This article was downloaded by: [University of Sydney] On: 14 March 2015, At: 04:27 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Research Quarterly for Exercise and Sport Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/urqe20
When Forgetting Benefits Motor Retention a
John B. Shea & David L. Wright
b
a
Motor Behavior Laboratory in the Department of Exercise and Sport Science , Pennsylvania State University , USA b
Department of Health and Physical Education , Texas A & M University , USA Published online: 26 Feb 2013.
To cite this article: John B. Shea & David L. Wright (1991) When Forgetting Benefits Motor Retention, Research Quarterly for Exercise and Sport, 62:3, 293-301, DOI: 10.1080/02701367.1991.10608726 To link to this article: http://dx.doi.org/10.1080/02701367.1991.10608726
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Research Quarterlyfor Exerciseand Sport © 1991 by the American Alliance for Health,
Physical Education, Recreation and Dance Vol. 62, No. 3, pp. 293·301
When Forgetting Benefits Motor Retention
Downloaded by [University of Sydney] at 04:27 14 March 2015
John B. Shea andDavid L. Wright Recent research (Lee & Weeks, 1987; Weeks, Lee, & Elliott, 1987) investigating the processes responsible for the contextual interference phenomenon has used a modified short-term motorretention paradigm to support the reconstruction explanation (Lee & Magill, 1985; Magill, 1989; Magill & HaU, 1990). The presentexperimentwas an extension ofthese experiments in whichforgetting ofan acquisition task was induced through performance of eithera similar or dissimilar distractortask during the intertrial interval. The effects of an extra practicetrial with the acquisition task as well as no activity during the intertrial interval were also investigated. In addition, forgetting ofthe acquisition task was assessed prior to a reconstruction trial, which immediately preceded a 2-min filled retention interval. Both similar and dissimilar distractortasks caused equivalent amounts offorgetting ofthe acquisition task prior to the reconstruction trial. However, retention of the acquisition task was significantly improved if its reconstruction occurred followingforgetting due to interferencefrom performance ofa similar distractortask. Thesefindings suggestforgetting and subsequent reconstruction alone are not sufficient for improved retention. Theseprocesses must occurin the context of a similar task for improved retention.
Key words: contextual interference, forgetting, reconstruction, retention
M
o tor learning researchers have become increasinglyin terested in con textual in terference (Battig, 1979). Contextual interference refers to the finding that practice of multiple tasks in a random (high contextual interference) practice schedule results in greater retention and transfer than when tasks are practiced in a blocked practice schedule (low contextual interference) in which practice ofone task is completed before practice on a different task is given (Lee & Magill, 1983; Shea & Morgan, 1979). Lee and Magill (1985) have proposed a reconstruction explanation for contextual interference. According to this explanation, a knowledge of results (KR) faction plan (Lee & Magill, 1985) or, more recently, action plan (Magill & Hall, 1990) for a particular task is forgotten as a result of intervening trials on other tasks in a random practice schedule. This forces the learner to engage in
John B. Sheais anassociate professor anddirector of theMotor Behavior Laboratory inthe Department of Exercise andSport Scienceat Pennsylvania State University. David L Wright is an assistant professor inthe Department of Health andPhysical Education at Texas A & M University. Address requests for reprints to John B. Shea, The Pennsylvania StateUniversity, Motor Behavior Laboratory, 131 White Building, University Park, PA 16802. Submitted:June 25, 1990 Revision accepted: December 18, 1990
RDES: September 1991
more effortful reconstructive processing to generate the action plan for subsequent performance of the task. Reconstructive processing is emphasized in this explanation because it is assumed to incorporate the cognitive activities similar to those required when performing retention and transfer tests (Lee, 1988). Magill and Hall (1990) have recently described the reconstruction process as the retrieval of the appropriate motor program representing an action and then adding to it the parameters specific to the constraints and goal of the task to be performed. In contrast to a random practice schedule, there is little opportunity for forgetting in a blocked practice schedule. The action plan is present in working memory and can be executed on successive trials with little or no reconstructive activity. Thus, any practice schedule that promotes complete or partial forgetting of a previously used action plan between repetitions of a goal directed movement will depress acquisition but promote retention performance relative to conditions in which action plan forgetting does not occur (Lee & Magill, 1985, p. 8; Magill & Hall, 1990). Support for the prediction that forgetting and reconstruction of an action plan enhances retention can be garnered from an experiment by Marshall, Wyatt, Moore, and Sigman (1975). A linear positioning task was used to examine the effect of either one or seven repeti tions of a criterion movement spaced at intervals of 5 or 60 s on subsequent recall following a 30-s retention interval. The 60-s spaced condition was assumed to have induced greater forgetting of previous presentations of the criterion movement than the 5-s spaced condition. Recall as measured by absolute error (AE) was more accurate for
293
Downloaded by [University of Sydney] at 04:27 14 March 2015
Shea andWright
the 60-s spaced condition than for the 5-s spaced condition. However, this finding was restricted to the sevenrepetition condition. If forgetting of previous presentations was the main determinan t ofsubsequent recall, this effect should also have been evidenced for the onerepetition condition. Furthermore, this finding was not obtained for constant error (CE) measures. In three subsequent experiments, Marshall,Jones, and Sheehan (1977) increased the number of test trials from that used in the Marshall et al. (1975) experiment. This enabled them to obtain evidence ofgreater retention for the 60-s spaced condition for both six and one repetition conditions. However, these effects were found for CE only but not for variable error (VE) or AE. Unfortunately, interfering activity that might have facilitated forgetting was not interpolated between repetitions or during the retention interval in these experiments. In addition, no independent test offorgetting was given between repetitions, making the assumption that forgetting occurred during these intervals a speculative one. Weeks, Lee, and Elliott (1987) investigated the effect different levels of initial forgetting had on subsequent recall test performance. Subjects made a self-selected criterion timing movement from three predetermined ranges within which they had previously been trained. An initial reproduction of this movement was made either immediately, after a 20-s unfilled interval, after a 20-s interval of counting backward by threes, or after a 20-s interval of counting backward by sevens. A second reproduction of the criterion movement was attempted following a 20-s filled retention interval. The findings offered some support for the notion that increased initial forgetting would subsequently lead to improved retention. VE scores generally revealed that recall performance on the second reproduction was less variable with increased error on the initial reproduction. However, response accuracy findings were notconsistentwith these VE findings. Absolute constant error (ICEI) was greater for the initial reproduction of the criterion movement when itwas made after the unfilled interval and counting backward by threes and sevens than when this reproduction was made immediately. However, the second reproduction of the criterion movement was more accurate than the immediate reproduction condition only after counting backward by threes and sevens. Further support for the reconstruction explanation was provided by Lee and Weeks (1987) in an experiment in which a linear positioning task was used. Subjects were presented a repetition of a criterion movement (by moving to a barrier at the criterion position) or were required to estimate the criterion movement either immediately or after 20 s of the same interfering activity. The results indicated that a delayed estimate led to larger error (VE and ICEI) than an immediate estimate. This led Lee and Weeks to assume that for either delayed repetition or estimation, more information about the criterion is for-
294
gotten than at the time of immediate repetition or estimation. This assumption is critical to their interpretation of subsequent recall performance following immediate and delayed repetition. This is because all repetitions were movements to a barrier, and an independent measure of forgetting could not be obtained for repetition conditions. The results of subsequent criterion movement recall indicated the delayed repetition condition had significantly less variability (VE) than the immediate repetition condition. However, no differences in accuracy (ICEI) were obtained between the delayed and immediate repetition groups. The reconstruction explanation clearly states it is the KR/action plan (Lee & Magill, 1985, p. 19) or action plan (Magill & Hall, 1990) that is forgotten and reconstructed. However, Weeks etal. (1987), and Lee and Weeks (1987) did not provide KRatanystage oftheir studies. In addition, subjects in the Weeks etal. (1987) experiment were never given additional experience with the criterion movement after forgetting occurred so they could reconstruct it before the retention interval. Furthermore, the procedure used to cause reconstructive processing by Lee and Weeks (1987) might have resulted in processing that was inappropriate for recall purposes. The assumption that movemen t to a barrier which defined the criterion position in the repetition condition caused subjects to reconstruct the forgotten action plan must be viewed with caution. It has been shown that constrained movements (movements in which the criterion position is defined by movement to a barrier) and unconstrained movements (movements in which the subject actively estimates the criterion position) do not share the same type ofprocessing (Kelso & Wallace, 1978). In the present case, this would suggest the reconstructive processing at the time of repetition of the criterion position might have been inappropriate for recall purposes in the repetition condition (Lee, 1988). Thus the procedures used byWeeksetal. (1987) andLeeand Weeks (1987) raise questions concerning the ability ofsubjects to correctly reconstruct a forgotten action plan. The present experiment consisted of a more definitive test of the reconstruction explanation (Lee & Magill, 1985; Magill & Hall, 1990). The paradigm was similar to the modified motor short-term memory paradigm used in earlier studies to investigate the reconstruction explanation (Lee & Weeks, 1987; Magill, 1989; Weeks et aI., 1987). According to Magill (1973), this kind of an approach (i.e., the modified motor short-term memory paradigm) might avoid the problems associated with reinforcement on every trial obscuring the mechanisms operating on each trial in multiple trial procedures. This approach, then, was ideally suited for the purpose of the present experimen t, which was restric ted to the investigation of the reconstruction process (Magill & Hall, 1990) associated with forgetting. No attempt was made to address the development of the reconstruction process over multiple practice trials with different tasks as is the
RUES: September 1991
Downloaded by [University of Sydney] at 04:27 14 March 2015
Shea and Wright
case in contextual interference experiments. It must therefore be emphasized that while the present experiment was inspired by the reconstruction explanation, its findings would not be expected to necessarily have complete congruence with typical contextual interference effects. The tasks used in the present experiment were similar to those used in the first experiments reporting contextual interference effects (Lee & Magill, 1983; Shea & Morgan, 1979). Subjects received two acquisition trials on which KR was provided before a retention interval. On the first trial an action plan for task performance was constructed. Components of the action plan forgotten as the result of interpolated task conditions during the intertrial interval could be reconstructed on the second trial. The importance of task similarity for retention benefits resulting from action plan reconstruction was also examined in the present experiment. This was accomplished by including conditions in which betweentrial forgetting was induced by performance of either a similar or dissimilar task during the intertrial interval. Shea and Zimny (1983, 1988) have implicated task similarity as a variable that limits the benefits of intertask interference experienced during acquisition on subsequent retention. This view holds that maximum retention benefits occur when similar rather than dissimilar tasks are practiced during acquisition. In contrast, the reconstruction explanation does not impart great importance to the nature of the intervening activity that causes forgetting between trials because the important issue for this hypothesis is that the learner reconstructs the forgotten action plan used for task performance (Lee & Magill, 1985; Magill & Hall, 1990).
