I
I
:
!
MOTOR SKILLS LEARNING AND THE
SPECIFICITY OF TRAINING PRINCIPLE
Mury L. Barnett, Diane Ross, Richard A. Schmidt,
and Bette Todd
The specificity of training principle from exercise physiology was tested
in a motor skills learning context. Two criterion conditions (either fatigued
or nonfatigued) for performing a movement-time task were defined on day
2 and subjects practiced 't'or this criterion under either latigued or nonfatigued conditions on day 1. Regardless ol whether the criterion day 2
performance was under fatigued or nonfatigued conditions, practice under
nonfatigued conditions produced more effective learning than did practice
under fatigued conditions, although the efiect was considerablv larger for
the nonlatigued criterion condition. This evidence was contrary to the
specificity of training principle lor motor skills situations.
Tnr spscIFIcITy of training principle, borrowed from the field of exercise
physiology, has at least potential relevance for the field of motor learning.
For physical exercise and training, it states that if an individual wants to
maximize criterion endurance performance in running or on the bicycle
ergometer, then he should train for this event at the same workload as the
criterion performance. The implication is that the systems that support the
exercise are developed in different ways for different intensities of training
and that training is maximally eftective for a criterion performance of the
same intensity. Presumably, training at a different intensity than that for the
criterion task results in changes that are appropriate for that particular
workload, but that are not appropriate for the specified criterion performance.
It is reasonable to ask whether the specificity of training principle could
be relevant for motor skills learning situations. Since many skills
in sports
MARY L. BARNETT teaches physical education at Marion High School,
Birmingham, Michigan. DIANE ROSS is assistant professor of physical education at California State College, Fullerton, California. RICHARD A. SCHI,IIDT
is a professor of physical education at The University of Michigan. Ann Arbor,
Michigan. BETTE TODD teaches physical education at Whitmer High School,
Toledo, Ohio.
440
Barnett, Ross, Schmidt and Todd
441
and industrial settings are performed under rather distinct environmental
conditions (e.g., noise, heat, spectators) and internal conditions (e.g.,
fatigue, loss of sleep), would it not be most beneficial for subjects to learn
these motor skills under the same environmcntal and internal conditions
as are present in the criterion performance?
A number of lines of evidence suggest that such might be the case. For
example, Henry's (3) specificity hypothesis sees motor skills as being quite
specific, bearing only superficial resemblance to each other. There is the
implication in his work that changing the motor task only slightly produces
a "new" motor task for which a new motor program must be iearned. The
evidence for this conclusion is based upon the low correlations typically
found among very similar-appearing motor skills, and the implications
about the changing motor program are not strong. Additionally, however,
it is known ( I ) that fatigue produces changes in the recruitment pattern
and intensity of motor units within a muscle. It is not unreasonable to
suggest that a motor skill performed under rested conditions would require
a different pattern and intensity of innervation (and hence, a different
motor program) than would the same task performed under fatigued conditions. If such is the case, then practicing the motor task under fatigue
conditions different from those in the criterion situation would result in
acquisition of a somewhat inappropriate motor program for the criterion
task, and learning would be more efficient if the practice conditions simulated exactly the criterion conditions.
A rival point of view to the principle just stated would be that, regardless
of the conditions under which the criterion task is to be performed, practice
should be under those conditions that produce the highest level of performance during practice trials; this idea might be termed the "optimal
conditions" view.
There has been little formal activity to test these two opposing points of
view within the realm of motor skills. An exception is a study by Singer (4)
providing some support for the specificity of training principle using two
spectator conditions with error scores on a mirror-tracing task, but speed
scores did not support this conclusion. Other than this mixed support,
however, there was no evidence found favoring either point of view. The
present experiment, therefore, was concerned with difterentiating among
them.
The "criterion" task was performed either under rested conditions or
under fatigued conditions, and subjects learned this task under either
rested or fatigued conditions. Support for the specificity of training view
would be provided if practicing under the same conditions as the criterion
task provided more efficient learning than practicing under the opposite
condition, for both criterion conditions. Support for the rival "optimal conditions" view would be provided if practice under the rested condition
always provided more efficient learning of the criterion task than did
practice under the fatigued condition.
