THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 1992.

44A (4) 627-644

Contextual Interference in Motor Learning:

Dissociated Effects
Due to the Nature of Task Variations
Timothy D. Lee
McMaster University, Hamilton, Ontario, Canada

Gabriele Wulf
Max-Planck-Inst. fur Psychologische Forschung, Munich, Germany

Richard A. Schmidt
University of California, Los Angeles, U.S.A.

The contextual interference effect in motor learning refers to the interference

that results from practising a task within the concept of other tasks in a
practice session. Several studies have shown that practice under conditions of
high contextual interference (i.e. with a random practice order) degrades
performance during acquisition trials, compared to low contextual
interference conditions (i.e. with a blocked order, where practice is
completed on one task before practice on another task is undertaken). In
contrast to acquisition performance, random practice usually leads to more
effective learning than blocked practice, as measured by retention and
transfer tests. One of the hypotheses regarding the effect suggests that a
random practice schedule induces more extensive planning operations during
practice than a blocked practice condition. If so, then differences between
these two conditions should emerge to the degree that the set of tasks
requires complete reconstruction of these planning operations on each trial.
To address this issue, we compared four groups of subjects: a blocked and
random group that practised three timing tasks that shared a common
characteristic (same relative timing), and a blocked and random group that
practised three tasks that each had different relative timing structures.
Subjects practised these tasks on each of two days, with a retention test and
two transfer tests that required either a relative timing structure that had
been practised previously or had not previously been practised. NO random/

Requests for reprints should be sent to Timothy Lee, Dept. of Physical Education,
McMaster University, 1280 Main St. West, Hamilton, Ontario, L8S 4K1,Canada. Funding
for this research was provided by a Natural Sciences and Engineering Research Council of
Canada operating grant to the first author.

01992 The Experimental Psychology Society

628 LEE, WULF, SCHMIDT

blocked differences occurred regardless of the relative timing of the patterns

during acquisition or retention. However, for both transfer tests, random
practice enhanced learning only for the group that had practised with tasks
that each had different relative timing during acquisition. Implications of
these results for an explanation of contextual interference are discussed.

A renewed interest in practice effects on the retention and transfer of skill

has contributed some important insights into the processes that subserve
motor learning (see Adams, 1987; Newell, 1991 for recent reviews). One
practice effect that has received considerable attention is the contextual
interference effect. Originally noted by Battig (1966) as an anomaly in
learning word lists, the effect has seen some attention in the learning of
procedural skills (Carlson, Sullivan, & Schneider, 1989; Carlson & Yaure,
1990) but has most extensively been studied in the learning of motor skills
(see Magill & Hall, 1990, for a recent review).
The contextual interference effect may be characterized as the influence
on performance when more than one task is practised during an experi-
mental session. According to Battig (1979), interference in the perform-
ance of a task may arise when it is performed in the context of other tasks.
This context effect can be created in two ways: interference may result
either from the order under which practice is structured, or from the
similarity of the tasks to be practised. The contextual interference effect
is considered a performance paradox because increases in interference lead
to degraded performance during practice, but to enhanced learning as
measured in retention and transfer tests.
Most empirical studies investigating contextual interference have
examined the impact of practice order and are typified by Shea and
Morgan’s (1979) experiment. Interference during the acquisition of three
patterns of arm movements was manipulated by organizing practice trials
according to either a blocked or a random order. Under blocked practice,
performance of all trials for one of the patterns was completed before
another task was practised. Under random practice, a different pattern was
practised on each trial. Shea and Morgan found that the greater inter-
ference caused by the random practice condition resulted in less effective
practice performance but in superior retention and transfer performance,
relative to the blocked practice condition.
According to Battig’s conceptualization, contextual interference effects
created by practice orders should interact with the relationship amongst
the set of tasks to be learned. Indeed, while Magill and Hall (1990)
reviewed many replications of the Shea and Morgan (1979) findings, they
noted a particular trend amongst failures to replicate. They found that
studies that failed to find differences between practice orders used a set of
tasks that shared the same generalized motor program (Schmidt, 1975,
 CONTEXTUAL INTERFERENCE 629

