Acta Psychologica 155 (2015) 92–100

Contents lists available at ScienceDirect

Acta Psychologica journal homepage: www.elsevier.com/ locate/actpsy

The impact of concurrent visual feedback on coding of on-line and pre-planned movement sequences Peter Leinen a, Charles H. Shea b, Stefan Panzer a,⁎ a b

Saarland University, Germany Texas A&M University, United States

a r t i c l e

i n f o

Article history: Received 12 September 2013 Received in revised form 18 December 2014 Accepted 20 December 2014 Available online 13 January 2015 PsycINFO classification: 2330 Motor Processes 2340 Cognitive Processes 2343 Learning & Memory Keywords: Control process Representation Visual feedback Motor learning

a b s t r a c t The purpose of this study was to determine the extent to which participants could effectively switch from on-line (OL) to pre-planned (PP) control (or vice versa) depending on previous practice conditions and whether concurrent visual feedback was available during transfer testing. The task was to reproduce a 2000 ms spatial–temporal pattern of a sequence of elbow flexions and extensions. Participants were randomly assigned to one of two practice conditions termed OL or PP. In the OL condition the criterion waveform and the cursor were provided during movement production while this information was withheld during movement production for the PP condition. A retention test and two effector transfer tests were administered to half of the participants in each acquisition conditions under OL conditions and the other half under PP conditions. The mirror effector transfer test required the same pattern of muscle activation and limb joint angles as required during acquisition. The non-mirror transfer test required movements to the same visual–spatial locations as experienced during acquisition. The results indicated that when visual information was available during the transfer tests performers could switch from PP to OL. When visual information was withdrawn, they shifted from the OL to the PP-control mode. This finding suggests that performers adopt a mode of control consistent with the feedback conditions provided during testing. © 2014 Elsevier B.V. All rights reserved.

1. Introduction How movement sequences are represented, and processed in the brain, and which neural bases are associated during the course of learning have stimulated the research since Lashley's (1951) seminal work on ‘The problem of serial order in behavior’ (Bapi, Miyapuram, Graydon, & Doya, 2006; Hikosaka, Nakamura, Sakai, & Nakahara, 2002; Keele, Ivry, Mayr, Hazeltine, & Heuer, 2003; Kirsch & Kunde, 2012; Korman, Raz, Flash, & Karnim, 2003; Nakahara, Doya, & Hikosaka, 2001; Shea & Kovacs, 2013; Shea & Wright, 2012; Tanaka & Watanabe, 2014). Theoretical frameworks of sequence learning have suggested that independent codes, coordinate systems or representations are responsible for sequence production and that several sources of information available during the course of practice assist the learning process (Keele et al., 2003). Further, these frameworks often argue that practice related changes occur. That is, the development of and the reliance on a particular representation changes during the course of practice depending on the practice conditions (Bapi et al., 2006; Dirnberger & Novak-Knollmueller, 2013).

⁎ Corresponding author at: Saarland University, Im Stadtwald B8.2, D-66041 Saarbrücken, Germany. Tel.: +1 49 681 302 2777; fax: +1 49 681 302 4901. E-mail address: [email protected] (S. Panzer).

http://dx.doi.org/10.1016/j.actpsy.2014.12.005 0001-6918/© 2014 Elsevier B.V. All rights reserved.

One theoretical model by Hikosaka which is based on behavioral and brain imaging data (Bapi, Doya, & Harner, 2000; Hikosaka et al., 1999, 2002) proposed that the learning of movement sequences involves both a fast developing, effector independent component represented in a visual–spatial coordinate system (e.g., spatial locations of the end effector and/or sequential target positions), and a slower developing effector dependent component represented in motor coordinates (e.g., sequence of activation patterns of the agonist/antagonist muscles and/or achieved joint angles). Note that both coordinate systems are thought to develop in parallel and representations in both coordinate systems are learned concurrently, but each of the processes operates in a single coordinate system. Hikosaka et al. (1999) suggested that the representation for a given sequence is distributed in the brain in different forms (visual–spatial and motor) with distinct neural networks supporting sequence production. In an initial pre-learning stage in which participants perform actions in a discrete, step-by-step manner by relying on the sensorimotor transformation for each action, where sequence production relies on serial sensorimotor processes for each action. By repeating the actions in a fixed order, new connections are formed between the mechanisms for individual actions, thus enabling the participants to perform the actions sequentially without relying on these step by step sensorimotor processes. At this stage of practice sequence production relies primarily on visual–spatial coordinates, but with increasing practice there is a shift in the reliance to the motor

P. Leinen et al. / Acta Psychologica 155 (2015) 92–100

coordinate system which is thought to be predominantly responsible for sequence execution later in practice (see also Bapi et al., 2000; Sakai et al., 1998). Recent behavioral studies using an inter-manual transfer design have provided empirical evidence that not only practice, but other factors play an important role in determining the more effective coordinate system available for sequence production (Keele, Jennings, Jones, Caulton, & Cohen, 1995; Kovacs, Han, & Shea, 2009; Kovacs, Muehlbauer, & Shea, 2009; Park & Shea, 2005; Shea, Kovacs, & Panzer, 2011 for an overview, Verwey & Clegg, 2005). It appears that task complexity (number of reversal in the sequence and/or sequence duration), and the availability of concurrent visual feedback during sequence production play an important role in determining which of the two representations is the more effective coding scheme for sequence production (Panzer, Krueger, Muehlbauer, Kovacs, & Shea, 2009; Shea et al., 2011). Kovacs, Han, and Shea (2009), for example, had participants practice either a relatively simple or a slightly more complex sequence of extension flexion movements for 99 trials during one practice session. The simple sequence involved 3 movement reversals and the movement duration was 1300 ms, while the more complex sequence involved five reversals and the movement duration was 2000 ms. It is important to note that both movement sequences were constructed by summing two sine-waves such that the first 1300 ms of the longer sequence was the same as the shorter sequence. The amplitudes of the two sine-waves were 45° for the first and 30° for the second. Following 99 practice trials with a right limb a delayed retention and two effector transfer tests (left limb) were administered. In one transfer test, the target locations were a mirror image (mirror transfer test) and required participants to perform the same pattern of muscle activation and limb joint angles as required during acquisition with the contralateral unpracticed effectors. The second transfer test (non-mirror transfer test) required movements to the same visual–spatial locations experienced during acquisition, however, because the contralateral limb was used new un-practiced patterns of muscle activation and joint angles were required to achieve the target locations. The results of the Kovacs, Han, and Shea (2009) experiment indicated that after 99 trials of practice the participants performance of the more complex sequence on the non-mirror transfer test, where the visual–spatial coordinates had been reinstated, was superior to performance on the mirror transfer test. Alternatively, mirror transfer performance of the simple sequence, which required fewer reversals, was superior compared to the non-mirror transfer. This later finding suggests that the simple sequence was more effectively coded in motor coordinates, which requires the same pattern of homologous muscle activation with the contra-lateral unpracticed limb, at this stage of practice. Indeed, the analysis of the kinematics for the simple sequences was consistent with those typically found for pre-planned movements (no on-line corrections) and the kinematics of the more complex sequence was consistent with on-line (feedback) control. That is, the kinematics of the more complex sequence with five reversals and duration of 2000 ms indicated that performers made iterative feedback based corrections during the progress of the movement, typically found in on-line controlled movements while the simple movement sequence was void of these types of corrections. This caused Kovacs and colleagues to hypothesize that the coordinate system used to code sequence information may be dependent on the control process used rather than the amount of practice or the sequence characteristics per se (Kovacs, Muehlbauer, & Shea, 2009; Verwey & Wright, 2004). Another assumption related to control processes is that visual feedback is used for on-line control (Elliott, Helsen, & Chua, 2001; Glover, 2004; Woodworth, 1899). To examine the relationships between the visual–spatial coordinate system, on-line control, and the contribution of concurrent visual feedback Kovacs, Boyle, Gruetzmacher, and Shea (2010) had participants perform a complex sequence with five reversals under two acquisition conditions. In one acquisition condition participants were provided before and during the movement a template