Method Subjects Subjects were 52 right-handed students at The Pennsylvania State University, who received class credit for their participation. Informed consentwas obtained from each subject prior to testing. Four subjects were excluded from the experiment because of their failure to follow instructions.
Apparatus The apparatus was similar to that described by Shea and Morgan (1979). It consisted of six freely movable barriers attached to the top surface of a platform. Three barriers were positioned on each side of the apparatus so that they faced the midline of the apparatus. A button switch, referred to as the start button, was located on the midline and front of the apparatus. Directly to the rear of
ROES: September 1991
the start button were two holes large enough to support a tennis ball but small enough to keep the ball from falling through. A ball rested in the first hole. A red warning light and a white stimulus light were mounted at the rear of the apparatus. A plywood barrier was located directly behind the lights on which the diagram of the task being performed could be displayed. Only one diagram was present at any time and was located directly above the white stimulus light. Two Hunter Klockcounters (model 220 C) were located behind the plywood barrier and were used to time components of the subject's performance. A control panel allowed the experimenter to present the warning light followed by the stimulus light for each trial. The two timers were initiated at the onset of the stimulus light. One timer stopped on release of the start button, and this constituted reaction time (RT). The other timer was stopped on placemen t of the tennis ball in the second hole, and this allowed recording of total time (TT).
Tasks Subjects performed three different tasks. Each task was performed with the right hand. A task consisted of releasing the start button in response to the onset ofthe white stimulus light, picking up the tennis ball, knocking down four barriers in the order designated by the task diagram, and placing the tennis ball in the second hole. The order in which the barriers were knocked down for each task were as follows: (a) left middle, right front, left rear, right rear (Acquisition TaskA); (b) right middle, left rear, right front, right rear (Acquisition Task B); and (c) right middle, left front, right rear, left middle (Acquisition Task C). In addition, there were three distractor tasks. The order in which the barriers were knocked down for these tasks were as follows: (a) left middle, right middle, left rear, rightfront (Distractor Task X); (b) right front, left rear, right rear, right middle (Distractor Task Y); and (c) left middle, right rear, left rear, right front (Distractor Task Z). All acquisition and distractor tasks were paired with the same white stimulus light.
Design andProcedures Subjects were randomly assigned to one of four groups. Groups were the same task, similar task, dissimilar task, and no task groups. The procedures used were designed to closely follow the sequence of psychological events underlying the reconstruction explanation as described by Lee and Magill (1985). The sequence of experimental events is shown in Figure 1. Each sequence consisted of an acquisition phase and a retention phase, and was repeated three times, once for each acquisition task. The order in which acquisition tasks were presented was counterbalanced across subjects in each group.
295
Downloaded by [University of Sydney] at 04:27 14 March 2015
SheaandWright
Acquisitionphase. The subject was seated facing the apparatus and close enough to it so as to easily pick up the ball, which was resting in the first hole and knock all of the barriers down with it. On Trial 1 the subject performed one ofthe three acquisition tasks (A, B, or C) with the diagram of this task present. In order to do this it was necessary for the subject to construct an action plan for task performance. Itwas emphasized that the task should be performed as quickly as possible without making errors. To signal the beginning of a trial, the experimenter said "ready," at which time the subject depressed the start button with the index finger of the right hand. Either 1, 2, or 3 s after the occurrence of the red warning light, the white stimulus light was presented, which signaled the initiation ofa trial. The subject performed the acquisition task as quickly as possible without making an error. KR with respect to total time (TT) in seconds, as well as correctness ofthe orderin which the barriers were knocked down, was provided immediately following this trial. This allowed modification ofthe already constructed action plan (Lee & Magill, 1985; Magill & Hall, 1990) for task performance on the next trial. The task diagram was removed from the view ofthe subject immediately followingKR. An interpolated task was performed by the subject after the completion ofTrial 1. The same task group was given an additional trial on the acquisition task with procedures identical to those of Trial 1. Instead of performing an additional trial on the acquisition task, the similar task group was given a trial on which the group performed one of the three distractor tasks (X, Y, or Z). KR was given after this trial. The rationale for this manipulation was to promote forgetting of the action plan for the acquisition task as a result of performing a task
similar to it. The dissimilar task group attempted to solve a jigsaw puzzle for 20 s instead of performing an additional trial on the acquisition task or on a distractor task. The length of this interval was the same as that necessary for the same task and similar task groups to complete their interpolated tasks. The dissimilar group was included to promote forgetting of the action plan for the acquisition task as a result ofperforming a task dissimilar to the acquisition task. This group was informed that their overall score in the experiment could be increased by putting as many puzzle pieces together as possible. The no task group rested for the 20 s instead of performing an interpolated task. This group could presumably engage in covert rehearsal of the action plan for the acquisition task during this interval. Following completion of the interpolated tasks, the subject was given a test to assess the status of the action plan in memory. This test will be referred to as the acquisition test. The procedures for this test were the same as those for Trial 1, with the exception that the acquisition task diagram was not present. In addition, KR was not provided after this test to reduce the likelihood that reconstruction of the action plan would take place. After the acquisition test, the task diagram was replaced above the white stimulus light. Trial 2 was administered using the same procedures as described for Trial L KR was provided after this trial. This procedure provided an opportunity to reconstruct the action plan for subjects who had forgotten the task (as evidenced by performance on the acquisition test). Retention-phase: Following Trial 2, the su bjectwas given a 2-min filled retention interval during which an object visualization task (Mitchell, 1983) was performed. This required the subject to imagine folding a diagram of a
Figure 1.Thesequence of experimental eventswas repeatedthree times, oncefor each acquisition task.
(c) Acquisition (d) Test Trial 2
(a) Trial 1 (b) Interpolated Tasks
..
Acquisition Phase
.. ..
(e) Retention Interval (2-min filled)
(f) (g) Retention Retention Test 1 Test 2
Retention Phase
..
Note. (a)Trial 1,on which the subjectformulated an action plan. The task diagram was present during thistrial, KR was provided, and the task diagram was removed from viewofthe subjectfollowing thistrial. (b) The Interpolated Tasks, which were intended to cause differential forgetting ofthe action plan. (c)The Acquisition Test, which assessed the amount offorgetting as a resultof performing the interpolated tasks. Thetask diagram was notpresent,and KR was notprovided -following this test. (d) Trial 2, on which the subject reconstructed the action plan. Thetask diagram was presentduring thistrial, KR was provided, and the task diagram was removed from viewofthe subjectfollowing this trial. (e)The 2-min Filled Retention Interval. (f) Retention Test 1,forwhich the task diagram was not present. KR was not provided following thistest. (g) Retention Test2,forwhich the task diagram was present. KR was not provided following thistest.
296
ROES: September 1991
Downloaded by [University of Sydney] at 04:27 14 March 2015
Shea and Wright
pattern that was presented and correctly select the final folded product from four alternatives available. The subject was informed that each alternative correctly identified would increase their overall score in the experiment. On completion of the object visualization task, the subject was given two retention tests. For Retention Test 1, the subject was required to perform the acquisition task as fast as possible in response to the white stimulus light without making an error. The task diagram was not present for this test, and KR was not provided. This test required the subject know task parameters that were the barriers comprising a task and the sequence in which these barriers were to be knocked down. It provided a measure of the speed with which an appropriate motor response could be performed strictly from memory. The testing procedures for Retention Test 2 were the same as for Retention Test 1 with the exception that the task diagram was present for this test. In this way, performance of Retention Test 2 was less dependent on knowledge of task parameters and more dependent on processes related to response execution (Adams, 1984; Magill & Hall, 1990).