1985). A generalized motor program is the memory representation for a

class of actions that share certain invariant motor control characteristics
(such as relative timing and relative force). According to this view,
generating a movement requires a two-stage process: (1) selecting the
appropriate generalized motor program for the goal, and (2) implementing
that program by specifying the parameters for movement (such as selecting
the absolute time and force as well as which muscles to use). Magill and
Hall’s review suggested that random/blocked practice order effects are
most prominent when the tasks to be learned each have different general-
ized motor programs, thereby requiring that both the selection and
parameterization process be undertaken by the random group on each
trial. Random/blocked differences were not observed, however, when all
of the tasks shared the same program, thereby requiring that the random
group only generate new movement parameters on each trial (see also
Wood & Ging, 1991).
Magill and Hall’s conclusion from their review of the literature argued
against Battig’s hypothesis regarding the interactive nature of practice
order and task similarity. Battig (1979) suggested that interference should
increase as tasks become more similar, resulting in an enhanced contextual
interference effect, rather than the diminished effect concluded by Magill
and Hall. Of course, these predictions are contingent upon how one
defines task similarity.’ From the generalized motor program view, a set of
tasks that differ only in terms of a variant feature are more similar than a
set of tasks that each differ in both variant and invariant features. Con-
sidering Battig’s perspective this way, the prediction is that random/
blocked differences would be large for a set of tasks that shared the same
generalized motor program as only the variant features of the task are
different. In contrast, tasks that differ both in the invariant and variant
features could be argued to be quite dissimilar, and random/blocked
differences would be predicted to be diminished. Clearly, the conclusion
made by Magill and Hall argued against Battig’s hypothesis.
In contrast, Magill and Hall’s conclusions are logically consistent with
the theoretical arguments regarding the random/blocked practice order
effects suggested by Lee and Magill (1983, 1985). Their position was based
‘Battig’s (1979) arguments regarding contextual interference were made with explicit
reference to memory for verbal materials. However, it is clear that the initial foundation for
Battig’s ideas were related to the learning of motor skills (e.g. Battig, 1966),and Battig later
returned to motor skills for an extension of his ideas regarding contextual interference (Battig
& Shea, 1980). Battig (1979) referred to inter-item similarity in terms of semantic and
orthographic relationships amongst the verbal items to be learned (e.g. Lauer, Streby, &
Battig, 1976). However, Battig did not define how similarity, amongst motor tasks might be
defined. Our operational definitions regarding similarity, and how the findings are interpreted
with respect to Battig’s conceptualization of contextual interference, should be viewed within
these limitations.
630 LEE, WULF, SCHMIDT

largely on Jacoby’s arguments regarding the spacing effect in memory for

verbal items (Cuddy & Jacoby, 1982; Jacoby, 1978). Lee and Magill
suggested that motor learning involves a problem-solving process, where the
problem is to find the best motor solution for the task goal (Adams, 1971;
Bernstein, 1967). Under a blocked order of practice, successive attempts
on a task do not require new “action planning” because the memory for
the previous plan is readily available for revision or reimplementation.
Under random practice conditions, however, the action plan for a previous
performance of the task is likely to have been forgotten due to the
interference of performing intervening tasks. The consequence of having
to reconstruct action plans throughout practice is a decrement to practice
performance but a later facilitation to retention and transfer.
A problem with this hypothesis, however, is that the precise nature of
the action plan was not specified by Lee and Magill (1985), and an
acceptable definition remains elusive (Newell, 1978). In terms of the
variables manipulated in the present experiment, we suggest that more
extensive planning would be required when a succeeding task required
both a different generalized motor program and parameterization, com-
pared to a succeeding task which required only a reparameterization of
the same program. There is ample evidence that changing a generalized
motor program requires more extensive operations than altering the
parameters of a program. For instance, Quinn and Sherwood (1983) found
that the EMG latencies in response to a stimulus were longer when a
reversal in arm direction was required than when an acceleration in the
direction of movement was required. Presumably, implementing a change
in the parameterization of the movement speed required a less extensive
operation than did changing the motor program (see also Roth, 1988).
Expanding Lee and Magill’s ideas, the extensiveness of the “recon-
struction” process can be considered in terms of the planning operations
that would be faced by a subject in a random practice condition. If a
subject were required to alter both the parameters and the generalized
motor program, there would be more extensive reconstruction required
than if only a reparameterization of the same program were required. The
resultant effect would be a dissociation as a function of task type, random/
blocked differences being larger when task variations involve different
generalized motor programs than when the set of tasks comprised variants
of the same program.
The predictions regarding randomhlocked differences as a function of
the nature of the tasks may be summarized, then, as follows: from the
Battig perspective, random/blocked differences should be larger when the
set of tasks practised are all variations of the same generalized motor pro-
gram, than when they each represent different programs; from the Lee and
Magill view, randodblocked differences would be larger when tasks are
 CONTEXTUAL INTERFERENCE 631

subserved by different generalized motor programs than by the same

program.
These predictions were tested in the present experiment by comparing
four groups of subjects: two groups of subjects (random vs. blocked
practice order) practised three trisegment tasks, each of which shared a
common relative timing; the other two groups of subjects (also random vs.
blocked practice orders) practised three tasks, each of which had a differ-
ent relative timing. The effects of these practice conditions were compared
over two days, where each day included an acquisition phase as well as a
retention test and two transfer tests.

