The effect of relative frequency of knowledge of results on the acquisition and retention of simple motor skills in the contextual interference paradigm
vi TABLE OF CONTENTS INTRODUCTION 1 Background of the Study. . . . . . . 1 Contextual Interference. . . . . . 2 Knowledge of Results. . . . . . 4 Relative Frequency of Knowledge of Results. . . 5 Contextual Interference and Relative Frequency of KR. . 7 Statement of the Problem. . . . . . . 11 Hypotheses. . . . . . . . . 11 Assumptions. . . . . . . . . 12 Delimitations. . . . . . . . . 12 Definition of Terms. . . . . . . . 13 REVIEW OF LITERATURE 15 Contextual Interference Effects in Motor Skill Learning. . . 15 Theoretical Interpretations of the Contextual Interference Effect. . 26 Action-Plan Reconstruction Hypothesis. . . . . 26 Elaboration and Distinctive Processing Hypothesis. . . . 31 Knowledge of Results (KR) and Motor Learning. . . . 37 Relative Frequency of Knowledge of Results. . . . 39 Guidance Hypothesis of Knowledge of Results. . . . 42 Combined Effects of Contextual Interference and Knowledge of Results 51
vii METHODOLOGY 58 Subjects. . . . . . . . . 58 Equipment. . . . . . . . . 58 Tasks. . . . . . . . . . 58 Design. . . . . . . . . 59 Procedures. . . . . . . . . 60 Data Treatment and Analysis. . . . . . . 62 RESULTS 64 Acquisition. . . . . . . . . 64 Retention. . . . . . . . . 67 DISCUSSION 70 APPENDICES. . . . . . . . . 77 REFERENCES. . . . . . . . . 98 CURRICULUM VITAE. . . . . . . . 102
LISTS OF TABLES, FIGURES AND APPENDICES Table 1. Experimental Design. . . . . . . 60 Table 2. Skewness and Kurtosis Statistics for raw |CE| and CE . 85 Table 3. Skewness and Kurtosis Statistics for transformed CE . 86 Table 4. Acquisition Analysis Between Subjects effects: A (KR) x B (Schedule) for log-CE . . . . . 88 Table 5. Acquisition Analysis Within Subjects effects: A (KR) x B (Schedule) x C (Time) x D (Block) for log-CE. . . 89 Table 6. Acquisition Analysis Between Subjects effects: A (KR) x B (Schedule) for log-VE. . . . . . 90 Table 7. Acquisition Analysis Within Subjects effects: A (KR) x B (Schedule) x C (Time) x D (Block) for log-VE. . . 91 Table 8. Retention Analysis Between Subjects effects: A (KR) x B (Schedule) x C (Retention Order) for log-CE. . . 92 Table 9. Retention Analysis Within Subjects effects: A (KR) x B (Schedule) x C (Retention Order) x D (Time) x E (Block) for log-CE. . . . . . . 93 Table 10. Retention Analysis Between Subjects effects: A (KR) x B (Schedule) x C (Retention Order) for log-CE. . . 95 Table 11. Retention Analysis Within Subjects effects: A (KR) x B (Schedule) x C (Retention Order) x D (Time) x E (Block) for log-CE. . . . . . . 96
Figure 1. Experimental Task . . . . . . 59 Figure 2. . Log-CE across acquisition and retention trial blocks as a function of KR and Schedule conditions. . . . . 65 Figure 3. Log-CE across acquisition and retention trial blocks as a function of KR conditions. . . . . . . 66 Figure 4. Log-CE across acquisition and retention trial blocks as a function of Practice Schedule conditions. . . . . 66 Figure 5. Log-CE across retention trial blocks of Practice Schedule-KR conditions as a function of Retention Order.. . . . 67 Figure 6. Log-CE across retention trial blocks of Practice Schedule-KR conditions on the Blocked-Random Retention Test. . . 68 Figure 7. Log-CE across retention trial blocks of Practice Schedule-KR conditions on the Random-Blocked Retention Test. . . 69 Figure 8. Quartile to Quartile Plot of raw CE data. . . . 84 Figure 9. Quartile to Quartile Plot of raw |CE| data. . . . 85 Figure 10. Quartile to Quartile Plot of transformed CE data. . . 86
Appendix A. Human Subjects Approval. . . . . . 78 Appendix B. Informed Consent. . . . . . . 80 Appendix C. Quartile to Quartile Plots. . . . . . 83 Appendix D. Analysis of Variance Summary Tables. . . . 87