Results TT and RT in seconds were recorded for Trials 1 and 2 and for the Retention Tests 1 and 2. These times were also recorded for the correctly performed tasks on the acquisition test. While TT was the total elapsed time to perform the task, RT was the time from the onset of the stimulus light to the release of the start button. Movement time (MT) scores were obtained by subtracting RT scores from TT scores. MT was therefore the time from the release of the start button to the completion of the task by placing the tennis ball in the second hole of the apparatus. RThas been taken as a measure ofthe amount of time required to prepare an action plan for execution, while MT has been used as an index of the time required for action plan implementation (Lee & Magill, 1983). In addition, the number of tasks correctly performed for the acquisition test was recorded. These scores were converted to percent correct measures for analysis. For all analyses, the region ofrejection was p < .05. The locus ofany significant effects was identified by a LSD post-hoc procedure.
reason, the findings of the acquisition test will be presented before the findings for Trial 1 and 2. Trial 1 findings reflected the processes responsible for the initial construction and execution of the action plan, while Trial 2 findings reflected the processes responsible for reconstruction and execution of the action plan. Acquisition test. A 4 x 3 (Groups x Task) analysis of variance with repeated measures on the second factor was conducted on the percent of tasks correctly performed for this test. This analysis showed that the effect ofgroups,F (3,44) = 16.53, P< .001, was significant. Subsequent post-hoc analysis indicated that the percent of correctly performed tasks for the dissimilar task (M = 14%) and similar task (M = 19%) groups did not differ significantly. The percent ofcorrectly performed tasks for these groups was significantly less than for the same task (M = 78%) and no task (M = 61 %) groups, ps < .01, however. In addition, the percen tages ofcorrectly performed tasks for the same task and no task groups were not significantly different from each other. The effects of task, F (2,88) = 1.15, and the Groups x Task interaction, F (6,88) = 1.89, were not significant. Separate one-way analyses ofvariance were performed on the median MT and RT measures for the tasks performed correctly for each subject. The data from subjects who did not respond correctly could not be used in these analyses. Table 1 presents the group MT and RT means, standard deviations, and the number of subjects in each group con tributing data for each of these analyses. It can be seen that these measures paralleled the percen t correct measures with the performance times for the similar task and dissimilar task groups being noticeably slower than for the same task and no task groups. However, the analyses of variance showed that the differences among the groups were not significant for RT, F (3,30) = .94, and MT, F (3,30) = 2.72. These findings for the acquisition test suggest that relative to the same task and no task groups, the action plan was almost completely forgotten by the similar and dissimilar task groups as a result of interpolated task performance. These groups therefore needed to engage in relatively greater reconstructive processing for performance on Trial 2. Table 1.Acquisitiontest group MT and RT means and standard deviations
Acquisition Phase The forgetting of an action plan and its subsequen t reconstruction are central to the reconstruction explanation as described by Lee and Magill (1985). It was therefore important to index the amount of action plan forgetting as a result ofinterpolated task performance to validate an interpretation of retention findings within the boundaries of the forgetting explanation. For this
ROES: September 1991
Measure MT
M
SO RT
M
SO
Different Task (n= 4)
3.66 .99 .52 .14
Groups Same Similar Task Task (n= 12) (n= 6)
3.77 2.57 .56 .29
2.39 1.13 .47 .22
No Task (n= 10)
2.57 .63 .40 .14
297
Downloaded by [University of Sydney] at 04:27 14 March 2015
Shea and Wright
Trials 1 and 2. Table 2 shows MT and RT group means and standard deviations for Trials 1 and 2. MT and RT were analyzed in a 4 x 3 x 2 (Groups x Task x Trials) multivariate analysis of variance (MANOVA) with repeated measures on the last two factors (Note 1). This analysis revealed significan t main effects ofgroups, Wi lks' s exact F (6,86) = 2.77, P < .05, task, F (4,174) = 2.96, P< .05, and trials, F (2,43) = 40.90, P< .05. None of the interactions were not significant. A subsequent univariate analysis of variance performed on MT measures revealed significant main effects of task, F (2,88) = 4.04, P < .05, and trials, F (1,44) = 83.05, p « .05. Post-hoc analysis indicated the MT for Task B (M = 3.27 s) was slower than the MT for Task A (M = 2.76 s), P< .05. In addition, MT fix Task C (M = 3.10 s) was not significantly different from MT for both Tasks A and B. Furthermore, MT for Trial 1 (M = 3.48 s) was significantly slower than MT for Trial 2 (M= 2.60 s). A univariate analysis of variance performed on RT revealed significantmain effectsofgroups,F (3,44) = 4.69, P < .01, task, F (2,88) = 3.65, P < .05, and trials, F (1,44) = 7.92, P< .01. Subsequent post-hoc analyses indicated that RT for the no task group (M = .38 s) was significantly faster than for the dissimilar task (M = .54 s), similar task (M= .50s), and same task (M= .48s) groups, ps < .05. The latter groups did not differ significantly from one another. RT for Task B (M = .52 s) was significantly slower than for Task C (M = .46 s) and Task A (M = .45 s), ps < .05. The difference in RT between Tasks C and A was not significant. In addition, RT was significantly faster for Trial 2 (M = .44 s) than for Trial I (M=.51s).
Retention Phase The acquisition test established thatagreateramount ofaction plan forgetting occurred for the similar task and dissimilar task groups than for the same task and no task groups between Trials 1 and 2. Moreover, the amount of action plan forgetting was not significantly different for the similar task and dissimilar task groups. These findings allowed a comparison of the similar task and dissimilar
task group retention test performances to determine the relative effectiveness of reconstructive processes when forgetting occurred as a result of performing similar and dissimilar interpolated tasks. In addition, a comparison of retention test performance of the similar task and dissimilar task groups with test performance of the same task and no task groups provided a determination of whether forgetting and reconstruction are necessary conditions for action plan retention. Retention Test 1. A 4 x 3 (Groups x Task) analysis of variance with repeated measures on the second factor was conducted on the percent of tasks correctly performed for this test. This analysis showed a significant main effect of groups, F (3,44) = 11.07, P< .01. Post-hoc analysis indicated the similar task (M = 75%), no task (M = 78%), and same task (M = 92%) groups correctly performed more tasks than the dissimilar task (M = 33 %) group, fr3 < .05. The differences among the similar task, no task, and same task groups were not significant. Separate one-way analysis ofvariance were performed on the median MT and RT measures for the tasks performed correctly for each subject. Table 3 present.s the MT and RT measures, standard deviations, and t.he number of subjects in each group that contribut.ed data t.o these analyses. It can be seen that the performance times for the dissimilar task and the similar task groups were slower than for the same task and no task groups. However, the analyses of variance showed that these differences among groups were not significant. for MT, F (3,39) = 1.09, and RT, F (3,39) = 1.97. Retention Test 2. A4x 3 (Groupsx Task) MAN OVA with repeated measures on the last factor was conducted on MT and RT measures for Retention Test 2. This analysis revealed a main effect of groups, Wilks's exact. F (6,86) = 2.46, P< .05. All ot.her main effect.s and interactions were not. significant. A subsequent. univariate analysis of variance performed on MT measures also revealed that the main effect of groups, F(3,44) = 3.16,p< .05,wassignificant.Post-hocanalysis indicated that MT for the dissimilar task group (M = 2.74 s) was significantly slower than for the similar task (M = 2.11), same task (M = 2.15 s), and no task (M = 2.20 s) groups, ps < .05. The latter groups did not
Table 2. Group MT and RT means and standard deviations for Trials 1and 2
Groups Dissimilar Task
Measure Trials MT
M
SO RT
M
SO 298
Similar Task
No Task
Same Task
T1
T2
M
T1
T2
M
T1
T2
M
T1
T2
M
3.55 1.08 .62 .27
3.04 .97 .47 .17
3.29 1.05 .54 .23
3.35 1.28 .51 .24
2.38 .55 .49 .21
2.86 1.09 .50 .23
3.46 1.41 .52 .21
2.31 .69 .44 .18
2.88 1.24 .48 .20
3.57 1.67 .38 .13
2.72 .98 .37 .12
3.17 1.42 .37 .12
RUES: September 1991
Shea andWright
differ significantly from one another. A univariate analysis performed on RT measures revealed no significant effects, all Fs < 1.