Method

Subjects
Undergraduate volunteers (32 females ant 16 males) from :Master
University participated as subjects. All subjects received credit in a
physical education course for participating in the experiment. Subjects
were assigned at random to one of four groups, with the only restriction
that each group be balanced for gender. None of the subjects had pre-
viously practised the task, and none was told the purpose of the experi-
ment.

Apparatus and Task

The apparatus and task were very similar to that used by Wulf and
Schmidt (1988; see their Figure 1, on p. 136, for an illustration of the
apparatus). The apparatus consisted of a wooden base upon which were
mounted four electromagnetic microswitches. The microswitches were
mounted in the shape of a diamond, such that the subject moved from the
front-centre (home) microswitch forward and to the left to contact the first
target, then forward and to the right to the second (centre-rear) target,
then towards the body and to the right for the final target. The distance
between each adjacent microswitch was 18 cm.
A 1 cm x 4 cm cylindrically shaped magnet was used as the stylus to tap
each target. Each switch was connected to a Lafayette Performance Pack,
which recorded the segment times to the nearest msec.
The subject’s task on each trial of the experiment was to move from the
resting home position (the centre-front microswitch) to the three succes-
sive targets according to specified goal movement times. A movement
segment was defined as the movement required between two targets.
Segment 1 was from the home position to the first target; Segment 2 was
from the first to the second target; and Segment 3 was from the second to
the third target. Each segment was assigned a goal movement time. The
632 LEE, WULF, SCHMIDT

three goal movement times for each task variation were printed on separate
cards and mounted on the wall behind the apparatus.

Procedures
A summary of the testing sequence is provided in Table 1. A factorial
arrangement of task variations (same- vs. different-phasing) and practice
conditions (random vs. blocked order), resulted in four independent
groups of subjects. Testing for each of the four groups of subjects took
place on two consecutive days. On each day the subjects completed four
phases of the experiment: 90 acquisition trials, 12 retention trials, 12
same-phasing transfer trials, and 12 different-phasing transfer trials. The
order of these phases was not the same on each day, however. On Day 1,
the subject first completed the 90 acquisition trials. Following the acquisi-
tion phase, the subject completed one of the transfer phases and then the
other transfer phase (counterbalanced across subjects within groups). The
final phase on the first day was retention. Performance on Day 2 began
with 12 retention trials, followed by 90 more acquisition trials. The transfer
trials were performed last, with the order of performance of these trials
reversed from the previous day.
Only the acquisition trials differentiated the four groups of subjects.
Every subject performed the same retention and transfer tests. A summary
of the details of the experimental design is provided in Table 2.

Acquisition Phase. The four groups of subjects comprised two random

and two blocked groups. One random and one blocked group practised
three movement variations that had the same relative timing (same-

TABLE 1
Summary of the Testing Sequence for All Groups

Trials

Day I 1. Acquisition I 90
2. Same-phasingtransfer I' 12
3. Different-phasingtransfer' 12
4. Retention 12

Day 2 1. Retention I1 12
2. Acquisition I1 90
3. Different-phasingtransfer 11' 12
4. Same-phasing transfer 11' 12

'Order counterbalancedacross subjects within groups.

20rder reversed from Day 1.
 CONTEXTUAL INTERFERENCE 633

TABLE 2
Summary of the Details of the Experimental Design:
Absolute Goal Movement Times, Goal Proportions (in parentheses), and
Overall Movement Times

Same-phasing Groups Different-phasing Groups

Acquisition
150-300-225 (0.22-0.44-0.33) 675 225-150-300 (0.33-0.22-0.44) 675
200-400-300 (0.22-0.44-0.33) 900 200-400-300 (0.22-0.44-0.33) 900
250-500-375 (0.22-0.44-0.33) 1125 500-375-250 (0.44-0.33-0.22) 1125