1 CHAPTER 1 INTRODUCTION Background of the Study Early theories of motor skill learning identified practice and feedback as crucial components of the learning environment (Adams, 1971; Schmidt, 1975). It is generally accepted that when all other aspects of performance are equal, more learning will occur if there are more practice trials, this phenomenon has been termed the power law of practice (Schmidt & Lee, 2005). Practice, logically, is essential if improvements in performance and long-term retention are desired, however, practice alone does not produce learning. A combination of practice and appropriate feedback are essential in the achievement of optimal learning. The theories of Adams (1971) and Schmidt (1975) are centralized around the interaction of feedback and the previous response to help guide the performer to the correct action. Bilodeau, Bilodeau & Schumsky (1959) elaborated on this position by suggesting that without knowledge of results (KR) no learning will occur at all. This highlights the importance of both components in the acquisition and retention of motor skills. The influence of the power law of practice on learning has been well documented in the early literature. Schmidt’s (1975) schema theory further developed the knowledge of the role of practice in learning by proposing the idea of variability of practice. When the number of practice trials were held constant, results showed subjects who performed the criterion task plus other tasks which surrounded the criterion task, performed poorer during acquisition but better in retention in comparison to subjects who only practiced the criterion task. This finding suggests that practice on tasks other than the criterion task
2 facilitated retention. Research has demonstrated that practice on a variety of tasks can positively influence the retention of a single task. However, how variable practice is scheduled during acquisition has also been shown to have substantial influence on the performance and learning of motor skills. Contextual Interference Contextual interference has received an abundance of attention in the motor learning literature (see Brady, 1998; Magill & Hall, 1990). The effects of contextual interference was first demonstrated in motor skills by Shea & Morgan (1979) using random and blocked practice schedules during the acquisition of three simple motor tasks. Contextual interference is produced by the order in which multiple tasks are presented during acquisition. High contextual interference is generated using a random practice schedule in which trials of the experimental tasks are interspersed unsystematically with each other (ABCBCACAB). Low contextual interference is produced using a blocked schedule where the tasks are presented in individual blocks with performance on each task being completely uninterrupted with any trials of the other tasks (AAA…BBB…CCC). The contextual interference effect is characterized by enhanced blocked group performance in comparison to the random group during acquisition. However, a reversal of these effects is observed during retention and transfer tests, with the random group outperforming the blocked group. Two main theoretical perspectives on the contextual interference effect have been thoroughly examined in motor skill learning. Lee & Magill (1983; 1985) proposed the action-plan reconstruction hypothesis, suggesting practice in a random schedule causes the subject to partially or completely forget the action-plans associated with the
3 alternative tasks not being performed. The extra processing requirement encountered by the random group during acquisition results in depressed acquisition performance. However, the forgetting of this information requires the subject to reconstruct the action- plan each time a different task than the previous one is to be performed. This reconstruction procedure increases the cognitive effort required and thus enhances learning and retention performance. Subjects in the blocked schedule are presented with the same task in distinct blocks of trials and hence the next trial is the same each time. Lee & Magill (1983; 1985) argue that this schedule does not require increased cognitive effort as subjects must only maintain the current action-plan in working memory and respond to the stimulus. Subjects do not undertake any active reconstruction of the tasks and thus this schedule produces depressed retention performance. An opposing viewpoint of the mechanisms