Downloaded by [University of Sydney] at 04:27 14 March 2015
Discussion The purpose of the present experimentwas to investigate the prediction of the reconstruction explanation (Lee & Magill, 1985; Magill & Hall, 1990) that motor retention can be facilitated by the reconstruction of an action plan that has been completely or partially forgotten. The supposition by Shea and Zimny (1983, 1988) that task similarity is a variable that limits the reten tion benefits of intertask interference, was also investigated. From the perspective of the reconstruction explanation (Lee & Magill, 1985; Magill & Hall, 1990), the nature of the task used to cause forgetting ofthe action plan should have little effect on retention. Findings of the present experiment showed performance on both Retention Tests 1 and 2 was facilitated when forgetting during acquisition occurred as the result of performance ofa similar distractor task (similar task group) rather than a dissimilar distractor task (dissimilar task group). These findings show forgetting and subsequent reconstruction of the action plan for a task is in itselfnot sufficient for improved reten tion (Lee & Magill, 1985) and suggest the mechanism responsible for improved recall following forgetting is something more than strengthening response compilation and execution processes. Shea and Zimny (l983, 1988) emphasized the importance of a context from which elaborative and distinctive processing can be performed, and the presence of a similar task in working memory at the time of reconstruction appears to have provided this context for the similar task group. Retention Test 1 emphasized knowledge of the barriers comprising a task and the sequence in which these barriers were to be knocked down, as well as the speed with which an appropriate response could be executed. The finding that the similar task group correctly performed more tasks than the dissimilar group is evidence Table 3. Retention TestGroup 1 MT and RT meansand standard
deviations Different Task Measure MT
M
SD RT
M
SD
Groups Similar Same Task Task
No Task
(n=4)
(n= 6)
(n= 12)
(n= 10)
2.58 .46 .44 .20
2.85 1.01 .51 .21
2.22 .77 .46 .13
2.42 1.00 .35 .07
ROES: September 1991
that reconstruction with the presence ofa similar task in working memory promoted a more complete encoding of individual task parameters. Retention Test 2 performance was less dependent on knowledge of task parameters and placed emphasis on processes related to response execution. MT was faster for the similar task group than for the dissimilar task group. This finding, in combination with the findings for Retention Test 1, suggests that the greater detailing of task parameters at the time of reconstruction (Trial 2) by the similar task group facilitated response execution. That is, fewer corrections during the response were necessary because of the relative accuracy of the initially selected response units. This in terpretation is consisten twith recen tfindings that while knowledge of task parameters is separate from mechanisms responsible for response execution, these can become blended to facilitate task performance (Willingham, Nissen, & Bullemer, 1989). The concurrent presence of two similar tasks in working memory was in itself not sufficient to engender a more complete processing of task characteristics. Ifthis had been the case, the similar task group should have benefited from the concurrent presence of the acquisition and distractor tasks in working memory prior to the acquisition test, and performance on the acquisition test should have been superior for the similar task than for the dissimilar task group. These groups did not differ on performance of this test, however. An explanation for this finding is that the acquisition test provided the impetus for subjects to engage in more extensive processing of the tasks on Trial 2. Performance on Retention Tests 1 and 2 for the same task and no task groups was not significantly different from performance of the similar task group, but was significantly better than the dissimilar task group. It is difficult to reconcile this finding with the reconstruction explanation (Lee & Magill, 1985; Magill & Hall, 1990). The reconstruction perspective would not predict these findings since the performance of the same task and no task groups demonstrated relatively less forgetting than the similar task and dissimilar task groups as indexed by the acquisition test. Thus, less reconstructive processing should have been performed by the same task and no task groups, which should have led to poorer retention performance. However, it is possible the provision of an additional trial of the criterion task (same task group) or the opportunity to covertly rehearse the criterion task (no task group) was sufficient to allow encoding of additional task related parameters, which increased the memorability of the task. This would be similar to subjects engaging in a form of intratask processing, which is an important component of the elaboration perspective, especially when addressing the initial stages oflearning a task (Limons & Shea, 1988). The discrepancy between these findings and others (Lee & Weeks, 1987; Weeks et aI., 1987) that have pro-
299
Downloaded by [University of Sydney] at 04:27 14 March 2015
Shea andWright
vided support for the forgetting explanation might be attributed to differences in experimental procedures as well as in the tasks used. Timing (Weeks et al., 1987) and linear positioning (Lee & Weeks, 1987) tasks might not be dependent on the same processes for performance as the tasks used in the present experiment. The present tasks were similar to those used in the first experiments reporting contextual interference effects (Lee & Magill, 1983; Shea & Morgan, 1979), and performance of these tasks is notably dependent on the retention of visual information that is centrally rehearsable (Posner, 1967). The present experimen t was designed as a test ofone fundamental process (i.e., reconstruction ofa previously forgotten action plan) thought to be responsible for the blocked and random group retention differences found in most contextual interference experiments (Lee & Magill, 1985; Magill & Hall, 1990). As such, this experiment should not be construed as providing a complete account for contextual interference findings. This point is underscored by the fact retention findings for the groups used in this experiment did not parallel those of their counterparts in contextual interference experiments. The no task and same task groups correspond to the blocked practice (low contextual interference) schedule groups and the dissimilar task, and similar task groups correspond to the random practice (high contextual interference) schedule groups. Iffindingsfor the present study had paralleled those for contextual interference experiments, retention performance for either the dissimilar task or similar task group should have been significantly better than for both the no task and same task groups. This was not the case, and the no task and same task groups performed as well as the similar task group and significantly better than the dissimilar group. Other processes in addition to those investigated in the present experiment would appear to be operating in contextual interference experiments. These processes, either singularly or in combination, might account for the discrepancies in findings from these experiments and the present one. The present experiment utilized a single-task/two-trial paradigm. This contrasts sharplywith the multiple-task/multiple-trial paradigm used in traditional contextual in terference experiments. The reduced complexity ofthe present experimental paradigm migh t have contributed to the retention performance of the same task and no task groups by removing detrimental effects of retroactive and/or proactive interference (Magill & Hall, 1990), which would be experienced by blocked practice schedule subjects in contextual interference experiments.
300
References Adams,]. A. (1984). Learning of movement sequences. Psychological Bulletin, 96, 3-28. Battig, W. F. (1979). The flexibility of human memory. In L. S. Cermak & F. 1. M. Craik (Eds.), Levels of processing and human memory (pp. 2344). Hillsdale, NJ.: Erlbaum. Kelso,]. A. S., & Wallace, S. A. (1978). Conscious mechanisms in movement. In G. E. Stelmach (Ed.), Information processing in motor control and learning (pp. 79-116). New York: Academic Press. Lee, T. (1988). Transfer-appropriate processing: A framework for conceptualizing practice effects in motor learning. In O. G. Meijer & K Roth (Eds.), Complex movement behavior: Themotor-actioncontroversy (pp. 201-215) .Amsterdam: NorthIiolland. Lee, T. D., & Magill, R. A. (1983). The locus of contextual in terference in motor-skill acquisition.JournalofExperimental Psychology: Learning, Memory, and Cognition, 9,730-746. Lee, T. D., & Magill, R. A. (1985). Can forgetting facilitate skill acquisition? In D. Goodman, R. B. Wilberg, & 1. M. Franks (Eds.) , Differing perspectives in motor learning, memory, and control (pp. 3-22). Amsterdam: North-Holland. Lee, T. D., & Weeks, D.]. (1987). The beneficial influence of forgetting on short-term memory retention of movement information. Human Movement Science, 6, 233-245. Limons, E., & Shea,]. B. (1988). Deficient processing in learning and performance. In A. M. Colley &]. R. Beech (Eds.), Cognition and action in skilled behatnor (pp. 333-347) . Amsterdam: North-Holland. Magill, R. A. (1973). The post-KR interval: Time and activity effects and the relationship of motor short-term memory theory. Journal ofMotor Behavior, 5, 49-56. Magill, R. A. (1989). Motor learning: Concepts and applications (3rd ed.). Dubuque, Iowa: Wm. C. Brown. Magill, R. A., & Hall, K. G. (1990). A review of the contextual interference effect in motor skill acquisition. Human Movement Science, 9,241-289. Marshall, P. H., Jones, M. '1'., & Sheehan, E. M. (1977). The spacing effect in short-term motor memory: The differential attention hypothesis. Journal ofMotor Behavior, 9, 119126. Marshall, P. H., Wyatt, S. L., Moore, S. A., & Sigman, S. E. (1975). Interrepetition interval in short-term memory. Perceptual and Motor Skills, 40, 535-538. Mitchell.], V. (Ed.) (1983). Tests in print III: An index to tests, test reviews, and the literature on specific tests. Lincoln: University of Nebraska Press. Posner, M. 1. (1967). Characteristics of visual and kinesthetic memory codes. Journal ofExperimental Psychology, 75, 103107. Shea,]. B., & Morgan, R. L. (1979). Contextual interference effects on the acquisition, retention and transfer of a motor skill.Journal ofExperimental Psychology: Human LearningandMemory,5,179-187.
RaES: September 1991
Shea andWright
Shea,j. B., & Zimny, S. T. (1983). Context effects in memory and learning movementinformation. InR. A. Magill (Ed.), Memory and control of action (pp. 345-366). Amsterdam: North-Holland. Shea,j. B., & Zimny, S. T. (1988). Knowledge incorporation in motor representation. In O. G. Meijer & K Roth (Eds.), Complex movement behavior: The motor-action controversy (pp. 289-314). Amsterdam: North-Holland. Weeks, D. J. Lee, T. D., & Elliott, D. (1987). Differential forgetting and spacing effects in short-term motor retention. Journal ofHuman Movement Studies, 13, 309-321. Willingham, D. B., Nissen, M. J. & Bullemer, P. (1989). On the development of procedural knowledge. Journal of Experimental Psycholbgy: Learning, Memary and Cognition, 15,
1047-1060.
Note 1. Wilks's criterion (A) was used for all MANOVA tests. Lambda was converted to exact Fs by the following formula: ExactF= (l-SQRT[A])/SQRT(A)' (NE+Q:P-1)/P with 2P and 2 (NE+Q-P-1 ) degrees of freedom, where NE= df for error, Q= df for hypothesis, and P= rank of hypothesis + error matrix.
Authors' Notes
Downloaded by [University of Sydney] at 04:27 14 March 2015
The authors thank Timothy D. Lee, Charles H. Shea, and two anonymous reviewers for their comments on earlier drafts of the manuscript.
