Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Contextual interference : single-task versus multi-task learning and influence of concurrent temporal… Maslovat, Dana 2002

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

Item Metadata

Download

Media
831-ubc_2003-0015.pdf [ 6.55MB ]
Metadata
JSON: 831-1.0077177.json
JSON-LD: 831-1.0077177-ld.json
RDF/XML (Pretty): 831-1.0077177-rdf.xml
RDF/JSON: 831-1.0077177-rdf.json
Turtle: 831-1.0077177-turtle.txt
N-Triples: 831-1.0077177-rdf-ntriples.txt
Original Record: 831-1.0077177-source.json
Full Text
831-1.0077177-fulltext.txt
Citation
831-1.0077177.ris

Full Text

C O N T E X T U A L INFERENCE: SINGLE-TASK VERSUS M U L T I - T A S K L E A R N I N G A N D I N F L U E N C E OF CONCURRENT T E M P O R A L I N T E R F E R E N C E by DANA MASLOVAT B.Sc, Simon Fraser University, 1996  A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF T H E REQUIREMENTS FOR THE D E G R E E OF M A S T E R OF SCIENCE in T H E F A C U L T Y OF G R A D U A T E STUDIES School of Human Kinetics  We accept this thesis as conforming to the required standard  UNIVERSITY OF BRITISH C O L U M B I A October 2002 © Dana Maslovat, 2002  In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department  or by his or  her  representatives.  It  is  understood  that  copying or  publication of this thesis for financial gain shall not be allowed without my written permission.  Department of The University of British Columbia Vancouver, Canada Date  DE-6 (2/88)  OQrXZJo^  Abstract Contextual interference (CI) is a learning effect whereby high interference practice conditions produce decreased acquisition performance yet increased retention and transfer performance. Thus, a more difficult practice environment, although initially detrimental to acquisition, actually benefits learning of the skill. Typical CI experimental paradigms involve the comparison of acquisition, retention and transfer performance of multiple tasks under a blocked acquisition schedule (low interference) versus a random acquisition schedule (high interference). Numerous studies have investigated contextual interference and it has been shown to be a stable, robust phenomenon. Two studies involving bimanual coordination were conducted to further examine the contextual interference effect. Experiment 1 involved comparison of acquisition, retention and transfer performance of a single task control group, two task blocked presentation group and a two-task random presentation group. Acquisition data showed both random and control groups outperformed the blocked group in performance of the coordination pattern. This was opposite to the expected CI effect and was attributed to the high number of acquisition trials providing enough time for the learning benefits of the interference to be realized. Retention data did show a typical CI effect for one dependent measure, with the random group significantly outperforming the blocked group. Neither two-task group significantly outperformed the control group, suggesting interference of a second task may be as beneficial to learning as extra practice on the initial task. No group effects were found during transfer performance, however there was a learning effect on the opposite, unpracticed coordination pattern.  Experiment 2 examined an alternate form of interference, requiring participants to concurrently verbalize a compatible or incompatible counting pattern while performing a bimanual coordination pattern, to determine if CI effects could be generalized to other forms of interference. No significant group effects were found in acquisition, retention or transfer performance. This was attributed to insufficient interference caused by the counting patterns perhaps due to anchoring strategies of the participants. Analysis of the retention data did provide weak support for a concurrent 2-count pattern providing more interference than a concurrent 4-count pattern. However more research in the area of concurrent temporal interference is required to determine possible interference effects. Scanning data did show a significant improvement in performance of the to-be-learned task as well as the symmetrical bimanual coordination pattern, in support of previous studies. Examination of the sound data provided information regarding anchoring strategies of participants.  iv  Table of Contents  Page Abstract  ii  Table of Contents  iv  List of Tables  vii  List of Figures  viii  Acknowledgment  xi  Introduction  1  Experiment 1  3  Introduction  3  Method  6  Participants  6  Apparatus and Task  7  Experimental Design  11  Instructions  11  Group Assignment  14  Scanning  15  Acquisition  16  Retention/Transfer  17  Dependent Measures and Analyses  17  Results  20 RMSE  20 Acquisition 90°  20  Acquisition 45°  25  Retention and Transfer  27  Scanning  27  Standard Deviation  27  Acquisition 90°  32  Acquisition 45°  32  Retention and Transfer  33  Discussion  33  Experiment 2  41  Introduction  41  Method  47  Participants  47  Task  48  Apparatus  49  Experimental Design  49  Instructions  49  Scanning  50  Acquisition  52  Retention/Transfer  52  Dependent Measures and Analyses  53  Results  '  54  RMSE  54  Standard Deviation  61  Sound Data  65  Discussion  72  Dependent Measures  72  Sound Data  75  Results Summary  92  Conclusion  94  References  96  Appendix A - Review of Literature  104  Origins  105  History  106  Theoretical Explanations of CI  110  Elaboration/Distinctiveness Theory  Ill  Action-Plan Reconstruction Theory  114  Retroactive Inhibition  118  Theory Summary  119  Factors and Considerations  120  Task Characteristics - Laboratory versus Applied (Field).... 121 Participant Characteristics - Age & Experience Level  124  Summary of Task Type, Age & Experience  128  Task Variations  128  Summary of Literature Review  131  Appendix B - Task Definition  133  Appendix C - Evaluation of Scanning Methods in Bimanual Coordination 137 Introduction  138  Method  144  Participants  144  Apparatus  145  Task  146  Experimental Design  147  Results  Orientation  147  Scanning  149  Acquisition  150  Dependent Measures and Analyses  150 152  RMSE Discussion  152 154  vii List of Tables Table 1.  Page A N O V A results from Experiment 1 for the dependent measure ofRMSE  2.  Mean and standard deviation values for the dependent measure o f R M S E from Experiment 1  3.  64  A N O V A results from Experiment 2 for the dependent measure of standard deviation for scanning trials  12.  63  Mean and standard deviation values for the dependent measure of standard deviation from Experiment 2  11.  60  A N O V A results from Experiment 2 for the dependent measure of standard deviation  10.  59  Mean and standard deviation values for the dependent measure o f R M S E for scanning trials from Experiment 2  9.  56  A N O V A results from Experiment 2 for the dependent measure o f R M S E for scanning trials  8.  55  Mean and standard deviation values for the dependent measure o f R M S E from Experiment 2  7.  31  A N O V A results from Experiment 2 for the dependent measure o f R M S E  6.  30  Mean and standard deviation values for the dependent measure of standard deviation from Experiment 1  5.  22  A N O V A results from Experiment 1 for the dependent measure of standard deviation  4.  21  66  Mean and standard deviation values for the dependent measure of standard deviation for scanning trials from Experiment 2  67  viii List of Figures Figure 1.  Page Schematic display of apparatus set-up and design including location of monitor, speakers and manipulanda  2.  9  Schematic display of apparatus set-up and design including movement amplitudes and location of table markers  10  3.  A schematic display of a Lissajous figure for 90°  12  4.  A schematic display of a Lissajous figure for 45°  13  5.  R M S E performance data for Expt. 1: All conditions, all groups  23  6.  R M S E performance data for Expt. 1: Acquisition 90°  24  7.  R M S E performance data for Expt. 1: Acquisition 45°  26  8.  R M S E performance data for Expt. 1: Retention & Transfer  28  9.  Scanning versus learning performance for Experiment 1: A l l groups combined  10.  29  Standard deviation performance data for Expt. 1: Retention (immediate & 1-week combined)  11.  R M S E performance data for Expt. 1: Acquisition 45° (1  34 st  18 acquisition trials) 12.  R M S E performance data for Expt. 2: All conditions, all groups  13.  69  Temporal location of acquisition counting peaks for Expt. 2: 3-count group  16.  62  Temporal location of acquisition counting peaks for Expt. 2: 2-count group  15.  57  R M S E performance data for Expt. 2: Scanning criteria, all groups combined  14.  37  70  Temporal location of acquisition counting peaks for Expt. 2: 4-count group  71  ix 17A.  Movement position during metronome pulses for Expt. 2: 2-Count group with anchoring (S23)  17B.  Movement position during metronome pulses for Expt. 2: 2- Count group without anchoring (S5)  18A.  153  R M S E performance data for scanning experiment: A l l groups, early versus late acquisition  C3.  91  R M S E performance data for scanning experiment: All groups, scanning versus acquisition  C2.  90  Movement position during metronome pulses for Expt. 2: Control group participant transfer to 4-count trials (SI6)  CI.  89  Movement position during metronome pulses for Expt. 2: Control group participant transfer to 3-count trials (SI6)  22C.  87  Movement position during metronome pulses for Expt. 2: Control group participant transfer to 2-count trials (SI6)  22B.  86  Movement position during metronome pulses for Expt. 2: Control group with anchoring (SI6)  22A.  85  Movement position during metronome pulses for Expt. 2: 3-Count participant transferred to 4-count trials (S21)  21.  83  Movement position during metronome pulses for Expt. 2: 3-Count participant transferred to 2-count trials (S21)  20B.  82  Movement position during metronome pulses for Expt. 2: 4-Count group with no anchoring (S28)  20A.  81  Movement position during metronome pulses for Expt. 2: 4- Count group with anchoring (SI)  19B.  80  Movement position during metronome pulses for Expt. 2: 3- Count group with no anchoring (S24)  19A.  79  Movement position during metronome pulses for Expt. 2: 3- Count group with anchoring (S21)  18B.  78  156  R M S E performance data for scanning experiment: Lissajous acquisition groups, late acquisition versus postscan  157  R M S E performance data for scanning experiment: Square acquisition groups, scan versus acquisition  Acknowledgement Many people assisted my efforts to complete this thesis. I would first like to thank my graduate advisor, Dr. Ian Franks for his wisdom, expertise and patience throughout my graduate degree. I would also like to thank my other thesis committee members, Dr. Romeo Chua and Dr. Tim Lee for their support, advice and constructive contributions to my research. I would also like to thank many of the students, staff and faculty at the University of British Columbia for their assistance. Mr. Paul Nagelkerke was responsible for the programming and data collection aspects of my thesis and pilot experiments. Ms. Shannon Bredin and Mr. Tony Carlsen guided me with their comments, questions and input throughout my thesis. In addition I would like to thank the people who volunteered their time to participate in these experiments. This acknowledgement would be incomplete without an expression of gratitude for the support and encouragement of friends, family and colleagues. I would especially like to thank Mr. Kevin Hanson for convincing me of pursuing a masters degree. Finally I cannot thank enough my loving wife, Tiffany, for her unending support, love and patience throughout this process.  1 Introduction Research in the area of motor learning encompasses a wide range of topics. One area that has been extensively studied is how practice conditions affect the acquisition, retention and transfer of a skill. The examination of the effect of practice conditions on skill development had led to the discovery of the motor learning phenomena known as contextual interference (CI). Contextual interference is the learning effect whereby high interference practice conditions for multiple tasks produce decreased acquisition performance yet increased retention and transfer performance. Thus, a more difficult practice environment, although initially detrimental to acquisition, actually benefits learning of the skill. Most research involving practice conditions and the contextual interference effect, involve learning of multiple tasks and manipulation of interference levels by altering the acquisition presentation schedule. In a blocked acquisition practice schedule all trials of one task are presented before the presentation of another task (e.g. A A A , B B B , CCC). Blocked presentation has been considered an example of low contextual interference because the task to be practiced was always presented in the same order and was thus both predictable and repetitious. Alternatively, a random acquisition practice schedule (e.g. A C B , B C A , CAB) would produce high contextual interference where the tasks are presented in both an unpredictable and non-repetitious order. A serial acquisition practice schedule produced moderate levels of contextual interference and was thought to be predictable yet non-repetitious (i.e. A B C , A B C , ABC). Typically a CI effect occurred when random acquisition produced improved performance on retention and transfer of the learned tasks, relative to blocked acquisition.  2 Practice conditions have been examined for many years. Berstein (1967, p. 362) stated: The process of practice towards the achievement of new motor skills essentially consists in the gradual success of a search for optimal motor solutions to the appropriate problems. Because of this, practice, when properly undertaken, does not consist in repeating the means of solution of a motor problem time after time, but is the process of solving this problem again and again by techniques which we changed and perfectedfrom repetition to repetition. It is already apparent here that, in many cases, practice is a particular type of repetition without repetition. Berstein's comments are consistent with current literature regarding the learning benefits of high interference practice conditions. He advised against "repeating the means of solution of a motor problem time after time", which may be considered akin to blocked practice, and suggested "a particular type of repetition without repetition", which may be considered akin to random practice. Determining what practice conditions maximize learning of a skill has very important implications for practitioners. The goal for most teaching situations is to maximize skill retention and allow for transfer of the skill to a new scenario. Implications of the contextual interference effect include increasing interference in practice conditions and ensuring acquisition or practice performance is not necessarily used as a gauge of learning.  3  Experiment 1 Introduction The typical contextual interference paradigm involved the learning of multiple tasks under varying practice conditions, with interference being controlled by the presentation of the tasks (see Appendix A, Brady, 1998; Magill & Hall, 1990 for a review). Acquisition, retention and transfer performance of the task was compared to determine if a CI effect is present. Improved performance during retention and transfer with a higher interference acquisition schedule when compared to a lower interference acquisition schedule usually led to the conclusion that interference benefits task learning (Brady, 1998). Alternately, the theory of practice specificity (Henry, 1968) suggested that maximal retention performance of a task was facilitated by practice conditions that mimic retention conditions. For the learning of a single task, it is unclear if performance is maximized by practicing the single task only (i.e. practice specificity) or by interference via acquisition of another task (i.e. contextual interference). In other words, it is unclear if the learning benefits of interference can be applied to single task performance or if the CI phenomenon is limited to multiple task learning. However to determine if interference is beneficial to learning, performance in high and low interference conditions must be compared to performance of a single task. This would provide a performance baseline for comparison to performance with interference to determine if interference actually benefited the learning of each task or if it is simply the case that high interference produced better performance than low interference when learning multiple tasks. Previous research in learning and interference has rarely involved the use of a single task control group. Three studies in recent CI literature have involved comparisons  of participants performing a single task versus multiple tasks. However, none of these studies have been designed to compare single task performance to multiple task performance and determine if interference truly does benefit learning of a task. In the first study involving a control group, Lee, Magill and Weeks (1985) used a barrier knockdown task to test the effects of random versus blocked variable practice on transfer tests. Participants either practiced four different knockdown tasks in a blocked or random order, or performed 4 times the number of trials of a single knockdown task (control group). Participants were then tested on an "inside" transfer task (within the range of practiced movement times) and an "outside" transfer task (outside the range of practiced movement times). Transfer results showed no difference in group performance during inside transfer tests, however the random group outperformed the blocked and control group during outside transfer, as measured by variable error. However this experiment did not perform any retention tests and thus conclusions regarding learning were difficult to assess. In addition, Lee et al. were unable to conclude if their results were better reconciled under schema theory or a contextual interference effect. The second study involving a control group was performed by Del Rey, Xiaoying and Simpson (1994). They performed an experiment using different key press patterns as the tasks to be learnt. Participants practiced the tasks in either a blocked or random acquisition schedule or were assigned to a single task control group that only practiced one of the key press tasks. No differences in retention was found between the random group, blocked group or control group. However, as the purpose of this study was to examine retroactive inhibition in a contextual interference paradigm, Del Rey et al. had the control group only perform one third the number of acquisition trials, reducing the effectiveness of a  5 comparison between random, blocked and single task control groups. In the third study involving a single task control group, Shewokis, Del Rey and Simpson (1998) used a coincident timing task performed at three speeds; slow, medium and fast. Groups performed the task at different speeds in either a random acquisition schedule, blocked acquisition schedule or only performed the task at one of the speeds (i.e. three single task control groups). Retention results indicated that at the slow speed, the blocked group was more variable and less accurate than either the random or control group. No difference was found between groups for the medium and slow speeds. However, again the control groups only had one third the number of acquisition trials compared to the other two groups. Furthermore, the experiment may not have been an accurate test of CI as coincident timing with different speed stimuli may be considered practicing the same task with different movement initiation times, which may not have an interfering effect. None of the previous studies involving a control group have been able to determine if single task learning is maximized via practicing the single task only or by interference via acquisition of another task. Thus, the purpose of the present study was to determine if interference is beneficial for single task learning. This was done by comparing performance during acquisition, retention and transfer of a single task control group to groups performing two tasks under either high or low interference conditions. A secondary purpose was to determine the effects of interference and single task learning on transfer of a learned task. Although practice specificity theory would predict retention performance would be more accurate in the control group, the CI effect would predict higher accuracy by both the random and blocked groups in transfer tests due to the learning benefits gained by increased interference. In addition, a typical contextual  6  interference effect of poorer performance of the random group, relative to the blocked group, during acquisition yet better performance during retention and transfer was also anticipated. The task chosen for this experiment was a bimanual coordination task, namely producing a specific relative phase pattern. A relative phase coordination task was chosen because it is a laboratory task that is thought to represent a global descriptor of bimanual coordination capabilities, and random and blocked practice of bimanual coordination patterns have been shown to elicit a contextual interference effect (Tsutsui, Lee & Hodges, 1998). Therefore, relative phase was a quantifiable measure of changes in a coordination pattern with practice. Furthermore, it has been shown that although only two coordination patterns can be performed skillfully without practice (0° and 180° relative phase), other patterns of bimanual coordination can be learned with sufficient practice (i.e. 45°, 90° and 135°) (Fontaine, Lee & Swinnen, 1997; Lee, Swinnen & Verschueren, 1995). Method Participants Thirty self-professed right-handed participants were randomly assigned to one of three groups (10 per group); two task blocked acquisition group (blocked), two task random acquisition group (random) and single task control group (control). Participants were inexperienced with the task and were naive to the purpose of the experiment. The study was conducted in accordance with the ethical guidelines of the University of British Columbia. All participants received a remuneration of $5 per session (for a total of $15)  7 and a completion bonus of $15. Participants were also informed that the best performer in the group would receive a performance bonus of $50. Apparatus and Task Participants were seated at a colour monitor ( V G A 640 x 480 pixels) measuring 27 cm in width and 20 cm in height (Zenith, Model #ZCM-1490). On either side of the monitor was two identical lightweight manipulanda that restricted arm movements to the elbow joint. Participants' arms were positioned such that the elbow joint was aligned with the axis of rotation and the hands were placed palm down on adjustable metal plates. The middle finger was secured between two vertical pins and velcro straps secured the forearms and hands. Amplitude was specified by computer feedback and markers on the table, specifying "in", "mid" and "out" positions for each arm. The required movement amplitude was 40° from the "in" to the "out" markers. A 40° movement translated to a 15 cm movement on the computer screen. Angular position was recorded using two optical encoders (Dynapar, E20-2500-130), one attached to the shaft of each manipulandum. Three-axis Quadrature Encoder interface cards (Advantech, PCL-833) were used to enable high-speed sampling of angular positions, giving a resolution of 1000 counts per revolution. Angular position was sampled at a rate of 1000 Hz. PC speaker output of the computer motherboard was used to create the audio metronome tones. The metronome signal was amplified by a speaker on each side of the monitor (Multi-Media, Model #EP691). During retention and transfer trials involving counting (control group only), participants wore headphones (Realistic Nova-10) to hear the metronome, and sound data was collected via a microphone (Sony Dynamic, F-V10T) next to the speaker. This sound data was amplified by a V C R (Panasonic, NV-8500) and the I V peak-to-peak output  8 from the V C R was sampled at a rate of 1000 Hz by an A-to-D converter (Techmar Labmaster, PGH). All collected sound data was full-wave rectified and filtered through a low-pass filter. Schematics of the apparatus set-up are shown in Figures 1 and 2. The tasks were to produce bimanual coordination patterns by manipulating two angular manipulanda. The movements of the manipulanda involved continuous flexion and extension movements about the elbow joint in the horizontal plane. By way of reference, when homologous muscles for both limbs flex and extend simultaneously, the coordination pattern is described as "in-phase", or 0° relative phase. When homologous muscles for both limbs flex and extend alternatively, the coordination pattern is described at "anti-phase" or 180° relative phase. In-phase and anti-phase patterns are intrinsic patterns of coordination, since they can be performed more accurately and with lower variability than other patterns of coordinative movement. Also, they are thought to represent natural human coordinative patterns (Kelso, 1984; Fontaine et al, 1997; Zanone & Kelso, 1992; see Kelso, 1995 for a review). Participants in this study either performed all trials on a single bimanual coordination pattern of 90° relative phase (left hand lagging one quarter cycle behind the right hand), or performed trials of two different bimanual coordination patterns of 90° and 45° relative phase (left hand lagging one eighth cycle behind the right hand). Augmented visual feedback was provided via a Lissajous figure projected on a computer screen, with participants' movement superimposed over the template. Specifically, movements of the right manipulandum produced horizontal movements of the cursor on the screen while movements of the left manipulandum produced vertical movements of the cursor on the screen. Each complete cycle of movement by the  9  Figure 1: Schematic display of apparatus set-up and design including location of monitor, speakers and manipulanda Computer Monitor Sp eak er  O  Participant  Sp eak er  10  11 participant produced one continuous plot over the Lissajous figure. The Lissajous figure for the 90° relative phase movement produced a circle template while the 45° relative phase produces an ellipse. Direction of the movement was shown at the top of the Lissajous figure by either the word "clockwise" and a right direction arrow or "counterclockwise" and a left direction arrow. A metronome (1 Hz) was used for all acquisition and retention trials with participants instructed to complete a full cycle for each "beep" of the metronome. Speed was monitored for correctness throughout the experiment. Each trial lasted for 12 seconds, with only the last 10 seconds of data analyzed. Schematic displays of a 90° and 45° Lissajous figure are shown in Figures 3 and 4. Experimental Design Instructions Participants were seated in front of the monitor with the two manipulanda on either side. All participants familiarized themselves with the task apparatus and were provided with general instructions. These alerted the participants to the goal of the task, that is, to learn how to move the arms in such a way as to produce the pattern displayed on the computer screen. Participants were informed that concurrent on-line feedback would be displayed on the monitor. This involved a continuous trace of their movement pattern in the form of a Lissajous figure throughout the duration of the trial. Participants were informed that each trial would last 12 seconds and they should try to produce 12 full cycles in that period. Participants were informed that an auditory metronome would sound throughout the trial, that would signal when to begin moving, and then "beep" every second after that (1 Hz) to signal the end / start of each cycle. All  12  Figure 3: A schematic display of a Lissajous figure for 90°  OUT  Left Hand Displacement  IN  OUT  IN  Right Hand Displacement  13  Figure 4: A schematic display of a Lissajous figure for 45°  OUT  Left Hand Displacement  IN  IN  OUT  Right Hand Displacement  14 participants were reminded to try and keep both their arms moving throughout the each trial. Participants were allowed a maximum of four practice trials without a metronome or Lissajous template to familiarize themselves with their arm movement and corresponding visual feedback provided on screen. Prior to the start of acquisition trials, all participants were given a minimum of 30 trials of experience with the task common to all groups (90° relative phase) to ensure a moderate level of experience with regards to bimanual coordination. The rationale for minimal experience level was to ensure participants have sufficient experience with the task such that interference will provide a benefit (see Appendix A for a review). Previous studies (Jams & Goverover, 1999; Del Rey, 1989; Del Rey, Wughalter, Whitehurst, 1982) have indicated that for inexperienced performers who are still trying to