behind the contextual interference effect were presented by Shea & Zimny (1983; 1988) who suggested that subjects in the random acquisition schedule undertake comparative and contrastive analyses of the tasks. Initially this extra processing impairs acquisition performance, but, produces enhanced mental representations of the tasks compared to the representations produced by blocked acquisition schedules, resulting in improved retention performance. This explanation suggests during the random schedule multiple tasks are held concurrently in working memory encouraging elaborative and distinctive processing within and between tasks. These processing strategies support the generation of multiple access points to task information in long term memory and therefore results in greater retention. Although an abundance of research has been carried out on practice schedule manipulations many have followed a very similar experimental design to the original
4 Shea & Morgan (1979) paper, with 54 acquisition trials (18 on each task), a 10-minute retention interval and 18 retention trials (6 on each task). However, others have manipulated the number of acquisition trials (Goode & Magill, 1986; Wulf & Lee, 1993; Albaret & Thon, 1998; Ste-Marie, Clark, Findlay & Latimer, 2004), retention trials (Lee & Magill, 1983; Del Rey, 1989; Carnahan, Van Erd & Allard, 1990; Shewokis, Del Rey & Simpson, 1998) and the retention interval (Lee & Magill, 1983; Goode & Magill, 1986; Wood & Ging, 1991; Wulf & Lee, 1993). Although an abundance of empirical research has been conducted on practice schedules, as outlined previously, it is not the sole determinant of successful motor skill learning. Knowledge of Results (KR), or feedback, has been identified as being equally important to the acquisition and retention of motor skills. Knowledge of Results (KR) Knowledge of Results (KR) can be defined as augmented information regarding the success of a movement relative to an assigned goal (Schmidt & Lee, 2005). KR can generally be easily verbalized, is often terminal in nature, i.e. post-response, and is provided with reference to the outcome or success of the movement. Historically, the role of KR has been identified as guiding the subject to the correct response. This belief stemmed from the idea that the more repetitions performed at or close to the correct response the stronger the memory representation for that movement and thus greater learning (Adams, 1971, Schmidt, 1975). This idea was originally proposed by Bilodeau, Bilodeau & Schumsky (1959) who suggested that KR is not only an important learning variable, but without KR no learning occurs at all. However, research has highlighted that although the research has demonstrated
5 the frequent provision of KR does indeed guide the subject to the correct response during acquisition, it appears that if provided repeatedly subjects then become dependent upon it and cannot perform on tests of learning during retention, where KR is not present (Salmoni, Schmidt & Walter, 1984; Sherwood and Lee; 2003). Understanding that the provision of KR may influence performance and learning differently questioned Schmidt’s (1975) schema theory which suggested that a higher relative frequency of KR during acquisition would enhance learning. However, further research has shown that a lower relative frequency may be more beneficial to motor skill learning. Relative Frequency of Knowledge of Results (KR) Relative frequency of KR refers to the percentage of trials on which KR is provided (Schmidt & Lee, 2005). Obviously, a 100% relative frequency represents the provision of KR after every trial. Many manipulations of this paradigm have been investigated with 50% (Goodwin, Eckerson & Voll, 2001; Lee, White & Carnahan, 1990) and 33% (Ho & Shea, 1978; Winstein & Schmidt, 1990) being common in the literature, with KR being presented every other trial or every third trial respectively. In the literature higher relative frequencies of KR outperform lower relative frequencies of KR during acquisition (Salmoni, Schmidt & Walter, 1984; Sparrow, 1995; Goodwin, Eckerson & Voll, 2001; Anderson, Magill, Sekiya & Ryan, 2005). However, during retention tests very different findings are observed. Many studies show the 100% relative frequency perform poorly during retention in comparison to their acquisition performance (Schmidt, Young, Swinnen & Shapiro, 1989; Yao, Fischman & Wang, 1994) and have a large difference score between the end of acquisition and the beginning of retention trials (Winstein & Schmidt, 1990; Winstein, 1991; Anderson, Magill, Sekiya & Ryan, 2005),