ROES: September 1991
301
Research Quarterly for Exercise and Sport Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/urqe20
When Forgetting Benefits Motor Retention a
John B. Shea & David L. Wright
b
a
Motor Behavior Laboratory in the Department of Exercise and Sport Science , Pennsylvania State University , USA b
Department of Health and Physical Education , Texas A & M University , USA Published online: 26 Feb 2013.
To cite this article: John B. Shea & David L. Wright (1991) When Forgetting Benefits Motor Retention, Research Quarterly for Exercise and Sport, 62:3, 293-301, DOI: 10.1080/02701367.1991.10608726 To link to this article: http://dx.doi.org/10.1080/02701367.1991.10608726
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Research Quarterlyfor Exerciseand Sport © 1991 by the American Alliance for Health,
Physical Education, Recreation and Dance Vol. 62, No. 3, pp. 293·301
When Forgetting Benefits Motor Retention
Downloaded by [University of Sydney] at 04:27 14 March 2015
John B. Shea andDavid L. Wright Recent research (Lee & Weeks, 1987; Weeks, Lee, & Elliott, 1987) investigating the processes responsible for the contextual interference phenomenon has used a modified short-term motorretention paradigm to support the reconstruction explanation (Lee & Magill, 1985; Magill, 1989; Magill & HaU, 1990). The presentexperimentwas an extension ofthese experiments in whichforgetting ofan acquisition task was induced through performance of eithera similar or dissimilar distractortask during the intertrial interval. The effects of an extra practicetrial with the acquisition task as well as no activity during the intertrial interval were also investigated. In addition, forgetting ofthe acquisition task was assessed prior to a reconstruction trial, which immediately preceded a 2-min filled retention interval. Both similar and dissimilar distractortasks caused equivalent amounts offorgetting ofthe acquisition task prior to the reconstruction trial. However, retention of the acquisition task was significantly improved if its reconstruction occurred followingforgetting due to interferencefrom performance ofa similar distractortask. Thesefindings suggestforgetting and subsequent reconstruction alone are not sufficient for improved retention. Theseprocesses must occurin the context of a similar task for improved retention.
Key words: contextual interference, forgetting, reconstruction, retention
M
o tor learning researchers have become increasinglyin terested in con textual in terference (Battig, 1979). Contextual interference refers to the finding that practice of multiple tasks in a random (high contextual interference) practice schedule results in greater retention and transfer than when tasks are practiced in a blocked practice schedule (low contextual interference) in which practice ofone task is completed before practice on a different task is given (Lee & Magill, 1983; Shea & Morgan, 1979). Lee and Magill (1985) have proposed a reconstruction explanation for contextual interference. According to this explanation, a knowledge of results (KR) faction plan (Lee & Magill, 1985) or, more recently, action plan (Magill & Hall, 1990) for a particular task is forgotten as a result of intervening trials on other tasks in a random practice schedule. This forces the learner to engage in
John B. Sheais anassociate professor anddirector of theMotor Behavior Laboratory inthe Department of Exercise andSport Scienceat Pennsylvania State University. David L Wright is an assistant professor inthe Department of Health andPhysical Education at Texas A & M University. Address requests for reprints to John B. Shea, The Pennsylvania StateUniversity, Motor Behavior Laboratory, 131 White Building, University Park, PA 16802. Submitted:June 25, 1990 Revision accepted: December 18, 1990
RDES: September 1991
more effortful reconstructive processing to generate the action plan for subsequent performance of the task. Reconstructive processing is emphasized in this explanation because it is assumed to incorporate the cognitive activities similar to those required when performing retention and transfer tests (Lee, 1988). Magill and Hall (1990) have recently described the reconstruction process as the retrieval of the appropriate motor program representing an action and then adding to it the parameters specific to the constraints and goal of the task to be performed. In contrast to a random practice schedule, there is little opportunity for forgetting in a blocked practice schedule. The action plan is present in working memory and can be executed on successive trials with little or no reconstructive activity. Thus, any practice schedule that promotes complete or partial forgetting of a previously used action plan between repetitions of a goal directed movement will depress acquisition but promote retention performance relative to conditions in which action plan forgetting does not occur (Lee & Magill, 1985, p. 8; Magill & Hall, 1990). Support for the prediction that forgetting and reconstruction of an action plan enhances retention can be garnered from an experiment by Marshall, Wyatt, Moore, and Sigman (1975). A linear positioning task was used to examine the effect of either one or seven repeti tions of a criterion movement spaced at intervals of 5 or 60 s on subsequent recall following a 30-s retention interval. The 60-s spaced condition was assumed to have induced greater forgetting of previous presentations of the criterion movement than the 5-s spaced condition. Recall as measured by absolute error (AE) was more accurate for
293
Downloaded by [University of Sydney] at 04:27 14 March 2015
Shea andWright
the 60-s spaced condition than for the 5-s spaced condition. However, this finding was restricted to the sevenrepetition condition. If forgetting of previous presentations was the main determinan t ofsubsequent recall, this effect should also have been evidenced for the onerepetition condition. Furthermore, this finding was not obtained for constant error (CE) measures. In three subsequent experiments, Marshall,Jones, and Sheehan (1977) increased the number of test trials from that used in the Marshall et al. (1975) experiment. This enabled them to obtain evidence ofgreater retention for the 60-s spaced condition for both six and one repetition conditions. However, these effects were found for CE only but not for variable error (VE) or AE. Unfortunately, interfering activity that might have facilitated forgetting was not interpolated between repetitions or during the retention interval in these experiments. In addition, no independent test offorgetting was given between repetitions, making the assumption that forgetting occurred during these intervals a speculative one. Weeks, Lee, and Elliott (1987) investigated the effect different levels of initial forgetting had on subsequent recall test performance. Subjects made a self-selected criterion timing movement from three predetermined ranges within which they had previously been trained. An initial reproduction of this movement was made either immediately, after a 20-s unfilled interval, after a 20-s interval of counting backward by threes, or after a 20-s interval of counting backward by sevens. A second reproduction of the criterion movement was attempted following a 20-s filled retention interval. The findings offered some support for the notion that increased initial forgetting would subsequently lead to improved retention. VE scores generally revealed that recall performance on the second reproduction was less variable with increased error on the initial reproduction. However, response accuracy findings were notconsistentwith these VE findings. Absolute constant error (ICEI) was greater for the initial reproduction of the criterion movement when itwas made after the unfilled interval and counting backward by threes and sevens than when this reproduction was made immediately. However, the second reproduction of the criterion movement was more accurate than the immediate reproduction condition only after counting backward by threes and sevens. Further support for the reconstruction explanation was provided by Lee and Weeks (1987) in an experiment in which a linear positioning task was used. Subjects were presented a repetition of a criterion movement (by moving to a barrier at the criterion position) or were required to estimate the criterion movement either immediately or after 20 s of the same interfering activity. The results indicated that a delayed estimate led to larger error (VE and ICEI) than an immediate estimate. This led Lee and Weeks to assume that for either delayed repetition or estimation, more information about the criterion is for-
294
gotten than at the time of immediate repetition or estimation. This assumption is critical to their interpretation of subsequent recall performance following immediate and delayed repetition. This is because all repetitions were movements to a barrier, and an independent measure of forgetting could not be obtained for repetition conditions. The results of subsequent criterion movement recall indicated the delayed repetition condition had significantly less variability (VE) than the immediate repetition condition. However, no differences in accuracy (ICEI) were obtained between the delayed and immediate repetition groups. The reconstruction explanation clearly states it is the KR/action plan (Lee & Magill, 1985, p. 19) or action plan (Magill & Hall, 1990) that is forgotten and reconstructed. However, Weeks etal. (1987), and Lee and Weeks (1987) did not provide KRatanystage oftheir studies. In addition, subjects in the Weeks etal. (1987) experiment were never given additional experience with the criterion movement after forgetting occurred so they could reconstruct it before the retention interval. Furthermore, the procedure used to cause reconstructive processing by Lee and Weeks (1987) might have resulted in processing that was inappropriate for recall purposes. The assumption that movemen t to a barrier which defined the criterion position in the repetition condition caused subjects to reconstruct the forgotten action plan must be viewed with caution. It has been shown that constrained movements (movements in which the criterion position is defined by movement to a barrier) and unconstrained movements (movements in which the subject actively estimates the criterion position) do not share the same type ofprocessing (Kelso & Wallace, 1978). In the present case, this would suggest the reconstructive processing at the time of repetition of the criterion position might have been inappropriate for recall purposes in the repetition condition (Lee, 1988). Thus the procedures used byWeeksetal. (1987) andLeeand Weeks (1987) raise questions concerning the ability ofsubjects to correctly reconstruct a forgotten action plan. The present experiment consisted of a more definitive test of the reconstruction explanation (Lee & Magill, 1985; Magill & Hall, 1990). The paradigm was similar to the modified motor short-term memory paradigm used in earlier studies to investigate the reconstruction explanation (Lee & Weeks, 1987; Magill, 1989; Weeks et aI., 1987). According to Magill (1973), this kind of an approach (i.e., the modified motor short-term memory paradigm) might avoid the problems associated with reinforcement on every trial obscuring the mechanisms operating on each trial in multiple trial procedures. This approach, then, was ideally suited for the purpose of the present experimen t, which was restric ted to the investigation of the reconstruction process (Magill & Hall, 1990) associated with forgetting. No attempt was made to address the development of the reconstruction process over multiple practice trials with different tasks as is the
RUES: September 1991
Downloaded by [University of Sydney] at 04:27 14 March 2015
Shea and Wright
case in contextual interference experiments. It must therefore be emphasized that while the present experiment was inspired by the reconstruction explanation, its findings would not be expected to necessarily have complete congruence with typical contextual interference effects. The tasks used in the present experiment were similar to those used in the first experiments reporting contextual interference effects (Lee & Magill, 1983; Shea & Morgan, 1979). Subjects received two acquisition trials on which KR was provided before a retention interval. On the first trial an action plan for task performance was constructed. Components of the action plan forgotten as the result of interpolated task conditions during the intertrial interval could be reconstructed on the second trial. The importance of task similarity for retention benefits resulting from action plan reconstruction was also examined in the present experiment. This was accomplished by including conditions in which betweentrial forgetting was induced by performance of either a similar or dissimilar task during the intertrial interval. Shea and Zimny (1983, 1988) have implicated task similarity as a variable that limits the benefits of intertask interference experienced during acquisition on subsequent retention. This view holds that maximum retention benefits occur when similar rather than dissimilar tasks are practiced during acquisition. In contrast, the reconstruction explanation does not impart great importance to the nature of the intervening activity that causes forgetting between trials because the important issue for this hypothesis is that the learner reconstructs the forgotten action plan used for task performance (Lee & Magill, 1985; Magill & Hall, 1990).