understand the movement, additional interference provided by a random task presentation may not provide any learning benefits and may actually hinder performance. Thus the positive effects of random practice may not be realized until some degree of expertise has been achieved. However, it is important to note that studies have shown the contextual interference effect with inexperienced participants using bimanual coordination (Tsutsui et al, 1998). Group assignment Participants were randomly assigned to one of three groups; two task blocked acquisition group (blocked), two task random acquisition group (random) and single task control group (control). The two task blocked and random acquisition groups practiced two bimanual coordination patterns, 90° and 45°, in either a blocked or random presentation schedule respectively. The single task control group practiced a bimanual  15 coordination pattern of 90°. The purpose of the single task control group was to determine a baseline value of performance for participants performing only one task, namely 90°. As both Experiment 1 and 2 require an identical single task control group, the data collected from this group was used for analysis in both experiments. Scanning Studies of the acquisition of bimanual coordination patterns have employed the use of a "scanning run" (Zanone & Kelso, 1992 & 1997 and more recently Hodges & Franks, 2000; McGarry, Hodges, Bredin, Franks, & Chua, 2000). This process typically involved requiring the participants to perform a variety of coordination patterns, sampled from the entire range of possible patterns, at various points in the learning process to systematically observe the participants' coordination tendencies. The rationale for a scanning procedure is that it provides an indication of a performer's inherent coordination tendencies at any one time and thus the experimenter can directly observe any changes to the learner's spectrum of coordination patterns at any time during the learning process and gauge what has been modified or acquired as a result of learning. These inherent tendencies have been considered a learner's attractor layout or coordination landscape (Zanone & Kelso, 1994) Scanning runs were performed for all participants prior to the beginning of acquisition trials (pre-acquisition), immediately following completion of the acquisition trials (post-acquisition), and immediately following the final day of testing (postretention). Scanning was performed via a discrete visual metronome. Two boxes (3 cm x 5 cm) were displayed at eye level in the center of a black computer screen, aligned horizontally, 5 cm apart. These two boxes served as a visual metronome whereby each  16 box flashed on (green) for 200 ms and off (black screen) for 800 ms at various phase relations. To manipulate relative phase, onset of the flash was controlled by a customized computer program, which allowed for the production of different phasing patterns. Participants were asked to continuously flex and extend their arms between the "in" and "out" markers such that they synchronized peak flexion of each arm with the onset of the respective flashing box. That is, when the right box flashed on (green), the right arm of the individual was in peak flexion; when the left box flashed on (green), the left arm of the individual was in peak flexion. Participants were required to coordinate peak flexion movements in 25 discrete 15-second trials where the lights flashed in one of 25 patterns of relative phase (i.e., 0°, 15°, 30°, 45°, 60°, 75°, 90°, 105°, 120°, 135°, 150°, 165°, 180°, 195°, 210°, 225°, 240°, 255°, 270°, 285°, 300°, 315°, 330°, 345°, 360°). All participants began each scanning run with the 360° relative phasing to familiarize themselves with the procedure, with the remaining 24 patterns randomly presented. Only the final 12 seconds of data collection for each scanning trial was analyzed. To reduce the potential effects of learning, no feedback was provided at any time during the discrete test trials. Acquisition All participants performed a total of 200 acquisition trials at 1 Hz over two consecutive days. Each day of acquisition participants completed 10 blocks of 10 trials with a maximum one-minute rest in between blocks. The control group performed 90° for all 200 acquisition trials. The blocked group performed 100 trials of 90° one day and 100 trials of 45° the other day, with order of task presentation counterbalanced across groups. The random group performed a random presentation of 90° and 45° over the two days of  17 acquisition with the only stipulation being an equal number of each task presented in each block of 10 trials. Retention/Transfer At the end of the second day of acquisition and one week following acquisition, all participants were given a retention test and two transfer tests. The retention test involved a block of 10 trials of 90°, with a one Hz metronome and augmented visual feedback as per the acquisition trials. Transfer tests included one block of 10 trials of 90° with a 1.25 Hz metronome and visual feedback (speeded transfer) and one block of 10 trials of 270° relative phasing at one Hz with visual feedback (opposite transfer), which represents the opposite pattern to 90° where the right hand lags one quarter of a cycle behind the left hand. As both Experiment 1 and 2 require an identical single task control group, the data collected from the control group was used for analysis in both experiments. Thus the control group performed additional immediate and one week retention tests to be used for Experiment 2. This included a block of 10 trials of 90° with concurrent verbalization of a 2-count, 3-count and 4-count pattern. In addition, the control group performed a block of 10 trials of 90° during which there was no metronome and participants could adapt whatever speed they felt most comfortable with (self paced). Dependent Measures and Analyses The final 10 seconds of data collection for acquisition, retention and transfer trial were analyzed, and the final 12 seconds of data collection for scanning trials were analyzed. The rationale for this is that the performance of the participant is thought to take a few seconds to become consistent and synchronize with the auditory metronome. Discrete measures of relative phase were calculated at a rate of 1000 Hz for all complete  18 cycles of movement within the final 10 or 12 seconds of each trial. Relative phase (RP) of the left hand in relation to the right was calculated for each point after the speed and position of the limbs was re-scaled to the interval [-1, 1]. The phase angles were calculated using the methods described by Scholz and Kelso (1989). Calculations were constrained such that all relative phases were converted to a value ranging between the criteria plus or minus 180° (ranged RP). That is, if a participant attempted to perform a relative phase of 0° and actually performed at a relative phase of 359°, this value would be converted to a relative phase of -1°. Relative phase provided a description of participants' performance within and across trials. From the relative phase values, the dependent measures of standard deviation and root mean square error were calculated. From each trial's calculated ranged RP values, a mean RP and standard deviation were calculated. Standard deviation represents the participant's variability within a trial, which has been shown to be an important performance variable, both at the start of practice indexing the exploration for new coordination patterns and the break from old coordination patterns and at the end of practice, indexing quality of learning. Root mean square error (RMSE) of each trial was considered a general measure of the participant's error. R M S E was calculated by first subtracting the observed relative phase from the required/criterion relative phase (constant error) for each calculated ranged relative phase. Each constant error value was squared and summed, then divided by the total number of points, and the square root of this value represented R M S E (see Franks, Wilberg & Fishburne, 1982).  RMSE =  19 This measure has been used successfully to capture group differences in previous experiments (e.g. Tsutsui et al., 1998) and should speak most clearly to the predictions made at the end of the introduction. Thus the standard deviation was the variation around the participant's produced RP value while R M S E was the variation from the criteria. Dependent measures from the acquisition period were subjected to either a 3 (group: control, random, blocked) x 10 (block) A N O V A with repeated measures on the last factor (90° acquisition) or 2 (group: random, blocked) x 10 (block) A N O V A with repeated measures on the last factor (45° acquisition). Dependent measures from the retention and transfer tests were subjected to a 3 (group: control, random, blocked) x 2 (time: immediate, one-week) A N O V A with repeated measures on the last factor. Dependent measures from the 90° scanning trials were subjected to a 3 (group: control, random, blocked) x 2 (type: scanning, learning) x 3 (time: pre-acquisition, postacquisition, post-retention) A N O V A with repeated measures on the last factor. The scanning analysis was used to determine if there were significant group differences as well as significant differences between the scanning trials prior to acquisition, at the end of acquisition and at the end of the experiment (one week later). Including the analysis of type of trial, allowed for comparison of performance during scanning trials and the corresponding 90° learning trials (either acquisition or retention trials). Learning trials included in this analysis were the trials of the 90° coordination pattern that occurred the closest in time to the respective scanning trials; pre-acquisition scanning trials were compared to the final practice trial, post-acquisition trials were compared to the final acquisition trial, and post-retention trials were compared to the last retention trial of the final day of testing. If the scanning trials accurately represented the participant's  20 coordination landscape at that moment, no type difference would be expected between the scanning trials and the corresponding learning trial. The alpha level for the entire experiment was set at .05 and the GreenhouseGeisser Epsilon factor was used to adjust the degrees of freedom for violation of the sphericity assumption (Greenhouse & Geisser, 1959). The Tukey H S D method (Tukey, 1953) was used for all post-hoc comparisons. In addition, as the first retention test was performed immediately after the completion of the acquisition period, this experimental design allowed for comparison of performance at the end of the acquisition period versus one-week retention performance. Results RMSE A summary of analyses performed on R M S E is shown in Table 1. Group means and standard deviations for the R M S E calculations are shown in Table 2. R M S E data plotted as a function of trial type is shown for all conditions in Figure 5, with conditions on the x-axis of the graph representing blocks of 10 trials. Acquisition 90° The R S M E group data for acquisition of the 90° coordination pattern, plotted as a function of block, is shown in Figure 6. The three groups significantly improved their performance of the 90° coordination pattern as evidenced by a significant main effect for block, F(4, 112) = 19.937, p < 0.001. Post hoc analyses revealed a significant difference 1  between block 1 and block 8. There was no significant main effect for group (p = 0.271), however the block x group interaction effect was significant, F(8, 112) = 2.806, p = 0.007. Post hoc analyses were conducted individually on each group to determine if a 1  Note that degrees of freedom in all F ratios have been adjusted by the Greenhouse-Geisser Epsilon Factor  21  Table 1: A N O V A results from Experiment 1 for the dependent measure of RMSE. Note the Greenhouse-Geisser Epsilon factor was used to adjust degrees of freedom for violation of the sphericity assumption. Effect  F-test  p-value  p<0.05  Acquisition 90° block group block x group block (Blocked group only) block (Random group only) block (Control group only)  E(4,112) == 19.937 F(2, 27) = 1.371 F(8, 112) == 2.806 F(4, 33) = 2.224 F(2, 19) = 12.212 F(4, 32) = 7.683  p < 0.001 p_ = 0.271 2 = 0.007 2 = 0.092 2 < 0.001 2 < 0.001  *  Acquisition 45° block group block x group  E(2, 42) = 42.520 F ( l , 18) = 0.111 F(2, 42) = 3.094  2 < 0.001 2 = 0.743 2 = 0.049  *  Retention time group time x group  F(l,27) = 0.296 F(2, 27) = 3.329 E(2, 27) = 1.732  2 = 0.591 2 = 0.051 2 = 0.196  Speeded Transfer time group time x group  F(l,27) = 2.119 E(2, 27) = 0.517 F(2, 27) = 0.356  2 = 0.157 2 = 0.602 2 = 0.704  270° Transfer time group time x group  F(l,27) = 16.453 E(2, 27) = 0.769 F(2, 27) = 1.784  2 < 0.001 2 = 0.473 2 = 0.187  * *  *  *  *  22  Table 2: Mean and standard deviation values for the dependent measure o f R M S E from Experiment 1. Condition Acquisition 90°  Blocked Group Block 1 Block 10  Random Group Block 1 Block 10  Control Group Block 1 Block 10  15.303 2.020  13.427 2.183  16.723 3.368  11.101 1.860  16.528 4.479  13.685 2.584  Acquisition 45°  Block 1  Block 10  Block 1  Block 10  Block 1  Block 10  M  27.174 7.867  15.020 4.481  33.262 12.466  12.256 4.147  M  SD  SD  —  —  Retention Immediate 1-Week Immediate 1-Week Immediate 1-Week 13.290 12.527 10.505 11.021 11.507 11.294 M SD 2.456 2.324 1.255 1.718 2.091 2.264  Speeded Immediate 1-Week Immediate 1-Week Immediate 1-Week Transfer M 13.344 13.507 12.491 11.926 12.959 12.538 SD 2.414 2.757 3.485 1.290 3.003 3.626  270° Immediate 1-Week Immediate 1-Week Immediate 1-Week Transfer 38.369 33.015 35.009 27.053 48.620 32.562 M SD 13.973 9.074 16.129 10.819 28.647 23.855  23  24  25 block effect was present. The two-task blocked presentation group did not show a significant block effect, F(4, 33) = 2.224, p = 0.092, however there was a significant block effect for both the two-task random presentation group, F (2, 19) = 12.212, p < 0.001 and the single task control group, F(4, 32) = 7.683, p < 0.001. Thus the random and control group significantly improved their performance (measured by RMSE) over the blocks of acquisition trials while the blocked group did not. Although both the random and control group showed a significant block effect, examination of extreme means revealed a greater rate o f R M S E improvement by the random group versus the control group. Acquisition 45° The R S M E group data for acquisition of the 45° coordination pattern, plotted as a function of block, is shown in Figure 7. Both groups were able to learn the 45° coordination pattern as evidenced by a significant main effect for block, F(2, 42) = 42.520, p < 0.001. This was shown by a significant difference between block 1 and blocks 4 through 10 inclusive. There was no significant main effect for group (p = 0.743), however the block x group interaction effect was significant, F(2, 42) = 3.094, p = 0.049. As with the acquisition of the 90° pattern, the block effect for the acquisition of the 45° pattern was different for the different groups. Post hoc analyses conducted individually on each group showed a significant block effect for both groups, however examination of extreme means revealed a greater rate o f R M S E improvement by the random group versus the blocked group.  26  27 Retention and Transfer Retention and both transfer performances (speeded 90° coordination pattern and 270° coordination pattern) showed no significant group effect or group x time interaction. The group effect just failed to reach significance, F(2, 27) = 3.329, p = 0.051 (Blocked, M = 12.908°, Random, M = 10.763°) for retention performance. However performance of the 270° coordination pattern did show an effect for time, F ( l , 27) = 16.453, p < 0.001, with performance significantly improving between testing immediately following acquisition ( M = 40.666°) and one week later ( M = 30.877°). Retention and transfer performance, plotted as one group, is shown in Figure 8. Scanning Analysis of R M S E for scanning trials did show an effect for time, F(2, 51) = 8.511, p = 0.001 and an effect for type, F ( l , 26) = 160.124, p < 0.001 but no main effect for group or interaction effects. Post hoc analyses on the main effect for time revealed both scanning trials (p = 0.024) and learning trials (p < 0.001) individually showed a time effect. However, post hoc analyses on type of trial showed a highly significant difference between scanning trials and learning trials. At all three time intervals, learning trials showed a highly significant lower R M S E than scanning trials. Data from all three groups combined for both scanning trials and learning trials is shown in Figure 9. Standard Deviation Standard deviation results closely mirrored R M S E results. A summary of analyses performed on standard deviation is shown in Table 3. Group means and standard deviations for the standard deviation calculations are shown in Table 4.  28  29  c CD  c  ID a: •  CD  E  o  0_  'C  CD  Q. X LLI CD O  c re E T3 CD  O t JQ CD  E  o  c E  a  re  to '3 cr o  o  a  <  a> _ a »  o  0_  3 == (0 < CD >  c "E c re o  V) (1)  o O" O <  1  CD  o o d 00  o o d  o o d co  o o d  o o d  o q d  CO  (Bap) 3SIAIU  o o  a.  1  V)  3  CD  a) E  H  30  Table 3: A N O V A results from Experiment 1 for the dependent measure of standard deviation. Note the Greenhouse-Geisser Epsilon factor was used to adjust degrees of freedom for violation of the sphericity assumption. Effect  F-test  p-valuep<0.05  Acquisition 90° block group block x group block (Blocked group only) block (Random group only) block (Control group only)  F(5, F(2, F(9, E(4, F(2, F(5,  125) == 13.892 27) = 1.389 125) == 2.477 34) = 2.004 20) = 8.200 45) = 6.311  p < 0.001 p_ = 0.267 £ = 0.012 £ = 0.118 £ = 0.002 £ < 0.001  *  Acquisition 45° block group block x group block (Blocked group only) block (Random group only)  F(4, F(l, F(4, E(3, F(3,  66) 18) 66) 26) 25)  £< £= £= £= £<  *  Retention time group time x group  E(l,27) = 0.039 E(2, 27) = 4.021 F(2, 27) = 2.418  £ = 0.832 £ = 0.030 £ = 0.108  Speeded Transfer time group time x group  F(l,27) = 2.035 F(2, 27) = 1.716 E(2, 27) = 0.301  £ = 0.157 £ = 0.199 £ = 0.734  270° Transfer time group time x group  F(l,27) = 16.347 E(2, 27) = 0.659 F(2, 27) = 2.073  £ < 0.001 £ = 0.525 £ = 0.145  = 28.499 = 0.259 = 3.307 = 5.341 = 32.915  0.001 0.617 0.018 0.006 0.001  * *  * * *  *  *  31  Table 4: Mean and standard deviation values for the dependent measure of standard deviation from Experiment 1. Condition Acquisition 90°  Blocked Group Block 10 Block 1  Random Group Block 10 Block 1  Control Group Block 10 Block 1  M  13.966 1.721  12.361 1.861  13.618 2.003  10.430 1.855  14.264 2.755  12.689 2.249  Acquisition 45°  Block 1  Block 10  Block 1  Block 10  Block 1  Block 10  19.557 7.347  12.941 2.989  23.093 5.587  10.487 2.459  SD  M  SD  —  —  Retention Immediate 1-Week Immediate 1-Week Immediate 1-Week 11.746 9.436 10.206 10.491 10.318 12.191 M 2.660 1.933 0.938 1.403 1.827 1.982 SD Speeded Immediate 1-Week Immediate 1-Week Immediate 1-Week Transfer 12.225 10.618 10.716 10.631 11.061 11.648 M 1.923 1.725 0.986 1.662 1.999 2.233 SD Immediate 1-Week Immediate 1-Week Immediate 1-Week 270° Transfer 30.218 30.614 24.054 43.140 28.254 35.004 M 15.162 9.794 27.488 22.544 9.382 12.691 SD  32 Acquisition 90° All three groups significantly improved their performance of the task during acquisition, shown by a main effect for block F(5, 125) = 13.892, p < 0.001. There was no significant main effect for group (p = 0.267), however the block x group interaction was significant, F(9, 125) = 2.477, p = 0.012. Post hoc analyses were conducted on each group individually to determine if a block effect was present. As with RMSE, no block effect was found for the two-task blocked presentation group, F(4, 34) = 2.002, p = 0.118, however there was a significant block effect for both the two-task random presentation group, F (2, 20) = 8.200, p = 0.002 and the single task control group, F(5, 45) = 6.311, p < 0.001. Thus the random and control group significantly decreased their variability over the blocks of acquisition trials while the blocked group did not. Examination of extreme means revealed a greater rate of decreased variability by the random group versus the control group. Acquisition 45° Both groups were able to significantly improve their performance of the 45° coordination pattern as evidenced by a significant main effect for block, F(4, 66) = 28.499, p < 0.001. There was no significant main effect for group (p = 0.617), however the block x group interaction effect was significant, F(4, 66) = 3.307, p = 0.018. Post hoc analyses conducted individually on each group showed a significant block effect for both groups: Blocked, F(3, 26) = 5.341, p = 0.006 and Random, F(3, 25) = 32.915, p < 0.001, however examination of extreme means revealed a greater rate of decreased variability by the random group versus the control group.  33 Retention and Transfer Analysis of the retention data showed no significant effect for time (p = 0.832) or time x group interaction (p = 0.108), however there was a significant group effect, F(2, 27) = 4.021, p = 0.030. Post hoc analyses of the group effect showed a significant difference between the Blocked ( M = 11.968°) and Random group ( M = 9.821°) but no significant difference between the Control ( M = 10.404°) and Blocked group ( M = 11.968°) or Control ( M = 10.404°) and Random group ( M = 9.821°). Retention data for all groups, for combined immediate and one-week performance is shown in Figure 10. Neither transfer performance (speeded 90° coordination pattern and 270° coordination pattern) showed a significant group effect or group x time interaction. However, performance of the 270° coordination pattern did show an effect for time, F ( l , 27) = 16.347, p < 0.001, with performance significantly improving between testing immediately following acquisition ( M = 36.253°) and one week later ( M = 27.509°). Discussion The purpose of the present study was to determine if interference is beneficial for single task learning. This was done by comparing performance during acquisition, retention and transfer of a single task control group to groups performing two tasks under either high (random presentation) or low (blocked presentation) interference conditions. A secondary purpose was to determine the effects of interference and single task learning on transfer of a learned task. It was predicted that the single task control group would outperform both two-task groups (random and blocked) during retention (due to practice specificity theory), while both two-task groups would outperform the single task control group during transfer (due to the contextual interference effect). In addition, a typical  34  O  O  a x  UJ  =o  o t - CD CO =  ca -9 •o E CD  0  O  O  = ^ co a)  "a  1 |  c  O T CD \ " Q . 06 C CD  ro  "co T:  E "° E T3 ^ i_ CO c CD  T3 C  CO  O  C  £ 6 £ CD 3  ii  CD .Id O  o m  o o  cd T—  o o T—  o o  -<fr  o o co  o o CN  T—  (Bap) Aaa JS  o o  o o ci  o o  o qCO  =  2  en  35 contextual interference effect of poorer performance of the random group during acquisition yet better performance during retention and transfer was also anticipated. Analysis of acquisition performance of both tasks showed an improvement by all three groups, yet at different rates. The random and control groups improved more than the blocked group during acquisition of the 90° coordination pattern. For the 45° coordination pattern, the random group improved more than the blocked group. These findings are opposite to what is normally seen in the contextual interference literature, where typically the blocked group outperforms the random group in acquisition. However, most laboratory contextual interference experiments involved inexperienced participants completing relatively few acquisition trials for each task. Previous CI studies have included 18 acquisition trials of three barrier knockdown tasks (Shea & Morgan, 1979) or 45 trials of three bimanual coordination tasks (Tsutsui et al., 1998) as compared to this experiment involving participants with a minimum of 30 trials of experience with coordination patterns and 100 trials of acquisition per task. Examination of the acquisition data from Shea and Morgan (1979) showed the blocked group initially outperforming the random group, however the random group eventually does perform at almost the same level as the blocked group after only 18 trials. Examination of the first two blocks of 10 acquisition trials for both 90° and 45° task from this experiment (Figures 6 and 7) showed a similar and typical CI effect, however the random group continued to improve over the next 80 trials and outperformed the blocked group. Thus this reversal of the typical CI acquisition effect may have simply been due to the number of acquisition trials providing enough time for the learning benefits of the interference to be realized. Examination of the R M S E of the first 18 acquisition trials of the task that  36 participants were not familiar with (45° coordination pattern) showed a more typical contextual interference effect (see Figure 11). As expected, the retention data did show a group effect, however only for one of the dependent measures (standard deviation). It was expected that the single task control group would show the best performance (lowest variability) of the three groups due to practice specificity. It was also expected the two-task random presentation group would outperform (lower variability) the two-task blocked presentation group due to the contextual interference effect. Although the contextual interference effect was shown by significantly better performance by the random group versus the blocked group in retention, neither the blocked nor random group performed differently than the control group. This is surprising as the control group received twice the number of acquisition trials when compared to either the blocked or random group. Thus it appears that the interference caused by practicing a second task is as beneficial to learning as extra practice on the initial task. Performance during the speeded transfer did not show any difference between groups or between the immediate and one week test. All groups had difficulty performing the opposite coordination pattern transfer (270°), shown by a much higher R M S E and standard deviation. However all groups improved between the immediate test and the one week transfer, likely due to learning effects. Although there was not a significant group effect, the data from the immediate opposite transfer condition does show both the random and blocked group performing with a lower R M S E and standard deviation than the control group. Performing two tasks during acquisition should benefit 270° transfer as  37  Q.  tn  UJ  .2  o c  « o  g .2 0)  o c  3  O  re  (0 oo  o t  0. o UJ  u o  IO  m  c  £I 3  O)  <  (Bap) 3SI/\IU  38 these participants would have had more opportunity to explore different coordination patterns. Results from the scanning analysis suggest that the method of scanning does not accurately represent the participant's ability to perform a given coordination pattern. Although both the scanning trials and learning trials showed the expected effect of significant improvement from pre-acquisition to post-acquisition and post-retention, the main effect for type of trial revealed the significant differences between scanning trials and learning trials. If the scan were accurately representing the performance of the participant at the given time, no difference in type of trial would be expected. The large differences in both R M S E and error bars (Figure 9) between scanning trials and learning trials at all three measured times in the experiment cast serious doubt on the validity of the scanning method. This may be explained by the fact that the method use to scan participants' coordination landscape was;yery different from the learning trials the participants experienced. The large differences in performance suggest little transfer of learning occurred between performance of a 90° coordination pattern via Lissajous figures with concurrent, on-line feedback and performance of the same pattern using coincident timing of the arms with two flashing squares. In summary, the acquisition data from this experiment did not show a typical contextual interference effect as the random group outperformed the blocked group. However this may be due to the experience level of the participants and the large number of acquisition trials. Retention data did show a contextual interference effect with respect to variability of performance and also showed the benefits of interference does affect single task learning and retention. Transfer performance to a new coordination pattern  39  showed a trend of increased performance by groups practicing two tasks versus the single task acquisition group, however more research would be needed to draw stronger conclusions in this area. The design of this experiment allowed for comparison of two stable motor learning phenomena; namely contextual interference and practice specificity. Coaches or practitioners often need to know the most efficient and effective method to teach an individual a single skill or task in a closed environment. Research in the area of practice specificity would suggest that coaches should have the learner only practice the to-belearned skill so that the practice environment closely resembles the retention environment. However, research in the area of contextual interference would suggest that retention performance is maximized with interference, usually from the practice of another task. Previous studies have not been able to compare single task learning to learning with interference, as either retention data were not collected (Lee et al., 1985) or the single task control group did not receive the same number of trials as the interference groups (Del Rey et al., 1994, Shewokis et al., 1998). Results from this study would suggest that extra practice on the single task is no more beneficial to learning than extra practice on a second, interfering task. Furthermore, in support of previous contextual interference findings, higher interference conditions (random practice) are more beneficial to retention performance than lower interference conditions (blocked practice). These results suggest that the importance of practice specificity may be less than originally thought. Rather than extra trials on the single task, equal benefit can be derived by the introduction of a second task. In addition, increased practice on a second task may also create learning benefits for transfer to other tasks. From a practical perspective, these  results suggest that practitioners can benefit from requiring their performers to practice multiple tasks, in a high interference schedule, even if single task retention is the primary concern. Thus it appears that contextual interference does provide benefits for single task learning and may be used rather than the principle of practice specificity in most learning environments.  