usedduring the practice session. With the exception that KR was not provided to the participants duringretention, all other procedures remained the same as those used for acquisition. In addition, twoeffector transfer tests without KR were administered in a counterbalanced order after the retentiontest (Fig. 1, top). The transfer tests were performed with the contralateral non-dominant left arm. Inthe ‘‘intrinsic transformation’’ test, the spatial sequence became mirrored so that the sequentialmovements remained the same when transferred to the non-dominant left arm (i.e., the same pat-tern of muscle activation and limb joint angles). In the ‘‘extrinsic transformation’’ test, the originalsequence was preserved but performed with the non-dominant left arm (i.e., the same goal move-ment pattern).

Note that participants in the CE and CC practice conditions did not perform the retention and trans-fer tests after the first practice session. However, all participants continued practice for an additionalacquisition phase (Acquisition 2). This practice phase involved the same experimental procedures as inAcquisition 1. After either a 10-min or 24-hr rest interval following the end of Acquisition 2, partici-pants assigned to the TE and CE practice conditions were then required to perform the 10-min testingsession while their TC and CC practice condition counterparts were administered the 24-hr testing ses-sion (retention and transfer tests). Note that the 10-min testing and 24-hr testing sessions were per-formed in the same manner as the initial testing session.




2.4. Data analysis and measurement

Data processing was performed using Matlab� (Mathworks�, Natick, MA). The individual trial timeseries were used to compute lever displacement. Angular displacement time series were filtered usinga second-order dual-pass Butterworth filter with a cut-off frequency of 10 Hz. The RMSE was com-puted to estimate the performance error in achieving the goal movement pattern. Such a measureis sensitive to both amplitude and timing errors in the produced movement pattern relative to the goalmovement pattern. To compute the RMSE, the difference between the criterion and the filtered actualmovement pattern was computed at each data point in the time series. Next, differences for each datapoint in the time series were squared, and means of the squared differences were computed on a trialbasis. Finally, the square root of the mean was computed for the final measure of RMSE. Values ofRMSE for individual trials were then averaged to yield a global estimate of RMSE for each block (ninetrials).

3. Results

3.1. Acquisition

A two-way repeated-measures analysis of variance (ANOVA) with practice condition (TE, TC, CE,CC) and block (Blocks 1–2 for Acquisition 1 and Blocks 3–17 for Acquisition 2) factors was applied sep-arately for Acquisition 1 and Acquisition 2. All significant effects were reported at p < .05 unless other-wise stated, and a Duncan’s multiple range test was used for post-hoc comparisons. Partial eta square(g2

p) was the effect size reported for all significant effects (Cohen, 1988). Outliers were removed fromthe analysis (±2 SD: approximately 3%). Mean RMSE values during the acquisition phases are displayedin Fig. 2.

3.2. Acquisition 1

The analysis revealed a main effect of block, F(1,44) = 87.90, g2p ¼ :66, indicating a higher RMSE on

block 1 (M = 100.46) compared to block 2 (M = 71.72). The main effect of practice condition,F(3,44) < 1, and the practice condition � block interaction, F(3,44) < 1, were not significant (Fig. 2,left).

Figure 2. Mean RMSE during Acquisition 1 (Blocks 1–2) and Acquisition 2 (Blocks 3–17) for each group (Day 1). Error barsreflect the standard error of the mean.




3.3. Acquisition 2

The analysis detected a main effect of block, F(14,616) = 18.01, g2p ¼ :29 (Fig. 2, right). The Duncan’s

multiple range test revealed a progressive RMSE decrement from block 3 (M = 62.25) to block 10(M = 46.90). From block 11 (M = 45.70) to block 17 (M = 43.74), the RMSE remained stable. The anal-ysis failed to detect either a main effect of practice condition, F(3,44) = 1.03, or a practice condi-tion � block interaction, F(42,616) < 1.