6 showing the effects observed during acquisition are transient in nature. Many studies have provided support for greater learning as a function of lower relative frequency of KR schedules (Lee, White & Carnahan, 1990, experiment 3; Sparrow & Summers, 1992, experiment 2; Vander Linden, Cauraugh & Greene, 1993; Weeks, Zelaznik & Beyak, 1993; Weeks & Kordus, 1998), which reinforces the important role of relative frequency of KR in the acquisition and retention of motor skills. Theoretically, it can be argued that the random practice schedule observed in contextual interference is very similar to the effects observed in the relative frequency of KR literature. In contextual interference subjects cannot use KR immediately as an intervening task is presented on the next trial. This may stimulate the subject to make comparative and contrastive analyses between tasks as suggested by Shea & Zimny (1983; 1988) as KR does not directly influence the subsequent task. In the relative frequency of KR literature, KR is provided on some trials but not on others. A similar phenomenon may exist in KR as observed in the contextual interference literature, as a lower relative frequency of KR should stimulate subjects to use KR provided on one task to aid the performance of the next task. The effect could be manifested by greater retention performance of lower relative frequency groups compared to their 100% counterparts. It can be proposed that the effect would be even more pronounced if a multiple task learning paradigm was used, as KR on one task would provide little information regarding the performance of the next, different task. Thus performance in retention of lower relative frequency groups could only be facilitated if comparative task analyses were taking place. In both paradigms the common findings demonstrate diminished performance in
7 acquisition but improved performance in retention for the random and lower relative frequency of KR groups. The purpose behind the contextual interference and relative frequency of KR paradigms is to facilitate the learning of motor skills. However, very little research has actually integrated the two paradigms together to create, what could possibly be the most beneficial learning conditions for the learning of motor skills. A modest collection of research has emerged where it appears that practice schedule and KR have been manipulated, however, much of this work has been unsystematic in nature and often not had the intention to investigate the relationship between the two concepts and their influence on motor skill learning. Contextual Interference and Relative Frequency of Knowledge of Results (KR) It is interesting to note that the main emphasis of the contextual interference paradigm is to facilitate the learning of motor skills. However, an abundance of the literature fails to manipulate the relative frequency of knowledge of results (KR) which has been shown to facilitate motor skill retention (Ho & Shea, 1978; Salmoni, Schmidt & Walter, 1984; Winstein & Schmidt, 1990; Schmidt, 1991a). The majority of the research employs a 100% relative frequency of KR (Shea & Morgan, 1979; Lee & Magill, 1983; Shea & Zimny, 1988; Immink & Wright, 1998; Wright, Magnuson & Black, 2005), meaning KR is provided after every trial during acquisition. Whereas the KR literature suggests a lower relative frequency during acquisition, for example, 50% (Goodwin, Eckerson & Voll, 2001) or 33% (Ho & Shea, 1978), improves retention performance and thus learning. Very few studies in the contextual interference literature have investigated the interaction of practice schedule and KR on the performance and learning of motor skills.
