Proceedings of the Iowa Academy of Science

Volume 69 Annual Issue Article 77

1962

The Value of Demonstration in Human Maze Learning

Richard A. Panman
State University of Iowa

Sidney J. Arenson
State University of Iowa

Milton E. Rosenbaum
State University of Iowa

Let us know how access to this document benefits you

Copyright ©1962 Iowa Academy of Science, Inc.

Follow this and additional works at: https://scholarworks.uni.edu/pias

Recommended Citation
Panman, Richard A.; Arenson, Sidney J.; and Rosenbaum, Milton E. (1962) "The Value of Demonstration in
Human Maze Learning," Proceedings of the Iowa Academy of Science, 69(1), 490-495.
Available at: https://scholarworks.uni.edu/pias/vol69/iss1/77

This Research is brought to you for free and open access by the IAS Journals & Newsletters at UNI ScholarWorks. It
has been accepted for inclusion in Proceedings of the Iowa Academy of Science by an authorized editor of UNI
ScholarWorks. For more information, please contact [email protected].

Offensive Materials Statement: Materials located in UNI ScholarWorks come from a broad range of sources and
time periods. Some of these materials may contain offensive stereotypes, ideas, visuals, or language.
 Panman et al.: The Value of Demonstration in Human Maze Learning

The Value of Demonstration 1n

Human Maze Learning 1
RICHARD A. PANMAN, SIDNEY J.
ARENSON, AND
MILTON E. RosENBAUM 2

Abstract. In a first experiment, forty subjects were assigned

to one of four groups. Each of these groups received one,
two or three demonstrations, or no demonstration at all of
a task on a bolt-head maze. It was found that number of
demonstrations and reduction of errors per trial were positively
related up to two demonstrations but no additional incre-
ment in performance appeared for three demonstrations. In a
second experiment, error making by a demonstrator was
contrasted with skilled demonstration. Three demonstrations
with errors resulted in significantly fewer errors per trial
than two demonstrations with errors. Three skilled errorless
demonstration was nonsignificantly superior to three demon-
strations with errors.

Deliberate demonstration may be considered as one among

many techniques employed in order to obtain learning of specific
responses. This method is frequently employed in school instruc-
tion at all levels and in training workers and servicemen in ma-
chine related and other performance, as well as in other less for-
mal social contexts. The use of demonstration as a technique
for learning is well known to educators in these areas but the ex-
perimental literature on demonstrational learning has thus far
been negligible. That is, little empirical knowledge is available
as to the conditions under which demonstration produces the de-
sired result or as to what variables facilitate or inhibit the demon-
strational learning process.
The purpose of this paper is to present two rather elementary
studies within a larger project, the aim of the project being to
provide a fuller empirical basis on which to evaluate demon-
strational procedures. We have quite deliberately chosen a task
and developed a procedure that would enable us to meet certain
minimal requirements of examining the demonstrational learn-
ing process. First we looked for a task that required minimal, if
any, development of sheer manipulative skill; a task in which
the motor responses can be considered to be present in the ac-
quired response repertory of the subjects. This choice was based
on our primary interest in the study of factors that influence
simple associations of spec.:ific cues and responses rather than
on more complex skills involving minute topographical details.
1 This research was supported by a grant from tbe National Science Foundation.
2 Department of Psychology, State University of Iowa, Iowa City.

490
Published by UNI ScholarWorks, 1962 1
 Proceedings of the Iowa Academy of Science, Vol. 69 [1962], No. 1, Art. 77

1962] HUMAN MAZE LEARNING 491

A second requirement was that the task would be one that per-
mits trial and error solution. We were not interested in a task in
which the principal contribution of a demonstrator would be to
reveal a trick to solution that would not be available to a naive
performer. Finally, the task we chose had to be flexible enough
to allow the investigation of a variety of independent variables
without drastic shifts in the procedure. Yet the task could not be
so simple that it could be learned completely during one skilled
demonstration. This would prohibit study of the factors that fa-
cilitate or inhibit demonstrational learning.
In this paper we present two studies. The first is concerned
with the reduction in error making as a function of the number
of skilled demonstrations presented. The question we raise is,
does performance continue to improve with an increase in the
number of demonstrations presented?
The second experiment deals with the reduction in error mak-
ing as a function of skilled versus unskilled demonstrations dur-
ing the acquisition process of an observer. The available litera-
ture provides contradictory suggestions concerning the effect of
error-making by a demonstrator.
Herbert & Harsh ( 1944) conducted an observational learning
experiment in which cats observed the puzzle solving activities
of other cats. They found that the observers profit more from
the demonstrator's trial and error performance during learning,
than they do from skilled performances. Klopfer ( 1959) used
birds who observed partners who were learning a discrimination
task. He found that the observers benefited more when the dem-
onstration was presented by a trained bird during acquisition
rather than by an unskilled bird. Research on individual human
learning ( McGeoch & Irion, 1952) suggests that performance of
inadequate responses during acquisition facilitates the acquisi-
tion process. The hypothesis can therefore be offered in the pres-
ent context that error making by a demonstrator will facilitate
learning in contrast to skilled demonstration.
EXPERIMENT I
Procedure
The subjects were 40 male undergraduates from a course in
elementary psychology.
The apparatus consisted of a simple bolt head maze, 12" by
14", mounted in a black, 1/4" plywood frame, 30" x 30'', which
stood in a vertical position on an ordinary 30" high table. The
maze was wired so that when a bolt was contacted with a metal
stylus a circuit was completed which included a light. The first
and last bolts on the maze path were encircled with white mark-
ers, and contact with these stops started and stopped a Stand-
https://scholarworks.uni.edu/pias/vol69/iss1/77 2
 Panman et al.: The Value of Demonstration in Human Maze Learning