Retention
200-400-300 (0.22-0.44-0.33) 900

Same-phasing Transfer
300-600-450 (0.22-0.44-0.33) 1350

Different-phasing Transfer
600-300-450 (0.44-0.22-0.33) 1350

phasing) requirements. The other random and blocked group practised

three variations that each had different relative timing requirements
(different-phasing).
The three tasks practised by the same-phasing groups all had the same
proportional segment movement goal times: the segment movement time
goals (in msec) for each of the variations were: 150-300-225, 200-400-
300, and 250-500-375; the proportions for the three segments relative to
the total movement time (sum of the three goal times) were 0.22-0.44-
0.33 for each of these variations. The three tasks practised by different-
phasing groups each had a different relative timing structure: the segment
movement times (in msec) were 225-150-300, 200-400-300, and 500-
375-250; the proportions were 0.33-0.22-0.44,O. 22-0.44-0.33, and 0.44-
0.33-0.22, respectively.
On each day 30 acquisition trials were performed on each of the task
variations (total = 90 acquisitions trials per day). Knowledge of results
(KR) in terms of the actual segment performance times was provided by
the experimenter on all acquisition trials.
The random practice groups performed the three task variations accord-
ing to a randomized procedure that had two restrictions: (1) that each task
variation be performed twice within each block of six trials, and (2) that no
more than two trials on the same task variation could be practised on
consecutive trials. On each day, the blocked practice groups performed all
30 trials of one variation before switching to another variation. The order
634 LEE, WULF, SCHMIDT

by which these variations were practised was counterbalanced across

subjects within groups according to a William’s square design. The same
order was never used for the same individual on the two days of acquisition.

Retention. The 12 retention trials were performed on the only task

variation that all subjects had practised (200-400-300). The retention
trials were conducted without KR. The rationale for conducting the
retention trials as the last phase on Day 1 and as the first phase on Day 2
was to assess contextual interference effects following both a brief and a
long delay, uninfluenced by performance on any of the other rasks.

Same-phasing Transfer. This transfer test (300-600-450) shared the

same relative timing requirements as the three versions practised by the
same-phasing groups and had the same relative timing as one of the
variations practised by the different-phasing groups (0.22-0.44-0.33).
However, the same-phasing transfer test required an absolute timing
parameterization that had a longer overall duration than any version that
had been practised. The 12 trials on each day of practice were performed
without KR.

Different-phasing Transfer. This transfer test (600-300-450) had both

absolute and relative timing requirements that differed from any task
variation that had been performed previously (0.44-0.22-0.33). The over-
all duration for this transfer test was the same as the overall duration for
the same-phasing transfer. The different-phasing transfer test was also
performed without KR.
To summarize, the four groups of subjects differed in terms of the nature
of task variations practised and the order by which they were practised
during the acquisition phases of the experiment. One blocked group and one
random group practised three task variations with the same relative timing
(but different overall duration), and a blocked and random group practised
three task versions with different phasings (and different overall dura-
tions). These four groups are called: (1) same-phasing/blocked-order, (2)
same-phasinghandom-order, (3) different-phasing/blocked-order , and (4)
different-phasinghandom-order. Practice was conducted on each of two
consecutive days, which also included a retention test and two transfer
tests.

Data Analyses
Absolute error (AE), calculated as the differences (without regard to
sign) between the actual times on each movement segment and the
respective goal times, was averaged over blocks of six trials for the purpose
 CONTEXTUAL INTERFERENCE 635

of statistical analyses (similar to the analyses performed by Wulf &

Schmidt, 1988).*Separate ANOVAs were performed on the same-phasing
and different-phasing groups for acquisition, retention, and transfer. The
acquisition data were analysed using a 2 (Group) x 3 (Task) x 2 (Day) x 5
(Trial Block) ANOVA with repeated measures on the last three factors.
The design for the retention and transfer data were 2 (Group) X 2 (Day) X
2 (Trial Block) ANOVAs, with repeated measures on the last two factors.
All significant effects are reported at the G0.05 level of significance.

Results
Acquisition
Same-phasing Practice. Under the same-phasing task variations, the
blocked group and random group performed very similarly during the
acquisition phases on both Day 1 and Day 2. Both groups showed
negatively decelerating performance curves on the first day and only a
small further reduction in AE on the second day (see Figure 1). The main
effect of group was not significant (F < 1.0). The only significant effects
found in the ANOVA were due to the typical benefits of practice with KR.
The main effects of day, F(1, 22) = 15.60, MSe = 2290.47, trial block,
F(4, 88) = 33.91, MSe = 331.81, and the Day X Block interaction,
F(4, 88) = 2.58, MSe = 356.66, were significant.

Different-phasing Practice. There was essentially no difference

between the blocked and random groups that practised the different-
phasing task versions on either Day 1 or Day 2. Both groups demonstrated
relatively greater improvements on Day 1 than on Day 2 of practice (see
Figure 1). The group effect was not significant. However, the effects of
day, trial block, and the Day x Block interaction were all significant, all
F’s > 10.0. In addition, performance of the 500-375-250 task was both
poorer overall and improved less rapidly than did the performance of the
other two task versions. The result was a significant task main effect,
F(2, 44) = 13.47, MSe = 4761.81, a Task x Day interaction, F(2, 44) =
3.89, MSe = 1778.50, and a Task X Trial Block interaction, F(8, 176) =
7.18, MSe = 449.98.