93

indicating the goal pattern and a cursor indicating the current position of their movement in an attempt to promote on-line control. In another condition the goal movement template was presented before responding but the template and cursor disappeared as soon as the movement began. This condition was designed to encourage participants to preplan the movement because extrinsic visual information was not available during response execution making on-line detection and correction of errors difficult. The results of this experiment provided clear evidence that participants in the on-line condition, where the template and cursor indicating the progress of the movement were available during movement execution, coded the sequence more effectively in visual–spatial coordinates. Participants in the pre-plan condition, where the template or cursor was not available during movement execution, coded the sequence more effectively in motor coordinates. These results led the researchers to conclude that the coding of movement sequences is at least to some extent dependent on the external information available during sequence production (concurrent visual feedback) and the different control processes (pre-planning/on-line control) used to produce the sequence. It is also interesting to note that pre-planning and on-line control of movement sequences have been shown to utilize different informations and rely on different neural pathways. Furthermore, pre-planning determines the initial kinematic characteristics of the movement including timing and velocity, while the on-line control monitors and occasionally adjusts movement progress “in flight” but these adjustments are limited to the spatial characteristics of the sequence (Glover, 2004). In other words, shorter duration movements with few elements rely predominantly on preplanning while longer duration movements with more elements have an initial pre-planned component after which movement control is gradually taken over by the on-line control mechanism. Research on the development of the two coordinate systems and the shift in the reliance for sequence production is pre-dominantly aimed at practice (Bapi et al., 2000; Hikosaka, Rand, Miyachi, & Miyashita, 1995; Kovacs, Han, & Shea, 2009) and, to a lesser extent, on information provided during response production. A related but yet unresolved question is whether or not participants can effectively switch between coding schemes and modes of control depending on the information provided. This question is of theoretical interest for two reasons. First, the Hikosaka et al. (1999) perspective also proposed that codes based on visual–spatial and motor coordinates developed in parallel, while the preference of one code for sequence production depends on the stage of practice (Sakai et al., 1998). In line with the theoretical framework of the ‘parallel neural network model’ (see Hikosaka et al., 1999), it also seems possible that either coordinate system can be accessed in any specific sequence production situation (see Kurata & Hoshi, 2002; Kurata & Wise, 1988). Thus, depending on the available extrinsic information, performers use the most salient code for sequence production independent of the stage of practice (see also Clegg, DiGirolamo, & Keele, 1998; Kovacs, Han, & Shea, 2009). This view supports the theoretical idea of more parallel processing in movement sequence learning, where the two coordinate systems cooperate with each other for sequence production. In addition this research question has the potential to shed some light on the ongoing theoretical debate that the two coordinate systems represent two different kinds of representations that form the ends of a continuum and not dichotomous states (Keele et al., 2003; see also Abrahmase, Ruitenberg, de Kleine, & Verwey, 2013). Second, an additional assumption of the planning–control model proposed by Glover (2004) is that participants can alternate between control modes on successive sequence productions, but attempt to choose the optimal control scheme given the nature of the sequence and characteristics of the performance environment. The view that pre-planning and on-line control complement one another was also developed by Woodworth (1899). The control of a movement was conceptualized as preprogrammed at the initial phase and corrected with visual feedback as the movement progressed, what Woodworth called

94

P. Leinen et al. / Acta Psychologica 155 (2015) 92–100

‘current control’ (see also Elliott et al., 2001; Thompson, McConnell, Slocum, & Bohan, 2007). The primary goal of the present experiment was to determine the extent to which performers can effectively alternate between the two control modes and the associated coordinate systems for sequence production when concurrent visual feedback is provided or withdrawn on a transfer test. Based on previous results from Kovacs, Muehlbauer, and Shea (2009), and the assumption that both coordinate systems developed in parallel and that both are available (Hikosaka et al., 1999; Kurata & Hoshi, 2002; Kurata & Wise, 1988), we hypothesized that participants, who are provided with concurrent visual feedback on the transfer test will perform the non-mirror transfer test better than the mirror transfer test independent of whether the previous practice promoted on-line or pre-plan control. The opposite pattern of transfer performance is expected for participants where concurrent visual feedback is not provided during the transfer tests. More specifically, if codes in both coordinate systems are available then we expected a shift in the reliance when concurrent visual feedback is available versus when it is withheld during the transfer test relative to when these conditions were available or not available during acquisition, retention and the transfer tests. This finding would be consistent with the hypothesis that testing feedback conditions available in the testing environment play a role in determining the most salient coding system. Note that in consideration of the early stage of practice, where the production of movement sequences is believed to rely more heavily on the visual– spatial coordinate system and the findings provided by Kovacs et al. (2010) that visual information facilitates perceptual-motor sequence learning (see also Elliott, Chua, Pollock, & Lyons, 1995), we hypothesized that sequence performance will be superior when concurrent visual feedback is provided compared to when visual feedback is withheld independent of whether the participants acquired the sequence with or without concurrent visual feedback. In addition and in line with Woodworth's assumption of ‘current control’, we expected that concurrent visual feedback enables performers to make effective iterative movement corrections to reduce errors during movement execution. A kinematic analysis is included to identify any systematic changes in the movement trajectories as a result of concurrent visual feedback. Therefore the index of harmonicity characterizes the subtle on-line adjustments during movement execution (Guiard, 1993). This type of correction would result in decreases in movement harmonicity resulting from the subtle changes in the movement trajectory. Therefore we expect a decrease in the index of harmonicity when concurrent visual feedback is available during sequence production relative to when concurrent visual feedback is withheld. 2. Method 2.1. Participants Undergraduate students (N = 36; mean age 24.5 years, SD = 2.87 years; 18 males and 18 females) participated for course credit in this experiment. Prior to completing the experiment each participant completed an informed consent form approved by the local ethic committee. All participants were right-hand dominant as determined by the Edinburgh Handedness Inventory (Oldfield, 1971) prior to the experiment and all were unaware of the goals of the experiment. 2.2. Apparatus The apparatus consisted of two horizontal levers supported at proximal end by a vertical axle that turned almost frictionless in a ballbearing support. The supports were fixed on the left and right sides of a table, allowing the lever to move in the horizontal plane over the table. At the distal end of each lever, a vertical handle was fixed. The handle's position could be adjusted so that, when grasping the handle, the participant's elbow could be aligned with the axis of rotation. A