442
The Research Quarterly, Vol.44, No.4
Method
Subiects. The subjects were 104 right-handed, female undergraduate and
graduate students from The University of Michigan. They volunteered to
take part and were not paid for their services.
Motor learning task. The learnin_q task consisted of a large wooden box
(20 X 12)X 4 in.) mounted on a standard 28-in. table. In one corner of
the box. a vertical axle was mounted in bushings and supported a horizontal
11-in. aluminum arm with a 5-in. hi_eh vertical wooden handle at the end.
The arm rotated freely in the horizontal plane through ncarlv 360o, but a
sturdv metal stop prevented continuous circular rotation. At the upper
right-hand corner of the box, a hinged masonite (4 Y.4 in.) barrier was
attached, and a slight tap to the right was sufficient to make the barrier fall.
When the handle and arm were aeainst thc stop so that only clockwise
movement was possible. a microswitch inside the box was "open," and
moving l/4 in. clockwise "closed" it causing a 0.01-scc. timer (Standard.
Type S-1) to run. When the hinged barrier was knocked over, contact
points were "opened" and the timer stopped.
The task was begun with the seateC subject graspine the handle with
the right hand, and the aluminum arm against the stop. The location of the
stop was such that the starting position was at "6 o'clock" with respect to
the subject, and the barrier approximately in the direction of "2 o'clock."
The subject moved the handle clockwise nearly one full revolution until it
hit the stop from the right, then reversed direction and rotated the handle
counterclockwise until the stop was struck from the left, then released the
handle and moved the hand rightward to knock over the barrier.
The score was the time (in .01 sec.) between the initial movement of
the handle from the stop until the barrier was struck, the movement being
performed as rapidly as possible. The task has been termed the Sigma task
because of the o-like movement required (2).
The fatigue task. The apparatus to induce muscular fatigue was an arm
ergometer with an 1l-in. arm and vertical handle mounted on a vertical
axle. so that the arm and handle rotated horizontally. The axle drove a
12-in. v-pulley with a nylon cord in the groove (under adjustable spring
tension) to provide frictional resistance to the hand movement. The
ergometer was mounted to the right of the Sigma task, with their respective
handles being the same length and height from the floor.
Design. The basic design involved 2 days of practice. On day 1, all
subjects practiced 20 trials of the Sigma task under one of two conditions.
The fatigued condition (F) involved 2 min. of ann ergometer cranking
before the first trial, with 14 sec. cranking during each of the subsequent
intertrial intervals. The rate was 60 rpm with a workload of 383 kgm/min.
A nonfatigued condition (NF) used the same intertrial interval as with
condition F, but subjects tapped a large "x" on the table top for 2 min.
(1 tap/sec. to a metronome) before trial 1 and for 14 sec. during each
Barnett, Ross, Schmidt and Todd
443
intertrial interval. The tapping was designed to minimize the difference
between conditions in terms of the attention demand during the intertrial
intervals. Subjects were assigned at random to one of two conditions on
day l, with the restriction that each condition have 52 subjects.
On day 2 (1 week +3 hr. after day 1) the subject returned for 10
additional trials on the Sigma task. Each of the two day 1 groups was
subdivided randomly into two groups of 26 subjects each, and these groups
performed either under the same or opposite conditions as on day 1. This
formed a 2 X 2 experimental design, with dimensions being day 1 conditions (either F or NF) and day 2 conditions (either F or NF), with 26
independent subjects per cell.