Method Subjects Subjects were 52 right-handed students at The Pennsylvania State University, who received class credit for their participation. Informed consentwas obtained from each subject prior to testing. Four subjects were excluded from the experiment because of their failure to follow instructions.
Apparatus The apparatus was similar to that described by Shea and Morgan (1979). It consisted of six freely movable barriers attached to the top surface of a platform. Three barriers were positioned on each side of the apparatus so that they faced the midline of the apparatus. A button switch, referred to as the start button, was located on the midline and front of the apparatus. Directly to the rear of
ROES: September 1991
the start button were two holes large enough to support a tennis ball but small enough to keep the ball from falling through. A ball rested in the first hole. A red warning light and a white stimulus light were mounted at the rear of the apparatus. A plywood barrier was located directly behind the lights on which the diagram of the task being performed could be displayed. Only one diagram was present at any time and was located directly above the white stimulus light. Two Hunter Klockcounters (model 220 C) were located behind the plywood barrier and were used to time components of the subject's performance. A control panel allowed the experimenter to present the warning light followed by the stimulus light for each trial. The two timers were initiated at the onset of the stimulus light. One timer stopped on release of the start button, and this constituted reaction time (RT). The other timer was stopped on placemen t of the tennis ball in the second hole, and this allowed recording of total time (TT).
Tasks Subjects performed three different tasks. Each task was performed with the right hand. A task consisted of releasing the start button in response to the onset ofthe white stimulus light, picking up the tennis ball, knocking down four barriers in the order designated by the task diagram, and placing the tennis ball in the second hole. The order in which the barriers were knocked down for each task were as follows: (a) left middle, right front, left rear, right rear (Acquisition TaskA); (b) right middle, left rear, right front, right rear (Acquisition Task B); and (c) right middle, left front, right rear, left middle (Acquisition Task C). In addition, there were three distractor tasks. The order in which the barriers were knocked down for these tasks were as follows: (a) left middle, right middle, left rear, rightfront (Distractor Task X); (b) right front, left rear, right rear, right middle (Distractor Task Y); and (c) left middle, right rear, left rear, right front (Distractor Task Z). All acquisition and distractor tasks were paired with the same white stimulus light.
Design andProcedures Subjects were randomly assigned to one of four groups. Groups were the same task, similar task, dissimilar task, and no task groups. The procedures used were designed to closely follow the sequence of psychological events underlying the reconstruction explanation as described by Lee and Magill (1985). The sequence of experimental events is shown in Figure 1. Each sequence consisted of an acquisition phase and a retention phase, and was repeated three times, once for each acquisition task. The order in which acquisition tasks were presented was counterbalanced across subjects in each group.
295
Downloaded by [University of Sydney] at 04:27 14 March 2015
SheaandWright
Acquisitionphase. The subject was seated facing the apparatus and close enough to it so as to easily pick up the ball, which was resting in the first hole and knock all of the barriers down with it. On Trial 1 the subject performed one ofthe three acquisition tasks (A, B, or C) with the diagram of this task present. In order to do this it was necessary for the subject to construct an action plan for task performance. Itwas emphasized that the task should be performed as quickly as possible without making errors. To signal the beginning of a trial, the experimenter said "ready," at which time the subject depressed the start button with the index finger of the right hand. Either 1, 2, or 3 s after the occurrence of the red warning light, the white stimulus light was presented, which signaled the initiation ofa trial. The subject performed the acquisition task as quickly as possible without making an error. KR with respect to total time (TT) in seconds, as well as correctness ofthe orderin which the barriers were knocked down, was provided immediately following this trial. This allowed modification ofthe already constructed action plan (Lee & Magill, 1985; Magill & Hall, 1990) for task performance on the next trial. The task diagram was removed from the view ofthe subject immediately followingKR. An interpolated task was performed by the subject after the completion ofTrial 1. The same task group was given an additional trial on the acquisition task with procedures identical to those of Trial 1. Instead of performing an additional trial on the acquisition task, the similar task group was given a trial on which the group performed one of the three distractor tasks (X, Y, or Z). KR was given after this trial. The rationale for this manipulation was to promote forgetting of the action plan for the acquisition task as a result of performing a task
similar to it. The dissimilar task group attempted to solve a jigsaw puzzle for 20 s instead of performing an additional trial on the acquisition task or on a distractor task. The length of this interval was the same as that necessary for the same task and similar task groups to complete their interpolated tasks. The dissimilar group was included to promote forgetting of the action plan for the acquisition task as a result ofperforming a task dissimilar to the acquisition task. This group was informed that their overall score in the experiment could be increased by putting as many puzzle pieces together as possible. The no task group rested for the 20 s instead of performing an interpolated task. This group could presumably engage in covert rehearsal of the action plan for the acquisition task during this interval. Following completion of the interpolated tasks, the subject was given a test to assess the status of the action plan in memory. This test will be referred to as the acquisition test. The procedures for this test were the same as those for Trial 1, with the exception that the acquisition task diagram was not present. In addition, KR was not provided after this test to reduce the likelihood that reconstruction of the action plan would take place. After the acquisition test, the task diagram was replaced above the white stimulus light. Trial 2 was administered using the same procedures as described for Trial L KR was provided after this trial. This procedure provided an opportunity to reconstruct the action plan for subjects who had forgotten the task (as evidenced by performance on the acquisition test). Retention-phase: Following Trial 2, the su bjectwas given a 2-min filled retention interval during which an object visualization task (Mitchell, 1983) was performed. This required the subject to imagine folding a diagram of a
Figure 1.Thesequence of experimental eventswas repeatedthree times, oncefor each acquisition task.
(c) Acquisition (d) Test Trial 2
(a) Trial 1 (b) Interpolated Tasks
..
Acquisition Phase
.. ..
(e) Retention Interval (2-min filled)
(f) (g) Retention Retention Test 1 Test 2
Retention Phase
..
Note. (a)Trial 1,on which the subjectformulated an action plan. The task diagram was present during thistrial, KR was provided, and the task diagram was removed from viewofthe subjectfollowing thistrial. (b) The Interpolated Tasks, which were intended to cause differential forgetting ofthe action plan. (c)The Acquisition Test, which assessed the amount offorgetting as a resultof performing the interpolated tasks. Thetask diagram was notpresent,and KR was notprovided -following this test. (d) Trial 2, on which the subject reconstructed the action plan. Thetask diagram was presentduring thistrial, KR was provided, and the task diagram was removed from viewofthe subjectfollowing this trial. (e)The 2-min Filled Retention Interval. (f) Retention Test 1,forwhich the task diagram was not present. KR was not provided following thistest. (g) Retention Test2,forwhich the task diagram was present. KR was not provided following thistest.
296
ROES: September 1991
Downloaded by [University of Sydney] at 04:27 14 March 2015
Shea and Wright
pattern that was presented and correctly select the final folded product from four alternatives available. The subject was informed that each alternative correctly identified would increase their overall score in the experiment. On completion of the object visualization task, the subject was given two retention tests. For Retention Test 1, the subject was required to perform the acquisition task as fast as possible in response to the white stimulus light without making an error. The task diagram was not present for this test, and KR was not provided. This test required the subject know task parameters that were the barriers comprising a task and the sequence in which these barriers were to be knocked down. It provided a measure of the speed with which an appropriate motor response could be performed strictly from memory. The testing procedures for Retention Test 2 were the same as for Retention Test 1 with the exception that the task diagram was present for this test. In this way, performance of Retention Test 2 was less dependent on knowledge of task parameters and more dependent on processes related to response execution (Adams, 1984; Magill & Hall, 1990).