3.4. Retention and transfer

3.4.1. TE and TC groups (Initial testing)To make sure that both testing groups equally performed and represented the motor skill during

early practice, we compared performances of the TE and TC groups at the initial testing (Fig. 3, left).Such an analysis was not of particular importance for the current study but was rather conductedas a prerequisite for subsequently analysing and interpreting the results at the 10-min and 24-hr test-ing. Thus, mean RMSE values on the retention and transfer tests were analysed in a 2 (practice condi-tions: TE, CE) � 3 (tests: retention, extrinsic transformation, intrinsic transformation) ANOVA withrepeated measures on the last factor. The analysis revealed a significant test effect, F(2,44) = 12.84,g2

p ¼ :36, with lower RMSE for the retention test (M = 60.44) compared to the extrinsic (M = 71.54)and intrinsic tests (M = 79.66), which are significantly different from each other. As expected, the anal-ysis failed to detect either a main effect of practice condition, F(1,22) < 1, or a practice condition � testinteraction, F(2,44) < 1.

3.4.2. TE and CE groups (10-min testing)To assess whether testing induces within-practice skill enhancements, we compared performances

of the TE and CE groups at the 10-min testing (Fig. 3, middle). Mean RMSE values on the retention andtransfer tests were analysed in a 2 (practice conditions: TE, CE) � 3 (tests: retention, extrinsic trans-formation, intrinsic transformation) ANOVA with repeated measures on the last factor. The analysisindicated a main effect of practice condition, F(1,22) = 8.27, g2

p ¼ :27, test, F(2,44) = 25.31, g2p ¼ :53,

and a significant practice condition � test interaction, F(2,44) = 6.83, g2p ¼ :23. A lower RMSE was

Figure 3. Mean RMSE at the initial testing, 10-min testing and 24-hr testing on the retention (R), extrinsic transformation (Et)and intrinsic transformation (It) tests for each group. Error bars reflect the standard error of the mean.




found for the TE group than for the CE group on the extrinsic and intrinsic transformation tests but noton the retention test. Specifically, for the TE group, a Duncan’s multiple range test indicated that theRMSE was significantly lower on the retention test (M = 43.51) than on the extrinsic (M = 54.43) andintrinsic transformation tests (M = 54.70), which did not differ from each other. However, for the CEgroup, the RMSE was significantly lower on the retention test (M = 44.72) compared with that onthe extrinsic (M = 67.70) and intrinsic transformation tests (M = 81.53), which also significantly dif-fered from each other.

3.4.3. TC and CC groups (24-hr testing)To determine whether the completion of the initial testing session positively influenced the long-

term representation of the motor skill, we analysed performance of the TC and CC groups at the 24-hrtesting (Fig. 3, right). Mean RMSE values on the retention and transfer tests were analysed in a 2 (prac-tice conditions: TC, CC) � 3 (tests: retention, extrinsic transformation, intrinsic transformation) ANO-VA with repeated measures on the last factor. The analysis indicated a marginally significant maineffect of practice condition, F(1,22) = 3.83, p = .06, a main effect of test, F(2,44) = 3.19, g2

p ¼ :12, anda significant practice condition � test interaction, F(2,44) = 3.48, g2

p ¼ :14. Lower RMSE values werefound for the TC group when compared to the CC group on the extrinsic and intrinsic transformationtests but not on the retention test. Post-hoc comparisons did not reveal any significant performancedifference for the TC group on the retention (M = 54.41), extrinsic transformation (M = 54.72) or intrin-sic transformation tests (M = 53.50). However, for the CC group, the RMSE was significantly lower onthe retention test (M = 53.51) compared with that on the extrinsic (M = 69.07) and intrinsic transfor-mation tests (M = 69.68), which did not differ from each other.