41  Experiment 2 Introduction Contextual interference (CI) is a motor learning phenomenon whereby high interference practice conditions for multiple tasks produce decreased acquisition performance yet increased retention and transfer performance relative to low interference practice conditions. Most contextual interference paradigms alter interference levels by manipulating presentation schedules during acquisition. A random presentation of tasks is thought to be a higher interference scenario than a blocked presentation of tasks due to its unpredictable and non-repetitious nature. However, it is possible to manipulate interference levels by other factors. In Battig's (1956) original work in verbal learning, interference was manipulated via intratask, or within task interference (see Appendix B for a review of task definition and what constitutes intratask interference versus intertask interference). Intratask interference involved having either a compatible or incompatible verbal cue associated with afinger-positioningtask. Incompatible cues were thought to produce higher interference level and facilitated performance of the task. Later, Battig (1979) expanded his definition of interference to include not only within task interference but also such factors as task complexity, practice presentation schedule, processing engaged by the learner, variability of the learner's response, environmental conditions and similarity/dissimilarity of multiple tasks (Magill & Hall, 1990). However, little work has been done to examine the effects of different sources of interference, as CI literature has focused on solely on various presentation schedules. One source of interference that has not been examined is concurrent temporal intratask interference. Most motor tasks have a temporal pattern or relative timing  42 associated with the individual movements. Relative timing has been thought to be an important, integral structure of the movement. Shapiro, Zernicke, Gregor and Diestel (1981) analyzed the relative timing associated with the step cycle phases of walking and running at different speeds. They discovered that, although the relative timing of walking and running step cycle phases was different, relative timing did not change when performing the walk or run at different speeds. Schmidt (1985) also provided numerous examples where relative timing of movements was invariant under different circumstances. Thus it appears that movements may have a unique relative timing, which is stable and integral to the movement itself (for a critical discussion of this topic see Gentner, 1987 and Franks & Stanley, 1991). Assuming movements do have a timing pattern that is integral to the movement, it would be possible to provide an external or augmented timing pattern that was either compatible or incompatible with the timing of the movement. Studies have shown that providing learners with a compatible external timing pattern can be beneficial to performance. Thaut, Mcintosh, Rice, Miller, Rathbun, and Brault (1996) showed cuing via rhythmically accentuated music increased effectiveness for a home based gait-training program for Parkinson's disease patients. Alternatively, incompatible stimulus-response (S-R) pairs have been shown to benefit learning as well. Battig (1956) found that incompatible verbal cues associated with a movement were more beneficial to learning as compared to compatible verbal cues. He attributed this benefit to the interference caused by the S-R incompatibility. Performance of a task with an external, incompatible timing pattern should produce a higher level of interference than a compatible timing pattern. However, it is unclear if this form of interference would produce typical contextual  43 interference results, namely decreased acquisition performance but increased retention and transfer performance. Thus the purpose of Experiment 2 was to examine an alternate form of interference to determine if CI can be generalized to different types of interference. Specifically, this experiment attempted to determine if concurrent temporal interference produced typical CI results. Studies have indirectly examined compatibility of movement and timing by exploring the performance effect of bimanual coordination patterns resulting from manipulation of the timing of the metronome pulse relative to the movement cycle. As rhythmic movements appear to have "anchor points" which are typically exhibited at points of maximal excursion, the dynamics of these anchor points likely depend on the relative temporal location of the metronome cue relative to the movement cycle (Carson, Byblow & Goodman, 1994). Byblow, Carson & Goodman (1994) performed a study where they required participants to perform in-phase and anti-phase bimanual coordination patterns but explicitly controlled the position of the metronome pulse relative to the movement cycle of each hand. Participants were required to coordinate their movement such that a given anchor point (either maximum pronation or supination) was synchronized with the metronome beat. They discovered that coordination of anchor points with the metronome decreased movement endpoint variability and this was further enhanced by coordination of the metronome with points of maximal pronation rather than supination. Fink, Foo, Jirsa & Kelso (2000), further examined this phenomenon of stabilization of a coordination pattern via coupling to task specific sensory information from the environment. Fink et al. (2000) defined "anchoring" as intentionally  44 synchronizing a particular point in a movement cycle with a stimulus. They examined the performance benefits of anchoring by requiring participants to synchronize peak flexion or extension of a finger during a bimanual coordination movement with either a single or double metronome. Participants performed an in-phase or anti-phase pattern with either the peak amplitude of right finger flexion coinciding with a single metronome or both peak flexion and extension of the right finger coinciding with a double metronome. They determined that anchored reversal points displayed lower spatial variation than unanchored reversal points and this variation was further enhanced in the double metronome condition. Further, they concluded that the stability of the global dynamics of the anti-phase coordination pattern (measured by transition times to in-phase) was enhanced by the metronome, again with additional benefits found in the double metronome condition. Thus, coordination of a movement anchor point with a metronome pulse appeared to increase stability and decrease variability of the movement. Performance of a bimanual coordination pattern involves four anchor points for each cycle of movement, as both limbs have two reversal points (i.e. right hand "in", left hand "in", right hand "out", left hand "out"). During an in-phase and anti-phase movement however, reversal points of one limb should correspond with reversal points of the opposite limb. For example, during an in-phase movement pattern, both limbs should reach the same reversal point at the same moment in time (right hand "in", left hand "in"). Alternately for an anti-phase movement pattern, the reversal point of one limb should correspond in time to the opposite reversal point of the opposite limb (right hand "in", left hand "out"). Thus, in-phase and anti-phase movement only have two times in each movement cycle at which limbs are at reversal points. However, performance of an  45 alternate coordination pattern would produce four distinct times in the movement cycle at which limbs are at reversal points, since reversal points of one limb would never correspond with reversal points of the opposite limb. For example, a 90° relative phase movement pattern involved the left hand lagging the right hand by one quarter of a cycle. Thus this pattern involved four distinct movement end points that are equally temporally spaced during the movement. As previous studies have shown that coinciding one reversal point with a metronome pulse decreased movement variability (Byblow et al., 1994) and coinciding two reversal points with a metronome pulse further decreased movement variability (Fink et al., 2000) it would seem logical that coinciding all four reversal points of a 90° with an external pacing signal should result in an even further decrease in movement variability. As the purpose of this experiment was to examine different forms of temporal interference, both compatible and incompatible with the movement, participants performed a 90° coordination pattern with either a 4-count, 3count or 2-count external pacing signal. For this experiment, the external pacing signal involved requiring the participants to verbalize a counting pattern in time to a metronome. Requiring the learner to verbalize a four-count pattern concurrently while performing a 90° coordination pattern should be compatible with the movement by providing assisting reference points for the learner. That is, the learner's hands would correspond to the "in" and "out" positions at each of the four numbers verbalized. For example a four-count may correspond to the right hand "in" ("one") the left hand "in" ("two"), the right hand "out" ("three") and the left hand "out" ("four"). Thus each count  46 of the verbal temporal pattern is associated with a stable anchor or end point of the movement, which should assist the learner during acquisition and decrease interference. Alternatively, verbalization of a two-count pattern concurrently while performing a 90° coordination pattern does not provide compatibility between the temporal verbalization and the relative timing of the movement. Two-count interference does not provide assisting reference points for both hands of the learner. Although two equally spaced counts may provide corresponding anchors for the "in" and "out" position of one hand, there are no counts to correspond to anchor points for the other hand. As proper execution of a 90° coordination pattern is dependent on the position of one hand relative to the other, this verbal temporal pattern will likely create the most interference and provide the most difficult acquisition condition. If a learner tries to use the counting pattern to anchor both hands' end points, this two-count strategy may "force" the learner into an in-phase or anti-phase pattern of coordination. Three-count verbalization may provide a moderate amount of interference. Learners would be able to use "one" to anchor one end point of one hand, but neither the "two" or "three" would correspond to another end point of either hand. However, certain musical dances, such as the waltz, have an underlying three-count temporal structure but still produce a fluid coordination pattern. Thus although specific reference points are not provided by three-count verbalization, more experienced learners or those with musical experience could find this pattern of verbalization compatible with their movement. In summary, it was predicted that the four-count verbalization group should be able to anchor all four reversal points, while the two count group should only be able to anchor one hand's reversal point, and the three count group should only be able to anchor  47 one reversal point. Thus the different types of counting should produce different amounts of interference for the learner, due to the differences in their compatibility. It was unclear if this form of interference would produce typical CI effects. The CI effect would predict that incompatible concurrent temporal verbalization (i.e. twocount) would provide a higher level of interference, thus producing decreased acquisition yet increased retention and transfer of a task relative to compatible concurrent temporal verbalization (i.e. four-count). If typical CI results were seen in this experiment, it could be inferred that concurrent temporal interference produces a similar result in performance as altering interference via different practice schedules (i.e. random versus blocked presentation). This would suggest that the contextual interference effect could be generalized to different forms of interference. The implication of such a finding would be that providing interference in a learning environment produces increased retention and transfer performance and the type of interference is immaterial. One possible confounding variable was that providing compatible concurrent temporal interference may provide the learner with a strategy to assist them during retention and transfer performance. Method Participants Thirty self-professed right-handed participants were randomly assigned to one of three groups (10 per group); two-count temporal interference (2-count), three-count temporal interference (3-count) and four-count temporal interference (4-count). Participants were naive to the purpose of the experiment and the study was conducted in accordance with the ethical guidelines of the University of British Columbia. All  48 participants received a remuneration of $5 per session (for a total of $15) and a completion bonus of $10. Participants were also informed that the best performer in the group would receive a performance bonus of $50. Task The task was to produce a 90° bimanual coordination pattern by manipulating two angular manipulanda, as in Experiment 1. However, each group had the additional task of concurrently verbalizing a timing pattern aloud in beat with a metronome while attempting to produce the 90° pattern. Participants repeated either a two-count, threecount or four-count cycle in one second, during the bimanual coordination movement. That is, while trying to produce a 90° bimanual coordination pattern, the two-count group continuously verbalized a "one-two" pattern with each number corresponding to a "beep" with a two Hz metronome; the three-count group continuously verbalized a "one-twothree" pattern with each number corresponding to a "beep" with a three Hz metronome; and the four-count group continuously verbalized a "one-two-three-four" pattern with each number corresponding to a "beep" with a four Hz metronome. The first "beep" and each corresponding first "beep" of the count cycle from the metronome had a different frequency to assist participants with distinguishing which "beep" corresponded to the "one" count. Each participant still tried to produce an entire bimanual coordination cycle in one second (one Hz) with the concurrent temporal interference. As participants completed as many as 150 trials within a session, participants were only required to count aloud every other trial, but were encouraged to count internally for the non-verbalization trials.  49 All groups were provided with augmented visual feedback via a Lissajous figure projected on a computer screen, with participants' movement superimposed over the template. Specifically, movements of the right manipulandum produced horizontal movements of the cursor on the screen while movements of the left manipulandum produced vertical movements of the cursor on the screen. Rather than showing on-line feedback of the entire 12-second trial (as per Experiment 1), participants only saw their current position and the previous 500 milliseconds of movement. The purpose of this was to try and increase the accuracy of results, as fewer distractions were present on the screen. Each trial lasted for 12 seconds, with only the last 10 seconds of data analyzed. Apparatus The apparatus and set-up was identical to that used in Experiment 1, except for the microphone and headphones used to collect sound data. Experiment 2 utilized a combined multimedia headset in the form of stereo headphones with a boom microphone attached (Certified Data, AP-850). Experimental Design Instructions Participants were seated in front of the monitor with the two manipulanda on either side. All participants familiarized themselves with the task apparatus and provided with general instructions. These alerted the participants to the goal of the task, that is, to learn how to move the arms in such a way as to produce the pattern displayed on the computer screen, while continuously verbalizing a counting pattern in time to a metronome. Participants were informed whether they were to perform a 2-count, 3-count or 4-count pattern and that each "count" should coincide with each "beep" of the  50 metronome. Participants were informed that the first "beep" of each count cycle of the metronome has a different frequency and they should try to synchronize their "one" count with the different frequency "beep". Participants were reminded that they were only required to verbalize the counting pattern aloud every other trial, with non-verbalization trials to be counted internally ("in their head"). Participants were informed that concurrent on-line feedback of the last 500 milliseconds of their movement would be displayed on the monitor, involving a trace of their movement pattern in the form of a Lissajous figure during the trial. Participants were also informed that each trial would last 12 seconds and they should try to produce 12 full cycles in that period. The criterion for one cycle was then explained and demonstrated to the participants. Participants were informed that they should try to produce a full cycle of the coordination pattern by the completion of one cycle of counting. All participants were reminded to try and keep both their arms moving throughout the each trial. Prior to any counting trials, participants were given a minimum of 15 practice trials of 90° relative phase, with full Lissajous feedback (all 12 seconds of trial), a 1 Hz metronome and no counting required. The rationale for minimal experience with the task was identical to that provided in Experiment 1. Participants were also allowed one counting practice trial with the respective metronome to familiarize them with the timing of the counting pattern. No arm movement was allowed during the counting practice trial. Speed and counting were monitored for correctness throughout the experiment. Scanning As per Experiment 1, a "scanning run" was used to systematically observe the learner's landscape as he or she progressed through the learning process. Scanning runs  51 were performed for all participants prior to the beginning of acquisition trials (preacquisition), immediately following completion of the acquisition trials (postacquisition), and immediately following the final day of testing (post-retention). Results from Experiment 1 provided evidence that scanning via a visual metronome may not be an accurate means of assessment of an individual's landscape for participants acquiring a coordination pattern via Lissajous feedback (see Appendix C for a full review). Thus, scanning for this experiment involved Lissajous figures as a criteria, and concurrent, on-line feedback. Each scanning trial was 15-seconds long at one of 12 different patterns of relative phase (0°, 30°, 60°, 90°, 120°, 150°, 180°, 210°, 240°, 270°, 300° and 330°). Patterns were randomly presented in a discrete manner, with the appropriate criteria and concurrent Lissajous feedback of the previous 500 milliseconds of movement displayed on the monitor. Participants were instructed to complete one full cycle of movement for every "beep" of a 1 Hz metronome, with direction of movement (clockwise or counterclockwise) displayed at the top of the computer screen. All scanning runs consisted of an initial trial of 90° relative phase, trials of all 12 different relative phase patterns, randomly presented, followed by another trial of 90° relative phase. Thus each scanning run required the participant to perform all 12 coordination patterns, with three trials of 90° collected. This assisted with orientation of the participants and served to collect additional data on the to-be-learned coordination pattern. In addition, during the scanning procedure, participants were asked to perform a series of "natural" or "intrinsic" patterns of coordination without feedback and a 1 Hz metronome. Both the in-phase (0°) and anti-phase (180°) are thought to be intrinsic  52 patterns as they can be performed more accurately and with lower variability than other patterns of coordinative movement and are thought to represent natural human coordinative patterns (Kelso, 1984; Fontaine et al, 1997; Zanone & Kelso, 1992; see Kelso, 1995 for a review). Each participant had the in-phase and anti-phase pattern explained and demonstrated to him or her and was asked to perform each pattern at 1 Hz. Prior to acquisition, each participant was also asked to perform a coordination pattern different from in-phase and anti-phase that they felt they would be able to perform consistently for a long period of time. Post acquisition and retention, participants were asked to perform a 90° pattern without feedback. These "intrinsic" trials provided a useful measure of how consistently and accurately the performer was able to produce an in-phase and anti-phase pattern and assisted in discovery of other natural coordinative patterns unique to each individual. Pre-acquisition, participants also performed the three "intrinsic" patterns (0°, 180° and any other pattern they felt they could perform consistently) with full feedback to assist with orientation of the apparatus and concurrent, on-line feedback. Acquisition All participants performed a total of 200 acquisition trials with their respective concurrent temporal interference over two consecutive days. Each day of acquisition participants completed 10 blocks of 10 trials with a maximum two-minute rest in between blocks. Retention/Transfer At the end of the second day of acquisition and one week following acquisition, all participants were given retention and transfer tests. All retention and transfer tests  53 were performed with concurrent augmented visual feedback of the previous 500 milliseconds, as per acquisition. The first block of 10 trials consisted of performing 90° at 1 Hz without any counting interference. The next three blocks of 10 trials included performing 90° at 1 Hz with each of the three temporal interference conditions: twocount interference with a two Hz metronome; three-count interference with a three Hz metronome; and four-count interference with a four Hz metronome. Order of presentation of these three retention/transfer blocks depended on group assignment. All groups first performed the temporal interference condition they practiced in the acquisition phase. Blocks of 10 trials of the remaining two interference conditions were randomly presented. Therefore, the two-count temporal interference group performed two-count interference retention first, followed by transfer to either three-count interference or fourcount interference. Finally, all groups performed a block of 10 trials of 90° during which there was no metronome and participants could adapt whatever speed they felt most comfortable with they thought they would perform most accurately (self paced). Dependent Measures and Analyses Mean relative phase and dependent measures of standard deviation and root mean square error (RMSE) were calculated as described in Experiment 1. As mentioned in Experiment 1, because both Experiment 1 and 2 required an identical single task control group, the data collected from this control group in Experiment 1 was used for analysis in Experiment 2. Dependent measures from the acquisition period were subjected to a 4 (group: control, 2-count, 3-count, 4-count) x 10 (block) A N O V A with repeated measures on the last factor. Dependent measures from the retention/transfer tests were subjected to a 4  54 (group: control, 2-count, 3-count, 4-count) x 2 (time; immediate, one-week) A N O V A with repeated measures on the last factor. Dependent measures from the scanning trials of the 0°, 90°, 180° and 270° coordination patterns were subjected to a 3 (group: 2-count, 3count, 4-count) x 3 (time: pre-acquisition, post-acquisition, post-retention) A N O V A . The control group scanning data was not analyzed in this experiment as it was collected via a different method (see Experiment 1). The alpha level for the entire experiment was set at .05 and the Greenhouse-Geisser Epsilon factor was used to adjust the degrees of freedom for violation of the sphericity assumption (Greenhouse & Geisser, 1959). The Tukey HSD method (Tukey, 1953) was used for all post-hoc comparisons. Results RMSE A summary of analyses performed on R M S E is shown in Table 5. Group means and standard deviations for the R M S E calculations are shown in Table 6. R M S E data plotted as a function of trial type are shown for all conditions in Figure 12, with conditions on the x-axis of the graph representing blocks of 10 trials. All four groups significantly improved their performance of the task during acquisition, shown by a main effect for block F(6, 207) = 21.259, p < 0.001. There was 2  no significant effect for group (p = 0.090) or group x block interaction (p = 0.174). Retention and transfer performance of 90° in all five conditions showed no significant group effect or group x time interaction. However, both 2-count (F(l, 36) = 15.433, p < 0.001) and 3-count (F(l, 36) = 6.275, p = 0.017) transfer/retention did show a significant effect for time (i.e. immediate versus one week retention). Post hoc analyses revealed a decrease in R M S E from immediate performance to one-week performance. Analysis of 2  Note that degrees of freedom in all F ratios have been adjusted by the Greenhouse-Geisser Epsilon Factor  55 Table 5: A N O V A results from Experiment 2 for the dependent measure o f R M S E . Note the Greenhouse-Geisser Epsilon factor was used to adjust degrees of freedom for violation of the sphericity assumption. Effect  .  