potentiometer was attached to the lower end of the axis to record the position of the lever and its output was sampled at 1000 Hz. A wooden cover was placed over the table to prevent participants from seeing the lever and their arm. A projector (100 Hz) was used to display the goal waveform (see Fig. 1A) and feedback on the wall facing the participant. The participants were seated at about 2 m from the wall and a 1.64 × 1.23 m image was projected in front of them on the wall. 2.3. Task, experimental groups and procedures Participants were seated in front of the apparatus on an adjustable height chair so that their lower arm was positioned at an approximately 85° angle to their upper arm in the starting position (Fig. 1G). At the beginning of each trial the goal waveform was displayed (Fig. 1A, D) and participants were asked to move the lever to the starting position (1° area at the beginning of the goal waveform). The goal waveform, a spatial–temporal waveform pattern of 2000 ms duration was created by summing two sine waves with different periods and amplitudes. The maximum amplitude in the goal waveform was 45° from the start position. One second after positioning the cursor in the start position, a tone indicated to the participant to begin his/her response when he/she was ready. Data collection was triggered by the movement of the lever. If the participant moved from the start position prior to presentation of the tone, the participant was required to move back to the start position for 1 s before the tone was presented again. This insured that the participants started in the correct position on the criterion sequence, but could initiate their response when they felt ready. The participants were instructed to move the lever with their dominant right arm through a sequential pattern of extension–flexion movements (5 reversals; changing the movement direction from extension to flexion or vice versa) in an attempt to produce the criterion spatial–temporal pattern displayed in front of them as accurately as possible. Approximately 1 s following the completion of the participant's response the criterion waveform and the movement pattern produced by the participant were overlaid in the display for 10 s (Fig. 1C, F). The time interval of 10 s was used to ensure that participants had enough time to process feedback information. During this time interval participants were not allowed to move the lever from start position. Prior to entering the testing room participants had been assigned to one of two acquisition conditions: an on-line (OL) or a pre-plan (PP) acquisition condition (see Table 1). In the OL condition, the participant was able to view the criterion waveform and the cursor indicating the position of their limb during movement production (Fig. 1B). In the PP condition, the criterion waveform and cursor were removed from the display during the production of the movement sequence (Fig. 1E). Note, that prior to the initiation of the movement and after movement production, both groups were provided identical information in the visual displays. The post-response display was considered knowledge of result (KR) because the actual pattern of movement produced by the participant (gray waveform) was overlaid on the goal movement pattern (black waveform). The limb used and the orientation of the criterion waveform during acquisition, retention, and the various transfer tests are depicted in Fig. 2. Acquisition practice under the online and preplanned conditions consisted of 11 blocks of 9 trials (Fig. 2A). Approximately 24 h after the completion of the acquisition phase, participants from each acquisition condition (OL and PP) returned for a retention test (Fig. 2B). All groups performed a retention test under the same conditions as experienced during acquisition except that the display of the criterion movement overlaid on the participant's movement (KR) was not provided as it had been during acquisition. After the retention test half of the participants from each acquisition condition were randomly assigned to an OL or a PP condition and administered “mirror” and “non-mirror” contra-lateral transfer tests (Fig. 2C, D). The two transfer tests (9 trials

P. Leinen et al. / Acta Psychologica 155 (2015) 92–100

95

Fig. 1. Schematic illustration of a sequence during a trial for online (A–C) and pre–plan (D–F) conditions during acquisition and retention. The left panels illustrate the display of the target waveform prior to movement initiation and the center panels illustrate the display during the movement. Note that in the on-line condition that the target waveform and cursor indicating the current position of the limb are provided while this information is withdrawn upon movement initiation in the pre-plan condition. The right panels illustrate the feedback provided by both conditions. This feedback includes the target waveform overlaid on the participant produced waveform. The participant's position on all acquisition and retention trials is also provided (G).

each) were performed with the contralateral limb (order counterbalanced). In the mirror transfer test the goal-waveform was mirrored on the vertical axis so that the same patterns of homologous muscle activation and limb joint angles were required as during acquisition and on the retention test. In the non-mirror transfer test the visual–spatial

Table 1 Experimental design. Day 1 Groups Pre-plan (PP) Online (OL)

Acquisition conditions

Day 2 Retention

N = 18

N = 18

N = 18

N = 18

Transfer M–NM PP–PP PP–OL OL–OL OL–PP

N=9 N=9 N=9 N=9

Note: “NM” and “M” indicate non-mirror transfer test and mirror transfer test, respectively.

locations of the target waveform used during acquisition and experienced during retention testing were reinstated but the participant used the contralateral limb. Note that in the mirror transfer test the motor coordinates used during acquisition were reinstated while the visual–spatial coordinates were changed. Alternatively, in the non-mirror transfer test the visual–spatial coordinates used during acquisition were reinstated but the motor coordinates were changed. In the OL–OL and PP–OL conditions online feedback was provided during the transfer test, while in the OL–PP and PP–PP conditions online feedback during movement production was withdrawn (see also Table 1). 2.4. Data analysis and measures Data analysis was performed using Matlab (Mathworks, Natick, MA; Version 2012a). The individual trial time series were used to compute lever displacement. To reduce noise the displacement time series was filtered with a 2nd order dual-pass Butterworth filter with a cutoff