8 Wulf & Lee (1993), although not specifically testing the interaction between the two, manipulated relative frequency of KR in a contextual interference paradigm, comparing a 100%, or every trial, relative frequency against a 3-trial summary KR schedule. No interaction between KR frequency and practice schedule were observed and thus no concrete findings regarding the influence of this relationship on motor skill learning can be put forth. However, closer analysis of the experimental KR manipulation may provide a possible explanation. A 100% relative frequency of KR schedule can be explained as receiving KR about every trial, with a lower relative frequency implying KR is provided after some trials but not after others. The design of Wulf & Lee’s experiment satisfied this criteria, with summary KR being provided after every third trial, thus two trials were performed without immediate paired KR. However, even though KR was presented in a summary format, KR was still provided about every trial and hence may be considered a variation of a 100% relative frequency. Additionally, although it has been shown that summary KR may be beneficial to motor skill acquisition and retention (Weeks & Sherwood, 1994), separating the presentation of KR from trial performance may actually hinder performance and learning, when compared to a more systematic presentation of a lower relative frequency of KR. Del Rey & Shewokis (1993) performed a similar study specifically looking at the relationship between contextual interference and KR. Their findings show a beneficial effect for 100% KR and 5-trial summary KR in a blocked acquisition schedule compared to a 10-trial summary KR condition. This suggests that subjects in the 10-trial summary condition were unable to use KR to positively influence their performance, and provides support for the proposition that separating the presentation of KR from trial performance
9 may actually hinder performance and learning. The authors also replicate the findings of Shea & Morgan (1979) showing better acquisition performance for the blocked group compared to the random group in a 100% KR schedule. However, no significant differences were observed during retention. Although some differences were observed, their contribution to the understanding of the relationship between practice schedule and KR and their influence on motor skill performance and learning is small. Similar to the study by Wulf & Lee (1993) the summary conditions used by Del Rey & Shewokis (1993) still afforded information regarding KR about every trial during acquisition, and thus do not shed any light onto the systematic application of relative frequency of KR on the contextual interference effect. It is also important to note that if the summary KR schedules were converted to standard relative frequencies (%), they would represent 20% and 10%, two frequencies which have limited empirical support for their use. This may suggest that the use of summary KR as an experimental variable may not be an effective method to investigate the relationship between relative frequency of KR and practice schedule. Wulf (1992) also integrated the concepts of practice schedule and KR frequency while investigating program and parameter learning relating to the general motor program (GMP), however, once again the interaction between contextual interference and the relative frequency of KR was not being directly investigated. Subjects were provided KR on two of the experimental tasks but not on the third, however, they did actually provide two KR pre-test trials on the task receiving no-KR. Although the authors manipulated both practice schedule and KR, their contribution to the knowledge base on the interaction of these two factors is limited as the study suffered from a few
10 methodological concerns. During acquisition, trial blocks for the random practice schedule contained performance of all three tasks, whereas trial blocks for the blocked practice schedule only contained performance of a single task. Also during immediate and delayed retention tests only the task which did not receive KR was tested, contrary to a classical contextual interference study. Another unusual factor in this design was the blocked group schedule. Subjects performed each task in blocks of 15 trials (C A B C A B) contrary to the common low contextual interference design of practicing all trials on one task before moving onto the next. This could have obviously affected both acquisition and retention performance of the blocked groups. This also creates a 66% relative frequency of KR for the blocked group, an interesting interpretation of the complex design. The study does provide some support for contextual interference as the blocked schedule acquisition performance was generally greater than the random schedule, plus the delayed retention test showed the random schedule had less error than their blocked group counterparts. Although some manipulation of both practice schedule and KR were present, the author did not examine the systematic application of relative frequency of KR in the contextual interference paradigm. Finally, Wulf and Schmidt (1994) examined the role of reminder feedback in the contextual interference paradigm with the attempt to induce response variability in the random condition. However, the design only included 100% feedback schedules with one group receiving reminder feedback about the previous trial prior to the next response. It is interesting to note that during each feedback presentation, two forms were provided, the trace of the last attempt superimposed on the actual trace to be replicated and the RMS error. If the subject is unaware of the meaning of RMS error, they may actually ignore it