3.4.4. Off-line improvements (from 10-min to 24-hr testing): testing and control groupsWe compared performance on tests conducted shortly after the end of practice (10-min testing)

and tests conducted following a 24-hr delay among participants who were assigned to the testingand control practice conditions (Fig. 4). The difference between these measures (10-min testing minus24-hr testing) highlighted the off-line development of the motor skill; a positive value reflects im-proved performance. To examine whether the off-line score significantly differed from zero for eachtest and within each group, single-sample t-tests were conducted. For the testing practice condition,t-tests revealed a significant performance deterioration on the retention test, t(11) = �2.49. No off-lineperformance improvements were found on the extrinsic and intrinsic transformation tests,

Figure 4. Off-line improvements are defined as the average RMSE difference (10-min testing minus 24-hr testing) on theretention (R), extrinsic transformation (Et) and intrinsic transformation (It) tests for the Testing groups (left panel) and theControl groups (right panel). Error bars reflect the standard error of the mean.




t(11) = �.05 and .27, respectively. For the control practice condition, a significant performance deteri-oration was also found on the retention test, t(11) = �2.27. The analysis revealed significant off-lineimprovements in performance on the intrinsic transformation test, t(11) = 2.19, but not on the extrin-sic transformation test, t(11) = 0.70.

4. Discussion

Recent research has provided evidence that testing positively influences the long-term representa-tion of motor skills, yielding enhanced transfer capabilities (Boutin et al., 2012a, 2012b). The presentstudy therefore sought to determine whether these testing benefits result from rapid, within-practiceimprovements and/or latent, between-practice skill improvements. We first confirmed our previousfindings that testing induces generalised long-term motor skill representations. However, the currentresults revealed that testing produced rapid but not latent enhancements of the motor skill. The per-formance improvements and the qualitative skill representational shifts observed at the 10-min test-ing were stabilised but did not extend during the period of consolidation. No testing benefits wereobserved during the latent development of the motor skill representation (i.e., from the 10-min tothe 24-hr testing).

4.1. Testing benefits encoding

Results for the acquisition phase indicated that with or without completion of an initial testing ses-sion, all groups similarly improved their performance during the first and second practice sessions(Acquisition 1 and Acquisition 2, respectively). This suggests that the performance improvements dur-ing Acquisition 2 were not testing-dependent or might have been masked by the knowledge of resultsprovided during the acquisition phase (see also Boutin et al., 2012b). Nonetheless, while initial testingdid not promote subsequent acquisition performance, our findings revealed different patterns of re-sults at the 10-min testing for the TE group when compared to the CE control group. Data indicatedthat the completion of an initial testing session during practice has positively influenced encodingprocessing in enabling the skill to be represented to a greater extent in both extrinsic and intrinsiccoordinates, yielding superior transfer capabilities at later retesting (i.e., 10-min testing).

Our results revealed an early development of an effector-dependent representation, as revealed byperformances at the initial testing session. Specifically, retention and transfer data for the TE and TCgroup indicated that practice induced a rapid effector-dependent representation (lower RMSE on theretention test compared to the transfer tests) that is represented in extrinsic coordinates (lower RMSEon the extrinsic than on the intrinsic transformation test). The current results are consistent with a pre-vious observation from Boutin et al. (2012a), who showed that practice induces rapid and non-trans-ferable (unpractised arm) motor skill improvements based on an extrinsic coding system. However,our findings interestingly showed different transfer patterns at the 10-min testing for the TE and CEpractice conditions. While no performance difference was observed for the retention test betweenthe TE and CE groups, a significant advantage for the TE group was found at the 10-min testing for boththe extrinsic transformation (i.e., the same spatial-temporal pattern) and the intrinsic transformationtests (i.e., the same pattern of muscle activation and/or limb joint angles). The current data indicate thatthe initial testing session promoted the rapid, within-practice development of a multiple coding repre-sentation of the motor skill (similar performance on the extrinsic and intrinsic transformation tests),which is normally based on an extrinsic coding system at this stage of practice (lower RMSE on theextrinsic than on the intrinsic transformation test for the CE group; see also Boutin et al., 2012a).

The current findings revealed that the initial testing session, which was performed in the TE prac-tice condition, amplified subsequent effector transfer as revealed by the performance observed at the10-min testing. This indicates that testing has the potential to induce within-practice skill improve-ments and yield enhanced transfer capabilities, without necessarily facilitating or interfering withthe learning of the motor task (similar performance for the TE and CE groups on the retention test).It should be noted, however, that the absence of testing benefits on the retention test are likely dueto the retrieval of information that is reduced or even eliminated during training when KR is provided




(e.g., Salmoni, Schmidt, & Walter, 1984). In this case, the mere completion of one extra block of retrie-val practice without KR on the criterion task during initial testing was presumably not sufficient toboost subsequent retention in comparison to the large number of practice blocks performed on thattask with KR. Additionally, because acquisition performance levelled off after only 11 practice blocks,one cannot rule out a possible ceiling effect on acquisition and retention performance, thus nullifyingthe potential learning benefits of testing.