96

P. Leinen et al. / Acta Psychologica 155 (2015) 92–100

Fig. 2. The display illustrating the orientation of the target waveform for each acquisition, retention, mirror, and non-mirror tests is provided (A–D, respectively). The limb used and the direction of the initial movement are also provided (E–H, respectively). Note that retention test was conducted approximately 24 h following the acquisition session. Following the retention test the two transfer tests (mirror and non-mirror) were conducted in a counter balanced order.

frequency of 10 Hz. Root Mean Square Error (RMSE) was computed to estimate the performance error in achieving the goal movement pattern. RMSE captures errors in both amplitude and time. Additionally, RMSE incorporates both the variability and bias of the performed movement pattern. RMSE was computed by the difference between the criterion and the actual movement pattern at each data point in the time series. Next, the differences for each data point in the time series were squared and the mean of the squared differences was computed on a trial basis. The final step in computing RMSE was to determine the square root of the mean of the squared differences. Values of RMSE for individual trials were then averaged to yield a global estimate of RMSE for each block (9 trials). Additionally an index of harmonicity (H) was computed as an indicator of subtle adjustments that might occur during movement production. For calculation windows between a pair of zero crossings in the displacement trace are defined in order to compute H (Guiard, 1993). Each non-overlapping time window comprises a single movement reversal. Within each time window, all deflections of the normalized acceleration trace are identified. When the acceleration trace is positive (negative displacement) within this window, H is computed as the ratio of minimum to maximum acceleration. Conversely, when the acceleration trace is negative (positive displacement) within this time window, H is computed as the absolute ratio of maximum to minimum acceleration. When a single peak (sinusoidal acceleration) occurs in the acceleration trace within this window the value of H is set to 1, indicating harmonic motion of the limb. If the acceleration trace crosses from positive to negative (or vice versa) within this window, the value of H is set to 0, indicating inharmonic motion. Finally, the individual harmonicity values of each time window for a trial are averaged yielding a global estimate of H. Adjustments were made for violations of homogeneity and sphericity (Winer, 1971). Partial eta square (ηp2) is the effect size reported for all significant effects (Cohen, 1988). Post-hoc comparisons of significant main effect were computed using the Duncan multiple range tests and paired t-tests. Detailed analyses of significant main effects, interaction effects, and post-hoc tests were followed by Bonferroni-corrected pairwise comparisons when necessary. 3. Results

with repeated measures on Block. The analysis of H was conducted to determine if there was an evidence of the participants to make adjustments during the sequence execution. Note, lower values of H would indicate increased number and/or magnitude of subtle on-line adjustments. 3.1.1. Acquisition: RMSE Mean RMSE and SEMs during acquisition and retention are displayed on Fig. 3A. The analysis detected an Acquisition condition × Block interaction, F(10,340) = 2.81, MSE = 4.19, p b .05, ηp2 = .08. Simple main effect analysis indicated that the Acquisition condition × Block interaction accrued from a faster decrease in RMSE for the OL condition on some of the early blocks (Blocks 1 and 3) relative to the PP condition. In addition the main effect of block, F(10,340) = 51.49, MSE = 4.19, p b .01, ηp2 = .60 reached significance. Duncan's multiple range test found that RMSE decreased through Block 6. No differences were detected for Blocks 7–11. The main effect of acquisition condition was also significant, F(1,34) = 156.65, MSE = 37.19, p b .01, ηp2 = .82, with smaller RMSE for the OL condition when concurrent visual information was available during sequence production. 3.1.2. Acquisition: H The analysis detected a main effect of block, F(10,340) = 16.47, MSE = 0.17, p b .01, ηp2 = .33. Duncan's multiple range test indicated that RMSE increased through Block 5. No differences were detected for Blocks 6–11. Both groups increased the H-value from block 1 to 5. The Acquisition condition × Block interaction was also significant, F(10,340) = 2.26, MSE = 0.01, p b .05, ηp2 = .06. Simple main effect analysis of the Acquisition condition × Block interaction indicated that the change was greater for the OL condition than for the PP condition. The main effect of acquisition condition failed to reach significance, F(1,34) b 1, MSE = 0.07, p N .05. 3.2. Transition from acquisition to retention: RMSE and H To analyze the effects of the transition from the end of the acquisition phase to the retention test, a 2 (acquisition condition: OL, PP) × 2 (phase: block 11 acquisition condition–retention test) ANOVA with repeated measures on phase was conducted for RMSE and H.

3.1. Acquisition: RMSE and H Mean RMSE and H on acquisition were analyzed in a 2 (acquisition condition: OL, PP) × 11 (block: 1–11) Analysis of Variance (ANOVA)

3.2.1. Transition from acquisition to retention: RMSE The analysis indicated a main effect of acquisition condition, F(1,34) = 20.23, MSE = 14.15, p b .01, ηp2 = .37. Participants in the

P. Leinen et al. / Acta Psychologica 155 (2015) 92–100

97

condition performed with lower RMSE (M = 5.52°, SEM = 0.21°) than participants for the PP acquisition condition (M = 10.97°, SEM = 0.69°).

3.3.2. Retention: H The analysis revealed a main effect for acquisition condition, F(1,34) = 4.59, MSE = 0.013, p b .05, ηp2 = .12 with lower H values for participants at the OL acquisition condition (M = 0.84, SEM = 0.04) than for participants at the PP acquisition condition (M = 0.92, SEM = 0.01).

Fig. 3. Mean RMSE for the on-line (OL) and pre-plan (PP) conditions during the acquisition phase and on the retention test (A), and on the mirror and non-mirror transfer tests (B). Note that the groups are coded according to the conditions experienced during acquisition and on the transfer test. For example, the OL–OL group experienced on-line conditions during both acquisition and on the transfer tests while the OL–PP experienced on-line conditions during acquisition but pre-planned conditions on the transfer tests. Error bars represent standard errors. The p-values represent the results of the paired t-tests (panel B).