492 IOWA ACADEMY OF SCIENCE [Vol. 69

ard Timer calibrated in milleseconds. The timer was on a table

to the right of the maze apparatus, and was shielded from S's
view by a masonite board. A 24 stop maze path was used in this
experiment.
Ten Ss were randomly assigned to each of 4 groups: no demon-
stration (ND), I demonstration (DI), 2 demonstrations ( D2),
and 3 demonstrations ( D3). When S came into the experimental
room, he was seated in a chair facing the maze and E read the
following instructions:
'This is a bolt head maze. Your objective is to learn the cor-
rect path from the upper left-hand corner to the lower right-
hand corner by moving from point to point with this stylus. The
beginning and ending points are marked with the white circles.
Please be sure to make contact with these circles when you be-
gin and when you finish.
vVhen you make a correct contact, the green light at the top
of the board will light up; if you make an incorrect contact, no
light will light up. In going from one contact to another always
move either horizontally or vertically, but never move diagonal-
ly; do not skip over any point.
Move on from point to point with the stylus unless you make
an error. If you make an error, go back to the last point at which
you found a green light and go on from there. Do not start a
trial over again because of errors. Finish it without going back
any further than necessary to get on the right path. You will
continue trials until you complete three errorless trials in a row,
that is, three complete trials in which the green light lights at
every point. There is only one path and it will not change.
As soon as you finish a trial, turn away from the maze and
face the wall over here for a 30 second rest interval. Do not look
at the maze until I instruct you to begin again. Do you have any
questions?"
E indicated by demonstration at appropriate points the var-
ious features of the procedure. The following additional instruc-
tions were given before each demonstration to the demonstra-
tion Ss and were omitted for the no demonstration Ss:
"Before you begin, I will demonstrate the correct path through
the maze."
The interval between demonstrations was the same as that
between trials, and E timed these with a stop watch. During this
interval E recorded the time for the demonstration or trial. Dur-
ing S's performances, E stood to his left rear and recorded er-
rors.
Published by UNI ScholarWorks, 1962 3
 Proceedings of the Iowa Academy of Science, Vol. 69 [1962], No. 1, Art. 77

1962] HUMAN MAZE LEARNING 493

Results
Three response measures were obtained; time, number of
trials, and number of errors. If time or trials are used as the de-
pendent measure, a reasonable correction to apply in order to
compare with trial and error performance is to add the time
spent in demonstrating or the number of demonstration trials to
the respective performance totals. When this is done the effect
of demonstration all but disappears. The time correction is also
dependent on the speed of the demonstration, and there is no
evidence available as to the effect of this variable. More im-
portant, an inspection of the data revealed that many Ss contin-
ued to make only one or two errors for extended trials. This
phenomenon is hidden when considering the time, trials, or total
errors measures. An errors per trial measure reflects an early
advantage most clearly, and the results of this measure are
presented here.
The mean errors per trial and standard deviations for each
of the 4 groups are presented in Table 1.
Table 1. Mean Error per Trial and Standard Deviations for Each of 4
Groups of Subjects.
Group Mean Standard Deviation
ND-- --~3.096-.--------~.684_ __