The testing benefits observed on both extrinsic and intrinsic transfer tests at the 10-min testing forparticipants assigned to the TE practice condition are suggested not to be attributed to the mere act ofretrieval practice but rather to the evoked degree of elaborative and distinctive processing, which de-pends on when retrieval takes place (see Boutin et al., 2012b). Indeed, it is theoretically assumed thatexperiencing an initial testing session during early practice offers learners the opportunity to engagein deeper elaborative and distinctive processing (the elaboration hypothesis, see Shea & Morgan, 1979;Shea & Zimny, 1983, 1988), such as inter-task comparisons of the to-be-learned and to-be-remem-bered tasks, which are thought to highlight the similarities and differences between the tasks (the cri-terion task and the new transfer tasks). This allows the learner to encode critical task-relatedinformation and leads to a stronger and more elaborate motor skill representation (Wright, Li, & Whit-acre, 1992). This provides theoretical explanation as to why within-practice retrieval processes do af-fect subsequent encoding processing. Therefore, we suggest that retrieval processing during taskencoding is critical for skill enhancements to occur. This agrees with Boutin et al. (2012b) who statedthat the scheduling of practice and testing sessions provides greater encoding variability and yieldsenhanced transfer capabilities. It is worth noting that prior exposure to the new transfer tasks in itselfis not responsible for the representational skill benefits observed at later retesting (Boutin et al.,2012b).

In addition, the benefits induced by deeper elaborative and distinctive processing during practicehave been shown to further extend beyond the end of training during the period of consolidation,yielding enhanced long-term retention and transfer performance (e.g., Boutin & Blandin, 2010a,2010b; Shea, Lai, Black, & Park, 2000). Thus, the magnitude of the testing effect observed during thefast, within-practice learning phase should last or extend during the period of consolidation (i.e., fromthe 10-min to the 24-hr testing).

4.2. Retrieval and consolidation processing

The results reported here suggest that the benefits of the initial testing session on the transfer per-formance observed at the 10-min testing are not compromised after a 24-hr period of rest, even in theabsence of additional practice. We observed better performance at the 24-hr testing for the group thatwas permitted to perform an initial testing session early in practice (TC practice condition) relative tothe group that did not (CC practice condition). Importantly, this demonstrates that the mere comple-tion of an initial testing session early in practice is crucial for maximising the long-term representa-tion of the motor task. The retention and transfer data for the TC practice condition at the 24-hr testingindicate that the initial testing session promoted the development of a multiple coding (similar per-formance on the extrinsic and intrinsic transformation tests) and effector-independent motor skillrepresentation (no performance difference between the retention and transfer tests). This suggeststhat both the extrinsic and intrinsic coding systems potentially contribute to retention and transferperformance. Conversely, for the CC practice condition, no effector transfer was observed for an extrin-sic and intrinsic transformation of the original pattern (similar performances on the extrinsic andintrinsic transformation tests). Note that the pattern of results observed for participants in the controlgroups at the 10-min and 24-hr testing replicated the previous findings of Boutin et al. (2012a), inwhich participants rehearsed the testing sessions (i.e., within-group design). The current resultsmay therefore not be explained by a positive influence during initial testing of embedded propriocep-tive feedback or the variability of practice induced by the transfer tasks on subsequent performance atlater retesting. This indirectly confirms the critical role of elaborative and distinctive processing in thewithin-practice skill enhancements observed in this study.