*Wulf and Schmidt also reported a proportional error score, which was the absolute
difference bstween actual duration of each segment relative to the totaI duration and the goal
of each segment relative to the total duration. This measure was considered an index of how
well the relative timing component of the generalized motor program was represented in
memory, as assessed by the retention and transfer tests. This proportional AE measure was
also computed and analysed in the present experiment. Similar to the findings of Wulf and
Schmidt, the results for this proportional AE measure paralleled the AE results very closely.
Details of the results of these analyses are available from the first author.
636 LEE, WULF, SCHMIDT

100 r IACQUISITION1

r-l
0 Same/Blocked
0 Same/Random
20 '
Different/Blocked
0 Different/Random

1 2 3 4 5 1 2 3 4 5
Day 1 Day 2
Blocks of 6 Trials
FIG. 1. Average absolute error during acquisition. Each data point represents the A E
averaged over six trials on each segment of the 200-400-300 task practised during acquisition.
The data are presented over five blocks of trials on each of two days.

Retention
Same-phasing Practice. Retention performance of the blocked and
random groups that practised the task versions with the same phasing are
shown at the left of Figure 2. The random group tended to have less AE
than the blocked group on Day 2. However, the main effect of Group and
all interactions with Group were not significant. In fact, no significant
differences at all were revealed in this analysis.

Different-phasing Practice. Performances of the blocked and random

groups with different-phasing practice are illustrated at the right of Figure
2. Although the random group tended to be more effective than the
blocked group on the first trial block of each day, the Group x Block
interaction failed to reach significance [ F ( l , 22) = 2.08, MSe = 196.32,
p = 0.161. The only retention effect to reach significance was the trial
block main effect, F(1, 22) = 5.23, MSe = 196.32. The second block of
trials was performed better than the first block of trials on both retention
tests.
 CONTEXTUAL INTERFERENCE 637

[RETENTION 1
a0

-z 70
0
w
-w.-
c

d,
>
U
60

50
"78 0 /
0

0
\
/. i
11 Different/Blocked 1
10 Different/Randorn I
1 2 1 2
Day 1 Day 2
Blocks of 6 Trials
FIG. 2. Average absolute error during retention. Each data point represents the AE
averaged over six trials on each segment of the 2004CK)-300 task. The data are presented
over two blocks of trials on each of two days.

Sam e-phasing Transfer

Same-phasing Practice. On the transfer test with the same relative
timing as the three task versions practised under same-phasing conditions,
the blocked and random groups performed similarly (see left side of
Figure 3). There were no effects of Group or interactions with Group (all
F's C 1.O). Both groups consistently improved their AE from the first block
of trials on Day 1 to the second block of trials on Day 2. These improve-
ments resulted in significant main effects for day, F(1, 22) = 8.87, MSe =
645.84, and trial block, F(1, 22) = 5.91, MSe = 636.82.

Different-phasing Practice. Same-phasing transfer performances of

the blocked and random groups that practised the phasing requirements of
this transfer task on only one task version (medium duration) are shown on
the right-hand side of Figure 3. The performance of the random group on
the first block of trials on each day was better than the performance of the
blocked group on these trials. These differences did not extend into the
638 LEE, WULF, SCHMIDT

JSAME-PHASING TRANSFER]
130

120
-
-.-
c
w
110
b
a 100 '

9
a
90 '

1 2 1 2 1 2 1 2
Day 1 Day 2 Day 1 Day 2
Blocks of 6 Trials
FIG. 3. Average absolute error during the same-phasing transfer test. Each data point
represents the AE averaged over six trials on each segment of the 300-600-450 task. The data
are presented over two blocks of trials on each of two days.

second block of trials, however. The result was a Group X Trial Block
interaction, F(1, 22) = 8.24, MSe = 724.80. No other signifficant effects
were found.

Different-phasing Transfer
Same-phasing Practice. Performances of the blocked and random
groups with same-phasing practice on the transfer task with new phasing
requirements are illustrated at the left of Figure 4. Consistent with the reten-
tion and same-phasing transfer performances, there were no differences
between the random and blocked practice orders on the different-phasing
transfer trials. Although there does appear to be an advantage for the
random group on Day 2 of transfer, the Group x Day interaction fell short
of significance [F(l, 22) = 3.06, MSe = 1640.28, p = 0.091. The only
significant effect to emerge in this analysis was an improvement in perform-
ance from the first to the second block of trials on both days, F(1, 22) =
15.10, MSe = 986.69.
 CONTEXTUAL INTERFERENCE 639

200
IDIFFERENT-PHASING TRANSFER I
0
180
\
-2
-,.-,,
U
c
160

140
: O i

\O
0

0
rn

a-
'. rn
9 120
\.
a
100
\ 0
a
\a
80

~
I , .I

1 2 1 2 1 2 1 2
Day 1 Day 2 Day 1 Day 2
Blocks of 6 Trials
FIG. 4. Average absolute error during the different-phasing transfer test. Each data point
represents the AE averaged over six trials on each segment of the 600-300-450 task. The data
are presented over two blocks of trials on each of two days.