OL condition performed with lower RMSE compared to participants of the PP condition. All other analysis failed to reach significance. 3.2.2. Transition from acquisition to retention: H The analysis revealed an Acquisition condition × Phase interaction, F(1,34) = 7.14, MSE = .01, p b .05, ηp2 = .17. Simple main effect analysis indicated that H-values from the end of practice to the retention test decreased for participants in the OL condition, while no difference was detected for the participants of the PP condition. All other analysis failed to reach significance. 3.3. Retention and transfer: RMSE and H Mean RMSE and H on the retention test were analyzed in an acquisition condition ANOVA. Mean RMSE and H of the transfer tests were analyzed in a 2 (acquisition condition: OL, PP) × 2 (test condition: OL, PP) × 2 (test: mirror, non-mirror) ANOVA with repeated measures on test. 3.3.1. Retention: RMSE The analysis detected a main effect of acquisition condition, F(1,34) = 56.26, MSE = 4.74, p b .01, ηp2 = .62. Participants in the OL

3.3.3. Transfer: RMSE Mean RMSEs and SEMs of the two transfer tests for each group are provided in Fig. 3B. The analysis detected a three-way Acquisition condition × Test condition × Test interaction, F(1,32) = 4.74, MSE = 5.40, p b .05, ηp2 = .13. Simple main effect analysis of the three-way interaction indicated that the RMSE in the test (mirror, non-mirror) was different for the test conditions (OL, PP). Further the analysis indicated for the mirror transfer tests on the PP test condition no differences between the OL–PP and PP–PP conditions. Subsequent post-hoc analysis with paired t-tests of the various test conditions revealed that the RMSE on the non-mirror transfer test for the OL–OL and PP–OL conditions differed significantly (p b .01) from the mirror transfer test (Table 2). The same analysis for the OL–PP and the PP–PP conditions revealed that RMSE on the mirror transfer test differed significantly for the OL–PP (p b .01) and for the PP–PP (p b .05) conditions from the non-mirror transfer test. In sum, it appeared that the interaction accrued, because the pattern of the results differed in the two transfer tests. That is, participants who received concurrent visual feedback during the transfer tests showed superior transfer performance on the non-mirror transfer-test compared to the mirror transfer test, while participants who were not provided concurrent visual information during movement production performed the mirror transfer test better compared to the non-mirror transfer test. Participants that were permitted to use concurrent visual information during acquisition performed the mirror transfer test under the PP test condition as well as participants who were not permitted to use visual information during acquisition. Additionally, the two-way interactions Acquisition condition × Test, F(1,32) = 5.41, MSE = 5.40, p b .05, ηp2 = .14, the Test condition × Test, F(1,32) = 38.19, MSE = 5.40, p b .01, ηp2 = .40, and the Acquisition condition × Test condition, F(1,32) = 7.41, MSE = 18.99, p b .01, ηp2 = .19 were significant. Simple main effect analysis of the Test condition × Test interaction indicated that participants who received concurrent visual feedback during the transfer test performed the non-mirror transfer test with lower RMSE. The participants who were not permitted to use concurrent visual feedback during the transfer tests showed lower RMSE at the mirror transfer test compared to the non mirror transfer test. This supported previous findings from the analysis of the three-way interaction. The main effects of test, F(1,32) = 7.16, MSE = 5.40, p b .05, ηp2 = .18, and test condition reached significance, F(1,32) = 38.98, MSE = 18.99, p b .01, ηp2 = .55. However the main effect of acquisition condition failed to reach significance, F(1,32) b 1, MSE = 18.99, p N .05. The main effect of test condition indicated that if concurrent visual feedback was available RMSE was lower, compared to the test condition when it was not available. Table 2 Means (M) and standard error of the mean (SEM) of the RMSE in deg (°) for the different test conditions at the mirror and non-mirror tests. Test conditions OL–OL

Mirror Non-mirror

OL–PP

PP–PP

PP–OL

M

SEM

M

SEM

M

SEM

M

SEM

8.91° 7.07°

0.44° 0.40°

13.54° 20.86°

0.96° 1.51°

13.48° 16.22°

0.81° 0.80°

12.42° 10.41°

0.49° 0.51°

98

P. Leinen et al. / Acta Psychologica 155 (2015) 92–100

3.3.4. Transfer: H The analysis of H failed to detect any differences between, test condition or test. In addition, all interactions failed to reach significance, too. The main effect of acquisition condition revealed a marginally significant difference, F(1,32) = 3.71, MSE = .022, p = .06, ηp2 = .10, with a higher H-value for the PP acquisition condition (M = .90, SEM = .035) compared to the OL acquisition condition (M = .84, SEM = .025). 4. Discussion The primary purpose of the present experiment was to determine if performers at an early stage of practice can effectively shift between pre-plan and on-line control and the associated coordinate systems (visual–spatial/motor) for sequence production when concurrent visual feedback is available or withheld on the transfer tests. As in the Kovacs et al. (2010) experiment, we hypothesized that providing salient visual error information during sequence production would increase the likelihood that participants would utilize the on-line control mode while withholding this information would increase the likelihood that performers would pre-plan the sequence (see also Shea et al., 2011). Based on the theoretical assumption that both coordinate systems are available (Kurata & Wise, 1988; Kurata & Hoshi, 2002, see also Hikosaka et al., 1999 for the “parallel view”, where both coordinate systems develop in parallel; or the multiple system idea for sequence learning provided by Clegg et al., 1998; Korman et al., 2003; Keele et al., 2003; Nakahara et al., 2001; Tanaka & Watanabe, 2014), we hypothesized that participants who were provided concurrent visual feedback during the effector transfer tests would perform the non-mirror transfer test better than the mirror transfer test where the visual–spatial coordinates remained the same as during acquisition but the motor coordinates changed independent of previous practice in an on-line or a pre-plan mode during acquisition. Conversely, we predicted that performers who were not provided with concurrent visual feedback during sequence production during the transfer tests would demonstrate superior effector transfer performance on the mirror transfer test, where visual–spatial coordinates were changed but the motor coordinates were reinstated, regardless of whether or not they had previous experience in an on-line or pre-plan condition, and the associated coordinate systems (visual–spatial or motor) during acquisition. Further, we hypothesized that providing visual information during sequence production would increase sequence performance (on-line advantage; see also Elliott et al., 1995). During the acquisition phase the participants of the OL and the PP conditions decreased their performance errors across the first 6 blocks of practice. After Block 6, RMSE for both groups leveled off. Participants in the OL condition, where concurrent visual feedback was provided, demonstrated a larger decrease in the RMSE in the first blocks compared to the PP condition and they showed consistently lower RMSE compared to participants in the PP condition, where concurrent visual feedback was withheld during acquisition. The difference between the two groups was apparent throughout acquisition and on the retention test. Additionally, the analysis of harmonicity on the acquisition phase and the retention and transfer tests produced some mixed results. Over acquisition both conditions increased their H-values with a larger increase for the OL condition. However one of the crucial features of concurrent visual feedback is the prescriptive role played by error information during acquisition in order to direct the learners response closer to the criterion response in an efficient and effective manner. The retention test indicated a significant difference between the OL and the PP conditions. The H-values for the OL-condition were lower compared to the PP-condition, which provided evidence that performers in the OL condition made more iterative feedback based corrections during the progress of the movement relative to the participants of the PP-condition. Feedback based corrections are typically found in the kinematics of on-line controlled movements (see Woodworth, 1899). The finding for the retention test is consistent with previous result