In comparison to the fast retrieval-based effects induced by testing during practice, our findings donot point to a critical contribution of system consolidation during the period of rest, at least in terms of




performance improvements. Indeed, the magnitude of the transfer performance at the 24-hr testingwas not enhanced in comparison to the transfer performance on tests performed shortly after theend of practice (i.e., 10-min testing). We found that the performance improvements observed at the10-min testing on both transfer tests were stabilised but did not extend during the off-line period. Thisis not the case for the control groups, where our results corroborate the previous findings of Boutinand colleagues (2012a) and indicate between-practice skill enhancements involving an off-line devel-oping effector-dependent component that is represented in intrinsic coordinates late in practice. Alto-gether, these findings provide strong evidence that testing does not influence the process ofconsolidation, suggesting that testing does not have relevance for skill consolidation.

Interestingly, participants in the TC group displayed an ability to switch between hands after a 24-hr period of rest without any performance deterioration (i.e., similar performances on the retentionand transfer tests), while their CC group counterparts were not able to do so (i.e., better performanceon the retention than on the transfer tests). Indeed, while the results indicated an effector-dependentrepresentation of the motor skill at the initial testing (TE and TC groups) and 10-min testing (TE and CEgroups), they revealed in contrast an effector-independent representation at the 24-hr testing for theTC group but not for the CC group. Our findings suggest that an effector-dependent representation of amotor skill is built quickly during practice and that the subsequent consolidation processes have thecapacity to restore the effector-independent representation of the skill and, in this way, contribute tomotor skill generalisation. One possible explanation is that during consolidation, the acute physiolog-ical effects induced by the lateralised rehearsal of the action potentials required for task productionduring training are eliminated. Indeed, practice might have induced an acute effector-dependent rep-resentation during encoding, which was then slowly recovered during the period of consolidation inthe absence of additional physical practice. This latent recovery occurred in both groups and yieldedperformance decrements on the retention test, but only resulted in an effector-independent represen-tation for the TC group; this is likely due to the lack of performance improvements on both transfertests for the CC group during encoding. Future research might therefore be conducted to determinewhether these latent restorative effects during consolidation are sleep- or time-dependent (see alsoSheth, Janvelyan, & Khan, 2008).

In conclusion, the present study extends the theoretical background of the testing effect on motorskill learning and transfer and provides a first step in understanding the origin of generalised motorrepresentations (see also Krakauer, Mazzoni, Ghazizadeh, Ravindran, & Shadmehr, 2006). We firstdemonstrate that the long-lasting testing benefits on effector transfer are essentially due to rapid,within-practice skill improvements. This corroborates and adds to the recent finding of Boutin et al.(2012a), suggesting that the reactivation and updating of existing consolidated motor representationsduring retrieval practice or after additional practice is critical for further modifications, a processknown as ‘‘reconsolidation’’ (e.g., Dudai & Eisenberg, 2004; Przybyslawski & Sara, 1997; Stickgold &Walker, 2007). In fact, reactivation provides the possibility to modify and update the existing repre-sentational trace of the skill with reference to newly encountered information (Dudai, 2006). In thiscase, if the newly encoded information matches the existing trace, then the skill representation islikely to be strengthened or even improved by a process of ‘‘re-encoding’’ (Moscovitch, Nadel, Wino-cur, Gilboa, & Rosenbaum, 2006; Walker, Brakefield, Hobson, & Stickgold, 2003) and to become moregeneralised if reactivation occurs in a different context (e.g., Gordon, 1981; see also Rasch & Born,2007, for a review). Together, these results indicate that testing contributes to the generalisation ofmotor skills during the process of encoding and reconsolidation (i.e., re-encoding) but not duringconsolidation.

5. Summary

In the present study, we showed that testing induces rapid, within-practice but not latent, be-tween-practice motor skill enhancements. During the fast learning phase, we found that the comple-tion of an initial testing session early in practice yielded superior immediate and long-term transfercapabilities. Conversely, we did not find any advantage of testing on the latent development of themotor skill representation during the off-line period (i.e., from the 10-min to the 24-hr testing). Thetesting benefits observed at the end of practice were stabilised but did not extend during the period




of consolidation. Our results indicated that testing contributes to the generalisation of motor skillsduring encoding but not consolidation.