Results TT and RT in seconds were recorded for Trials 1 and 2 and for the Retention Tests 1 and 2. These times were also recorded for the correctly performed tasks on the acquisition test. While TT was the total elapsed time to perform the task, RT was the time from the onset of the stimulus light to the release of the start button. Movement time (MT) scores were obtained by subtracting RT scores from TT scores. MT was therefore the time from the release of the start button to the completion of the task by placing the tennis ball in the second hole of the apparatus. RThas been taken as a measure ofthe amount of time required to prepare an action plan for execution, while MT has been used as an index of the time required for action plan implementation (Lee & Magill, 1983). In addition, the number of tasks correctly performed for the acquisition test was recorded. These scores were converted to percent correct measures for analysis. For all analyses, the region ofrejection was p < .05. The locus ofany significant effects was identified by a LSD post-hoc procedure.
reason, the findings of the acquisition test will be presented before the findings for Trial 1 and 2. Trial 1 findings reflected the processes responsible for the initial construction and execution of the action plan, while Trial 2 findings reflected the processes responsible for reconstruction and execution of the action plan. Acquisition test. A 4 x 3 (Groups x Task) analysis of variance with repeated measures on the second factor was conducted on the percent of tasks correctly performed for this test. This analysis showed that the effect ofgroups,F (3,44) = 16.53, P< .001, was significant. Subsequent post-hoc analysis indicated that the percent of correctly performed tasks for the dissimilar task (M = 14%) and similar task (M = 19%) groups did not differ significantly. The percent ofcorrectly performed tasks for these groups was significantly less than for the same task (M = 78%) and no task (M = 61 %) groups, ps < .01, however. In addition, the percen tages ofcorrectly performed tasks for the same task and no task groups were not significantly different from each other. The effects of task, F (2,88) = 1.15, and the Groups x Task interaction, F (6,88) = 1.89, were not significant. Separate one-way analyses ofvariance were performed on the median MT and RT measures for the tasks performed correctly for each subject. The data from subjects who did not respond correctly could not be used in these analyses. Table 1 presents the group MT and RT means, standard deviations, and the number of subjects in each group con tributing data for each of these analyses. It can be seen that these measures paralleled the percen t correct measures with the performance times for the similar task and dissimilar task groups being noticeably slower than for the same task and no task groups. However, the analyses of variance showed that the differences among the groups were not significant for RT, F (3,30) = .94, and MT, F (3,30) = 2.72. These findings for the acquisition test suggest that relative to the same task and no task groups, the action plan was almost completely forgotten by the similar and dissimilar task groups as a result of interpolated task performance. These groups therefore needed to engage in relatively greater reconstructive processing for performance on Trial 2. Table 1.Acquisitiontest group MT and RT means and standard deviations
Acquisition Phase The forgetting of an action plan and its subsequen t reconstruction are central to the reconstruction explanation as described by Lee and Magill (1985). It was therefore important to index the amount of action plan forgetting as a result ofinterpolated task performance to validate an interpretation of retention findings within the boundaries of the forgetting explanation. For this
ROES: September 1991
Measure MT
M
SO RT
M
SO
Different Task (n= 4)
3.66 .99 .52 .14
Groups Same Similar Task Task (n= 12) (n= 6)
3.77 2.57 .56 .29
2.39 1.13 .47 .22
No Task (n= 10)
2.57 .63 .40 .14
297
Downloaded by [University of Sydney] at 04:27 14 March 2015
Shea and Wright
Trials 1 and 2. Table 2 shows MT and RT group means and standard deviations for Trials 1 and 2. MT and RT were analyzed in a 4 x 3 x 2 (Groups x Task x Trials) multivariate analysis of variance (MANOVA) with repeated measures on the last two factors (Note 1). This analysis revealed significan t main effects ofgroups, Wi lks' s exact F (6,86) = 2.77, P < .05, task, F (4,174) = 2.96, P< .05, and trials, F (2,43) = 40.90, P< .05. None of the interactions were not significant. A subsequent univariate analysis of variance performed on MT measures revealed significant main effects of task, F (2,88) = 4.04, P < .05, and trials, F (1,44) = 83.05, p « .05. Post-hoc analysis indicated the MT for Task B (M = 3.27 s) was slower than the MT for Task A (M = 2.76 s), P< .05. In addition, MT fix Task C (M = 3.10 s) was not significantly different from MT for both Tasks A and B. Furthermore, MT for Trial 1 (M = 3.48 s) was significantly slower than MT for Trial 2 (M= 2.60 s). A univariate analysis of variance performed on RT revealed significantmain effectsofgroups,F (3,44) = 4.69, P < .01, task, F (2,88) = 3.65, P < .05, and trials, F (1,44) = 7.92, P< .01. Subsequent post-hoc analyses indicated that RT for the no task group (M = .38 s) was significantly faster than for the dissimilar task (M = .54 s), similar task (M= .50s), and same task (M= .48s) groups, ps < .05. The latter groups did not differ significantly from one another. RT for Task B (M = .52 s) was significantly slower than for Task C (M = .46 s) and Task A (M = .45 s), ps < .05. The difference in RT between Tasks C and A was not significant. In addition, RT was significantly faster for Trial 2 (M = .44 s) than for Trial I (M=.51s).
Retention Phase The acquisition test established thatagreateramount ofaction plan forgetting occurred for the similar task and dissimilar task groups than for the same task and no task groups between Trials 1 and 2. Moreover, the amount of action plan forgetting was not significantly different for the similar task and dissimilar task groups. These findings allowed a comparison of the similar task and dissimilar
task group retention test performances to determine the relative effectiveness of reconstructive processes when forgetting occurred as a result of performing similar and dissimilar interpolated tasks. In addition, a comparison of retention test performance of the similar task and dissimilar task groups with test performance of the same task and no task groups provided a determination of whether forgetting and reconstruction are necessary conditions for action plan retention. Retention Test 1. A 4 x 3 (Groups x Task) analysis of variance with repeated measures on the second factor was conducted on the percent of tasks correctly performed for this test. This analysis showed a significant main effect of groups, F (3,44) = 11.07, P< .01. Post-hoc analysis indicated the similar task (M = 75%), no task (M = 78%), and same task (M = 92%) groups correctly performed more tasks than the dissimilar task (M = 33 %) group, fr3 < .05. The differences among the similar task, no task, and same task groups were not significant. Separate one-way analysis ofvariance were performed on the median MT and RT measures for the tasks performed correctly for each subject. Table 3 present.s the MT and RT measures, standard deviations, and t.he number of subjects in each group that contribut.ed data t.o these analyses. It can be seen that the performance times for the dissimilar task and the similar task groups were slower than for the same task and no task groups. However, the analyses of variance showed that these differences among groups were not significant. for MT, F (3,39) = 1.09, and RT, F (3,39) = 1.97. Retention Test 2. A4x 3 (Groupsx Task) MAN OVA with repeated measures on the last factor was conducted on MT and RT measures for Retention Test 2. This analysis revealed a main effect of groups, Wilks's exact. F (6,86) = 2.46, P< .05. All ot.her main effect.s and interactions were not. significant. A subsequent. univariate analysis of variance performed on MT measures also revealed that the main effect of groups, F(3,44) = 3.16,p< .05,wassignificant.Post-hocanalysis indicated that MT for the dissimilar task group (M = 2.74 s) was significantly slower than for the similar task (M = 2.11), same task (M = 2.15 s), and no task (M = 2.20 s) groups, ps < .05. The latter groups did not
Table 2. Group MT and RT means and standard deviations for Trials 1and 2
Groups Dissimilar Task
Measure Trials MT
M
SO RT
M
SO 298
Similar Task
No Task
Same Task
T1
T2
M
T1
T2
M
T1
T2
M
T1
T2
M
3.55 1.08 .62 .27
3.04 .97 .47 .17
3.29 1.05 .54 .23
3.35 1.28 .51 .24
2.38 .55 .49 .21
2.86 1.09 .50 .23
3.46 1.41 .52 .21
2.31 .69 .44 .18
2.88 1.24 .48 .20
3.57 1.67 .38 .13
2.72 .98 .37 .12
3.17 1.42 .37 .12
RUES: September 1991
Shea andWright
differ significantly from one another. A univariate analysis performed on RT measures revealed no significant effects, all Fs < 1.