Dl 2.362b .679
D2 1.405° .488
D.'3 1.041< .628
Note: Means with a letter in common are not significantly different from
each other.
A simple analysis of variance was performed, and the obtain-
ed F-Value of 19.77 was significant beyond the .01 level. The
"critical difference" (Lindquist, 1953, p. 93) was computed. The
Dl group did significantly better than the ND group, and the D2
group did significantly better than the Dl group. Three demon-
strations failed to yield significantly fewer errors per trial than
two demonstrations.
The results show that 1, 2, or 3 demonstrations, when compar-
ed with the trial and error learning of a no demonstration condi-
tion, have a significant effect on the acquisition of a simple task.
Increase in demonstrations beyond 2 did not produce any sig-
nificant decrement in errors per trial. If the trend indicated by
these results continues, it appears that the effect of additional
demonstrations beyond two would progressively diminish. It is
conceivable that a large number of demonstrations would result
in a near errorless performance by S on his first trial. However,
further research may establish that point at which the incre-
ment in performance produced by additional demonstrations
would not be commensurate with the time and effort expended
https://scholarworks.uni.edu/pias/vol69/iss1/77 4
 Panman et al.: The Value of Demonstration in Human Maze Learning

494 IOWA ACADEMY OF SCIENCE [Vol. 69

on these demonstrations. This optimum number of demonstra-

tions would necessarily be dependent upon the complexity of the
task and the purposes of the educator.
EXPERIMENT II
Procedure
The subjects were 30 male and 30 female undergraduates from
a course in elementary psychology.
The apparatus was identical with that used in Experiment I,
except that the maze path consisted of 36 stops and a green light
indicated correct responses while a red light indicated errors.
Ten male and ten female Ss were randomly assigned to each of
these three groups: 2 demonstrations with errors ( E2), 3 demon-
strations with errors ( E3), and 3 skilled demonstrations ( S3).
In the error demonstrations, E made errors at 9 stops. These
stops were selected on the basis of the locations of the most fre-
quent errors committed by Ss in an essentially similar study
employing a 36 stop maze path. Nine errors were used to equate
the total number of responses in the E2 and S3 conditions. Each
error involved 2 additional responses-one off the path and one
back to the path, giving a total of 54 responses in error demon-
strations. Thus, 108 responses occurred in both the E2 and S3
demonstrations. The E3 and S3 conditions were equated with re-
spect to number of demonstrations.
In this experiment, S was instructed that in addition to the
green light on correct contacts, a red light at the top of the
board would light up if he made an incorrect contact. The S was
not informed that errors would be made in the demonstrations.
The remainder of the instructions and the procedure were iden-
tical with those in Experiment I.
Results
The mean errors per trial and standard deviations for each of
the 3 groups are presented in Table 2.
Table 2. Mean Errors per Trial and Standard Deviations for Each of 3
Groups of Subjects
Group Mean Standard Deviation
E2 S.15• l.S05
ES 2.52b 1.066
SS 2.IOb .711
See note below Table 1.
An analysis of variance was performed and the effect of treat-
ment was found to be significant at the .05 level. There was no
significant effect of sex. The "critical difference" was comput-
ed. The E3 and S3 groups both did significantly better than the
Published by UNI ScholarWorks, 1962 5
 Proceedings of the Iowa Academy of Science, Vol. 69 [1962], No. 1, Art. 77

1962] HUMAN MAZE LEARNING 495

E2 group. Three skilled demonstrations did not yield significant-

ly fewer errors per trial than 3 demonstrations with errors.
The hypothesis that demonstration with errors, when compar-
ed with skilled demonstrations, would produce superior learn-
ing, was not confirmed.
The present results agree with those of Klopfer obtained with
birds, and disagree with those obtained by Herbert & Harsh
with cats. The studies cited by McGeoch & Irion found that per-
formance of inaccurate responses facilitated maze learning in
humans. One relevant respect in which many of these latter
studies differed from the present one is that the Ss actually per-
formed and repeated errors and did not merely observe the per-
formance of inaccurate responses.
Such discrepancies in findings can be only be resolved by the
further study of the specific conditions under which demonstra-
tion with errors is superior to skilled demonstration. Differential
instruction may prove to be one of these specific, important var-
iables. A replication of the present experiment with an addition-
al condition in which the S is instructed to avoid the errors he
observes in the demonstration may produce results imilar to
those found when the S actually performed inaccurate respons-
es. Such a study is now being conducted in the Iowa Laborator-
ies.
J.iterature Cited
Herbert, J. J. & Harsh, C. M. Observational learning by cats. J. Comp.
Psychol., 1944, 17, 81-95.
Klopfer, P. H. Social interactions in discrimination learning with special
reference to feeding behavior in birds. Behaviour, 1959, 14, 282-299.
Lindquist, E. F. Design and Analysis of Experiments in Psychology and
Education. Boston: Houghton Miffiin, 1953.
McGeoch, J. A. & Irion, A. L. The Psychology of Human Learning. New
York: Longmans, Green & Co., 1952.

https://scholarworks.uni.edu/pias/vol69/iss1/77 6