F-test  p-valuep<0.05  Acquisition block group blockxgroup  F(6, 207) = 21.259 F(3,36) = 2.338 F(17,207) = 1.331  p_< 0.001 p_ = 0.090 £ = 0.174  Retention (No Count, I Hz) time group time x group  F ( l , 36) = 0.126 F(3,36) = 1.202 F(3,36) = 0.542  P = 0.725 £ = 0.323 2 = 0.657  2- Count Transfer time group time x group  F ( l , 36) = 15.433 F(3,36) = 0.605 F(3,36) = 0.877  £ < 0.001 £ = 0.616 p_ = 0.462  *  3- Count Transfer time group time x group  F ( l , 36) = 6.275 F(3,36)= 1.114 F(3,36) = 0.181  £ = 0.017 £ = 0.356 p = 0.908  *  4- Count Transfer time group time x group  F ( l , 36) = 2.737 F(3,36) = 0.914 F(3,36) = 0.919  £ = 0.107 £ = 0.444 £ = 0.442  0- Count (Self-paced) Transfer time F ( l , 36) = 0.074 group F(3,36)= 1.142 time x group F(3,36) = 1.408  £ = 0.787 £ = 0.345 £ = 0.256  Immediate Performance (All trial types) type F(3, 108)= 10.021 group F(3,36) = 0.802 type x group F(9, 108) = 0.553  £ < 0.001 £ = 0.501 £ = 0.833  *  1- Week Performance (All trial types) type F(3,92) = 3.825 group F(3,36) = 1.240 type x group F(8,92) = 0.892  £ = 0.017 £ = 0.309 £ = 0.524  *  *  56 Table 6: Mean and standard deviation values for the dependent measure of R M S E from Experiment 2. Control Group Acquisition Block Block 20 1 16.526 11.662 M 2.240 4.479 SD Condition  Retention Immed 11.507 2.090  M SD  2-Count Group Block Block 20 1 24.714 13.051 10.390 2.576  1-Week Immed 11.294 2.264  11.242 1.824  3-Count Group Block Block 20 1 18.299 13.331 6.388 4.756  4-Count Group Block Block 1 20 21.185 14.046 8.916 3.094  1-Week Immed  1-Week Immed  1-Week  11.940 2.851  12.671 2.607  13.499 2.754  13.359 5.806  12.560 2.142  2-Count Immed Transfer 13.699 M 4.847 SD  1-Week Immed  1-Week Immed  1-Week Immed  1-Week  11.445 2.64  11.476 2.532  12.587 3.38  13.224 3.367  3-Count Immed Transfer 11.834 M 2.581 SD  1-Week Immed  1-Week Immed  1-Week Immed  1-Week  10.890 2.729  10.591 1.725  12.137 2.013  11.940 2.352  4-Count Immed Transfer 11.256 M 1.682 SD  1-Week Immed  1-Week Immed  1-Week Immed  1-Week  11.190 3.089  10.656 1.950  11.103 1.430  12.574 2.152  O-Count Immed Transfer 10.844 M 2.223 SD  1-Week Immed  1-Week Immed  1-Week Immed  1-Week  11.322 3.756  10.270 1.609  10.272 1.612  12.737 5.405  Condition Immediate M SD 1-Week M SD  Retention (No Count) 12.167 3.369 Retention (No Count) 12.351 2.659  12.266 2.483  11.274 2.060  11.597 2.598  10.291 1.979  13.638 4.521  12.692 3.727  12.492 3.545  11.959 2.584  14.493 4.066  13.143 3.016  12.549 2.485  12.005 2.236  Trial Type (Retention/Transfer) 4-Count 3-Count 2-Count 13.524 4.004  12.236 2.892  11.974 2.619  2-Count  3-Count  4-Count  12.183 2.988  11.390 2.250  11.381 2.269  0-Count (Self-Paced) 11.275 2.300 0-Count (Self-Paced) 11.150 3.499  57  58 immediate performance of all five retention/transfer conditions showed a significant effect for type of retention/transfer F(3,108) = 10.021, p < 0.001 but no significant group effect (p = 0.501) or group x type interaction (p = 0.833). The main effect for type was found to be due to the difference in extreme means, with the highest R M S E for 2-count transfer and lowest R M S E for self paced (0-count) transfer. Analysis of performance one week post acquisition of all five retention/transfer conditions also showed a significant effect for type F(3, 92) = 3.825, p = 0.017 but no significant group effect (p = 0.309) or group x type interaction (p = 0.524). The main effect for type was found to be due to the difference in extreme means, with the highest R M S E for performance of 90° with a 1 Hz metronome (no counting) and lowest R M S E for self paced (0-count) transfer. A summary of analyses performed on R M S E for the scanning trials is shown in Table 7. Group means and standard deviations for the R M S E calculations are shown in Table 8. Analysis of the scanning trials of the intrinsic coordination patterns of 0° and 180° did not show a significant effect for time (p = 0.456, p = 0.629), group (p = 0.381, p = 0.995) or time x group interaction (p = 0.188, p = 0.876). However scanning performance of both 90° and 270° showed a significant effect for time (F(l, 34) = 49.091, p < 0.001 and F(2, 50) = 8.902, p = 0.001) but no effect for group (p = 0.957, p = 0.772) or group x time interaction (p = 0.406, p = 0.896). Post hoc analyses of the 90° scan showed a significant decrease in R M S E between pre-acquisition scanning and both post-acquisition and post-retention scanning. The main effect for time for the 270° scan was found to be due to the difference in extreme means, with highest R M S E during the pre-acquisition  59  Table 7: A N O V A results from Experiment 2 for the dependent measure o f R M S E for scanning trials. Note the Greenhouse-Geisser Epsilon factor was used to adjust degrees of freedom for violation of the sphericity assumption. Effect  F-test  p-valuep<0.05  F(l,39) =  0.717 F(2, 26) = 1.000 F ( 3 , 39) = 1.678  p = 0.456 2 = 0.381 2 = 0.188  Scan 90° time group time x group  F(l,34) = 49.091 F(2, 27) = 0.050 F ( 3 , 3 4 ) = 0.970  2 < 0.001 2 = 0.957 2 = 0.406  Scan 180° time group time x group  F(2, 46) = 0.417 F(2, 27) = 0.005 F(3,46) = 0.260  2 = 0.629 2 = 0.995 2 = 0.876  Scan 270° time group time x group  F(2, 50) = 8.902 F(2, 27) = 0.261 E(4, 50) = 0.249  2 = 0.001 2 = 0.772 2 = 0.896  Scan 0° time group time x group  *  *  60 Table 8. Mean and standard deviation values for the dependent measure o f R M S E for scanning trials from Experiment 2. Condition 2-Count Group Scan0° PrePost- PostAcq Acq Ret M 11.480 10.210 9.070 SD 6.115 3.870 3.509  Scan 90° M SD  Scan 180° M SD  Scan 270° M SD  3-Count Group 4- Count Group PrePost- Post- PrePost- PostAcq Acq Ret Acq Acq Ret 12.430 9.650 14.390 10.756 11.030 9.570 5.306 2.637 9.588 3.516 4.014 2.810  PrePost- Post- PrePost- Post- PrePostAcq Acq Ret Acq Acq Ret Acq Acq 21.630 10.770 9.400 21.830 9.810 9.860 19.320 12.420 7.371 2.813 1.664 8.473 3.022 2.229 6.399 2.805  PostRet 11.070 5.165  PrePost- Post- PrePost- Post- PrePost- PostAcq Acq Ret Acq Acq Ret Acq Acq Ret 18.830 17.260 17.130 18.190 19.660 16.320 20.720 15.680 16.990 12.514 5.904 12.791 9.487 17.725 9.367 17.298 5.018 7.876  PrePost- Post- PrePost- Post- PrePostAcq Acq Ret Acq Acq Ret Acq Acq 58.630 42.100 40.440 49.620 41.510 29.480 49.590 38.790 32.461 24.907 37.826 24.613 34.659 34.853 25.236 28.587  PostRet 29.620 22.747  61 scan and lowest R M S E during the post retention scan. Performances of all four coordination patterns, with data from all four groups combined, are shown in Figure 13. Standard Deviation A summary of analyses performed on standard deviation is shown in Table 9. Group means and standard deviations for the standard deviation calculations are shown in Table 10. Standard deviation results closely mirrored R M S E results. All four groups significantly improved their performance of the task during acquisition, shown by a main effect for block F(8, 273) = 23.801, p < 0.001. There was no significant effect for group (p = 0.120) or group x block interaction (p = 0.157). Retention and transfer performance of 90° in all five conditions showed no significant group effect or group x time interaction. Unlike R M S E , 2-count (p = 0.051) and 3-count (p = 0.087) retention/transfer did not show a significant effect for time, however 4-count transfer/retention did show a significant effect for time F ( l , 36) = 4.205, p < 0.048. Post hoc analyses revealed a decrease in standard deviation from immediate performance to one-week performance. Analysis of immediate performance of all five retention/transfer conditions showed a significant effect for type of retention/transfer F(3,l 11) = 5.007, p = 0.002 but no significant group effect (p = 0.584) or group x type interaction (p = 0.868). The main effect for type was found to be due to the difference in extreme means, with the highest standard deviation for 2- count transfer and lowest standard deviation for self paced (0count) transfer. Analysis of performance one week post acquisition of all five retention/transfer conditions also showed a significant effect for type F(3, 108) = 6.023, p = 0.001 but no significant group effect (p = 0.227) or group x type interaction  62  63 Table 9: A N O V A results from Experiment 2 for the dependent measure of standard deviation. Note the Greenhouse-Geisser Epsilon factor was used to adjust degrees of freedom for violation of the sphericity assumption. Effect  F-test  p-valuep<0.05  Acquisition block group block x group  F(8, 273) = 23.801 F(3, 36) = 2.084 F(23, 273)= 1.315  p < 0.001 p = 0.120 £ = 0.157  *  Retention (No Count, 1 Hz) F ( l , 36) = 0.024 time F(3, 36) = 0.554 group F(3, 36)= 1.841 time x group  £ = 0.877 £ = 0.346 £ = 0.649  2-Count Transfer time group time x group  F ( l , 36) = 4.091 F(3, 36) = 0.704 F(3, 36)= 1.472  £ = 0.051 £ = 0.556 £ = 0.238  3-Count Transfer time group time x group  F ( l , 36) = 3.089 F(3, 36) = 1.198 F(3, 36) = 0.278  £ = 0.087 £ = 0.324 £ = 0.841  4-Count Transfer time group time x group  F ( l , 36) = 4.205 F(3, 36) = 0.706 F(3, 36) = 0.925  £ = 0.048 £ = 0.555 £ = 0.438  O-Count (Self-paced) Transfer F ( l , 36) = 0.014 time F(3, 36) =1.126 group F(3, 36)= 1.148 time x group  £ = 0.907 £ = 0.351 £ = 0.343  Immediate Performance (All trial types) type F(3, 111) = 5.007 group F(3, 36) = 0.6562 typexgroup F(9, 111) = 0.512  £ = 0.002 £ = 0.584 £ = 0.868  *  1-Week Performance (All trial types) type F(3, 108) = 6.023 group F(3, 36) =1.516 type x group F(9, 108) = 1.040  £ = 0.001 £ = 0.227 £ = 0.414  *  *  64 Table 10: Mean and standard deviation values for the dependent measure of standard deviation from Experiment 2. Control Group Acquisition Block Block 20 1 14.264 10.962 M 2.221 2.755 SD Condition  Retention Immed  M  SD 2-Count Transfer  M SD  2-Count Group Block Block 20 1 16.345 11.757 3.497 2.109  1-Week Immed 10.248 1.676  10.491 1.827  10.318 1.982  Immed  1-Week Immed  11.058 1.959  10.001 1.877  10.898 1.780  3-Count Immed Transfer 10.215 M 2.24 SD  1-Week Immed  4-Count Immed Transfer 10.030 M 1.081 SD  1-Week Immed  0-Count Immed Transfer 10.142 M 2.116 SD  1-Week Immed  Condition Immediate  M  SD 1-Week  M SD  9.593 1.958  9.828 1.876  10.207 2.392  Retention (No Count) 10.850 2.749 Retention (No Count) 10.913 1.870  10.012 1.541  10.355 2.299  9.248 1.452  3-Count Group Block Block 1 20 14.487 12.172 3.305 3.963  4-Count Group Block Block 1 20 16.776 11.927 5.605 1.176  1-Week Immed  1-Week Immed  1-Week  10.186 1.589  11.290 1.95  11.858 1.662  11.741 4.868  10.918 1.208  1-Week Immed  1-Week Immed  1-Week  10.106 1.983  10.995 1.794  11.705 2.574  11.894 3.514  1-Week Immed 9.642 1.963  11.017 2.494  1-Week Immed 9.341 1.738  10.997 2.687  1-Week Immed 9.468 1.448  10.676 1.886  11.258 1.96  1-Week Immed  1-Week  10.934 1.68  10.728 2.036  11.330 2.268  1-Week Immed  1-Week  10.328 1.423  10.623 1.515  10.650 1.348  1-Week Immed 9.724 1.75  10.365 1.33  Trial Type (Retention/Transfer) 4-Count 3-Count 2-Count 11.277 2.345  10.644 2.152  10.508 1.925  2-Count  3-Count  4-Count  10.702 2.119  10.224 1.940  10.030 1.659  1-Week 10.892 2.089  0-Count (Self-Paced) 10.108 1.743 0-Count (Self-Paced) 10.073 1.954  65 (p = 0.414). The main effect for type was found to be due to the difference in extreme means, with the highest standard deviation for performance of 90° with a 1 Hz metronome (no counting) and lowest standard deviation for 4-count retention/transfer. A summary of analyses performed on standard deviation for the scanning trials is shown in Table 11. Group means and standard deviations for the standard deviation calculations are shown in Table 12. Analysis of standard deviation of the scanning performance also closely mirrored the analysis o f R M S E . Scanning trials of both intrinsic coordination patterns of 0° and 180° did not show a significant effect for time (p = 0.151, p = 0.438), group (p = 0.123, p = 0.912) or time x group interaction (p = 0.153,p = 0.781). However scanning performance of both 90° and 270° showed a significant effect for time (F(l, 34) = 95.017, p < 0.001 and F(2, 54) = 4.001, p = 0.024) but no effect for group (p = 0.697, p = 0.790) or group x time interaction (p = 0.485, p = 0.578). Post hoc analyses of the 90° scan showed a significant decrease in standard deviation between pre-acquisition scanning and both post-acquisition and post-retention scanning. The main effect for time for the 270° scan was found to be due to the difference in extreme means, with highest R M S E during the pre-acquisition scan and lowest R M S E during the post retention scan. Sound Data The filtered, full-wave rectified sound data was analyzed to determine the peak output of the participants' counting. The peak of each count was considered to be the point where the slope of the sound data changed from positive to negative, with the restriction that the next peak be a minimum of 100 msec later in time. Once each peak was determined, the time during the trial at which the peak occurred was established.  66 Table 11: A N O V A results from Experiment 2 for the dependent measure of standard deviation for scanning trials. Note the Greenhouse-Geisser Epsilon factor was used to adjust degrees of freedom for violation of the sphericity assumption. Effect  F-test  p-valuep<0.05  Scan 0° time group time x group  E(2, 40) = 2.058 E(2, 26) = 2.278 F(3,40) = 1.849  E = 0.151 P = 0.123 p = 0.153  Scan 90° time group time x group  F(l,34) = 95.017 F(2, 27) = 0.365 E(3, 34) = 0.798  p_< 0.001 p_ = 0.697 P = 0.485  Scan 180° time group time x group  E(2, 46) = 0.801 F(2, 27) = 0.093 E(3, 46) = 0.398  2 = 0.438 2 = 0.912 2 = 0.781  Scan 270° time group time x group  F(2, 54) = 4.001 F(2, 27) = 0.238 E(4, 54) = 0.725  2 = 0.024 2 = 0.790 2 = 0.578  *  *  67 Table 12: Mean and standard deviation values for the dependent measure of standard deviation for scanning trials from Experiment 2. Condition 2-Count Group Post- PostPreScan 0° Acq Ret Acq 6.730 9.500 8.290 M 3.031 1.434 4.933 SD  Scan 90° M SD  PostPreAcq Acq 16.670 9.040 2.258 2.402  PostRet 8.370 2.085  4- Count Group 3-Count Group Post- PostPost- Post- PrePreRet Acq Acq Acq Ret Acq 10.830 7.270 11.820 8.611 7.490 6.750 1.925 4.898 8.695 2.676 2.634 2.141 PostPreAcq Acq 18.290 9.100 6.326 2.767  PostRet 9.180 2.323  Post- PostPreRet Ac q Acq 16.47C 10.030 9.460 3.724 1.551 1.211  Post- PostPost- Post- PrePost- Post- PreScan 180° PreRet Acq Acq Acq Ret Acq Acq Ret Acq M 17.690 15.240 15.680 17.200 18.260 15.350 19.360 12.630 14.330 SD 12.858 5.874 12.737 9.768 16.721 8.803 16.537 4.210 6.594  Scan 270° M SD  PrePost- Post- PrePost- Post- PrePostAcq Acq Ret Acq Acq Ret Acq Acq 49.230 33.050 35.260 41.390 39.080 27.820 35.370 33.920 31.052 20.516 34.215 25.017 34.562 34.825 16.026 24.913  PostRet 25.960 22.888  68 This absolute time was converted into a relative time within each 1 Hz metronome cycle. That is, for all counting conditions, the "one" beep of the metronome occurred every 1000 msec during the trial. Each peak of the sound data was converted to a time between 0-999 msec, relative to the temporal position of the peak output of the metronome "one" beep. Thus the peak of every count by the participants was converted into a relative cycle time between 0-999 msec. If a participant counted perfectly in time with the metronome, this would result in sound peaks at 0 and 500 msec for the 2-count group; 0, 333 and 667 msec for the 3-count group; and 0, 250, 500 and 750 msec for the 4-count group. The full 1000 msec metronome cycle was divided into 10 msec intervals and the frequency of each count's relative cycle time was determined. The relative temporal location of counting peaks (and metronome peaks) for the first 10 acquisition trials (early acquisition) and last 10 acquisition trials (late acquisition), separated by group, are shown in figures 14-16. In addition to the relative timing of the counting peaks, it was also determined where in the movement participants were for each metronome pulse. The hand positions for each metronome pulse were converted into an X - Y coordinate and the locations were plotted into an X - Y grid for early acquisition (first 10 trials), late acquisition (last 10 trials) as well as transfer trials to the 2-count, 3-count and 4-count conditions. These plots provided important information regarding the participants' strategy for completing the task, including the use of anchoring movement positions to the metronome pulses. As no quantitative analyses were performed on these plots, they are discussed in the discussion portion of this paper.  69  70  71  72 Discussion The purpose of this experiment was to examine an alternate form of interference and its effect on learning. Participants were required to verbalize different counting patterns while performing a 90° bimanual coordination task to determine if typical contextual interference effects were observed. Concurrent 4-count verbalization was thought to have high compatibility with the task and thus be low interference, while concurrent 2-count verbalization was thought to have low compatibility with the task and thus provide high interference. If concurrent temporal interference showed the same performance effects as varying the presentation schedule (blocked versus random practice), it was predicted that the 4-count group would outperform the 2-count group during acquisition, while the 2-count group would outperform the 4-count group during retention and transfer. Dependent Measures Analysis of both standard deviation and R M S E during the acquisition period showed all groups improved in performance during the 20 blocks of trials. This was also reflected in the scanning data, as there was a significant decrease in both R M S E and standard deviation from the pre-acquisition scan to the post-acquisition scan of 90°. However, although there was significant improvement during acquisition there was no significant group effect or interaction effect in acquisition, retention or any transfer tests. This may suggest that concurrent temporal interference does not follow the typical contextual interference results or that the various counting patterns did not provide sufficiently differential interference. Since none of the counting groups were significantly  73  different than the control group (no counting), it appears that the counting in and of itself was not interfering. In addition, providing the participants with a minimum of 15 practice trials without counting may have masked any interference effects of the different counting patterns. Allowing participants trials to achieve a level of experience with the task, enabled them to get the idea of the movement before the counting interference was introduced. However, it was likely during these practice trials that the cognitive demands of the task were greatest (Gentile, 1972), and thus the time during which interference of concurrent counting would be greatest. Once the participants understood the basics of the task, addition of the counting patterns may not have provided sufficient interference to detect performance differences. Requiring participants to perform the concurrent counting patterns from the first acquisition trials would likely have been more interfering and may have caused group differences in performance during acquisition, retention and transfer. Although no group differences were found in acquisition, retention or transfer, results from this study did provide some support for different amounts of interference with different counting patterns. All participants performed five sets of retention/transfer trials (no counting 1Hz, 2-count, 3-count, 4-count, no counting self paced) immediately following acquisition and one week later. Analysis of these five conditions showed a significant effect for the type of trial. During the immediate retention/transfer testing, all participants performed with lowest R M S E and standard deviation during self paced trials with no counting and performed with highest R M S E and standard deviation during 2count trials. It was not surprising that the self-paced condition was performed with the  74 least error and variation as there was no counting interference and participants could adapt their preferred speed. The highest error and variation in the 2-count condition provided support for the prediction that the 2-count pattern would produce the highest amount of interference for the participants. During the one week retention/transfer testing, all participants again performed with lowest R M S E during the self paced condition, however R M S E was highest during trials with the 1 Hz metronome and no counting. Standard deviation results were lowest during 4-count trials and highest during trials with the 1 Hz metronome and no counting. The result of lowest standard deviation during 4-count trials, one week after acquisition, provided weak support for the prediction that 4-count trials would produce the lowest amount of interference for the participants. The highest R M S E and standard deviation in the no counting 1 Hz condition, one week following acquisition, was not expected as there was no counting to interfere with the performance of the coordination pattern. One possible explanation for this result was that performance of acquisition trials with counting patterns may have provided participants with a strategy to help them perform the task. Removal of the counting pattern may not have allowed the participants to employ their learned strategy. Analysis of the five types of retention and transfer conditions, both immediate and one week following acquisition, did not show a group effect or group by type interaction. That is, there was no significant difference in the way the groups performed each of the five conditions. This was surprising, as the practice specificity theory would predict increased performance in the condition under which each group practiced during acquisition. For example, the group that learned the coordination pattern with 2-count interference would be expected to perform with lower error and variation than the other  75 groups when tested in the 2-count retention/transfer condition. The results of this study do not support the practice specificity hypothesis as no group performed significantly better during retention/transfer in the condition under which they acquired the coordination pattern. As with Experiment 1, this may suggest that the importance of practice specificity may be less than originally thought. Analysis of the scanning performance for both dependent measures produced expected results with regards to pre and post acquisition performance of 0°, 90°, 180° and 270°. It was expected that acquisition of a new coordination pattern would not result in a significant change to performance of the two intrinsic coordination patterns of 0° and 180°. Neither 0° or 180° showed a significant difference in performance between preacquisition, post-acquisition or post-retention scans. The significant difference between pre-acquisition and post-acquisition scans of 90° supports the result of a significant block effect during acquisition. Previous studies (Zanone & Kelso, 1997) have shown that acquisition of a new coordination pattern also produces positive transfer to the symmetrical partner. For example, acquisition of a 90° coordination pattern (right hand leading left hand by one quarter cycle) also improves performance of a 270° coordination pattern (left hand leading right hand by one quarter cycle) that has not been practiced. Analysis of the scanning data of the 270° pattern supports Zanone and Kelso's findings, as there was a significant decrease in R M S E and standard deviation before and after acquisition of the 90° pattern. Sound Data Inspection of the sound data provided information regarding participants' strategies during performance of the bimanual coordination movement and concurrent  76 counting. Examination of the temporal location of counting peaks showed high frequency of counts in time with the metronome (Figures 14-16). The highest peaks of the distribution of counting, coincided with the "beeps" of the metronome for two, three and four-count patterns. Thus it appeared that the participants were able to accurately count in time with the metronome at a consistent, stable cadence. Position of movement ( X - Y location) at metronome pulses during early acquisition (first 10 trials), late acquisition (last 10 trials) and all three transfer conditions (2-count, 3-count, 4-count) were examined. Examination of where in the movement participants were for each metronome pulse provided information regarding anchoring. Anchoring can be defined as intentionally synchronizing a particular point in a movement cycle with a stimulus, such as a metronome pulse, and has been shown to decrease spatial and temporal variability at the anchored points (Fink et al., 2000). The task for this experiment, production of a 90° coordination pattern, involved a left limb lag by one quarter of a cycle and thus included four anchor points (two reversal points for each hand) equally spaced in time. The prediction of this experiment was that different groups would employ different anchoring strategies and this would result in different group performances during acquisition, retention and transfer. Performance of the 90° coordination pattern with a 4 Hz metronome was expected to produce acquisition performance with the lowest RMSE. This was predicted because participants would be able to anchor all four reversal points to the pulse of the metronome, thus lowering variability of movement. Alternately, performance of the 90° coordination pattern with a 2 Hz metronome was expected to produce acquisition performance with the highest R M S E as participants would only be  77 able to anchor reversal points for one hand. It was unclear how participants would perform the coordination pattern with a 3 H z metronome as only one anchor point was compatible with the metronome pulse, however a three count pattern was though to be beneficial for producing fluid coordination movements due to anecdotal evidence regarding music and dance patterns. Examination of the movement position relative to the metronome pulses did not provide support for the predictions of this experiment. Within each group, different individual employed different strategies to perform the same task. During late acquisition trials, once strategies should be ingrained, some participants did appear to use an anchoring strategy, as all 10 trials were similar with respect to where they were in the movement during the metronome pulses. Other participants, however, varied greatly where they were in the movement during the metronome pulse and did not appear to employ any anchoring strategy. Furthermore, each group had this variability with some participants anchoring to the metronome and others not anchoring. Examples from each group of participants who anchored and did not anchor during trials in late acquisition are shown in Figures 17-19. It was predicted that anchoring would be a beneficial strategy for participants, resulting in reduced variability of movement and therefore, lower RMSE. However, there did not appear to be a difference in R M S E between participants who did and did not anchor. Thus, anchoring of movement points to an outside stimulus may not be beneficial to a 90° coordination pattern. Previous studies examining the phenomenon of anchoring only required participants to perform in-phase or anti-phase movements. During in-phase or anti-phase movement patterns, reversal points of one limb corresponded to a reversal  78  79 Q)  O  8£  mo  (saaj6ap) pueH U^l  80  81  (saajBep) pueH wan  o  o CO  <  <'"  G  •  0) %  < ^O  1*  •D co D)  • •<°o <o 8  #  w a) eft  n  £><o  ^  OO  0  * o*  o o  o_ < o •pi  o  < <  •o <  < < <  Oo  <3  o CO  P<<  <  «p  I?9-  eft  <o  82  (soajBap) pueH yen o  CM  +J Q. X HI  o  o  CN  o (0  0  CD  o°  tn  ° o^ o <>o_  0  l#  3  a—  < o<><° ° •  < <x <  O Oo  «  |o ?L O)  I <  c c c  <  2  a £g O CO  .E x: «-  +2  •i * C  3  .2 i_§  < <  W 0) CO o> a> 2,  o m  •a c ro  o +;  ^3  Q. C  c o CD  O  E 4 CD >  o <  CD 3  <<  o  •g o o  ft  o  l  8rib  C P Q ^ O  "0~£> O O C T  o  o<9  ft  6  o  0<  <  83 a) o £ 5 £ 5 .c o O I— I— U• O < O  (saejBep) pueH uan  CM  +J  Q.  X LU  W  <  o w ^  o £  CO Q . CM 3  o  I °>  1 = 0)  <  U  sg  w o> a> k_  C5^  a>  o  o  c 5 .2 Q.  •o c (0  X  wo  c o Cd O) V > 'tf o  £ 3  G)  •S  5  CM  ^  o  84 point for the alternate limb as well. Thus, anchoring during these coordination patterns may also have functioned to strengthen the coupling between the limbs, resulting lower variability. However, during a 90° coordination pattern, reversal points of one limb corresponded to a midpoint of the other limb. Thus, anchoring during these movements may not have produced the same effect of lowering variability. During transfer to different counting trials, those participants that adopted an anchoring strategy during late acquisition, continued to employ an anchoring strategy. For example, participant 21 who used an anchoring strategy during 3-count acquisition also employed a similar anchoring strategy during transfer to 2-count and 4-count trials. The movement position plots for these transfer trials for participant 21 are shown in Figure20A&20B. In addition, movement position relative to metronome pulse was plotted for the control group who acquired the coordination pattern without counting and a 1 Hz metronome. These participants were not expected to use an anchoring during acquisition as the metronome was simply a pacing instrument. Again, as with the counting groups, some participants appeared to adopt an anchoring strategy, even though no instructions were given regarding position of movement at metronome pulses. Although the grouping of metronome pulses was not as consistent as the counting groups, there is an unequal distribution of pulses throughout the movement cycle, suggesting a coinciding of movement position with the metronome pulse. A sample plot of an individual from the control group that appeared to use anchoring strategy is shown in Figure 21. As with the counting groups, individuals that employed an anchoring strategy during acquisition, continued this strategy during transfer to counting trials. Results from an anchoring  85 CD  O  8 I (sa8j6ap) puBH yen  ft  86  (se&iBap) puBH yen  <  ft  87 <D  c O  (saajBsp) pue|4j ua~]  V) CD o  CO 2,  T3 E  re  ft  88 control participant, performing transfer trials to 2-count, 3-count and 4-count are shown in Figure 22A, 22B and 22C: In summary, participants in all four groups employed different strategies to perform the task. Some participants in all groups did adopt the predicted anchoring strategy throughout acquisition, retention and transfer, while others did not. This variability within the groups may partially explain why there were no group differences observed during acquisition, retention and transfer. Predictions of this experiment were based on participants using a particular type of strategy to perform the task. The fact that numerous participants did not employ the expected strategy would help explain why the predicted results were not seen. In this experiment, the movement position relative to the metronome pulse was not explicitly controlled through instructions to the participants. Participants were instructed to perform a bimanual coordination pattern while concurrently verbalizing a counting pattern. Instructions involved the suggestion of anchoring movement positions to the metronome but this was not required of the participants. To ensure the different groups used the various anchoring strategies, instructions could have been given with required hand placements at given metronome counts. For example, the 4-count group could have been instructed that the right hand was to be at the "in" position at the one count, the left hand was to be "in" at the two count, the right hand was to be "out" at the three count and the left hand was to be "out" at the four count. This would have ensured that all participants used the expected strategy of anchoring movement positions to metronome pulses. Although this change to instructions would have ensured participants employed the predicted anchoring strategy, it is still unclear if this strategy would have produced  89 CD  O •  O H O  (saajBap) PUBH  ft  90 <D <D  O  o  O  2  h o  f <  (S33j6ap) puBH y a i  (0  o> o>  TO 2.  •o c re X  D)  ft  91  (saajBap) puei-i ua~|  ft  92 the predicted results. During acquisition, it was predicted that anchoring of all four reversal points would produce lower R M S E than anchoring of only two reversal points. This was hypothesized due to previous literature showing decreased variability of anchored movements (Byblow et al., 1994 & Fink et al., 2000). However, results from this experiment did not show that those individuals that performed the acquisition trials with an anchoring strategy performed the movement pattern with lower R M S E than those that did not. As previous literature on anchoring only examined in-phase and anti-phase movements it may be that anchoring of movement positions is not beneficial for performance of a 90° coordination task. During in-phase and anti-phase movement patterns, reversal points of one limb corresponded to a reversal point for the alternate limb as well. However, during a 90° coordination pattern, reversal points of one limb corresponded to a midpoint of the other limb. Thus, the assumption that increased anchoring during a 90° coordination pattern produces decreased variability may not be a valid one. Summary In summary, results of this experiment were not as predicted as there were no group differences during acquisition, retention and transfer. Although all four groups learned a new bimanual coordination pattern, the different counting patterns did not appear to provide sufficient interference for any group differences to materialize. Scanning performance results were as predicted as neither intrinsic coordination pattern (0° or 180°) showed any difference before and after acquisition, yet both the to-belearned pattern (90°) and the symmetrical partner (270°) showed significantly changed pre and post acquisition. Examination of where participants coincided their movement  93 with the metronome pulse revealed participants in all groups adopted different strategies to perform the movement. In all groups, some individuals adopted an anchoring strategy, while others did not. This may have also contributed to the lack of main effect for group during acquisition, retention and transfer. However, those participants that used an anchoring strategy appeared to use this strategy during acquisition, retention and transfer. As compatible and incompatible concurrent temporal patterns have not been previously examined, further research in this area may provide more information regarding different counting patterns while trying to perform a coordination task. Suggestions for further research include requiring participants to anchor their movements to all counts in their verbal patterns/metronome pulse or incorporating the counting pattern before participants have a chance to practice and become experienced with the task. However, it is still unclear if an anchoring strategy would be beneficial to the learning of a 90° bimanual coordination pattern all anchoring studies to date have focussed solely on in-phase (0°) and anti-phase (180°) movements.  94 Conclusion Two experiments were conducted to investigate the motor learning phenomenon known as contextual interference. Results from Experiment 1 showed that rather than extra trials on a single task, additional practice on a second task could be beneficial to learning, even if the desired outcome is only to learn a single task. In addition, it appeared that the learning benefits of interference could manifest themselves during acquisition, if sufficient trials are employed. Experiment 2 had limited results regarding interference, as the concurrent temporal interference did not appear to provide sufficient interference for group differences to emerge. However it did provide information pertaining to anchoring strategies used when participants performed a bimanual coordination task while concurrently counting in time with a metronome. Neither study provided support for the practice specificity effect. In both studies, group that performed under identical conditions did not outperform other groups during retention testing. This suggested that the notion of practice specificity in motor learning may be overstated. Rather, it seems that practice variability and introduction of contextual interference are important tools to maximize learning and performance of a motor task. The movement position data relative to metronome pulse from Experiment 2 illustrated how participants are able to use different strategies to achieve the same outcome. In addition, participants who employed a more variable movement strategy, by not anchoring specific positions to the metronome, were able to perform the task without any apparent performance decrement. Thus it is important for practitioners to ensure that learners practice either a single task in a variety of situations (practice variability) or  provide contextual interference during the learning process, typically by introducing multiple tasks presented in an unpredictable manner. Overall, the idea of introducing interference does not seem to negatively affect performance and may have significant positive effects. Further research in this area should allow insight into the effect of different forms of interference and the most efficient way to introduce this interference to maximize performance and learning.  96 References Battig, W. F. (1956) Transfer from Verbal Pretraining to Motor Performance as a Function of Motor Task Complexity. Journal of Experimental Psychology, 51(6), 371-378. Battig, W. F. (1966). Facilitation and interference. I n E . A. Bilodeau (Ed.), Acquisition of skill (pp. 215-244). New York: Academic Press. Battig, W. F. (1979). The flexibility of human memory. In L. S. Lermak & F. I. M . Craik (Eds.), Levels of processing in human memory (pp. 23-44). Hilldale, NJ: Erlbaum. Bernstein N . A . (1967). The co-ordinated and regulation of movement. In H . T. A . Whiting, (Ed.), Human Motor Actions: Bernstein Reassessed (1984) (pp. 362365). North Holland: Amsterdam. Bortoli, L., Robazza, C , Durigon, V . , & Carra, C. (1992). Effects of contextual interference on learning technical sport skills. Perceptual and Motor Skills. 75. 555-562. Brady, F. (1997). Contextual interference and teaching golf skills. Perceptual and Motor Skills. 84. 347-350. Brady, F. (1998). A theoretical and empirical review of the contextual interference effect and the learning of motor skills. Quest 50, 266-293. Bransford, J. D., Franks, J. J., Morris, C. D., & Stein, B. S. (1979). Some general constraints on learning and memory research. In L. S. Cermak & F. I. M . Craik (Eds.), Levels of processing and human memory (pp. 331-354). Hillsdale, NJ: Lawrence Erlbaum. Brown, J. (1958). Some tests of the decay theory of immediate memory. Quarterly Journal of Experimental Psychology. 10, 12-21. Byblow, W.D., Carson, R.G., & Goodman, D. (1994). Expressions of asymmetries and anchoring in bimanual coordination. Human Movement Science, 13, 3-28. Carson, R.G., Byblow, W.D., & Goodman, D. (1994). The Dynamic Substructure of Bimanual Coordination. In S P . Swinner, H . Heuer, J. Massion & P. Casaer (Eds.), Interlimb Coordination: Neural, Dynamical & Cognitive Constraints (pp. 319-337). Orlando, FL: Academic Press. Cassio, M . , Meira, J. R., & Tani, G. O. (2001). The Contextual Interference Effect in Acquisition of Dart-Throwing Skill Tested on a Transfer Test with Extended Trials. Perceptual and Motor Skills. 92. 910-918.  97 Del Rey, P. (1989). Training and contextual interference effects on memory and transfer. Research Quarterly for Exercise and Sport. 60. 342-347. Del Rey, P., Whitehurst, M . , & Wood, J. M . (1983). Effects of experience and contextual interference on learning and transfer by boys and girls. Perceptual and Motor Skills. 56. 581-582. Del Rey, P., Wughalter, E. H . , & Whitehurst, M . (1982). The effects of contextual interference on females with varied experience in open sport skills. Research Quarterly for Exercise and Sport. 53. 108-115. Del Rey, P., Xiaoying, L . , & Simpson, K. J. (1994). Does retroactive inhibition influence contextual interference effects? Research Quarterly for Exercise and Sport. 65, 120-126. Fink, P.W., Foo, P., Jirsa, V . K . , & Kelso, J.A.S. (2000). Local and global stabilization of coordination by sensory information. Experimental Brain Research. 134. 9-20. Fleishman, E. A . & Quaintance, M . K. (1984). Taxonomies of Human Performance. Orlando, FL: Academic Press. Fontaine, R. J., Lee, T. D., & Swinnen, S. P. (1997). Learning a New Bimanual Coordination Pattern: Reciprocal Influences of Intrinsic and To-Be-Learned Patterns. Canadian Journal of Experimental Psychology. 51(1). 1-9. Franks, I. M . , & Stanley, M . L . (1991). Learning the Invariants of a Perceptual Motor Skill. Canadian Journal of Psychology, 45(3), 303-320. Franks, I. M . , Wilberg, R. B., & Fishburne, G. J. (1982). Consistency and error in motor performance. Human Movement Science, 1(2), 109-123. French, K. E., Rink, J. E., & Werner, P. H . (1990). Effects of contextual interference on retention of three volleyball skills. Perceptual and Motor Skills. 71. 179-186. Gabriele, T. E., Hall, C. R., & Lee, T. D. (1989). Cognition in motor learning: Imagery effects on contextual interference. Human Movement Science, 8, 227-245. Gentile, A. M . (1972). A working model of skill acquisition with application to teaching. Quest. 17. 3-23. Gentner, D. R. (1987). Timing of Skilled Motor Performance: Tests of the Proportional Duration Model. Psychological Review. 94(2), 255-276. Goodwin, J. E. & Meeuwsen, H . J. (1996). Investigation of the contextual interference effect in the manipulation of the motor parameter of over-all force. Perceptual and Motor Skills. 83, 735-743.  98 Green, S., & Sherwood, D. E. (2000). The Benefits of Random Variable Practice for Accuracy and Temporal Error Detection in a Rapid Aiming Task. Research Quarterly for Exercise and Sport. 71(4). 398-402. Greenhouse, S.W., & Geisser, S. (1959). On methods in the analysis of profile data. Psvchometrika. 24. 95-112. Haken, H . , Kelso, J. A . S., & Bunz, H . (1985). A theoretical model of phase transitions in human hand movements. Biological Cybernetics. 51, 347-356. Hall, C , Bernoties, L . , & Schmidt, D. (1995). Interference effects of mental imagery on a motor task. British Journal of Psychology. 86. 181-190. Hall, K. G., Domingues, D. A., & Cavazos, R. (1994). Contextual interference effects with skilled baseball players. Perceptual and Motor Skills. 78. 835-841. Hall, K. G. & Magill, R. A . (1995). Variability of Practice and Contextual Interference in Motor Skill Learning. Journal of Motor Behavior. 27(4), 299-309. Hebert, E. P., Landin, D., & Solmon, A. (1996). Practice schedule effects on the performance and learning of low- and high-skilled students: A n Applied Study. Research Quarterly for Exercise and Sport. 67. 52-58. Henry, F. M . (1968). Specificity vs. Generality in Learning Motor Skill. In R. C. Brown & G. S. Kenyon. Classical Studies on Physical Activity, (pp. 328-331). Englewood Cliffs, NJ: Prentice Hall. Hodges, N.J., & Franks, I.M. (2000). Attention focusing instructions and coordination bias: Implications for learning a novel bimanual task. Human Movement Science. 19, 843-867. Hodges, N . J. & Lee, T. D. (1999). The role of augmented information prior to learning a bimanual visual-motor coordination task: Do instructions of the movement pattern facilitate learning relative to discovery learning? British Journal of Psychology. 90, 389-403. Immink, M . A . , & Wright, D. L . (1998). Contextual Interference: A Response Planning Account. The Quarterly Journal of Experimental Psychology. 51A(4). 735-754. Immink, M . A . , & Wright, D. L . (2001). Motor Programming During Practice Conditions High and Low in Contextual Interference. Journal of Experimental Psychology: Human Perception and Performance. 27, 423-437. Inui, N . (1996). Transfer to changed serial position in a task of tracking a sequence of light stimuli. Perceptual and Motor Skills. 82. 799-802.  99 Jams, T., & Goverover, Y . (1999). Effects of contextual interference and age on acquisition, retention, and transfer of motor skill. Perceptual and Motor Skills, 88. 437-447. Jams, T., Wughalter, E. H . , & Gianutsos, J. G. (1997). Effects of contextual interference and conditions of movement task on acquisition, retention, and transfer of motor skills by women. Perceptual and Motor Skills. 84. 179-193. Jenkins, J. J. (1979). Four point to remember: A tetrahedral model of memory experiments. In L . S. Cermak & F. I. M . Craik (Eds). Levels of processing in human memory (pp. 429-446). Hillsdale, NJ: Lawrence Erlbaum. Kelso, J. A . S. (1984). Phase transitions and critical behavior in human bimanual coordination. American Journal of Physiology: Regulatory. Integrative and Comparative Physiology. 15. R1000-R1004. Kelso, J. A . S. (1995). Dynamic Patterns: The self organization of brain and behavior. Cambridge, M A S S : MIT Press. Landin, D., & Hebert, E. P. (1997). A Comparison of Three Practice Schedules Along the Contextual Interference Continuum. Research Quarterly for Exercise and Sport. 68f4), 357-361. Lee, T. D. & Magill, R. A . (1983a). Activity during the post-KR interval: Effects upon performance or learning? Research Quarterly for Exercise and Sport. 54. 340-345. Lee, T. D., & Magill, R. A . (1983b). The locus of contextual interference in motor-skill acquisition. Journal of Experimental Psychology, 9. 730-746. Lee, T. D., & Magill, R. A. (1985). Can forgetting facilitate skill acquisition? In D. Goodman, R. B. Wilberg, & I. M . Franks (Eds.), Differing Perspectives in motor learning, memory, and control (pp. 3-22). Amsterdam: North Holland. Lee, T.D., Magill, R.A., & Weeks, D.J. (1985). Influence of Practice Schedule on Testing Schema Theory Predictions in Adults. Journal of Motor Behavior, 17(3), 283-299. Lee, T. D., Swinnen, S. P., & Verschueren, S. (1995). Relative Phase Alterations During Bimanual Skill Acquisition. Journal of Motor Behavior, 27(3), 263-274. Lee, T. D., Wishart, L . R., Cunningham, S., & Carnahan, H . (1997). Modeled timing information during random practice eliminates the contextual interference effect. Research Quarterly for Exercise and Sport, 68, 100-105. Lee, T. D., Wulf, G., & Schmidt, R. A. (1992). Contextual interference in motor learning: Dissociated effects due to the nature of task variations. The Quarterly Journal of Experimental Psychology. 44A, 627-644.  100 L i , Y . , & Wright, D. L . (2000). A n assessment of the attention demands during randomand blocked-practice schedules. The Quarterly Journal of Experimental Psychology. 53A(2). 591-606. Magill, R. A. (2001). Motor Learning: Concepts and Applications (6 Edition). New York, N Y : McGraw Hill. th  Magill, R. A . , & Hall, K. G. (1990). A review of the contextual interference effect in motor skill acquisition. Human Movement Science. 9. 241-289. McGarry, T., Hodges, N . J., Bredin, S. S. D., Franks, I. M . , & Chua, R. (2000). Stability of the scanning procedure for mapping dynamical landscapes. Presented at N A S P S P A Conference, San Diego. Peterson, L . R. (1959). Short-term retention of individual verbal items. Journal of Experimental Psychology. 58. 193-198. Peterson, L . R., & Peterson, M . J. (1960). The effect of spacing repetitions on short-term retention. American Psychologist. 15. 450. Peterson, L . R., Wampler, R., Kirkpatrick, M . & Saltzman, D. (1963). Effect of spacing presentation on retention of a paired associate over short intervals. Journal of Experimental Psychology. 66. 206-209. Pigott, R. E., & Shapiro D. C. (1984). Motor schema: The structure of the variability session. Research Quarterly for Exercise and Sport. 55. 41-45. Pollatou, E., Kioumourtzoglou, E., Agelousis, N . , & Mavromatis, G. (1997). Contextual interference effects in learning novel motor skills. Perceptual and Motor Skills. 84, 487-496. Pollock, B. J., & Lee, T. D. (1997). Dissociated contextual interference effects in children and adults. Perceptual and Motor Skills. 84. 851-858. Rose, D. J. (1997). A Multilevel Approach to the Study of Motor Control and Learning. Needham Heights, M A : Allyn & Bacon. Schmidt, R. A. (1975). A schema theory of discrete motor skill learning. Psychological Review. 82. 225-260. Schmidt, R. A . (1985). The 1984 C H . McCloy Research Lecture: The Search for Invariance in Skilled Movement Behavior. Research Quarterly for Exercise and Sport. 56(2). 188-200. Schmidt, R. A . (1988). Motor control and learning: A behavioral emphasis (2nd edition). Champaign, JX: Human Kinetics.  101 Schmidt, R. A., & Lee, T. D. (1999). Motor control and learning: A behavioral emphasis (3rd edition). Champaign, DL: Human Kinetics. Schmidt, R. A . , & Wrisberg, C. A. (2000). Motor learning and performance (2nd edition). Champaign, EL: Human Kinetics. Scholz, J. P., & Kelso, J. A . S. (1989). A Quantitative Approach to Understanding the Formation and Change of Coordinated Movement Patterns. Journal of Motor Behavior. 21(2). 122-144. Schoner, G., Zanone, P. G., & Kelso, J. A . S. (1992). Learning as Change of Coordination Dynamics: Theory and Experiment. Journal of Motor Behavior, 24(1), 29-48. Sekiya, H . , Magill, R. A . , & Anderson, D. I. (1996). The contextual interference effect in parameter modifications of the same motor program. Research Quarterly for Exercise and Sport. 67. 59-68. Sekiya, H . , Magill, R. A . , Sidaway, B., & Anderson, D. I. (1994). The contextual interference effect for skill variations from the same and different generalized motor programs. Research Quarterly for Exercise and Sport, 65, 330-338. Shapiro, D. C , Zernicke, R. F., Gregor, R. J., & Diestel, J. D. (1981). Evidence for Generalized Motor Programs Using Gait Pattern Analysis. Journal of Motor Behavior. 13(1), 33-47. Shea, C. H . , & Kohl, R. M . (1991). Composition of practice: Influence on the retention of motor skills. Research Quarterly for Exercise and Sport, 62, 187-195. Shea, C. H . , Lai, Q., Wright, D. L., Immink, M . , & Black, C. (2001). Consistent and Variable Practice Conditions: Effects on Relative and Absolute Timing. Journal of Motor Behavior. 33(2). 139-152. Shea, J. B., & Morgan, R. L . (1979). Contextual interference effects on the acquisition, retention, and transfer of a motor skill. Journal of Experimental Psychology. 5. 179-187. Shea, J. B., & Titzer, R. C. (1993). The influence of reminder trials on contextual interference effects. Journal of Motor Behavior. 25, 264-274. Shea, J. B., & Wright, D. L . (1991). When forgetting benefits motor retention. Research Quarterly for Exercise and Sport. 62. 293-301. 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. 345-366). Amsterdam: North Holland.  102 Shewokis, P. A. (1997). Is the contextual interference effect generalizable to computer games? Perceptual and Motor Skills. 84. 3-15. Shewokis, P. A . , Del Rey, P., & Simpson, K. J. (1998). A Test of Retroactive Inhibition as an Explanation of Contextual Interference. Research Quarterly for Exercise and Sport. 69(1). 70-74. Smith, P. J. K . (1997). Attention and the contextual interference effect for a continuous task. Perceptual and Motor Skills. 84. 83-92. Smith, P. J. K., & Rudisill, M . E. (1993). The influence of proficiency level, transfer distality, and gender on the contextual interference effect. Research Quarterly for Exercise and Sport. 64. 151-157. Thaut, M . H . , Mcintosh, G. C , Rice, R. R , Miller, R. A . , Rathbun, J., & Brault, J. M . (1996). Rhythmic Auditory Stimulation in Gait Training for Parkinson's Disease Patients. Movement Disorders. 11(2). 193-200. Tsutsui, S., Lee, T. D., & Hodges, N . J. (1998). Contextual interference in learning new patterns of bimanual coordination. Journal of Motor Behavior. 30. 151-157. Tukey, J. W. (1953). The problem of multiple comparisons. Unpublished Manuscript. Princeton University. Tuller, B. & Kelso, J. A . S. (1989). Environmentally-specified patterns of movement coordination in normal and split brain subjects. Experimental Brain Research. 75, 306-316. Turnbull, S. D. & Dickinson, J. (1986). Maximizing Variability of Practice: A Test of Schema Theory and Contextual Interference Theory. Journal of Human Movement Studies. 12, 201-213. Van Merrienboer, J. J. G., de Crook, M . B. M . , & Jelsma, O. (1997). The transfer paradox: Effects of contextual interference on retention and transfer performance of a complex cognitive skill. Perceptual and Motor Skills, 84. 784-786. Wallace, S. A. (1996). Dynamic Pattern Perspective of Rhythmic Movement: An Introduction. In Zelaznik, H . N . (Eds), Advances in Motor Learning and Control (pp. 155-194). Champaign: Human Kinetics. Wegman, E. (1999). Contextual interference effects on the acquisition and retention of fundamental motor skills. Perceptual and Motor Skills. 88. 182-187. Whitall, J. (1996). On the interaction of concurrent verbal and manual tasks: Which initial task conditions produce interference. Research Quarterly for Exercise and Sport. 67. 349-354.  103 Wood, C. A . , & Ging, C. A . (1991). The Role of interference and task similarity on the acquisition, retention, and transfer of simple motor skills. Research Quarterly for Exercise and Sport. 62. 18-26. Wright, D. L . , & Shea, C. H . (2001). Manipulating Generalized Motor Program Difficulty During Blocked and Random Practice Does Not Affect Parameter Learning. Research Quarterly for Exercise and Sport. 72(1). 32-38. Wright, D.L., L i , Y . , & Coady, W. (1997). Cognitive Processes Related to Contextual Interference and Observational Learning: A Replication of Blandin, Proteau and Alain (1994). Research Quarterly for Exercise and Sport. 68(1). 106-109. Wright, D. L . , L i , Y . , & Whitacre, C. (1992). The contribution of elaborative processing to the contextual interference effect. Research Quarterly for Exercise and Sport. 63, 30-37. Wrisberg, C. A . , & Liu, Z. (1991). The effect of contextual variety on the practice, retention, and transfer of an applied motor skill. Research Quarterly for Exercise and Sport. 62. 406-412. Wulf, G. & Lee, T. D. (1993). Contextual interference in movements of the same class: Differential effects on program and parameter learning. Journal of Motor Behavior. 25. 254-263. Wulf, G. & Schmidt, R. A. (1988). Variability of Practice: Facilitation in Retention and Transfer Through Schema Formation or Context Effects? Journal of Motor Behavior. 20(2), 133-149. Yamanishi, J., Kawato, M . , & Suzuki, R. (1980). Two Coupled Oscillators as a Model for the Coordinated Finger Tapping by Both Hands. Biological Cybernetics. 37, 219225. Zanone, P. G., & Kelso, J. A . S. (1992). Evolution of Behavioural Attractors With Learning: Nonequilibrium Phase Transitions. Journal of Experimental Psychology. 18(2). 403-421. Zanone, P. G., & Kelso, J. A. S. (1994). The Coordination Dynamics of Learning: Theoretical Structure and Experimental Agenda. In Swinnen, S. P., Heuer, H . , Massion, J., & Casaer, P. (Eds.), Interlimb Coordination: Neural. Dynamical, and Cognitive Constraints (pp. 461-490). New York: Academic Press. Zanone, P. G., & Kelso, J. A. S. (1997). Coordination Dynamics of Learning and Transfer: Collective and Component Levels. Journal of Experimental Psychology: Human Perception and Performance. 23. 1454-1480.  Appendix A  R E V I E W OF L I T E R A T U R E  105 Origins Contextual interference (or intratask interference effect as it was originally called) has its origins in the verbal learning work of William Battig. Battig (1956) studied participants' performance on "paired associate lists" which included learning a finger positioning task with associated verbal responses. Participants performed various positioning tasks with one to four fingers and were provided with a stimulus to dictate which task they were to perform. Participants were also provided with either a relevant verbal response such as a verbal description of finger positions, or an irrelevant verbal response such as a nonsense string of letters corresponding to the number of finger positioned (i.e. " Z O N T " for a four finger positioning task). Battig studied the situation where a relevant verbal response was provided, which he called "relevant S-R," and the situation where an irrelevant verbal response was provided, which he called "relevant S." Battig discovered, somewhat counter intuitively, that with a complex finger-positioning task such as four finger positioning, the relevant S scenario actually facilitated performance or learning of the finger positions. Battig (1966, chap. 5) related this discovery to the role of intratask and intertask processing. Intratask, or within task processing would involve comparisons within the paired associate task (i.e. between the finger positioning and verbal response). Intertask, or between task processing would involve comparisons between the different paired associate tasks. It was assumed that more intratask interference would be present in the relevant S condition than the relevant S-R condition due to decreased stimulus-response compatibility of an irrelevant verbal cue. Thus the results of Battig's 1956 study suggested, "intertask facilitation is produced  106 by intratask interference" (Battig, 1966, p. 227). This statement formed the basis for the description of the contextual interference effect. This discovery opposed the prevailing view at the time that interference always had a negative effect on learning. Battig (1979, chap. 2) eventually changed the name of his discovery from intratask interference to contextual interference (CI) as this represented an expansion to a broader and more general conceptualization of the interference present in practice. This interference included not only the interference of the task and materials but all factors extraneous to the task including all the changes across trials in both the experimental and processing contexts. The entire practice context, including the task, practice schedule and processing engaged by the learner are potential sources of interference that could enhance learning (Magill & Hall, 1990, p. 244). History As Battig's research was predominately in the verbal learning area, he could not extend this CI phenomenon to other areas of learning. However, Shea & Morgan (1979) performed an experiment to test the CI effect in the motor learning field. In this study, participants performed a rapid movement task that involved reacting to a stimulus by grasping a tennis ball and then knocking down a series of three barriers in a prescribed order. Three different knockdown orders or "tasks" were used for acquisition and retention. Each task was practiced 18 times for a total of 54 acquisition trials. Shea & Morgan altered the amount of contextual interference by presenting the tasks in the acquisition phase in either a blocked or random order. Shea & Morgan's experiment also had both a 10-minute and 10-day retention test, where the two acquisition groups (blocked or random) were tested on all three tasks in either a blocked or random fashion.  