References

Boutin, A., Badets, A., Salesse, R. N., Fries, U., Panzer, S., & Blandin, Y. (2012a). Practice makes transfer of motor skills imperfect.Psychological Research, 76, 611–625.

Boutin, A., & Blandin, Y. (2010a). Cognitive underpinnings of contextual interference during motor learning. Acta Psychologica,135, 233–239.

Boutin, A., & Blandin, Y. (2010b). On the cognitive processes underlying contextual interference. Contributions of practiceschedule, task similarity and amount of practice. Human Movement Science, 29, 910–920.

Boutin, A., Panzer, S., Salesse, R. N., & Blandin, Y. (2012b). Testing promotes effector transfer. Acta Psychologica, 141, 400–407.Boutin, A., Fries, U., Panzer, S., Shea, C. H., & Blandin, Y. (2010). Role of action observation and action in sequence learning and

coding. Acta Psychologica, 135, 240–251.Brainard, D. H. (1997). The psychophysics toolbox. Spatial Vision, 10, 433–436.Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hilsdale, NJ: Erlbaum.Colby, C. L., & Goldberg, M. E. (1999). Space and attention in parietal cortex. Annual Review of Neuroscience, 22, 319–349.Criscimagna-Hemminger, S. E., Donchin, O., Gazzaniga, M. S., & Shadmehr, R. (2003). Learned dynamics of reaching movements

generalize from dominant to nondominant arm. Journal of Neurophysiology, 89, 168–176.Dudai, Y. (2006). Reconsolidation: The advantage of being refocused. Current Opinion in Neurobiology, 16, 174–178.Dudai, Y., & Eisenberg, M. (2004). Rites of passage of the engram: Reconsolidation and the lingering consolidation hypothesis.

Neuron, 44, 93–100.Gordon, W. C. (1981). Mechanisms of cue-induced retention enhancement. In N. E. Spear & R. R. Miller (Eds.), Information

processing in animals: Memory mechanisms (pp. 319–339). Hillsdale, NJ: Erlbaum.Hikosaka, O., Nakahara, H., Rand, M. K., Sakai, K., Lu, X., Nakamura, K., et al (1999). Parallel neural networks for learning

sequential procedures. Trends in Neuroscience, 22, 464–471.Karni, A., & Bertini, G. (1997). Learning perceptual skills: Behavioral probes into adult cortical plasticity. Current Opinion in

Neurobiology, 7, 530–535.Karni, A., Meyer, G., Jezzard, P., Adams, M. M., Turner, R., & Ungerleider, L. G. (1995). Functional MRI evidence for adult motor

cortex plasticity during motor skill learning. Nature, 377, 155–158.Kleiner, M., Brainard, D., & Pelli, D. G. (2007). What’s new in Psychtoolbox-3? Perception, 36, ECVP Abstract Supplement.Korman, M., Raz, N., Flash, T., & Karni, A. (2003). Multiple shifts in the representation of a motor sequence during the acquisition

of skilled performance. Proceedings of the National Academy of Sciences of the United States of America, 100, 12492–12497.Kovacs, A. J., Muehlbauer, T., & Shea, C. H. (2009). The coding and effector transfer of movement sequences. Journal of

Experimental Psychology: Human Perception and Performance, 35, 390–407.Krakauer, J. W., Ghilardi, M. F., & Ghez, C. (1999). Independent learning of internal models for kinematic and dynamic control of

reaching. Nature Neuroscience, 2, 1026–1031.Krakauer, J. W., Mazzoni, P., Ghazizadeh, A., Ravindran, R., & Shadmehr, R. (2006). Generalization of motor learning depends on

the history of prior action. PLoS Biology, 4, 1798–1808.Krakauer, J. W., & Shadmehr, R. (2006). Consolidation of motor memory. Trends in Neurosciences, 29, 58–64.Lange, R. K., Godde, B., & Braun, C. (2004). EEG correlates of coordinate processing during intermanual transfer. Experimental