Different-phasing Practice. Different-phasing transfer performances

of the groups that experienced different phasing patterns during practice
are illustrated at the right of Figure 4. The random group performed with
less AE than the blocked group, resulting in a significant group main
effect, F(1, 22) = 6.21, MSe = 3303.98. This difference between groups
did not interact with any other factor.-
The only other significant effects found were due to the consistent
performance improvements from the first block of trials on Day 1 to the
second block of trials on Day 2 (see Figure 4). This reduction in AE
resulted in main effects for day, F(1, 22) = 7.38, MSe = 1954.82, and trial
block, F(1, 22) = 8.52, MSe = 784.48.

Discussion
Key ingredients that are particular to motor learning phenomena include
subject and task variables, practice, feedback, and criterion tests used for
assessing learning. It should not be surprising that these factors interact in
640 LEE, WULF, SCHMIDT

a complex manner when combined in different ways. Indeed, based upon

the recent review of the contextual interference literature by Magill and
Hall (1990), it seems unlikely that a simple theoretical formulation of even
this learning phenomenon is forthcoming. Nevertheless, the results of the
present experiment appear to clarify some of the issues regarding the
interaction of practice schedules with the nature of the set of tasks
practised.
Based on the theoretical formulation of Battig (1979), tasks that are
highly similar should produce more interference under random practice
conditions than tasks that are not similar. These efects would be mani-
fested by large randomblocked learning differences for highly similar
tasks, but small random/blocked differences for tasks of low similarity.
Based on Lee and Magill’s (1985) arguments, however, there should be
more interference between tasks of low similarity than between highly
similar tasks. Considered in terms of the variables manipulated in the
present experiment, Battig’s perspective predicts that tasks that differed
only in terms of overall duration, and not relative timing, should have
produced larger randomiblocked differences than tasks that differed in
both the overall duration and the relative timing. In contrast, Lee and
Magill’s hypothesis predicted that larger random-blocked differences would
occur when the task revisions differed in both the overall duration and the
relative timing than when only the overall duration was different.
The results of both the acquisition data and the retention data were
equivocal in terms of the predictions made based on both Battig’s and Lee
and Magill’s hypotheses. For acquisition and retention, the differences
between the random and blocked groups were not significant. One possib-
ility for this result could be the nature of the task. The multiple goal
timing task used here is considerably more difficult than the movement
pattern tasks, duration-goal timing tasks, and coincident timing tasks that
have often been used in contextual interference studies (see Magill & Hall,
1990, for a review). Indeed, the absence of acquisition differences between
blocked and random practice orders has also been noted for subjects
learning computer games (Lee & White, 1990).
Another possibility is that the concept of a “typical” contextual inter-
ference effect is incorrect. Two findings support this contention. In one
study, segmental training (of the type reported here) resulted in better
transfer than practice on a single movement duration (Langley & Zelaznik,
1984). This finding was suggested to be a within-movement type of
contextual interference that influenced acquisition performance, but not
transfer (Carnahan & Lee, 1989). Thus, the differences between acquisi-
tion groups might have been “typical”, although transfer was not. Another
study by Pollock and Lee (submitted) showed a different trend to the
typical contextual interference effect. Random and blocked groups of
adults and seven-year old children were compared. The performance of the
 CONTEXTUAL INTERFERENCE 641