107 Thus there were four different retention groups; blocked acquisition - blocked retention, blocked acquisition - random retention, random acquisition - random retention and random acquisition - blocked retention. There was also two transfer tests to new tasks or knockdown orders, one of similar complexity (3 barriers) and one of greater complexity (5 barriers). Similar to Battig's work, Shea & Morgan (1979) found different performance in acquisition, retention and transfer for the groups with different interference levels. During acquisition, the blocked group performed better than the random group (although approximately equal at the end of the 54 practice trials). For the 10-minute and 10-day retention tests, the random acquisition - blocked retention group performed best, and the blocked acquisition - random retention group performed worst for both tests. The 10minute retention test also supported the CI theory as the random acquisition - random retention group outperformed the blocked acquisition - blocked retention groups although the blocked - blocked group did outperform the random - random group on the 10-day retention test. Results of the transfer test were again consistent with CI theory as the random practice group outperformed the blocked group on both transfer tasks, with a larger advantage for the more complex movement. Thus as predicted by Battig's (1966, chap. 5) work, higher interference practice conditions (i.e. random acquisition) facilitated learning of subsequent tasks, shown by increased retention and transfer. Shea & Morgan's (1979) results showed total time (including both reaction and movement time) as the dependent variable, which may suggest the group differences in acquisition are due to increased reaction time in the random group rather than changes in actual movement time. However, Shea & Morgan analyzed movement times and reaction  108 times individually and the results for acquisition, retention and transfer paralleled those of total time. In addition, Lee & Magill (1983b) performed an experiment to determine if the effect seen in Shea & Morgan's study resulted from practice schedule manipulation or the reaction time paradigm (as participants in the random group are presented with a choice reaction time condition while participants in the blocked group are presented with a simple reaction time paradigm) or an interaction of the two. Lee & Magill duplicated Shea & Morgan's experiment but controlled for reaction time by using blocked and random conditions that were either cued, where a waning light cued the upcoming signal to respond, or non-cued. Lee & Magill's results led them to conclude that elevated reaction times in acquisition were likely due to the reaction paradigm, whereas elevated movement times in acquisition were affected by both reaction paradigm and practice schedule manipulation. However, the retention data clearly supported Shea & Morgan's contention that the benefits of high interference practice (random schedule) versus low interference practice (blocked schedule) are due to the manipulation of practice schedules and not the effects of reaction paradigm or the interaction of practice schedule with reaction paradigm. Shea & Morgan's (1979) experiment was a landmark study in the development of our understanding of contextual interference. It showed the CI effect could be generalized from the verbal learning environment to the motor skill learning environment, showing the stability of CI. It also showed the benefits of CI can only be seen in learning tests which occur after the practice has ended, implying that practice performance can be a misleading basis for determining learning effects. CI effects were also seen for retention and transfer to a similar and more complex task, therefore providing benefits to  109 performing the same task at a later time as well as a new task or novel variation of the same task. Lastly, interference does not necessarily have a negative effect on learning, as high interference conditions produced better retention and transfer of a skill than low interference conditions. Thus interference must somehow improve not only retention of the practiced skill but also transfer to a similar skill. Determining a formal definition for contextual interference may prove difficult. CI has been defined as "The effect on learning of the degree of functional interference found in a practice situation when several tasks must be learned and are practiced together" (Magill & Hall, 1990, p. 244). However, this definition may need further explanation. Interference levels are likely related to attentional demands, as factors requiring more attention are presumed to produce more interference due to the concept of a limited capacity of attention (Schmidt & Lee, 1999, p. 62-79). During practice of a task, many contextual factors can potentially contribute to this interference, including task complexity (i.e. within task incompatibility), practice schedule (i.e. less predictability), processing engaged by the learner, variability of learner's response, environmental conditions (i.e. crowd noise, lighting) and similarity or dissimilarity of tasks. Magill & Hall's above definition of CI also involves the learning and practicing of "several tasks" which may also be difficult to strictly define. As will be discussed later, different researchers have varying definitions of what constitutes different tasks to produce the CI effect. Schmidt (1975) relates different tasks to motor programs with different invariant characteristics, but this is certainly a current topic of debate in motor learning. However, a practical statement of contextual interference in the motor skill realm can be restated by the expected results of this effect. That is, a phenomenon where multiple tasks presented  110 in high contextual interference level condition (i.e. random presentation of tasks) will show decreased performance during acquisition yet increased performance in retention and transfer relative to low contextual interference levels (i.e. blocked presentation of tasks). Theoretical Explanations of CI Since Shea & Morgan's (1979) landmark motor skill study, many experiments have shown CI to be a stable, robust phenomenon. However, an explanation for this effect is still elusive. It appears that CI must influence the cognitive processing operations of motor learning. Wright, L i and Coady (1997) had participants observe models perform three key press tasks in either a random or blocked practice schedule. A CI effect was shown not only for the models but also for the observers, as those participants observing the random practice models exhibited better performance on a retention test than those observing the blocked practice models. Thus Wright et al. (1997) concluded that at least part of the CI phenomenon must be involved in processing movement planning and organization rather than movement execution. Gabriele, Hall and Lee (1989) found random imagery practice caused as much interference as random physical practice, implying the CI effect has at least some basis in the cognitive processing level. They also found typical CI effects with mental practice, as the retention data revealed a benefit of random imagery practice versus blocked imagery practice. In a study by Hall, Bernoties and Schmidt (1995), it was concluded that imagery practice could create interference effects similar to those of physical practice, again stressing the cognitive basis of contextual interference. These studies suggest CI must have some interaction with the cognitive processes of motor learning, however what is actually happening at the  Ill  cognitive level is unclear. There are presently, three main explanations for CI and each have both support and criticisms. These include the elaboration/distinctiveness theory, the action-plan reconstruction or "forgetting" theory and the retroactive inhibition theory. Elaboration/Distinctiveness Theory In Battig's (1966, chap. 5) work in verbal learning, he spoke of the importance of elaboration and distinctiveness for effective memory. Shea & Morgan (1979) also proposed greater elaboration of tasks as a tentative explanation for the CI effect. Although memory representation has been linked to both elaboration and distinctiveness (Bransford, Franks, Morris & Stein, 1979, chap. 15) (Jenkins, 1979, chap. 20), how these concepts specifically relate to learning is unclear. Shea & Zimny (1983, p. 345-363) proposed the elaboration theory, suggesting that distinctive processing enhances memorability by emphasizing the contrastive value of the information, which, in turn, creates a more unique memory representation. Thus the more information a performer has to compare and contrast with the current task, the more distinct and elaborate the memory representation will be, leading to better subsequent performance on that task. Contrastive information is thought to be related to intratask and intertask processing, although the definitions are slightly different than those used in Battig's work in the verbal learning area. Intratask processing consists of individual task analysis that excludes reference or comparison to other tasks being acquired. In contrast, intertask processing also incorporates between task analyses and facilitates identifying similarities and differences between the tasks to be learned. It is proposed that blocked or low CI conditions limit an individual to intratask analysis since only one task is in working memory. However random or high CI  112 conditions, the learner can engage in both intratask and intertask processing since multiple tasks reside in working memory. This intertask processing results in comparisons between tasks and an increased level of elaborative processing. This increased elaborative processing leads to an increase in distinctiveness and thus a stronger representation in memory. This is the basis of the elaboration/distinctiveness view. Note that this explanation predicts that tasks must have some similarity to produce the CI effect, as this would be required for meaningful intertask comparison. Dissimilar tasks would not produce as great of learning benefits as similar tasks, because similar tasks would allow for more intertask comparisons and thus greater distinctiveness. A practical example of the elaboration/distinctiveness theory could involve an individual trying to produce two different movement patterns. The individual may remember that one is in the shape of an "L" and the other in the shape of a "T" by comparing the similarity (both can be represented by letters) and the differences (two different letters) of the movements. These comparisons may help the individual to remember the two movements by making them more distinct due to the more elaborate or detailed processing involved. This would result in better retention and transfer than an individual who only had one of the movements in working memory and thus could only use intratask processing. There is empirical evidence to support this explanation of CI. Wright, L i and Whitacre (1992) hypothesized that if this explanation was correct, augmenting the interpresentation interval in a blocked practice schedule with intertask processing should increase performance in the retention phase to be similar to random practice. They found, as predicted, that providing additional intertask processing to blocked practice did indeed  113 increase performance to the same level as random practice. Further, additional intratask processing to the blocked practice group did not produce any improvement, nor did additional intratask or intertask processing presented to the random group. Thus they concluded that intertask processing is at least partially responsible for the benefits of random practice in retention. Shea & Kohl (1991) had participants learn only one task with either an unfilled inter-trial period or a filled inter-trial period with performance of either a similar or dissimilar task. They hypothesized that if intertask processing was responsible for CI effects, the group with a similar distractor in the inter-trial period would outperform the other two groups (unfilled or dissimilar distractor). Their results supported their hypothesis with increased retention in the similar distractor group and no improvement or difference in retention between the dissimilar distractor and unfilled group. Again, they concluded that intertask processing is related to the expected CI effects. Other support is provided by verbal reports of participants following blocked and random practice. Verbal reports of the participants in the random schedule group suggested they used several different strategies to perform the task and made comparisons between tasks. This was not evident from the verbal reports of the blocked practice schedule group. However, there are some criticisms of the elaboration/distinctiveness theory of contextual interference. One major criticism is that although this explanation can explain the retention and transfer effects seen in high contextual interference conditions, it fails to explain why acquisition performance is poorer in these conditions. It is also argued that the concepts of elaboration and distinctiveness are difficult to define and impossible to measure as there is no independent measure of these qualities (Lee & Magill, 1985, p. 3-  114 22). Some empirical evidence also contradicts the expected results of this theory, as shall be apparent in the discussion of the action-plan reconstruction theory. Action-Plan Reconstruction Theory As an alternative to the elaboration theory, Lee & Magill (1985, p. 3-22) have proposed the action-plan reconstruction or "forgetting" theory. This proposed theory is in response to their perceived shortfalls of Shea & Zimny's (1983) elaboration theory. The basis of this theory is that rather than increasing elaboration or distinctiveness, high CI conditions increase effortful processing because previously encoded information pertaining to the skill has been either partially or completely forgotten. That is, continuous presentation of new tasks in a random practice schedule constantly purges working memory, which results in continual interchange of task information into working memory from trial to trial. This interchange of information requires the performer to constantly reconstruct the movement plan of the task from long term memory and the information currently available, which results in stronger task processing and memory representation. In contrast, blocked presentation allows information to remain in working memory throughout the trials, so less reconstruction is needed. Thus, "forgetting" decreases acquisition performance but actually benefits retention performance via reconstruction of the movement plan. Transfer performance is also improved in random practice because of the similarity between processing demands of a repeated task and a new, similar task. This explanation of contextual interference does explain the typical interference effects on acquisition, retention and transfer results in both a blocked and random practice schedule. The action-plan reconstruction theory also relates interference to  115 attentional demands, as interference with two tasks will only occur if both occupy the attentional demands of the individual (Whitall, 1996). If one or both of the tasks are automatic, no interference will occur, leading to less forgetting and thus decreased benefit in retention and transfer. Note that this explanation, contrary to the elaboration theory, does not predict performing similar tasks will be beneficial to learning. The best result for retention and transfer is a practice condition that maximizes interference and thus forgetting and processing reconstruction, which is likely a dissimilar task. A practical example outside of the motor skill area may help clarify the action-plan reconstruction theory. Imagine an individual whose task is to perform a multiplication in a variety of ways; adding the number three nine times, adding the number nine three times and multiplying three by three by three. If these multiplications were presented in a blocked order, likely the individual would solve the problem the first time but simply repeat the known answer on subsequent trials. However in a random order, the individual would need to constantly solve the problem over and over again, assuming enough interference between presentations of the problem. This solving of the multiplication time and time again would likely increase the memory representation of the cognitive processes associated with solving the problems and likely help the individual solve other multiplications quicker than practicing in the blocked practice schedule. Lee & Magill propose the action-plan reconstruction hypothesis based on some of the research done on spacing effects and forgetting in cognitive psychology. Originally, Brown (1958) and Peterson (1959) presented a list of verbal items, followed by a filled retention interval to prevent rehearsal activity, followed by a retention test. They found that increasing the retention interval decreased the amount of recall. This loss of  116 information was attributed to both the amount of time between presentation and recall, and the interference caused by the rehearsal preventing activities. This experiment was later repeated, however the list of verbal items was re-presented before the recall test (Peterson & Peterson, 1960 and Peterson, Wampler, Kirkpatrick & Saltzman, 1963), with the interval between presentations filled with rehearsal preventing activity. Decreased recall occurred again as the interval from the second presentation to recall (retention interval) was increased. However recall performance improved, to a maximum value at 816 seconds of interval time, as the filled interval between presentation of the items (interpresentation interval) was increased. Because it can be assumed from the first experiment that forgetting had occurred during this inter-presentation interval, this second experiment suggested, somewhat paradoxically, that forgetting actually facilitated retention. There is empirical evidence to support the action-plan reconstruction theory. L i & Wright (2000) studied the amount and time course of attentional demands of blocked and random practice environments. They hypothesized reconstruction activity cannot take place until the learner receives the necessary information pertaining to what plan needs to be reconstructed. Thus the pre-response interval should incur large attentional demands for the random compared to the blocked practice. Although it is not clear when attentional demands would be greatest for the elaboration theory, L i & Wright suggested the inter-trial interval as a possible locus. Results from the study indicated greater attentional demands for random versus blocked practice and that these demands were greatest during the pre-response rather than the inter-trial interval, supporting the actionplan reconstruction theory. However as the time course for attentional demands is unclear  117 for the elaboration theory, this study does not oppose the elaboration theory. Immink & Wright (1998) allowed participants to self-select the amount of time used to plan an upcoming movement, suggesting increased study time would represent greater reconstructive activity. Their findings indicated random practice participants spent more time planning their movement during acquisition than blocked practice participants, thus supporting the action-plan reconstruction theory. However, as study time could also involve intertask comparisons, these results also do not oppose the elaboration theory. Lee, Wishart, Cunningham andCarnahan (1997) provided a modeled template of temporal and spatial information about a movement just prior to performance. This was hypothesized to introduce relevant, action planning information into working memory, thus reducing the effectiveness of random practice. As predicted, the random group with the modeled template performed similarly to the blocked group with decreased retention yet increased acquisition, presumably because much of the problem solving processing normally required for the action were not undertaken. Also, Wood & Ging (1991) used participants who practiced either three high or low similarity tasks in either a random or blocked order. They found typical CI effects for the low similarity tasks, but no difference in practice schedule for the high similarity tasks. This again provided support for the action-plan reconstruction theory and opposes the elaboration viewpoint. There are, however, criticisms of this theory and some of the empirical evidence does not support the predictions from action-plan reconstruction. Shea & Wright (1991) tested participants with an unfilled inter-trial period, a similar distractor task or a dissimilar distractor task and found retention test performance was best in the similar distractor group. Shea & Wright concluded that forgetting and subsequent reconstruction  118 of the cognitive processes associated with a task is not in itself sufficient for improved retention. Critics of this theory also question what exactly is being forgotten to aid performance. Lee & Magill (1985, p. 17-19) argue that if at least some of the information about a previous task is forgotten, then some or all of the cognitive processing analyses must be undertaken again. They also view actions as being guided by so-called "actionplans" of which a part or whole is forgotten during interference and must be reconstructed upon subsequent repetition of the task. However the notion of an "actionplan" is open to debate and a controversial issue, as its precise nature was not specified by Lee & Magill, nor has an acceptable definition been proposed (Lee, Wulf & Schmidt, 1992). Retroactive Inhibition A third explanation of contextual interference involves the concept of retroactive inhibition. The retroactive inhibition theory attributes the observed differences of practice schedules to the disadvantage of blocked practice rather than the advantages of random practice. Retroactive inhibition is the forgetting of a task caused by practice imposed between the original learning and the retention test (Schmidt & Lee, 1999 p. 397-398). That is, for blocked acquisition practice, performance decreases on a retention test because of the intervening tasks between the practicing of a task and the retention test. For example if A , B & C represent three tasks to be learned and are practiced in a blocked order (AAA, B B B , CCC), during a retention test of task A, tasks B and C will retroactively inhibit the recall of task A. This would explain high acquisition yet poor retention performance in a blocked practice schedule.  119 Again, there is empirical evidence to support this explanation. Shewokis et al. (1998) changed the amount of retroactive inhibition present by altering the number of intervening trials between acquisition and retention. As retroactive inhibition was decreased, the differences in retention between blocked and random practice disappeared. These results caused Shewokis et al. to question Battig's hypothesis regarding better learning in random compared to blocked practice due to enhanced processing during random practice. In a similar study, Del Rey et al. (1994) varied the amount of retroactive inhibition by testing only one task in retention and thus tried to isolate the retroactive inhibition effects. As retroactive inhibition increased (with more trials in between the selected task's final presentation and the recall test), recall performance decreased. Shea & Titzer (1993) provided a reminder trial for each task at the end of acquisition in blocked practice to try and remove the expected retroactive inhibition effects. As predicted, the blocked group with a reminder trial performed equal to the random group in both an immediate and delayed retention test. This evidence supports the notion of retroactive inhibition negatively affecting the retention of a task acquired in a blocked practice schedule. Theory Summary All three proposed explanations for contextual interference emphasize the role of cognitive processes yet the exact mechanism is unclear. Much work has been done to try and separate these theories and prove which is the "correct" explanation for CI, although no clear support has emerged for one theory. Many view these theories as mutually exclusive and competing against each other due to their differences in proposed mechanisms; however there is certainly some congruity between all three. Both the  120 elaboration and action-plan reconstruction theory outline that performers in a random practice setting engage in more active or effortful processing and emphasize that this processing increases the strength of the representation of the task in memory. This, in turn, increases the retention of the task as well as transfer to a similar action. Furthermore, the different mechanisms proposed to explain CI may also manifest themselves at different times in the learning process. Early in skill acquisition when performers are trying to understand the idea of the desired movement, action-plan reconstruction may provide benefit. Later in the acquisition phase when the performer is more interested in fine-tuning their response and preparing for new situations, elaboration and distinctiveness via intertask comparisons may be responsible for benefits to learning. Immink & Wright (2001) suggest a more reconciliatory approach towards the action-plan reconstruction and elaboration theory and suggest both theories may assume a role in the process of learning. The action-plan reconstruction concept may promote the refinement of response programming processes while the elaboration concept may benefit the memory quality of the response. Thus, future research into theoretical explanations for the CI effect may involve studying how the proposed theories complement and interact with each other rather than determining which theory warrants more merit. Factors and Considerations Since Shea & Morgan's (1979) original work in motor learning, many studies have examined contextual interference. Evidence from these studies has shown CI to be a robust phenomenon in that it has been demonstrated with numerous tasks, populations and environments. However, some studies have not elicited a CI effect, which questions the generalizability of CI. Understanding where and when CI occurs would be very useful  121 in determining its application to the teaching environment as well as providing further insight into the cognitive processes involved in this phenomenon. A number of areas have been examined in an attempt to determine the generalizability, or lack thereof, of CI. These areas have included characteristics of the task, including laboratory task effects versus applied task effects; characteristics of the participants, including age and experience levels; and task variations. Each topic will be discussed in turn. Task Characteristics - Laboratory versus Applied (Field) The presence of CI in laboratory tasks is well documented and not limited to simple, discrete, closed tasks. CI effects have also been observed for the acquisition of new bimanual coordination patterns (Tsutsui et al., 1998), novel motor tasks (Pollatou, Kioumourtzoglou, Agelousis & Mavromatis, 1997), continuous lab tasks (Smith, 1997), complex cognitive tasks (Van Merrienboer, de Crook & Jelsma, 1997), open tasks (Jarus, Wughalter & Gianutsos, 1997), serial positioning tasks (Inui, 1996) and such varying tasks as force production, computerized maze, computer keyboard, linear positioning, pursuit rotor and tracking and probe reaction time (Brady, 1998). Results of these experiments have been incredibly consistent in demonstrating that the CI effect does indeed have a broad base of support and CI can be considered a stable and dependent principle of motor learning in the lab setting. However, lab tasks typically involve comparatively simple tasks with minimal degrees of freedom. Perhaps prematurely, Schmidt (1988, p. 497), based on the stability of the CI effect in the lab, assumed the applications of CI for practical learning situations. However determining if the CI effect does indeed occur in non-laboratory or field research has proved to be a difficult task. Cassio, Meira and Tani (2001) examined random and  122 blocked practice for inexperienced participants throwing darts at varying distances with two different dart grips, and found no CI effects. French, Rink and Werner (1990) studied different practice schedules with relatively inexperienced high school PE students learning three different volleyball skills and found no learning effects based on practice conditions. Brady (1997) also found no practice schedule effects for university-aged novice golfers learning four different golf skills in either a blocked or random order. Thus, none of these studies found any of the predicted CI effects with differing practice schedules. Bortoli Robazza, Durigon and Carra (1992) used novice volleyball players in grade nine and three different skills in a random, blocked, serial or serial with high interference practice schedules. They found that the random and serial practice groups performed significantly better in one of the two transfer tests for only one of the skills, namely the serve. They concluded these results provided support for the CI effect in the field although at best this can be thought of as weak support as the acquisition and retention performance showed no difference, and expected results were only seen in one skill in one transfer test. Wegman (1999) studied grade four girls practicing ball rolling, racket striking and ball kicking in either a random, blocked or