provided by Kovacs et al. (2010) who found that concurrent visual feedback during sequence production resulted in a superior performance compared to a condition where concurrent visual information was withheld. Also consistent with Kovacs et al. (2010) were the kinematic findings, which suggest that when concurrent visual feedback is provided that participants use primarily an on-line control mode to produce the movement sequence. In terms of the transfer performance, participants of the OL–PP and PP–PP conditions performed the mirror transfer test, where the motor coordinates were reinstated, better compared to the non-mirror transfer test, where the visual–spatial coordinates were reinstated. Note, in these two conditions concurrent visual feedback was withheld during the two transfer tests. In contrast, the participants of the PP–OL and OL–OL conditions exhibited superior performance on the non-mirror transfer test compared to the mirror transfer test. In these two conditions concurrent visual information was provided during the two transfer tests. Thus the results of the present experiment provide clear support for our initial predictions related to the transfer pattern. That is, when concurrent visual information was provided during the transfer tests participants exhibited superior transfer performance on the nonmirror transfer test compared to the mirror transfer test. Alternatively, withdrawing concurrent visual information resulted in deteriorated transfer performance on the non-mirror transfer test, but increased transfer performance on the mirror transfer test, independent of whether or not the participant practiced the sequence during the acquisition phase with or without concurrent visual information. Therefore after one practice session and the same amount of practice performers can shift from an on-line to a pre-plan control and from a pre-plan to an on-line control mode (Glover, 2004). This finding is consistent with the hypothesis that control processes invoked under the different feedback conditions rather than the stage of practice or sequence difficulty are the prime determiner of the most salient coding system (Kovacs et al., 2010). Note, in the current experiment the task for both acquisition conditions was the same, only the visual feedback provided during response production was different. These findings extended previous results from Kovacs et al. (2010) that performers can alternate between the control modes or reference system for successful sequence production (see also Glover, 2004). It should be noted that to the best of our knowledge this experiment represents one of the few studies that provides evidence that performers can shift between coordinate systems for response production. These findings are partially consistent with the theoretical perspective of Hikosaka et al. (1999) which proposed that codes based in visual–spatial and motor coordinates develop in parallel but codes based on visual–spatial coordinates are dominant early in practice with less reliance at this stage of practice on codes based in motor coordinates becoming dominant later in practice. However, by taking a more “modest” view of the Hikosaka model, one might argue that performers have access to codes based in both coordinate systems (Kurata & Hoshi, 2002; Nakahara et al., 2001; see also Keele et al., 2003) and the effectiveness of transfer can be dependent on the control processes mediated by availability of concurrent visual feedback during sequence production. We did observe that in the on-line acquisition and the on-line retention conditions performers outperformed participants that practiced under the pre-plan acquisition condition. This also seems true for the transfer tests, even if the performers had previous experience in the on-line or pre-plan condition. As proposed, the dominant representation developed under the online condition was based on visual–spatial coordinates and the dominant representation developed for the pre-plan condition was based on motor coordinates, with the visual–spatial representation more effective at this stage of practice. If performers have the opportunity to use concurrent visual feedback, the codes based in the visual–spatial coordinate system are utilized for sequence production, even though they have previous experience with the pre-plan control mode, which is associated with the motor coordinate system (see Elliott et al., 1995;

P. Leinen et al. / Acta Psychologica 155 (2015) 92–100

Saunders & Knill, 2003; Desmurget & Grafton, 2000 for a review). Withholding visual information forces performers to utilize codes based in the motor coordinate system for sequence production even though they had previous experience in an on-line control mode where codes based in visual–spatial coordinates are primarily responsible for sequence production. On the one hand, if concurrent visual information is available performers use this kind of information and reinstated the visual–spatial coordinate system for response production even after previous practice without concurrent visual information. On the other hand if visual information is not available performers utilized motor information and reinstated the motor coordinate system for sequence execution. This finding provided evidence for the flexibility of the two coordinate systems developed during acquisition, because the acquired visual motor representation includes rules or procedures to process visual or motor information (Elliott et al., 1995) and that the two coordinate systems represent ends of continuums which operate together for successful task performance (Nakahara et al., 2001; see also Kovacs, Han, & Shea, 2009). The present findings are also in line with the results provided by Zirngibl and Koch (2002). Based on the distinct assumptions that sequences are (a) learned by the development of an association between two responses, (R–R learning; Nattkemper & Prinz, 1997) or (b) by the response and the resulting effect in this case the next stimulus (R–S learning; Ziessler & Nattkemper, 2001) they showed that both learning mechanisms are developed during the course of learning and that both are important for sequence learning. Zirngibl and Koch (2002) conclude that ‘there is a more gradual transition in the dominance of one mechanism over the other’ (Zirngibl & Koch, 2002, p. 161). The larger increase of the H-values for the OL condition at the acquisition phase is consistent with assumptions based on classical motor learning theories. According to these theories (e.g., closed-loop theory, Adams, 1971; schema theory, Schmidt, 1975), learning can be optimized when feedback is provided more frequently, immediately, and/ or precisely. Adams (1971), for instance, proposes a guidance role for feedback. According to Adams, feedback provides information to solve the motor problem by guiding the movement to the target on successive trials. He suggested that feedback does not produce learning directly, rather, it creates the appropriate situation (i.e., being on target) so that the actual learning mechanisms can operate (i.e., the movement's feedback producing an increment in “strength” for the perceptual trace). This source of information is thought to serve as a basis for error correction on subsequent trials and thus can be used to achieve more effective performance as practice continues. In the current experiment participants of the OL condition used the concurrent visual feedback information during acquisition, and they reduced adaptive on-line corrections on the basis of concurrent error information used from previous trials. The “guiding” effect during acquisition of feedback when it is presented in a concurrent fashion is very powerful, but could not be maintained after a rest interval of 24 h. The results of the retention test are also consistent with models of goal directed aiming. The two component model of limb control presented by Woodworth (1899) proposed that aimed movements were comprised of an initial preprogrammed phase which moves the limb toward the target where visual and proprioceptive feedback are used if necessary, to make adaptive corrections to achieve the target. The present experiment required participants to produce a movement sequence that matched as closely as possible a target waveform with or without concurrent visual feedback. When visual feedback was available during movement production feedback based adjustments of an ongoing movement were possible (see also van Doorn & Unema, 2004; Saunders & Knill, 2003). Thus participants used this information to make adaptive on-line corrections during movement execution (Glover, 2004). This resulted in reduced RMSE but also resulted in reduced harmonicity (lower H-values), because performers made subtle corrections during the progress of the movement (see also Kovacs et al., 2010). For participants who were not permitted to use concurrent