adult subjects conformed to expectations; blocked practice better than

random during acquisition, and random better than blocked during reten-
tion and transfer. However, there were no differences during acquisition
trials between the seven-year old, blocked and random groups. Never-
theless, the children who had practised under a random order later
performed better than the blocked group on tests of retention and transfer.
It is clear that there is no “typical” contextual interference effect, and
factors such as task and subject variables interact in complex ways to affect
performance on acquisition, retention and transfer tests (Jenkins, 1979;
Shea & Zimny, 1983).
Retention and transfer place quite separate and distinct processing
demands on the subject. Retention tests require the subject to perform a
task that had previously been practised, while transfer tests require the
subject to extrapolate from their previous experiences in an attempt to
achieve novel task demands. The interaction of processing demands con-
strained by the test for learning and the processing demands encouraged by
the acquisition conditions are prime determinants of performance (Lee,
1988). Thus, it should not be surprising when retention and transfer data
provide incongruous results (e.g. Magill, Meeuwsen, Lee, & Mathews,
unpublished manuscript).
Despite the incompatibility of the retention and transfer results, the
conclusions regarding the theoretical predictions on transfer were
relatively straightforward. Based on Battig’s ideas, the randomhlocked
differences for the same-phasing groups should have been the largest. The
present findings provided no support for this prediction in any of the
retention or transfer tests. Based on the Lee and Magill hypothesis,
however, the random/blocked differences were predicted to be the largest
for the different-phasing groups. Though no differences were found during
retention, the results for both the same-phasing and different-phasing
transfer tests did support the prediction. On the same-phasing transfer test
the random group performed more effectively than the blocked group on
the first trial block of each day. On the-different phasing transfer test the
random group was more effective overall than the blocked group.
The reconstruction hypothesis (Lee & Magill, 1983, 1985) provides a
reasonably good explanation for some of the present results. The advan-
tage for learning under random practice conditions (compared to blocked
practice) occurred when successive trials required that a different relative
timing structure be executed. However, no advantage was found when
successive trials required only that a different overall duration, with the
same relative timing, be executed. The advantage for random practice
conditions, then, appeared to depend upon the relationship of the
invariant, and not the variant, feature of successive trials. When a different
phasing was required on a subsequent trial, the subject had to abandon
the relative timing structure that had been executed on the preceding trial
642 LEE, WULF, SCHMIDT

and construct a new phasing. However, when a trial necessitated only a

different overall duration for the same relative timing as the preceding
trial, short-term retention of the previously executed invariant structure
obviated a more complete reconstruction. The abandonment and sub-
sequent reconstruction of relative timing, and not overall duration, resulted
in the contextual interference effects that later emerged on the transfer
tests.
The results of the present experiment also-help to clarify the relation-
ship between contextual interference effects and variability of practice tests
of schema theory (Schmidt, 1975; Shapiro & Schmidt, 1982). Previously,
the suggestion was made that schema theory could not explain the effects
of random versus blocked practice orders when the amount of practice
variability was equated (Lee, Magill, & Weeks, 1985). However, as
schema threory is closely tied to the concept of the generalized motor
program, this limitation would be true only if randomhlocked effects were
found using task variations of the same motor program. The results of the
present experiment and the conclusions of Magill and Hall (1990) do not
support this limitation to schema theory, as practice order effects were
found only for task variations of different generalized motor programs.
The effects of practice on motor performance and learning is often
discussed under the term “repetition” (e.g. Bartlett, 1932). The present
experiment revealed that the impact of practice orders on motor learning
was different if the task variations were considered in terms of a motor
control characteristic (i.e. phasing). Thus, what is repeated on consecutive
trials may assume a different meaning, depending on the level of analysis
of the task. Closer attention to task analysis may help to identify not only
what is repeated in a repetition, but what it is important to repeat and not
to repeat when attempting to enhance motor learning.

REFERENCES
Adams, J.A. (1971). A closed-loop theory of motor learning. Journal of Motor Behavior, 3,
111-150.
Adams, J.A. (1987). Historical review and appraisal of research on the learning, retention
and transfer of human motor skills. Psychological Bulletin, 101, 41-74.
Bartlett, F.C. (1932). Remembering: A study in experimental and social psychology. Cam-
bridge: Cambridge University Press.
Battig, W.F. (1966). Facilitation and interference. In E.A. Bilodeau (Ed.), Acquisition of
skill (pp. 215-244). New York: Academic Press.
Battig, W.F. (1979). The flexibility of human memory. In L.S. Cermak & F.I.M. Craik
(Eds.), Levels of processing in human memory (pp. 23-44). Hillsdale, NJ: Lawrence
Erlbaum Associates, Inc.
Battig, W.F., & Shea, J.B. (1980). Levels of processing of verbal materials: An overview. In
P. Klavora & J. Flowers (Eds.), Motor learning and biomechanical factors in sport
(pp. 24-33). Toronto: University of Toronto.
Bernstein, N . (1967). The coordination and regulation of movements. London: Pergamon.
 CONTEXTUAL INTERFERENCE 643