combined practice schedule. Acquisition performance for all three skills was significantly better in the blocked group but in the retention phase, only the racket striking skill was improved in the random group. Again, this experiment at best can provide only partial support for CI in the non-lab setting. Wrisberg & Liu (1991) altered practice schedules for universityaged novice badminton players learning two service variations. They found no difference in acquisition between random and blocked groups, yet higher retention of the short serve  123 only and higher transfer of the both short and long serves with the random group. Thus the transfer data and part of the retention data supported the expected CI results, yet acquisition performance did not. Shewokis (1997) considered computer games as the non-lab task and found university students showed no significant difference between blocked and random practice schedules except for the transfer tests where the random practice group outperformed the blocked practice group. Once again, only part of this experiment's results is consistent with contextual interference's expected effects. One study has shown relatively strong support for the presence of CI in the field. Hall, Domingues and Cavazos (1994) used skilled collegiate baseball players taking additional batting practice in either a blocked or random order as well as a control group who received no additional practice. Although acquisition performance did not differ between groups, the results from the transfer test were very significant, when compared to the pretest. The random practice group improved 56.7%, the blocked group 24.8% and the control group 6.2%. However, it is unclear if this study was an appropriate test of contextual interference. CI involves the effects of interference on learning of multiple tasks and it may be argued that batting practice with different pitches does not require different tasks, but instead involves the same task of a baseball swing with the timing of the initiation of the swing altered. Thus, results from CI studies in the field have been contradictory and equivocal. One reason for this ambiguity of the experimental data may be related to the inability to limit interference and decrease variability of performance in the field. Tasks performed in a laboratory setting are typically simpler tasks with fewer degrees of freedom and distractions. The increase in degrees of freedom in the field makes it very difficult to  124 produce an identical or invariant action. Thus, performers in the field may create their own interference and variability, regardless of the practice schedule. True blocked practice may not be feasible in a non-lab setting due to participants' inherent variability in attempting to produce a relatively complex skill. Compounding this interference in an applied setting is the experience level of the performer. Inexperienced or novice participants, first learning a skill in a non-laboratory setting, are likely to create their own variability regardless of whether they are in a blocked or random practice schedule. Increasing the interference by presenting tasks in a random order may not provide any additional benefits as CI may have a ceiling effect of interference (Wright et al., 1992). Therefore blocked practice may be as effective as random for the inexperienced performer in the applied setting. Once a performer acquires a level of proficiency with the task, actions will naturally become more consistent and less variable; thus additional interference may be necessary to see the expected CI benefits. Furthermore, this experience factor would not be expected in the lab as even an inexperienced participant could likely produce a relatively invariant performance due to the simplicity and reduced degrees of freedom of a laboratory task. Thus, rather than assuming the CI phenomenon does not occur outside the lab, it may be more useful to determine what is causing the interference in the field setting and what the optimal level of interference is for the particular task. These considerations may explain some of contradictions in the empirical evidence. Participant Characteristics - Age & Experience Level. Interacting with the task characteristics in any experiment are the characteristics of the participant or performer. As previously mentioned, age and experience level of the  125 performer may also affect CI as response variability and interference levels may be higher for younger, less experienced performers. Again, additional interference provided by a random task presentation may not provide any learning benefits and may actually hinder their performance. High interference practice conditions should have a higher difficulty level than low interference practice conditions, which may pose problems for novice performers who do not have a clear idea of the skill. Thus the positive effects of random practice may not be realized until some degree of expertise has been achieved. Alternatively, more experienced performers may require the additional difficulty associated with a random schedule. Therefore the most effective practice schedule for beginners may be introduction of the skill in blocked order until some level of expertise is achieved, followed by random presentation. Note that this prediction is consistent with Gentile's (1972) description of skill acquisition that includes a two-stage model. Stage one is getting the idea of the movement where an individual organizes a motor pattern to solve a problem while stage two is fixation and diversification where an individual has acquired a general concept of the motor pattern that appears effective, and focuses on refinement and consistency. This is also consistent with the guidance hypothesis, which states that guidance is most effective early in practice when the task is unfamiliar (Schmidt & Wrisberg, 2000, p. 212). Too much knowledge of results (KR), too late in the learning process has been shown to decrease retention, while less K R has actually facilitated retention (Lee & Magill, 1983a). Empirical evidence seems to support this proposal of the interaction of participant's experience and interference. Jarus & Goverover (1999) used beanbag throwing with a group of 5, 7 and 11 year olds in a blocked, random or combined practice  126 schedule and found no difference in practice schedule with the 5 and 11 year olds but better acquisition and retention with the 7 year olds in the blocked condition. They concluded, in support of the proposed interaction of age and practice schedule, that high contextual interference may not be needed for children to learn a skill, as the random condition did not provide additional benefit for any age group. Del Rey et al. (1982) and Del Rey (1989) studied the effect of different practice schedule on trained and untrained females performing an anticipation task. Classification of trained and untrained was based on a questionnaire outlining experience in similar tasks. Results showed that the experienced group performed better in retention tests when acquiring a skill in a random schedule while the inexperienced group performed better in retention tests when acquisition was in a blocked order. Pigott & Shapiro (1984) studied beanbag tossing of various weights in children six to eight years old in either a random, blocked (in increasing weight) or randomly blocked condition (blocked trials, randomly assigned) and found the randomly blocked condition best for transfer. However the randomly blocked condition would not provide much different interference than the blocked condition as trials were still completed for one weight before the next was presented. Thus this study does seem to support less interference is beneficial for children learning a task. Del Rey, Whitehurst and Wood (1983) used an anticipation task with 6 to 11 year olds in either a blocked or random condition and found the blocked group outperformed the random group in both acquisition and transfer. These results again provided support for at least initial blocked practice for children may be beneficial not only for acquisition, but transfer and presumably retention as well.  127 Other studies have provided less or no support for the notion of experience and age relating to CI. Hebert, Landin and Solmon (1996) found with high and low skilled university-aged tennis players that practice schedule effect was influenced by ability. Low skilled players showed more improvement on the post-test when acquiring the skill in a blocked order rather than random order. However, no difference of schedule was seen on the post-test for the high skilled players. The lack of significance in the skilled players may be at least partially due to the increased variability of using an applied skill. Smith & Rudisill (1993) used an anticipation lab task with groups split into high and low proficiency based on pre-test scores. When presented the tasks in random or blocked order, no difference was found for acquisition or transfer. However, pre-test scores may not be an valid measurement of participant skill proficiency or experience level and may have confounded the results. Pollock & Lee (1997) used a ballistic aiming task with seven year olds in a blocked and random presentation and found no difference in acquisition yet typical CI effects shown by increased retention and transfer in a random versus blocked practice. Tsutsui et al. (1998) also found a typical CI effect for a novel bimanual coordination task although this task was performed in a lab setting which, as previously mentioned, tends to have a lower level of interference and variability. Overall it does appear that the age and experience level of the performer do affect the effectiveness of a given practice schedule and is a point of consideration for application of interference to developing a skill. It is likely that this experience level effect also depends on the complexity of the task and is likely a greater consideration in the field rather than in the lab.  128 Summary of Task Type. Age & Experience Controlling of interference levels may be difficult in applied task settings or with performers who are inexperienced with the task. True blocked practice may not be feasible in these situations due to the inherent variability and increased degrees of freedom associated with a new or complex skill. With inexperienced performers or in an applied setting, additional interference provided by a random practice schedule may not benefit and may in fact hinder performance. The amount of interference for optimal learning may vary along the CI continuum with age and experience level of the performer, task characteristics and complexity and environmental conditions. Landin & Hebert (1997) examined the effect of three practice schedules along the contextual interference continuum on the acquisition and learning of a basketball shooting task and discovered the moderate CI group exhibited best results on the retention tests. This result may suggest the need to explore schedules along the CI continuum rather than focusing only on the extremes of high and low interference conditions. Maximal learning may also be facilitated by beginning acquisition of a skill with low interference conditions (blocked practice) until performers understand the basics of the movement. Once the performer gets the idea of the movement, higher interference conditions (random practice) may be required to continue to increase learning during the "fine-tuning" stage of skill acquisition. Finally, the combined effects of task type and participant characteristics may explain some of the equivocal empirical results. Task Variations. Another area examined by researchers is analysis of the different tasks to be learned within a practice environment. Magill & Hall (1990) proposed a two-part  129 hypothesis based on contextual interference and Schmidt's (1975) idea of a movement class. According to Schmidt, a movement class or generalized motor program (GMP), is defined by the invariant or unchangeable characteristics of actions such as relative timing or relative force. Any action that changes an invariant must also change the movement class. However, parameters such as overall movement duration or overall force are free to vary from one performance to another yet still be classified as the same movement class. Magill & Hall hypothesized that parameter modification alone would not provide enough interference to see the CI effect. However, invariant modification would provide enough interference to see the CI effect. Thus when multiple tasks to be learned are from different "movement classes," different levels of contextual interference created by practice schedule manipulations would lead to the expected different retention and transfer effects. However when multiple tasks to be learned are from the same "movement class," the contextual interference effect will either not be found, or a combined practice schedule of blocked practice followed by random practice will be most advantageous. A number of studies have investigated this proposal by modifying task parameters and invariants. Goodwin & Meeuwsen (1996) manipulated overall force in a golf putting experiment involving university-aged novice golfers, by varying the distance of the putt, a parameter modification. They found no difference between the blocked and random practice schedule and attributed this equal performance to manipulation within the same movement class. Wulf & Lee (1993) used university-aged participants performing three movement patterns that had the same relative timing (invariant feature) but different overall durations (parameters). No differences were found in retention and transfer,  130 leading them to the conclusion that no CI effects occur if movements of the same class are used. Sekiya, Magill, Sidaway and Anderson (1994) had participants perform three tasks with either altered relative timing (invariant modification) or total movement time (parameter modification) in either blocked or serial order. For the changed relative timing, they found typical CI effects, supporting the one prediction of the movement class hypothesis. However, they also found typical CI effects for the changed overall movement time, which does not support the second part of Magill & Hall's hypothesis. In a study similar to their 1994 study, Sekiya, Magill and Anderson (1996) changed the overall force of the movement and, similar to their earlier experiment, still found a CI effect, concluding that parameter modification may indeed be sufficient to create interference to increase retention and transfer. In addition, Green & Sherwood (2000) did find a CI effect for parameter modification of overall movement time, which is again in opposition to Magill & Hall's hypothesis. Therefore, the literature does not provide unequivocal support for Magill & Hall's proposal of movement class and contextual interference. Recently, Shea, Lai, Wright, Immink and Black (2001) studied the effects of blocked and random practice on both absolute and relative timing. Participants practiced three tasks with identical relative timings but different absolute timing, in either a random or blocked presentation order. Results indicated blocked practice decreased relative timing errors, which they related to movement class or GMP learning, while random practice decreased absolute timing errors, which they related to parameter learning. Shea et al. concluded these findings provided additional evidence for the dissociation of memories supporting the GMP performance and parameter performance, and cautioned  131 that outcome measures in CI studies may affect the results. Studies that use relative timing performance, or possibly other invariants, as an outcome measure may not see a definitive CI effect as blocked practice appears to be more beneficial for this aspect of learning. Wright & Shea (2001) also examined relative and absolute timing and suggested blocked practice may be beneficial early in training to focus the learner's attention on acquiring invariants and strengthening the motor program, however once parameter scaling becomes the emphasis of learning, switching to random practice may show increased learning performance. Thus, as discussed with performer experience and age, blocked and random practice may affect different areas of the "motor program" concept and thus may benefit performance at different times in the learning process. However, in the consideration of task variations, it is important to note that Schmidt's description of a class of movements or GMP with invariants and parameters is debatable and subject to many criticisms. Summary of Literature Review The phenomenon of contextual interference has been studied extensively since its original discovery by Battig (1966, chap. 5) in the verbal learning realm and Shea & Morgan's (1979) discovery in the motor domain. Contextual interference refers to the interference found in a practice situation when several tasks are learned and practiced together. Increasing interference, often by manipulation of schedule of task presentation, typically has the effect of decreasing acquisition performance but increasing retention and transfer performance. Theories have been presented for the explanation for CI and this effect is thought to involve cognitive processing however the exact mechanism is unclear. Although the CI effect has shown to be stable and predictable in many laboratory  tasks, limited applications have been seen in the field, likely due to inherent interference from the variability and experience of the performer and difficulty of the task. Determining the optimal amount of interference for maximal retention and transfer of a skill may involve considerations of all the factors affecting the CI effect. However, the applications of determining this optimal environment are almost limitless. The CI effect also implies that practitioners should use caution when judging performers' learning by practice performance, as retention and transfer can be increased by conditions that decrease acquisition performance. It is also important for practitioners to understand that interference can have a positive effect and benefit learning. It is interesting to note how prophetic Bernstein (1967, p. 362-365) was when he discussed optimal achievement of new motor skills by practice as a particular type of repetition without repetition.  Appendix B  T A S K DEFINITION  134 The motor learning phenomenon known as contextual interference has its origins in the early work in the field of verbal learning by William Battig. Results from Battig's 1956 study led him to the conclusion that "intertask facilitation is produced by intratask interference" (Battig, 1966, p. 227, see Appendix A for a full review). This statement from Battig is often cited in studies involving contextual interference. Battig (1966, chapter 5) considered intratask interference as any interference that occurred within a task. Intertask facilitation was the increased performance found between tasks due to this interference. That is, performance on subsequent tasks was increased when interference was present in the learning of the original task. This "intratask interference" was later expanded to include interference anywhere in the entire practice context and was thus renamed as contextual interference (Magill & Hall, 1990, p. 224). Paramount in any discussion involving contextual interference is a functional definition of a "task" to accurately determine what is involved in "intratask" interference and how it benefits "intertask" performance. In addition, studies involving contextual interference often also involve the notion of practice variability (Hall & Magill, 1995, Lee et al., 1992, Wulf & Schmidt, 1988, and Turnbull & Dickinson, 1986). Both contextual interference and practice variability influence skill learning during practice, but involve different aspects of the practice environment. Contextual interference involves the effect on learning provided by interference in a practice situation (usually via presentation schedule) when several tasks must be learned (Magill & Hall, 1990, p. 244). Alternately, practice variability involves the effects on learning provided by practicing a task in a variety of situations or practicing a number of variations of a given class of tasks (Schmidt & Wrisberg, 2000, p.  135 243-245 and Rose, 1997, p. 238-241). For both these motor learning phenomena, the notion of a task is an important consideration. Again, a functional definition of task is required. Although "task" is a commonly used term in every day life, there appears to be a lack of uniform conception or definition of a task (Fleishman & Quaintace, 1984). The definition that seems most appropriate for the investigation of motor learning and examination of practice conditions, is that provided by Magill (2001, p. 3) when describing a motor skill or action (which can be considered synonymous with a task): "a goal-directed activity that consists of body and/or limb movements". Magill lists several characteristics that are common to all motor skills; there must be a goal to achieve, skills must be performed voluntarily, skills require body and/or limb movement, and need to be learned. This task definition can be used to examine the tasks associated with the experiments outlined in this paper. In Experiment 1, participants in the single task control group had only one task to perform, or goal to achieve; movement of the limbs in a 90° relative phase coordination pattern to create a circular Lissajous figure on a computer screen. Participants in the two-task blocked group or two-task random group were asked to perform two tasks, or achieve two goals; movement of the limbs in a 90° relative phase coordination pattern to create a circular counterclockwise Lissajous figure on a computer screen and movement of the limbs in a 45° relative phase coordination pattern to create a elliptical Lissajous figure on a computer screen. Transfer to the speeded condition (90° relative phase at 1.25 Hz) would classify as performance of a variation of the same task (intratask transfer) as the goal of the movement is the same, as is the limb movement,  only it is performed at a faster rate. However transfer to the opposite coordination pattern (270° relative phase) would classify as performance of a new task (intertask transfer) as the goal is now to create a clockwise Lissajous figure and the limb movements are now different. In Experiment 2, each group performed a unique single task. Each group's task was to move the limbs in a 90° relative phase coordination pattern to create a circular counterclockwise Lissajous figure, while concurrently verbalizing a counting pattern aloud in beat with a metronome. Intratask interference was adjusted by manipulating the compatibility of the counting pattern and coordination pattern. A counting pattern that was thought to be compatible with the coordination pattern was expected to have a low level of intratask interference while a counting pattern that was thought to be incompatible with the coordination pattern was expected to have a low level of intratask interference. The relationship between the counting patterns and movement pattern is considered intratask interference as both aspects are involved in the single task or goal of the action. As with Battig's (1956) original work, it was expected that higher intratask interference would result in facilitation of intertask performance during transfer tests to new tasks. In summary, the notion of a task is an important, yet difficult concept to define. However, the most appropriate definition for the experiments associated with this paper involves: a goal to achieve, body and/or limb movement, and a voluntary, learned movement. A functional definition of a task allows for improved understanding of such motor learning concepts as practice variability, contextual interference, intratask and intertask transfer, and intratask interference.  137  Appendix C  E V A L U A T I O N OF S C A N N I N G METHODS IN B I M A N U A L COORDINATION  Note: This experiment was conducted in conjunction with other researchers (S.S.D. Bredin, S.J. Clark, R. Chua, I.M. Franks) who will be referenced in future citations.  138 Introduction In traditional motor learning studies, participants are monitored and examined from the initial acquisition phases of learning through to retention or transfer testing. Learning is typically inferred through indirect measures, such as response outcome (e.g., reaction time, error scores) and/or response production (e.g., kinematics, electromyography). Once performance measures are acquired, a general methodological practice is to group the data and plot the averaged performance as a function of practice; hence, producing the familiar performance curve. In this type of approach, little can be said as to the influence of individual differences (i.e., the personal history or experiences that the learner brings to the learning environment) on the acquisition of a new form of skilled behaviour (e.g., Zanone & Kelso, 1992). However, it is generally accepted that the acquisition of a new motor skill is profoundly influenced by the capabilities and knowledge that an individual is afforded as a result of a learner's history or past experiences. From a dynamics perspective, the learning process may be thought of as an interplay between intrinsic dynamics and environmental or task requirements (i.e., behavioural information). Accurate and consistent performance of relative phase patterns, not intrinsic to the human motor system, must be learned. In simplest terms, learning is the process of acquiring a skill through experience or practice against the backdrop of existing coordination tendencies. In many respects, the learning process affords the individual an opportunity to expand his or her limited behavioural repertoire, passing from one organized state to another (Kelso, 1995).  