Brain Research, 159, 161–171.McGaugh, J. L. (2000). Memory – a century of consolidation. Science, 287, 248–251.Moscovitch, M., Nadel, L., Winocur, G., Gilboa, A., & Rosenbaum, R. S. (2006). The cognitive neuroscience of remote episodic,

semantic and spatial memory. Current Opinion in Neurobiology, 16, 179–190.Panzer, S., Krueger, M., Muehlbauer, T., Kovacs, A. J., & Shea, C. H. (2009). Inter-manual transfer and practice. Coding of simple

motor sequences. Acta Psychologica, 131, 99–109.Pelli, D. G. (1997). The VideoToolbox software for visual psychophysics: Transforming numbers into movies. Spatial Vision, 10,

437–442.Przybyslawski, J., & Sara, S. J. (1997). Reconsolidation of memory after its reactivation. Behavioural Brain Research, 84, 241–246.Rasch, B., & Born, J. (2007). Maintaining memories by reactivation. Current Opinion in Neurobiology, 17, 698–703.Robertson, E. M. (2009). From creation to consolidation: A novel framework for memory processing. PLoS Biology, 7, 11–19.Robertson, E. M., Pascual-Leone, A., & Miall, R. C. (2004). Current concepts in procedural consolidation. Nature Reviews

Neuroscience, 5, 576–582.Robertson, E. M., Pascual-Leone, A., & Press, D. Z. (2004). Awareness modifies the skill-learning benefits of sleep. Current Biology,

14, 208–212.Salmoni, A. W., Schmidt, R. A., & Walter, C. B. (1984). Knowledge of results and motor learning: A review and critical reappraisal.

Psychological Bulletin, 95, 355–386.Shadmehr, R., & Brashers-Krug, T. (1997). Functional stages in the formation of human long-term motor memory. Journal of

Neuroscience, 17, 409–419.Shadmehr, R., & Holcomb, H. H. (1997). Neural correlates of motor memory consolidation. Science, 277, 821–825.Shea, C. H., Lai, Q., Black, C., & Park, J.-H. (2000). Spacing practice sessions across days benefits the learning of motor skills.

Human Movement Science, 19, 737–760.Shea, J. B., & Morgan, R. L. (1979). Contextual interference effects on the acquisition, retention, and transfer of a motor skill.

Journal of Experimental Psychology: Human Learning and Memory, 5, 179–187.Shea, J. B., & Zimny, S. T. (1983). Context effects in memory and learning movement information. In R. A. Magill (Ed.), Memory

and control of action (pp. 145–366). Amsterdam: North-Holland.Shea, J. B., & Zimny, S. T. (1988). Knowledge incorporation in motor representation. In O. G. Meijer & K. Roth (Eds.), Complex

movement behavior: ‘‘The’’ motor-action controversy (pp. 289–314). Amsterdam: North-Holland.


http://refhub.elsevier.com/S0167-9457(12)00120-0/h0005

































































Sheth, B. R., Janvelyan, D., & Khan, M. (2008). Practice makes imperfect: Restorative effects of sleep on motor learning. PLoS One,3, e3190.

Soechting, J. F., & Flanders, M. (1989). Errors in pointing are due to approximations in sensorimotor transformations. Journal ofNeurophysiology, 62, 595–608.

Stickgold, R., & Walker, M. P. (2007). Sleep-dependent memory consolidation and reconsolidation. Sleep Medicine, 8, 331–343.Wagner, U., Gais, S., Haider, H., Verleger, R., & Born, J. (2004). Sleep inspires insight. Nature, 427, 352–355.Walker, M. P. (2005). A refined model of sleep and the time course of memory formation. Behavioral and Brain Sciences, 28,

51–64.Walker, M. P., Brakefield, T., Hobson, J. A., & Stickgold, R. (2003). Dissociable stages of human memory consolidation and

reconsolidation. Nature, 425, 616–620.Witt, K., Margraf, N., Bieber, C., Born, J., & Deuschl, G. (2010). Sleep consolidates the effector-independent representation of a

motor skill. Neuroscience, 171, 227–234.Wright, D. L., Li, Y., & Whitacre, C. (1992). The contribution of elaborative processing to the contextual interference effect.

Research Quarterly for Exercise and Sport, 63, 30–37.

















Documents

Retrieval practice in motor learning