99

visual feedback during sequence execution it was not possible to use visual information to make adaptive corrections during sequence production (Elliott, Binsted, & Heath, 1999). Therefore participants cannot effectively detect some movement errors or effectively modify their movement using visual feedback information from the perceptual system (Flowers, 1976; Prablanc, Desmurget, & Grea, 2003; Searle & Taylor, 1948) resulting in less precise movement execution (see Meyer, Smith, & Wright, 1982). The results are consistent with the general tenets of the two component model (see also Elliott et al., 2001; Woodworth, 1899). Interesting is that present experimental findings are also consistent with the research on haptic perception. For example Kappers (1999) was interested in the role of processing of haptic sensory feedback information. In an experiment she asked blindfolded participants to match the spatial position of the right hand with the position of the left hand either in a mirrored version, or a non-mirrored version. The results of the reported experiments demonstrated that participants showed a strong bias for the mirrored posture of the hand positions. This suggested a tendency to use motor coordinates to match both hands when only haptic information is available to the participants (see also Kappers & Koenderink, 1999; Kappers, 2002). 5. Conclusions The results of the present experiment indicated that the effectiveness of transfer of a movement sequence with a spatial–temporal pattern of 2000 ms duration and 5 reversals was partially dependent on the availability of concurrent visual feedback provided during sequence execution. When concurrent visual feedback was not available, inter-manual transfer was enhanced when the transfer test involved the same motor coordinates as experienced during acquisition. Conversely when visual information was available during sequence production, inter-manual transfer performance was superior when the visual– spatial coordinates were reinstated. Note that the presence of this visual information prompted participants to engage on-line control. This pattern of transfer occurred independent of whether previous practice involved pre-plan or on-line control conditions. The finding of the present experiment including the spatial temporal movement sequence showed that performers can alternate between codes based in visual– spatial and motor coordinate systems depending on the type of transfer test administered and the availability of on-line visual feedback. This indicates the existence of two parallel coordinate systems (visual–spatial, motor) and that both coordinate systems are available after one day of practice and performers consider the information available to select the most salient code for sequence production. The results are in line with the notion that codes governing the control of movement sequences are based on different sources of sensory information from different dimensions including visual–spatial and motor information (see also Robertson & Pascual-Leone, 2001; Clegg et al., 1998), and support the more parallel view of sequence production where two coordinate systems cooperate to form codes based in visual–spatial and motor coordinates, so that costs of processing the relevant information are minimized (see Hikosaka et al., 1999). To give an example, whenever the visual–spatial coordinate system is less likely to succeed at an early stage of learning while visual information is not available then sequence production relies primarily on the motor coordinate system. For future research it seems important to increase our understanding of the flexibility of the perceptual motor system when performance conditions change (Blandin, Toussaint, & Shea, 2008; Verwey, Shea, & Wright, in press). Acknowledgments This work was supported by a grant from the German Research Foundation (PA 774/10-1).

100

P. Leinen et al. / Acta Psychologica 155 (2015) 92–100

References Abrahmase, E.L., Ruitenberg, M.F.L., de Kleine, E., & Verwey, W. (2013). Control of automated behavior: Insights from the discrete sequence production task. Frontiers in Human Neuroscience. http://dx.doi.org/10.3389/fnhum.2013.00082. Adams, J.A. (1971). A closed-loop theory of motor learning. Journal of Motor Behaviour, 3, 111–150. Bapi, R.S., Doya, K., & Harner, A.M. (2000). Evidence for effector independent and dependent representations and their differential time course of acquisition during motor sequence learning. Experimental Brain Research, 132(2), 149–162. Bapi, R.S., Miyapuram, K.P., Graydon, F.X., & Doya, K. (2006). fMRI investigation of cortical and subcortical networks in the learning of abstract and effector-specific representations of motor sequences. Neuroimage, 32, 714–727. Blandin, Y., Toussaint, L., & Shea, C.H. (2008). Specificity of practice: Interaction between concurrent sensory information and terminal feedback. Journal of Experimental Psychology: Learning, Memory and Cognition, 34, 994–1000. Clegg, B., DiGirolamo, G., & Keele, S. (1998). Sequence learning. Trends in Cognitive Science, 2, 275–281. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Desmurget, M., & Grafton, S. (2000). Forward modeling allows feedback control for fast reaching movements. Trends in Cognitive Science, 4, 423–431. Dirnberger, G., & Novak-Knollmueller, J. (2013). Motor and perceptual sequence learning: Different time course of parallel processes. Neuroreport, 24, 578–583. Elliott, D., Binsted, G., & Heath, M. (1999). The control of goal-directed limb movements: Correcting errors in the trajectory. Human Movement Science, 18, 121–136. Elliott, D., Chua, R., Pollock, B.J., & Lyons, J. (1995). Optimizing the use of vision in manual aiming: The role of practice. Quarterly Journal of Experimental Psychology, 48, 72–83. Elliott, D., Helsen, W.F., & Chua, R. (2001). A century later: Woodworth's (1989) two component model of goal-directed aiming. Psychological Bulletin, 127, 342–357. Flowers, K.A. (1976). Visual “closed loop” and “open loop” characteristics of voluntary movement in patients with Parkinsonism and intention tremor. Brain, 99, 269–310. Glover, S. (2004). Separate visual representations in the planning and control of action. Behavioral Brain Sciences, 27, 3–78. Guiard, Y. (1993). On Fitts's and Hooke's laws: Simple harmonic movement in upper-limb cyclical aiming. Acta Psychlogica, 82, 139–159. Hikosaka, O., Nakahara, H., Rand, M.K., Sakai, K., Lu, X.F., Nakamura, K., et al. (1999). Parallel neural networks for learning sequential procedures. Trends in Neuroscience, 22(10), 464–471. Hikosaka, O., Nakamura, K., Sakai, K., & Nakahara, H. (2002). Central mechanisms of motor skill learning. Current Opinion in Neurobiology, 12, 217–222. Hikosaka, O., Rand, M.K., Miyachi, S., & Miyashita, K. (1995). Learning of sequential movements in the monkey: Process of learning and retention of memory. Journal of Neurophysiology, 74, 1652–1661. Kappers, A.M. (1999). Large systematic deviations in the haptic perception of parallelity. Perception, 28(8), 1001–1012. Kappers, A.M. (2002). Haptic perception of parallelity in the midsagittal plane. Acta Psychologica, 109(1), 25–40. Kappers, A.M., & Koenderink, J.J. (1999). Haptic perception of spatial relations. Perception, 28(6), 781–795. Keele, S.W., Ivry, R., Mayr, U., Hazeltine, E., & Heuer, H. (2003). The cognitive and neural architecture of sequence representation. Psychological Review, 110, 316–339. Keele, S.W., Jennings, P., Jones, S., Caulton, D., & Cohen, A. (1995). On the modularity of sequence representation. Journal of Motor Behavior, 27, 17–30. Kirsch, W., & Kunde, W. (2012). Impact of hand orientation on bimanual finger coordination in an eight-finger tapping task. Human Movement Science, 31, 1399–1408. Korman, M., Raz, N., Flash, T., & Karnim, A. (2003). Multiple shifts in the representation of a motor sequence during the acquisition of skilled performance. PNAS, 100, 12492–12497. Kovacs, A.J., Boyle, J., Gruetzmacher, N., & Shea, C.H. (2010). Coding of on-line and preplanned movement sequences. Acta Psychologica, 133, 119–126. Kovacs, A.J., Han, D. -W., & Shea, C.H. (2009). The representation of movement sequences is related to task characteristics. Acta Psychologica, 132, 54–56. Kovacs, A.J., Muehlbauer, T., & Shea, C.H. (2009). The coding and effector transfer of movement sequences. Journal of experimental psychology — Human perception & performance, 35, 390–407.