Procedures. On day 1, the Sigma task was explained and the subject
was allowed to perform one very slow response to determine if instructions
had been understood. Then those subjects under condition F were given
arm ergometer instructions (including one slow revolution) while those
under condition NF were told how to tap on the "x." Actual testing began
with 2 min. of either cranking or tapping, and this was ended by the
command "Stop." This was the signal to move to the Sigma task, to prepare
to grasp the handle in the starting position and, upon the command "Start,"
to begin the Sigma task trial. Subjects did not use "Start" as a stimulus as
in reaction time situations, as they were free to wait a moment or so before
they desired. After completing the movement, the subject
responding
moved quickly to the secondary apparatus and immediately began either
tapping or cranking. The next "Stop" was presented 15 sec. after the
command "Start," with cranking or tapping time being approximately 14
sec. Knowledge of results in.01 sec. (e.g., 1.23 sec.) was provided while
the subject was tapping or cranking. This procedure resulted in 20-sec.
intertrial intervals, including 14 sec. cranking or tapping, and 6 sec. for
moving to the appropriate apparatus.
Day 2 procedures were identical to those on day 1 for a given condition,
except that no slow Sigma task trial was provided. Instructions were given
for the new interpolated task for those subjects who changed conditions
from day 1 to day 2, and a slow tapping or cranking trial was presented.
if
Results
Day 1 perfornidnce. The mean movement times for the two day 1 practice conditions are presented as a function of trials in Figure 1. Subjects in
both conditions improved considerably with practice, and the trials efiect
in the day 1 conditions X trials ANOVA w:$ significant with
F(19,1938) - 183.0, P <.01. Condition F appeared to lengthen the
movement time approximately .15 sec. relative to condition NF over all
day 1 trials, and thiseffect was significantwith F(1,102) - 10.2, P <.01.
The groups were remarkably "parallel" except for the slightly greater
improvement for condition NF over trials 1-6, but this difference was
The Research Quarterly, Vol.44, No. 4
444
2.2
:o
2.t
zo
o
IJ
F
=
F
zu.r
1.9
t.8
l'7
1.6
=,
Io
r.5
2
e
t.4
r.3
lrJ
=
1.2
to
TRIALS
I
practice
apparently sufficiently large to produce a significant day 1 condition
X trials
Figure
l.
Mean movement times over trials as a lunction
ol
the day
conditions.
interaction, wirh F(1,9,1938) - 3.3, P <.01. Thus, the day 1 analysis
indicated that subjects in both conditions improved considerably during
day 1, but that the ergometer cranking depressed performance considerably
on all practice trials.
Day 2 performance. The mean movement times for all four groups
created on day 2 are shown in Figure 2. All groups appeared to demonstrate continued improvement throughout day 2 trials, and the trials eftect
was significant in a day 1 y day 2 X trials ANOVA, with F(9,900)
52.37, P < .01.
Again, as on day 1, the presence or absence of fatigue on day 2 appeared
to have a large effect on performance, with condition F producing approximately.12 sec. greater movement times than condition NF, and this day 2
conditions efiect was significant with F(1,000) - 9.21, P <.01. Therefore, the presence of fatigUe had nearly the same depressing effect on day 2
as it did on day 1.
However, the variable of chief interest was the day I conditions effect,
that is, the eftect of the type of practice on day 1 on the level of day 2
performance. Figure 2 shows that. when the criterion task was performed
under condition NF, day 1 practice under the opposite conditions (i.e.,
condition F) produced longer movement times than when day 1 practice
was NF. This effect has been found before (2), and it indicates that it is
Barnett, Ross, Schmidt and Todd
44s
F-F
F- NF
NF-F
d
(l)
3
NF- NF
t.4
trJ
F
F
A
l-3
t.z
'e--+--o--
*-{
-+--{--'--G-__G__O._
-!
lrJ
A
r.r
=
e
r-o
IJ
z
3
5
7
I
I
9
I
o
TRIA LS
Figure 2. Mean day 2 movement times over trials as a ioint function
and day 2 practice conditions.
ol the day I
more beneficial (for a nonfatigued criterion task) to practice the task
under condition NF. The findings appeared to be somewhat difterent in
the case where the criterion task was to be performed under condition F.