Downloaded by [University of Sydney] at 04:27 14 March 2015
Discussion The purpose of the present experimentwas to investigate the prediction of the reconstruction explanation (Lee & Magill, 1985; Magill & Hall, 1990) that motor retention can be facilitated by the reconstruction of an action plan that has been completely or partially forgotten. The supposition by Shea and Zimny (1983, 1988) that task similarity is a variable that limits the reten tion benefits of intertask interference, was also investigated. From the perspective of the reconstruction explanation (Lee & Magill, 1985; Magill & Hall, 1990), the nature of the task used to cause forgetting ofthe action plan should have little effect on retention. Findings of the present experiment showed performance on both Retention Tests 1 and 2 was facilitated when forgetting during acquisition occurred as the result of performance ofa similar distractor task (similar task group) rather than a dissimilar distractor task (dissimilar task group). These findings show forgetting and subsequent reconstruction of the action plan for a task is in itselfnot sufficient for improved reten tion (Lee & Magill, 1985) and suggest the mechanism responsible for improved recall following forgetting is something more than strengthening response compilation and execution processes. Shea and Zimny (l983, 1988) emphasized the importance of a context from which elaborative and distinctive processing can be performed, and the presence of a similar task in working memory at the time of reconstruction appears to have provided this context for the similar task group. Retention Test 1 emphasized knowledge of the barriers comprising a task and the sequence in which these barriers were to be knocked down, as well as the speed with which an appropriate response could be executed. The finding that the similar task group correctly performed more tasks than the dissimilar group is evidence Table 3. Retention TestGroup 1 MT and RT meansand standard
deviations Different Task Measure MT
M
SD RT
M
SD
Groups Similar Same Task Task
No Task
(n=4)
(n= 6)
(n= 12)
(n= 10)
2.58 .46 .44 .20
2.85 1.01 .51 .21
2.22 .77 .46 .13
2.42 1.00 .35 .07
ROES: September 1991
that reconstruction with the presence ofa similar task in working memory promoted a more complete encoding of individual task parameters. Retention Test 2 performance was less dependent on knowledge of task parameters and placed emphasis on processes related to response execution. MT was faster for the similar task group than for the dissimilar task group. This finding, in combination with the findings for Retention Test 1, suggests that the greater detailing of task parameters at the time of reconstruction (Trial 2) by the similar task group facilitated response execution. That is, fewer corrections during the response were necessary because of the relative accuracy of the initially selected response units. This in terpretation is consisten twith recen tfindings that while knowledge of task parameters is separate from mechanisms responsible for response execution, these can become blended to facilitate task performance (Willingham, Nissen, & Bullemer, 1989). The concurrent presence of two similar tasks in working memory was in itself not sufficient to engender a more complete processing of task characteristics. Ifthis had been the case, the similar task group should have benefited from the concurrent presence of the acquisition and distractor tasks in working memory prior to the acquisition test, and performance on the acquisition test should have been superior for the similar task than for the dissimilar task group. These groups did not differ on performance of this test, however. An explanation for this finding is that the acquisition test provided the impetus for subjects to engage in more extensive processing of the tasks on Trial 2. Performance on Retention Tests 1 and 2 for the same task and no task groups was not significantly different from performance of the similar task group, but was significantly better than the dissimilar task group. It is difficult to reconcile this finding with the reconstruction explanation (Lee & Magill, 1985; Magill & Hall, 1990). The reconstruction perspective would not predict these findings since the performance of the same task and no task groups demonstrated relatively less forgetting than the similar task and dissimilar task groups as indexed by the acquisition test. Thus, less reconstructive processing should have been performed by the same task and no task groups, which should have led to poorer retention performance. However, it is possible the provision of an additional trial of the criterion task (same task group) or the opportunity to covertly rehearse the criterion task (no task group) was sufficient to allow encoding of additional task related parameters, which increased the memorability of the task. This would be similar to subjects engaging in a form of intratask processing, which is an important component of the elaboration perspective, especially when addressing the initial stages oflearning a task (Limons & Shea, 1988). The discrepancy between these findings and others (Lee & Weeks, 1987; Weeks et aI., 1987) that have pro-
299
Downloaded by [University of Sydney] at 04:27 14 March 2015
Shea andWright
vided support for the forgetting explanation might be attributed to differences in experimental procedures as well as in the tasks used. Timing (Weeks et al., 1987) and linear positioning (Lee & Weeks, 1987) tasks might not be dependent on the same processes for performance as the tasks used in the present experiment. The present tasks were similar to those used in the first experiments reporting contextual interference effects (Lee & Magill, 1983; Shea & Morgan, 1979), and performance of these tasks is notably dependent on the retention of visual information that is centrally rehearsable (Posner, 1967). The present experimen t was designed as a test ofone fundamental process (i.e., reconstruction ofa previously forgotten action plan) thought to be responsible for the blocked and random group retention differences found in most contextual interference experiments (Lee & Magill, 1985; Magill & Hall, 1990). As such, this experiment should not be construed as providing a complete account for contextual interference findings. This point is underscored by the fact retention findings for the groups used in this experiment did not parallel those of their counterparts in contextual interference experiments. The no task and same task groups correspond to the blocked practice (low contextual interference) schedule groups and the dissimilar task, and similar task groups correspond to the random practice (high contextual interference) schedule groups. Iffindingsfor the present study had paralleled those for contextual interference experiments, retention performance for either the dissimilar task or similar task group should have been significantly better than for both the no task and same task groups. This was not the case, and the no task and same task groups performed as well as the similar task group and significantly better than the dissimilar group. Other processes in addition to those investigated in the present experiment would appear to be operating in contextual interference experiments. These processes, either singularly or in combination, might account for the discrepancies in findings from these experiments and the present one. The present experiment utilized a single-task/two-trial paradigm. This contrasts sharplywith the multiple-task/multiple-trial paradigm used in traditional contextual in terference experiments. The reduced complexity ofthe present experimental paradigm migh t have contributed to the retention performance of the same task and no task groups by removing detrimental effects of retroactive and/or proactive interference (Magill & Hall, 1990), which would be experienced by blocked practice schedule subjects in contextual interference experiments.
300
References Adams,]. A. (1984). Learning of movement sequences. Psychological Bulletin, 96, 3-28. Battig, W. F. (1979). The flexibility of human memory. In L. S. Cermak & F. 1. M. Craik (Eds.), Levels of processing and human memory (pp. 2344). Hillsdale, NJ.: Erlbaum. Kelso,]. A. S., & Wallace, S. A. (1978). Conscious mechanisms in movement. In G. E. Stelmach (Ed.), Information processing in motor control and learning (pp. 79-116). New York: Academic Press. Lee, T. (1988). Transfer-appropriate processing: A framework for conceptualizing practice effects in motor learning. In O. G. Meijer & K Roth (Eds.), Complex movement behavior: Themotor-actioncontroversy (pp. 201-215) .Amsterdam: NorthIiolland. Lee, T. D., & Magill, R. A. (1983). The locus of contextual in terference in motor-skill acquisition.JournalofExperimental Psychology: Learning, Memory, and Cognition, 9,730-746. Lee, T. D., & Magill, R. A. (1985). Can forgetting facilitate skill acquisition? In D. Goodman, R. B. Wilberg, & 1. M. Franks (Eds.) , Differing perspectives in motor learning, memory, and control (pp. 3-22). Amsterdam: North-Holland. Lee, T. D., & Weeks, D.]. (1987). The beneficial influence of forgetting on short-term memory retention of movement information. Human Movement Science, 6, 233-245. Limons, E., & Shea,]. B. (1988). Deficient processing in learning and performance. In A. M. Colley &]. R. Beech (Eds.), Cognition and action in skilled behatnor (pp. 333-347) . Amsterdam: North-Holland. Magill, R. A. (1973). The post-KR interval: Time and activity effects and the relationship of motor short-term memory theory. Journal ofMotor Behavior, 5, 49-56. Magill, R. A. (1989). Motor learning: Concepts and applications (3rd ed.). Dubuque, Iowa: Wm. C. Brown. Magill, R. A., & Hall, K. G. (1990). A review of the contextual interference effect in motor skill acquisition. Human Movement Science, 9,241-289. Marshall, P. H., Jones, M. '1'., & Sheehan, E. M. (1977). The spacing effect in short-term motor memory: The differential attention hypothesis. Journal ofMotor Behavior, 9, 119126. Marshall, P. H., Wyatt, S. L., Moore, S. A., & Sigman, S. E. (1975). Interrepetition interval in short-term memory. Perceptual and Motor Skills, 40, 535-538. Mitchell.], V. (Ed.) (1983). Tests in print III: An index to tests, test reviews, and the literature on specific tests. Lincoln: University of Nebraska Press. Posner, M. 1. (1967). Characteristics of visual and kinesthetic memory codes. Journal ofExperimental Psychology, 75, 103107. Shea,]. B., & Morgan, R. L. (1979). Contextual interference effects on the acquisition, retention and transfer of a motor skill.Journal ofExperimental Psychology: Human LearningandMemory,5,179-187.
RaES: September 1991
Shea andWright
Shea,j. B., & Zimny, S. T. (1983). Context effects in memory and learning movementinformation. InR. A. Magill (Ed.), Memory and control of action (pp. 345-366). Amsterdam: North-Holland. Shea,j. B., & Zimny, S. T. (1988). Knowledge incorporation in motor representation. In O. G. Meijer & K Roth (Eds.), Complex movement behavior: The motor-action controversy (pp. 289-314). Amsterdam: North-Holland. Weeks, D. J. Lee, T. D., & Elliott, D. (1987). Differential forgetting and spacing effects in short-term motor retention. Journal ofHuman Movement Studies, 13, 309-321. Willingham, D. B., Nissen, M. J. & Bullemer, P. (1989). On the development of procedural knowledge. Journal of Experimental Psycholbgy: Learning, Memary and Cognition, 15,
1047-1060.
Note 1. Wilks's criterion (A) was used for all MANOVA tests. Lambda was converted to exact Fs by the following formula: ExactF= (l-SQRT[A])/SQRT(A)' (NE+Q:P-1)/P with 2P and 2 (NE+Q-P-1 ) degrees of freedom, where NE= df for error, Q= df for hypothesis, and P= rank of hypothesis + error matrix.
Authors' Notes
Downloaded by [University of Sydney] at 04:27 14 March 2015
The authors thank Timothy D. Lee, Charles H. Shea, and two anonymous reviewers for their comments on earlier drafts of the manuscript.
ROES: September 1991
301