11 and concentrate solely on the comparison of the two traces, which is more representative of knowledge of performance information rather than knowledge of results, and thus provides little contribution to the understanding of the interaction of these two factors. Although substantial research has been completed on both contextual interference and the relative frequency of knowledge of results individually, no single study has identified the possible interaction of contextual interference and the systematic application of relative frequency of KR. Based upon the research findings in both sets of literature, it could be argued that both phenomena may be controlled by similar, if not the same mechanisms/processes, as both paradigms have continually exhibited very similar results, with diminished acquisition performance but enhanced retention performance for the random practice schedule and lower relative frequency of KR groups. Statement of Problem The present study investigated whether the contextual interference and relative frequency of knowledge of results paradigms were controlled by the same mechanisms/processes. The study also examined the influence of relative frequency of knowledge of results (KR) on the contextual interference paradigm, specifically, whether a lower frequency of KR would positively or negatively influence the contextual interference effect. Hypotheses For the purposes of this study, the following research hypotheses will be tested: 1. Acquisition performance will be superior for groups who perform the tasks in a blocked order compared to groups who perform the tasks in a random order. 2. Retention performance will be superior for groups who performed the tasks in
12 a random acquisition order compared to groups who performed the tasks in a blocked order during acquisition. 3. Acquisition performance will be superior for groups who receive 100% KR compared to groups who receive 33% KR. 4. Retention performance will be superior for groups who received 33% KR during acquisition compared to groups who receive 100% KR during acquisition. 5. Acquisition performance will be superior in the blocked practice schedule for groups that receive 100% KR compared to groups that receive 33% KR. 6. Retention performance will be superior in the blocked practice schedule for groups that receive 33% KR compared to groups that receive 100% KR. 7. Acquisition performance will be superior in the random practice schedule for groups that receive 100% KR compared to groups that receive 33% KR. 8. Retention performance will be superior in the random practice schedule for groups that receive 33% KR compared to groups that receive 100% KR. Assumptions In the present study the following assumptions were made: 1. Subjects had no previous knowledge of the experimental protocol or the tasks used. 2. Subjects performed to the best of their ability throughout the study. Delimitations 1. All subjects were right-handed. 2. All subjects were university students from Indiana University Bloomington.
13 3. All subjects participated voluntarily. 4. The study was conducted in a laboratory setting. Definition of Terms Acquisition
A set of practice trials where a learner first acquires a new motor skill. Blocked Practice Schedule
A practice sequence in which all the trials on one task are completed together, uninterrupted by practice on any of the other tasks; low contextual interference (Schmidt & Lee, 2005). Contextual Interference
The interference in performance and learning that arises from practicing one task in the context of other tasks (Schmidt & Lee, 2005). In motor skill learning, contextual interference is controlled through the manipulation of practice schedules in a multiple task learning situation (Shea & Morgan, 1979). Knowledge of Results (KR)
Augmented feedback related to the nature of the result produced in terms of the environmental goal (Schmidt & Lee, 2005). Random Practice Schedule
A practice sequence in which the tasks being practiced are ordered (quasi-)randomly across trials; high contextual interference (Schmidt & Lee, 2005). Relative Frequency of Knowledge of Results
The percentage of trials for which KR is provided; the absolute frequency divided by the number of trials (Schmidt & Lee, 2005).
A set of trials administered after a retention interval to assess learning. During in these trials it is common that no feedback or KR is provided (Schmidt & Lee, 2005).
15 CHAPTER 2 REVIEW OF LITERATURE
Research in motor learning has covered a vast array of research topics; however, the major theoretical perspectives in this area (Adams, 1971; Schmidt, 1975) have commonly focused on the role of practice schedule and feedback in the learning of motor skills. Both theories emphasize the importance of practice and feedback as central components in the formation of memory representations of skilled behavior. The following review of literature will describe the experimental evidence supporting the contextual interference effect and the theoretical perspectives which have developed as a result. The review will also present the research findings describing the role of relative frequency of knowledge of results in the learning of motor skills and the associated theoretical contributions. Finally, the review will identify the literature which has manipulated both practice schedule and relative frequency of knowledge of results and provide a brief synopsis of their findings. Contextual Interference Effects in Motor Learning Contextual interference has been the most extensively researched topic in motor learning for the past twenty five years (See Brady, 1998; Magill & Hall, 1990, for reviews). The first demonstration of the contextual interference in motor learning was provided by Shea & Morgan (1979) by manipulating practice schedule to provide two different levels of contextual interference in the learning of three simple motor skills. This influence of this early work on contextual interference research is unparalleled, with the majority of future studies using practice schedule to manipulate the levels of