In this case, there appeared to be little effect of the type of day 1 practice
on the performance under condition F on day 2, and even a slight tendency
for day 1 condition NF to result in superior day 2 performance relative to
day 1 condition F, indicating that it might be preferable to practice the
fatigued criterion task under nonfatigued conditions. If the specificity of
training view is correct, there should be an interaction between day I and
day 2 dimensions, but this interaction appeared not to be present and was
not significant, with F(1,100) - 0.81, P >.05. However, if the "optimal
conditions" view is correct, we should see an eftect of day I practice conditions on day 2 performance. There was a tendency for the condition NF
on day 1 to produce superior performance on day 2 for both day 2 conditions F and NF, but this effect just failed significance, with F(1,100)
2.73, P .10. These data provided rather strong evidence against the
- of training view, and some support for the "optimal conditions"
specificity
view for skills learning.
The Research Quarterly, VoI. 44, No. 4
Discussion
The specificity of training principle, borrowed from the literature dealing with cardiovascular and strength training. seemed a likely candidate
for application in motor skills contexts because of the rather strong support
for it in the exercise areas. However. the present data offered no support
for it. If the view were correct for skills. the day 2 data should have shown
that those treatment groups that had the same conditions on day 1 and
day 2, regardless of the day 2 conditions, would out perform those groups
that had a shift in conditions from day 1 to day 2. Such a trend was found
for condition NF on day 2, in which practicing on day I under condition
F produced less learning of the criterion NF task than did practicing under
condition NF. This finding supported others (e.g., 2) in the literature
dealing with physical fatipue and learning. showing that it is more efficient
to learn a task to be performed under NF conditions by practicing under
NF rather than F conditions. However, this trend was not present for the
situation in which the criterion task was to be performed under condition F.
In this situation, it seemed to make little difterence under which conditions
the task was learned on day 1, with the subjects who practiced under
condition NF on day t having a slight day 2 advantage over those subjects
with F practice on day 1, which was contrary to the predictions of the
specificity of training principle.
The more attractive possibility in the light of the present data is the
"optimal conditions" view which states that regardless of the conditions
under which the criterion task is to be performed, it is more efficient to
practice the task under those conditions that produce the optimal performance in practice. The predictions from this hypothesis are that practicing
under NF conditions on day 1 should lead to better performance than
practicing under day 1 condition F, regardless of the day 2 conditions. This
pattern of day 2 findings was found in the present data, but with a much
larger effect when the criterion task was performed under condition NF
than when under condition F. Statistically. the prediction of this optimal
conditions hypothesis is a significant main effect of day 1 practice condition
on the day 2 performance, and this effect was nearly statistically significant
(P
.10) in the present data. Therefore, with some desree of caution
because of the somewhat elevated alpha level for the statistical test, there
is support for the "optimal conditions" view.
While there was little support for the specificity of training principle in
the present data. there may be other siruations in which this view has merit.
Consider a motor task where the specific stimuli to which specific responses
are learned change markedly as the environmental conditions change. For
example, if a task involves responses to rather weak auditory stimuli,
addine a noisy environment could have the effect of masking the auditory
stimuli so that other cues (e.g., visual) become dominant. In such a case,
it would seem more efficient to practice the task under those conditions
that most resemble the conditions for criterion performance. The chief
Barnett, Ross, Schmidt ancl I
oclct
++t
difference between this hypothetical situation and the one in the present
data seems to be that the environmental conditions cause a change in the
cues used for proper performance, a condition apparently not present with
the manipulation of fatigue.
While it is still possible that the specificity of training principle might be
supported in subsequent work with other experimental performance
variables, this does not deny the fact that for fatigue as a perforrnance
variable there was no support for the principle and some support for an
"optimal conditions" alternative.
References
t. BrsuuteN, J. V. Muscles alive-their functions revealed by electromyography.
Baltimore: Williams and Wilkins, 1962.
,,
GoDwrN, M. A., and ScHptror, R. A. "Muscular fatigue and discrete motor
learning." Research Quarterly 42:37 4-82, 197 l.
J. HENRv, F. M. "Specificity vs. generality in learning motor skills." Proceedings of
the College Physical Education Association, Washington, D.C., 1958.
4. SrNcrR, R. N. "Effect of an audience on performance of a motor task." Iournal
ol Motor Behavior,2:88-95, 1970.