Kurata, K., & Hoshi, E. (2002). Movement-related neuronal activity reflecting the transformation of coordinates in the ventral premotor cortex of monkeys. Journal of Neurophysiology, 88, 3118–3132. Kurata, K., & Wise, S.P. (1988). Premotor and supplementary motor cortex in rhesus monkeys: neuronal activity during externally- and internally-instructed motor tasks. Experimental Brain Research, 72, 237–248. Lashley, K.S. (1951). The problem of serial order in behavior. In L.A. Jeffress (Ed.), Cerebral mechanisms in behavior: The Hixon symposium (pp. 112–136). New York: Wiley. Meyer, D.E., Smith, K.J.E., & Wright, C. (1982). Models for the speed and accuracy of aimed movements. Psychological Review, 89(5), 449–482. Nakahara, H., Doya, K., & Hikosaka, O. (2001). Parallel cortico-basal ganglia mechanisms for acquisition and execution of visuomotor sequences: A computational approach. Journal of Cognitive Neuroscience, 13, 626–646. Nattkemper, D., & Prinz, W. (1997). Stimulus and response anticipation in a serial reaction time task. Psychological Research, 60, 98–112. Oldfield, R.C. (1971). The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia, 9, 97–113. 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. Park, J.H., & Shea, C.H. (2005). Sequence learning: Response structure and effector transfer. Quarterly Journal of Experimental Psychology, 58, 387–419. Prablanc, C., Desmurget, M., & Grea, H. (2003). Neural control of on-line guidance of hand reaching movements. Progress in Brain Research, 142, 155–170. Robertson, E.M., & Pascual-Leone, A. (2001). Aspects of sensory guidance in sequence learning. Experimental Brain Research, 137, 336–345. Sakai, K., Hikosaka, O., Miyauchi, S., Takino, R., Sakaki, Y., & Putz, B. (1998). Transition of brain activation from frontal to parietal areas in visuo-motor sequence learning. Journal of Neuroscience, 18, 1827–1840. Saunders, J.A., & Knill, D.C. (2003). Humans use continuous visual feedback from the hand to control fast reaching movements. Experimental Brain Research, 152, 341–352. Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225–260. Searle, L., & Taylor, F. (1948). Studies of tracking behavior. Rate and time characteristics of simple corrective movements. Journal of Experimental Psychology, 38, 615–631. Shea, C.H., & Kovacs, A.J. (2013). Complex movement sequences: How the sequence structure affects learning and transfer. In E. Day, W. Bennett, A. Portray, & W. Arthur (Eds.), The state of science of individual team skill decay — Retention and its implications for practice. Taylor & Francis. Shea, C.H., Kovacs, A.J., & Panzer, S. (2011). The coding and inter-manual transfer of movement sequences. Frontiers in Psychology, 2. http://dx.doi.org/10.3389/fpsyg. 2011.00052. Shea, C.H., & Wright, D.L. (2012). The representation, production, and transfer of simple and complex movement sequences. In M. Williams, & N. Hodges (Eds.), Skill acquisition in sport: Research, theory, & practice (pp. 131–149). Taylor & Francis. Tanaka, K., & Watanabe, K. (2014). Implicit transfer of spatial structure in visuomotor sequence learning. Acta Psychologica, 153, 1–12. Thompson, S.G., McConnell, D.S., Slocum, J., & Bohan, M. (2007). Kinematic analysis of multiple constraints on a pointing task. Human Movement Science, 26, 11–26. van Doorn, R.R.A., & Unema, P.J.A. (2004). Influence of different modes of real time visual information on single aimed movements. Acta Psychologica, 116, 309–326. Verwey, W.B., & Clegg, B.A. (2005). Effector dependent sequence learning in the serial RT task. Psychological Research, 69, 242–251. Verwey, W.B., Shea, C.H., & Wright, D.L. (in press). A cognitive framework for explaining serial processing and sequence execution strategies. Psychological Bulletin Review. Verwey, W.B., & Wright, D.L. (2004). Effector-independent and effector-dependent learning in a discrete sequence production task. Psychological Research, 68, 64–70. Winer, B.J. (1971). Statistical principles in experimental design. New York: McGraw-Hill. Woodworth, R.S. (1899). The accuracy of voluntary movement. Psychological Review, 3, 1–114. Ziessler, M., & Nattkemper, D. (2001). Learning of event sequences is based on response– effect learning: Further evidence from a serial reaction task. Journal of Experimental Psychology: Learning, Memory and Cognition, 27, 595–613. Zirngibl, C., & Koch, I. (2002). The impact of response mode on implicit and explicit sequence learning. Experimental Psychology, 49(2), 153–162.