Carnahan, H., & Lee, T.D. (1989). Training for transfer of a movement timing skill. Journal
of Motor Behavior, 21,48-59.
Carlson, R.A., Sullivan, M.A., & Schneider, W. (1989). Practice and working memory
effects in building procedural skill. Journal of Experimental Psychology: Learning,
Memory and Cognition, 15,517-526.
Carlson, R.A., & Yaure, R.G. (1990). Practice schedules and the use of component skills in
problem solving. Journal of Experimental Psychology: Learning, Memory and Cognition,
16,484-496.
Cuddy, L.J., & Jacoby, L.L. (1982). When forgetting helps memory: An analysis of
repetition effects. Journal of Verbal Learning and Verbal Behavior, 21,451-467.
Jacoby, L.L. (1978). On interpreting the effects of repetition: Solving a problem versus
remembering a solution. Journal of Verbal Learning and Verbal Behavior, 17,649-667.
Jenkins, J.J. (1979). Four points to remember: A tetrahedral model of memory experi-
ments. In L.S. Cermak & F.I.M. Craik (Eds.), Levels of processing in human memory
(pp. 429-446). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Langley, D.J., & Zelaznik, H.N. (1984). The acquisition of time properties associated with a
sequential motor skill. Journal of Motor Behavior, 16,275-301.
Lauer, P.A., Streby, W.J., & Battig, W.F. (1976). The effects of alphabetic organization on
on and delayed retention of semantically similar words. Journal of Experi-
mental Psychology: Human Learning and Memory, 2, 182-189.
Lee, T.D. (1988). Transfer-appropriate processing: A framework for conceptualizing prac-
tice effects in motor learning. In O.G. Meijer & K. Roth (Eds.), Human movement
behaviour: The “motor-action” controversy (pp. 201-215). Amsterdam: North-Holland.
Lee, T.D., & Magill, R.A. (1983). The locus of contextual interference in motor-skill
acquisition. Journal of Experimental Psychology: Learning, Memory and Cognition, 9,
730-746.
Lee, T.D., & Magill, R.A. (1985). Can forgetting facilitate skill acquisition? In D. Good-
man, R.B. Wilberg, & I.M. Franks (Eds.), Differing perspectives on motor memory,
learning and control (pp. 3-22). Amsterdam: North-Holland.
Lee, T.D., Magill, R.A., & Weeks, D.J. (1985). Influence of practice schedule on testing
schema theory predictions in adults. Journal of Motor Behavior, 17, 283-299.
Lee, T.D., & White, M.A. (1990). Influence of an unskilled model’s practice schedule on
observational motor learning. Human Movement Science, 9, 34%367
Hall, K.G. (1990). A review of the contextual interference effect in motor
on. Human Movement Science, 9, 241-289.
Magitl, R.A., Meeuwsen, H., Lee, T.D., & Mathews, R.C. (unpublished manuscript). Is
the contextual interference in motor skill learning a spacing effect?
Newell, K.M. (1978). Some issues on action plans. In G.E. Stelmach (Ed.), Information
processing in motor control and learning (pp. 41-54). New York: Academic Press.
Newell, K.M. (1991). Motor skill acquisition. Annual Review of Psychology, 42,213-237.
Pollock, B.J., & Lee, T.D. (submitted). Contextual interference in motor learning:
Dissociated effects due to the age of the subject.
Quinn, J.T. Jr., & Sherwood, D.E. (1983). Time requirements of changes in program and
parameter variables in rapid ongoing movements. Journal of Motor Behavior, 15, 163-
178.
Roth, K. (1988). Investigations on the basis of the generalized motor programme hypo-
thesis. In O.G. Meijer & K. Roth (Eds.), Human movement behaviour: The “motor-
action” controversy (pp. 261-288). Amsterdam: North-Holland.
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological
Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research
Quarterly for Exercise and Sport, 56, 188-200.
644 LEE, WULF, SCHMIDT

Shapiro, D.C., & Schmidt, R.A. (1982). The schema theory: Recent evidence and develop-
mental implications. In J.A.S. Kelso & J.E. Clark (Eds.), The development of movement
control and co-ordination (pp. 113-150). New York: Wiley.
Shea, J.B., & Morgan, R. (1979). Contextual interference effects on the acquisition, reten-
tion, and transfer of a motor skill. Journal of Experimental Psychology: Human Learning
and Memory, 5,179-187.
Shea, J.B., & Zimny, S.T. (1983). Context effects in memory and learning movement
information. In R.A. Magill (Ed.), Memory and control of action (pp. 345-366). Amster-
dam: North-Holland.
Wood, C.A., & Ging, C.A. (1991). The role of interference and task similarity on the
acquisition, retention and transfer of simple motor skills. Research Quarterly for Exercise
and Sport, 62, 18-26.
Wulf, G., & Schmidt, R.A. (1988). Variability in practice: Facilitation in retention and
transfer through schema formation or context effects? Journal of Motor Behavior, 20,
133-149.

Revised manuscript received 26 July I991