139 When a new spatiotemporal coordination pattern is acquired, the entire layout of a system's coordination dynamics is modified to incorporate the new phasing pattern into the existing landscape (Schoner, Zanone, & Kelso, 1992). As such, a to-be-learned pattern becomes part of a system's underlying dynamics through the establishment of a new stable state within the existing pattern repertoire. Learning then, develops under the influence of pre-existing ordered patterns and involves a relatively permanent change in behaviour towards the pattern required by the environment. Prior to learning, two coordination patterns emerge as "attractors". Attractors are patterns of high stability and can be imagined as the bottom of a well on a landscape (Wallace, 1996). Any motion of a ball on a landscape towards the attractor will tend to force the ball into the bottom of the well. The initial attractors are pre-existing modes of coordination, namely in-phase and anti-phase (Haken, Kelso, & Bunz, 1985). In-phase and anti-phase patterns are intrinsic patterns of coordination, since they can be performed more accurately and with lower variability than other patterns of coordinative movement. Also, they are thought to represent natural human coordinative patterns (Kelso, 1984; Fontaine et al, 1997; Zanone & Kelso, 1992; see Kelso, 1995 for a review). By way of reference, when homologous muscles for both limbs flex and extend simultaneously, the coordination pattern is described as "in-phase", or 0° relative phase. When homologous muscles for both limbs flex and extend alternatively, the coordination pattern is described at "anti-phase" or 180° relative phase. Theory posits that early in learning, acquisition of the coordination task should be relatively difficult (Wallace, 1996). It is predicted that a strong behavioural attraction towards these pre-existing modes of coordination (i.e., in-phase, anti-phase) will emerge  140 within initial pattern phasing attempts. When a required phasing pattern does not correspond to a pre-existing attractor, the intrinsic dynamics of the system and the to-belearned task are thought to compete against one another (Wallace, 1996). Subsequently, fluctuations and instability are postulated to occur in the behaviour pattern as it is drawn towards the intrinsically stable attractor. As learning progresses, it is further predicted that environmental information becomes memorized and interference from pre-existing modes of coordination will be gradually overcome. As strength of the memorized information increases, the intrinsic landscape is modified to reveal a new behavioural attractor that corresponds to the required pattern. Modifications in the attractor landscape serves to reduce competition within the system (Wallace, 1996). Due to this competition of intrinsic dynamics and acquisition of a new task, one common methodological concern pertains to how instinctive patterns interfere with the learning of new skills. In an attempt to eliminate the influence of unwanted differences between individuals, typical experimental protocol has been to employ a motor task that is completely novel to the participants involved in the study. The rationale for this approach is that differences between individuals should be minimized since each participant will sufficiently lack previous practice or exposure to the learning task. In addition, the required learning task is usually unique in nature and therefore, the influence of previously acquired movements should be negligible (Lee, Swinnen, & Verschueren, 1995). Further, the dynamics approach provides an environment, which affords the experimenter an opportunity to examine complex, coordinative movements. Such a trend has become evident within the literature since complex coordinative skills are thought to be more representative of the tasks found in a real-world environment  141 (Hodges & Lee, 1999). Further, when learners are asked to attempt the acquisition of a highly complex task, ceiling effects generally do not surface as a result of inherent limitations within the measurement procedure. To examine bimanual coordination from a dynamics perspective, participants are generally required to learn a task involving the rhythmic movement of two limbs in a coordinated phasing pattern. One of the most common to-be-learned patterns utilized has been the 90-deg relative phase pattern. That is, participants are routinely asked to coordinate the movement so that one limb lags the other by a quarter of a cycle in time. Theoretically, before a learning episode occurs, 90-deg relative phase is considered to be an unstable fixed point half-way between the stable attractor states of in- and anti-phase (Zanone & Kelso, 1992). Use of the 90-deg relative phase pattern has involved the flexion-extension of a variety of effectors such as index fingers (e.g., Zanone & Kelso, 1992) and forearms (e.g., Lee et al., 1995). Examination into the learning of a new coordination pattern is made possible by first identifying the initial state of the learner, and then observing the changes in the learner's landscape as he or she progresses through the acquisition process. It is, therefore, essential to identify the layout of the attractor state prior to learning so that the experimenter can gauge what has been modified or acquired as a result of learning. One technique that has been employed by researchers to systematically procure the evolving attractor layout throughout the entire learning process is the scanning run or probe. The origin of this procedure can be traced back to two sets of experiments (see Tuller & Kelso, 1989; Yamanishi, Kawato, & Suzuki, 1980), and has been utilised quite extensively by Zanone and Kelso (e.g., 1992, 1997), and more recently by others (e.g.,  142 Hodges & Franks, 2000) in the motor behaviour field. The rationale behind use of the scanning procedure is that it provides an indication of an individual's inherent coordination tendencies at any one time (e.g., before, during, and after a learning episode). As such, the experimenter can observe any changes to the learner's landscape as he or she progresses through acquisition, as well as gauge what has been modified or acquired as a result of learning. Probing the entire attractor layout provides valuable insight into the evolution of behavioural change as a function of practice. In a 'probe' or scanning run of an attractor landscape, the participant is asked to produce a large number of different phasing patterns. The purpose of which, is to identify both preferred and less preferred patterns among the movement components of a system. Within the literature, the scanning procedure has been designed so that it is continuous in nature. That is, participants are guided through all the required phasing patterns by memory, paced by a metronome (e.g., Yaminishi et al., 1980), or more commonly, guided by visual metronomes (e.g., Tuller & Kelso, 1989). Studies involving visual L E D metronomes manipulate the required relative phase by varying the time delay between the onset of the two LED's. For each relative phasing, participants are required to synchronize peak flexion of a limb with the onset of the respective flashing square L E D . It is imperative that the scanning procedure provides an accurate picture of an individual's intrinsic landscape. Recently, studies have attempted to examine the inherent variability of the scanning procedure in probing an individual's intrinsic dynamics (McGarry, Hodges, Bredin, Franks, & Chua, 2000). This study examined the reproducibility or stability of the dynamic landscape obtained from repeated scanning trials via a continuous or discrete method. Continuous scanning consisted of a single trial  143 of many different coordination pattern plateaus, presented one after another without any pause in the trial. Discrete scanning consisted of many discrete trials of the various coordination pattern plateaus, presented one after another. Results of this study indicated that a continuous scanning procedure produced significantly more within subject variability when compared to a discrete scanning procedure. This variability was attributed to the nature of the scan rather than an indication of learning. Thus it appeared a discrete method of scanning may be more reliable than a continuous method. More recent studies (see Experiment 1) have questioned the validity of using a visual metronome scanning procedure when the method of acquisition is very different. Thus to ensure validity, it may be necessary to employ the same scanning method as the participants use during acquisition. The purpose of this study was to examine and evaluate two different discrete methods of scanning and determine their validity based on two different acquisition methods. The first scanning method evaluated was the use of two visual L E D metronomes in the form of "flashing squares". Two squares on a computer screen were flashed on and off at various phase relations, with participants required to synchronize peak flexion of each arm with the onset of the respective flashing box. The second scanning method evaluated was the use of a "Lissajous" figure template and augmented visual feedback of the participant superimposed over the template. Specifically, movements of the right manipulandum produced horizontal movements of the cursor on the screen while movements of the left manipulandum produced vertical movements of the cursor on the screen. Each complete cycle of movement by the participant produced one continuous plot over the Lissajous figure template. For example, production of a 90°  144 relative phase pattern (right limb lead by one quarter cycle) produced a circular Lissajous figure in a counter-clockwise direction while production of a 270° relative phase pattern (left hand lead by one quarter cycle) produced a circular Lissajous figure in a clockwise direction. To determine if validity of scanning method depended on the method participants acquired a new coordination pattern, both scanning methods were evaluated under similar and dissimilar acquisition methods. That is, participants involved in the visual metronome (flashing square) scanning method acquired a new coordination pattern either via visual metronomes or via Lissajous figures. Alternately, participants involved in the Lissajous scanning method acquired a new coordination pattern either via Lissajous figures or via visual metronome. It was predicted that scanning trials performed in similar conditions to acquisition would produce high validity. However it was unclear if either type of scanning method performed with participants who performed acquisition trials in different conditions, would prove valid. Evaluation of these methods should provide valuable information for experimenters regarding use of scanning tests to determine and assess changes in a performer's landscape. Method Participants Informed consent was received from twenty self-professed right-handed individuals from a university population. None of the participants had previous exposure to the to-be performed task, and were naive to the purpose of the experiment. Participants were randomly assigned to one of four groups (five per group) based on the method of acquisition of the to-be performed task and method of the scanning run. These included  145 Lissajous acquisition, Square scanning (Lissajous-Square or L-S); Lissajous acquisition, Lissajous scanning (Lissajous-Lissajous or L-L); Square acquisition, Square scanning (Square-Square or S-S); and Square acquisition, Lissajous scanning (Square-Lissajous or S-L). The study was conducted in accordance with the ethical guidelines of the University of British Columbia. All participants received a remuneration of $10 upon completion of the experiment. Apparatus Participants were seated at a colour monitor ( V G A 640 x 480 pixels) measuring 27 cm in width and 20 cm in height (Zenith, Model #ZCM-1490). On either side of the monitor was two identical lightweight manipulanda that restricted arm movements to the elbow joint. Participants' arms were positioned such that the elbow joint was aligned with the axis of rotation and the hands were placed palm down on adjustable metal plates. The middle finger was secured between two vertical pins and velcro straps secured the forearms and hands. Amplitude was specified by computer feedback and markers on the table, specifying "in", "mid" and "out" positions for each arm. The required movement amplitude for each arm was 40° from the "in" to the "out" markers. A 40° movement translated to a 15 cm movement on the computer screen. Angular position was recorded using two optical encoders (Dynapar, E20-2500-130) one attached to the shaft of each manipulandum. Three-axis Quadrature Encoder interface cards (Advantech, PCL-833) were used to enable high-speed sampling of angular positions, giving a resolution of 10,000 counts per revolution. Angular position was sampled at a rate of 1000 Hz. PC speaker output of the computer motherboard was used to create the audio metronome tones. The metronome signal was amplified by a speaker on each side of the monitor  146 (Multi-Media, Model #EP-691). A MS-DOS computer, running proprietary software for data collection and analysis controlled the equipment. Task Participants were required to produce various bimanual coordination patterns by continuously flexing and extending the arms between the I N and OUT boundary markers. During acquisition, the goal of the participant was to learn how to correctly move the arms in the horizontal plane to produce a coordination pattern of 90° relative phase (left hand lagging one quarter cycle behind the right hand). During scanning, participants were required to perform discrete trials on one of 12 different patterns of relative phase (0°, 30°, 60°, 90°, 120°, 150°, 180°, 210°, 240°, 270°, 300° and 330°). Trials performed under the Lissajous condition involved augmented, concurrent visual feedback (limited only by the screen refresh rate of 60 Hz), via a Lissajous figure template projected on the computer screen with participants' movement superimposed over the template. Feedback of the movement, except where otherwise noted was only for the previous 500 milliseconds of movement. Each complete cycle of movement by the participant produced one continuous plot over the Lissajous figure. Direction of the movement was shown at the top of the Lissajous figure by either the word "clockwise" and a right direction arrow or "counterclockwise" and a left direction arrow. A metronome (1 Hz) was used for all acquisition and scanning trials with participants instructed to complete a full cycle for each "beep" of the metronome. Each trial lasted for 15 seconds. Trials performed under the flashing square condition involved a discrete visual metronome and no feedback. Two boxes (3 cm x 5 cm) were displayed at eye level in the  center of a black computer screen, aligned horizontally, 5 cm apart. These two boxes served as a visual metronome whereby each box flashed on (green) for 200 ms and off (black screen) for 800 ms at various phase relations. To manipulate relative phase, onset of the flash was controlled by a customized computer program, which allowed for the production of different phasing patterns. Participants were asked to continuously flex and extend their arms between the "in" and "out" markers such that they synchronized peak flexion of each arm with the onset of the respective flashing box. That is, when the right box flashed on (green), the right arm of the individual was in peak flexion; when the left box flashed on (green), the left arm of the individual was in peak flexion. Each trial lasted for 15 seconds. Experimental Design Orientation Participants were seated in front of the monitor with a manipulandum on either side. All participants familiarized themselves with the task apparatus and were provided with general instructions. These alerted the participants to the goal of the task, that is, to move the arms in such a way to perform a certain pattern on the computer screen. After instructions were given, a number of trials with various criteria and feedback were given to increase participants understanding of the apparatus and help orient participants as to what was expected of them. All participants were given two trials of practice moving the manipulanda without a criterion (i.e. the screen was blank). During these trials for all groups but Square-Square, concurrent on-line feedback of the entire trial was displayed on the monitor, involving a continuous trace of their movement pattern in the form of a  148  Lissajous figure. The Square-Square group performed these trials with no feedback, as they would not receive on-line feedback during any acquisition or scanning trials. Participants were next asked to perform a series of "natural" or "intrinsic" patterns of coordination without feedback and a 1 Hz metronome. Each participant had the in-phase (0°) and anti-phase (180°) pattern explained and demonstrated to him or her and was asked to perform each pattern at 1 Hz. Each participant was also asked to perform a coordination pattern different from in-phase and anti-phase that they felt they would be able to perform consistently for a long period of time. These "intrinsic" trials provided a useful measure of how consistently and accurately the performer was able to produce an in-phase and anti-phase pattern and assisted in discovery of other natural coordinative patterns unique to each individual. To orient participants regarding what type of trials they would be performing during acquisition and scanning, participants were given group specific orientation trials as well. To avoid learning effects, orientation trials consisted of natural or "intrinsic" coordination patterns. The Lissajous-Square group repeated the three "intrinsic" trials previously described (0°, 180° and any different pattern) with feedback throughout the entire trial to allow them experience with concurrent on-line feedback and expose them to how these patterns were represented with Lissajous feedback. In addition, participants performed trials of 0° and 180° in the flashing square condition to orient them with regards to these types of trials. Participants in the Lissajous-Lissajous group also repeated the three "intrinsic" trials with full feedback. As this group would not be required to perform trials in the flashing square condition, this group did not require orientation of the flashing squares. However, to ensure that all participants were exposed to an equal  149 number of trials, participants were shown Lissajous feedback of a participant performing a consistent in-phase pattern (0°) and anti-phase pattern (180°). During these two trials, participants were instructed to only watch the monitor and not move their hands. Conversely the Square-Square group did not require orientation of the Lissajous condition, but again needed to be exposed to an equal number of trials. Thus this group watched (no movement) three trials of flashing square patterns that included a 0°, 180° and one other pattern randomly selected. To orient themselves with the flashing square condition, participants performed two trials in this condition, 0° and 180°. Finally the Square-Lissajous group performed identical orientation trials as the Lissajous-Square group, however the order was reversed with this group performing the two flashing square trials first and the three Lissajous trials second. In summary, all group performed five orientation trials to assist them with understanding the expectations of the task and conditions they would be exposed to. All orientation trials consisted of patterns that were assumed to be intrinsic such that there would be no learning effect. Scanning All participants performed a scanning run prior to the beginning of acquisition trials (pre-scan) and immediately following completion of the acquisition trials (postscan). Depending on group assignment, scanning was either performed via a discrete visual metronome (Square condition) or via Lissajous figures as a criterion (Lissajous condition). Scanning via the Lissajous condition involved presentation of the appropriate criteria and concurrent Lissajous feedback displayed on the monitor. Participants were instructed to complete one full cycle of movement for every "beep" of a 1 Hz metronome, with direction of movement (clockwise or counterclockwise) displayed at the  150 top of the computer screen. No feedback was provided for scanning via the Square condition. Each scanning trial was 15 seconds long at one of 12 different patterns of relative phase (0°, 30°, 60°, 90°, 120°, 150°, 180°, 210°, 240°, 270°, 300° and 330°). A l l scanning runs consisted of an initial trial of 90° relative phase, discrete trials of all 12 different relative phase patterns, randomly presented, followed by another trial of 90° relative phase. Thus, each scanning run required the participant to perform all 12 coordination patterns, with three trials of 90° collected. Acquisition All participants performed a total of 75 acquisition trials of a 90° coordination pattern. This included five blocks of 15 trials with a maximum two-minute rest in between blocks. Depending on group assignment, acquisition trials were either performed via a discrete visual metronome (Square condition) or via Lissajous figures as a criterion (Lissajous condition). Acquisition trials via the Lissajous condition involved presentation of the appropriate criteria and concurrent Lissajous feedback displayed on the monitor. Participants were instructed to complete one full cycle of movement for every "beep" of a 1 Hz metronome, with direction of movement displayed at the top of the computer screen. N o feedback was provided for scanning via the Square condition. Dependent Measures and Analyses Discrete measures of relative phase were calculated for all complete cycles of movement within each trial. Relative phase (RP) of the left hand in relation to the right was calculated for each point after the speed and position of the limbs was re-scaled to the interval [-1, 1]. The phase angles were calculated using the methods described by  151 Scholz and Kelso (1989). Calculations were constrained such that all relative phases were converted to a value ranging between the criteria plus or minus 180° (ranged RP). That is, if a participant attempted to perform a relative phase of 0° and actually performed at a relative phase of 359°, this value would be converted to a relative phase of-1°. Relative phase provided a description of participants' performance within and across trials. From the relative phase values, the dependent measure of root mean square error was calculated. From each trial's calculated ranged RP values, a mean RP and standard deviation were calculated. Root mean square error (RMSE) of each trial was considered a general measure of the participant's error. R M S E was calculated by first subtracting the observed relative phase from the required/criterion relative phase (constant error) for each calculated ranged relative phase. Each constant error value was squared and summed, then divided by the total number of points. The square root of this value represented R M S E (see Franks, Wilberg & Fishburne, 1982). This measure has been used successfully to capture group differences in previous experiments (e.g. Tsutsui et al., 1998).  RMSE = The dependent measure o f R M S E was subjected to a 4 (group: Lissajous-Square, Lissajous-Lissajous, Square-Square, Square-Lissajous) x 4 (time: pre-scan, early acquisition, late acquisition, post-scan) A N O V A , with repeated measure on the last factor. Each value in the A N O V A consisted of a participant's average R M S E performance during three trials. Specifically, "pre-scan" consisted of the three scanning  trials of 90° prior to acquisition, "early acquisition" consisted of performance on the first three trials of 90° during acquisition, "late acquisition" consisted of performance on the final three trials of 90° during acquisition and "post-scan" consisted of the three scanning trials of 90° after acquisition. The alpha level for the entire experiment was set at .05 and the GreenhouseGeisser Epsilon factor was used to adjust the degrees of freedom for violation of the sphericity assumption (Greenhouse & Geisser, 1959). The Tukey HSD method (Tukey, 1953) was used for all post-hoc comparisons. Results RMSE R M S E data, plotted as a histogram as a function of condition is shown in Figure C I . Results of the repeated measures A N O V A showed a significant effect for time F(2, 38) = 24.216, p < 0.001 and a significant time x group interaction F(7, 38) = 6.172, p < 0.001. Post hoc analyses of each time condition revealed the following results. No group differences were found during pre-scan or early acquisition trials. In late acquisition, both Lissajous acquisition groups (L-S and L-L) performed with significantly less R M S E than both Square acquisition groups (S-S and S-L). However, the L-S group was not significantly different than the L - L group and the S-S group was not significantly different than the S-L group. In the post-scan trials, the L-S performed with significantly more R M S E than all three other groups and the L - L group performed with significantly less R M S E than the S-L group. Post hoc analyses of each group revealed the following results. The L-S group did not show a significant difference from pre-scan to early acquisition but did show significantly less R M S E from early acquisition to late  153 CD  i_  CO  3  CD  i—  CO  CT  3 CT  CO CO  CO  CO  op CD  _ 1 i_ w • 1 CO CO CO 3 CT  Li Li w H El  •  CO CO  Li  1 CD i  CO  3 CT  w  •  c Qi  E 0 a S c  N  c o  c .2  |co '3  C  3 CO O"  o o tn co  •  CT  o  <  8 | E § o "£ »  CO  c o  w  CD  a  Q. 3  UJ  2  *  5  u  CD "S o w .2  s  o 3  ii CO  o  (73  O O  d CN  o o 0  o (Bap) 3SIAIU  CD  E  H  154 acquisition. However this group performed with significantly higher R M S E from both early and late acquisition to post-scan. The L - L group performed with significantly higher R M S E in pre-scan than all other times and also showed significantly less R M S E between early acquisition and late acquisition. In addition, no significant difference was found between late acquisition and post scan. Both square acquisition groups (S-S and SL) did not show significant differences across the analyzed times. Discussion The purpose of this experiment was to assess two scanning methods used in bimanual coordination learning studies; visual metronomes in the form of flashing square L E D ' s and Lissajous figures with concurrent on-line feedback. Examining trials during a pre-acquisition and post acquisition scan, as well as trials early in acquisition and late in acquisition allowed for evaluation of the scanning methods. Performance in preacquisition scans was expected to be similar to early acquisition trials and performance in post-acquisition scans was expected to be similar to late acquisition trials. In addition, the experimental conditions also allowed for assessment of the two types of acquisition trials, namely flashing square L E D ' s and Lissajous figures. Results showed both Lissajous acquisition groups (L-S and L - L ) were able to significantly improve their performance of the new coordination pattern, shown by a significant decrease in R M S E from early acquisition to late acquisition. Both Square acquisition groups (S-S and S-L) did not appear to acquire the new coordination pattern, as there was no significant difference in R M S E from early acquisition to late acquisition. This is likely due to the fact that the Square acquisition groups did not receive any  155 feedback regarding their performance, whereas the Lissajous acquisition groups received continuous, concurrent feedback. These results are illustrated in Figure C2. The Lissajous scan did appear to accurately reflect the performance of the Lissajous acquisition group. Although there was a significant decrease in R M S E between the pre-acquisition scan and early acquisition trials, this can be explained by a learning effect from the various Lissajous scanning trials and the first few acquisition trials. In addition, there was no significant difference in R M S E between late acquisition and the post-acquisition scan. Although performance for this group dramatically improved during acquisition, the post-scan does reflect this change. The Square scanning method does not appear to be an accurate method for evaluating participants that performed their acquisition trials via Lissajous figures. Examination of performance of the L-S square group showed no significant difference between pre-acquisition scanning and post-acquisition scanning, even though the group did show significant improvement in performance during the acquisition period. In addition, the post-acquisition scanning performance had significantly higher R M S E that the late acquisition performance. This implies that the Square scanning method does not accurately reflect the change in performance of participants that have learned a new coordination pattern via Lissajous figures with concurrent feedback. These results are illustrated in Figure C3. The results of this experiment make it very difficult to evaluate either scanning method for the Square acquisition groups. Both Square acquisition groups showed no significant difference in R M S E between pre-acquisition scanning and early acquisition trials as well as no significant difference between late acquisition trials and post-  156 CD CD i— CO 3  co c/>  CT W Li• ( /1 )  t/> c/>  CT t/> Ui Li cb 03  CD i 03 3  C/) cr CT Li Li co (0 U  •  •  C/5  CT O < CD  U  o w cs CD  E  H  c o [to '3  cr o < 1—  CD  LU  (Bap) 3SIAIcJ  157  158 acquisition scanning. However, as neither Square acquisition group significantly improved their performance of the new coordination pattern between early and late acquisition, it is impossible to ascertain if the scanning method accurately reflects the participant's coordination landscape. That is, although the scanning trials are no different than the appropriate acquisition trials, we are unable to evaluate the scanning methods as the participants did not acquire a new bimanual coordination pattern (i.e. their landscape was unchanged after the acquisition period). These results are illustrated in Figure C4. In summary, results from this experiment suggest that visual flashing square metronomes, without movement-related feedback, do not provide sufficient information to acquire a new bimanual coordination pattern after 75 trials. However, Lissajous figures, with concurrent on-line feedback do provide sufficient information for acquisition of a new bimanual coordination pattern after 75 trials. Results also imply that for participants learning a new coordination pattern via Lissajous figures, Lissajous scanning does accurately evaluate their coordination landscape, while flashing squares without feedback do not accurately evaluate their coordination landscape. Thus it is suggested that researchers that are using Lissajous figures for acquisition of a new coordination pattern employ Lissajous scanning procedures rather than flashing squares, to ensure accurate measurement of the participant's ability to perform various coordination patterns. Suggestions for further research in this area include a flashing square acquisition group that does receive feedback, similar to the Lissajous acquisition group, to facilitate increased acquisition performance. Incorporating into this experiment a group showing improvement during acquisition via flashing squares, would allow for evaluation of both  159 CD L_  CO  3  CO CO  CT  CO cb i  •_J  CD I  TO CO 3 3 CT  CT  CO CO  n  •  CO  o CO to o  a.  o  E c o c  a £ x .52 CD  D  c o o  s i w a)  'to '3 CT O < CD CO  i- > o _  re w "O w o a 3  w  c re 2 E o)  II a> Si  o U—• |co  m '= 5 re  <  a  * ..  o o  w  '3 CT  o  1_ CO  <D >-  LU  re  ° 3  O)  ii  c CO  o CO  o q  d  CM  o q 0 o  o o 0 00  o q  d co  (Bsp) 3SlAld  o o  d  o o  d c\i  o o  CD  E  160 scanning methods (Lissajous and Squares) for participants learning via flashing squares. This would allow for recommendations regarding scanning procedures to researchers that use flashing squares for acquisition of a new coordination pattern.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0077177/manifest

Comment

Related Items