2. 1044 ChUd Development
those tasks which present a reahstic challenge ion Three tasks m {jarticular—dichotomous
to a child, relative to his or her cognitive level, sorting, class inclusion, and combinatorial rea-
are likely to engage persistent interest (Harter soning—are developmentally ordered both in
1978b) If this IS true, the problem of pre- terms of the percentage of children correctly
dictmg and manipulating intrinsic motavation performing each task at successive age levels
becomes one of matching task demands to cog- and in terms of the order of attainment of
nitive levels these skills by mdividual children over time
Implicit m this formulation is the familiar This clearly defined sequence of task diflB-
hypothesis that children are attracted to quot;mod- culty affords the opportunity to define more
erately discrepantquot; or quot;moderately complexquot; precisely what is meant by an optimally chal-
stimuli Unfortunately, this intuitively appeal- lengmg task Deci's (1975) model of intrinsic
ing hypothesis has been difficult to test be- motivation suggests that children seek out chal-
cause of problems m specifying degrees of dis- lenging tasks m order to increase feelings of
crepancy and complexity from the subject's competence and self-determination, while Pia-
point of view (Flavell 1977, Kessen, Haith, & get's equilibration model of cognitive growth
Salapatek 1970) A stringent test of the hy- implies that such tasks can only be defined in
pothesis requires a set of stimuli or tasks with relation to the child's cognitive level (Piaget
a clearlv estabhshed developmental sequence of 1977) A combination of these two approaches
difficulty so that one can determine where indi- leads to the prediction that children will choose
vidual children stand m the sequence and how to work on and find most interesting those tasks
much discrepancy there is between their level which are just ahead of them m a developmen-
of understandmg and the cognitive require- tally ordered sequence of tasks
ments of each task Fortunately, the develop-
Method
ment of classification skills provides just such
Subjects —^A total of 90 children (45 boys
a task sequence It is this sequence which was
and 45 girls) from kindergarten and grades 1,
used to predict and manipulate intrinsic moti-
2, and 4 served as subjects They were selected
vation in the present study
from a larger group of 117 children from a
The two major aims of the study were (a) rural Midwestern community on the basis of
to test the hypothesis that intrinsic motivation their performance on a battery of classification
depends on the match between cognitive level tasks (see below) The resulting sample ranged
and task demands and (b) to assess the effects in age from 4-5 to 10-6
of extrinsic rewards and praise on several theo- Procedure —Each child was tested indi-
retically defined levels of intrinsic motivation vidually by the first author on the following
The lnveshgabon was conducted m two parts three tasks (1) Dichotomous sorting sorbng
In experiment 1, children who differed in per- 22 cards with colored shapes on them into two
formance on a pretest battery of classification piles A correct sort is one in which all of the
tasks were provided with a choice of tasks re- cards are sorted along a single dimension, that
quiring different levels of understanding of IS, color, shape, or number (2) Glass inclu
classification The particular tasks the children sion comparing the number of items in a su
selected to work on, the amount of time they perordmate class ( e g , triangles) with the
spent on the tasks, and their ratings of task number of items m a subclass ( e g , red tri-
interest were used as indices of intrinsic moti- angles) (3) Gombinatorial reasoning systemi-
vation In expenment 2, children received cally generating all possible pairs of colored
either rewards, praise, or no rewards for work- chips when given sets of eight different colors
ing on a type of classification task which was
either at, above, or below their predicted level Assessment procedures described by Hoo-
of classification interest In this way, an assess- per, Brainerd, and Sipple (Note 2) were fol
ment could be made of the effects of rewards lowed Based on this assessment, 15 male and
and praise on a wide range of lmtial levels of 15 female subjects were assigned to each of
mtrmsic motivation three cognitive ability groups group 1 (N =
30)—subject has successfully produced one
dichotomy and failed all subsequent tasks
Experiment 1
(mean age = 5-5), group 2 (N = 30)—sub
Both logical analysis (Flavell 1972) and ject has successfully produced three dichoto
empirical results (Kofsky 1966, Hooper, Swin- mies and failed class mclusion and combina-
ton, & Sipple, Note 1) indicate that classifica- torial reasoning (mean age = 6-5), and group
tion skills develop m a strikingly ordered fash- 3 (N = 30)—subject has passed all tasks
3. Danner and Lonkj 1045
through class mclusion and failed combina- should try to work at the center which was
torial reasoning (mean age = 8-8) Of the 117 just right for them At this point, the experi-
subjects assessed, only three individuals did not menter left the room and went to a one-way
follow the proposed difficulty sequence and vision room where he could unobtrusively re-
were not involved further in the study This cord the amount of time spent by the child at
result strongly supports our assumption that each center After the free-choice period, the
these tasks are developmentally ordered The child was asked to rate each center on a five-
remammg 24 subjects ehminated from further point scale of interest and a five-point scale of
consideration had all failed class inclusion, but difficulty
group 2 had already been assigned its full
Results and Discussion
complement of 30 subjects
Table 1 presents the mean time spent at
Approximately 1 week after the pretesting, each center and the rabngs of each center by
each child was brought individually to a mo- the three cognitive ability groups
bile experimental classroom where tasks m
It IS clear that each cognitive ability group
three separate quot;centersquot; were demonstrated by
spent more time with and rated as most inter-
the second author These tasks provided ex-
esting the center involving tasks which were
ploratory and manipulative experience with
one step ahead of the group's pretest level of
sorting (center 1), class inclusion (center 2),
classificabon skill Since the amount of time
and combinatorial reasoning (center 3) The
spent at each center was not independent of
order of introduction of each center was ran-
the tune spent at the other two centers, non-
domized for each child Materials in all centers
parametnc procedures were used m the anal-
were identical, only the instructions describing
ysis of the time-in-center data Three Fried-
what was to be worked on (le , sorting, class
man ANOVA tests were used to compare the
inclusion, or combmatonal reasoning) varied
time spent m each of the centers by each cog-
Materials consisted of three-dimensional objects
nitive ability group The first test mdicated
varying in size, shape, weight, density, volume,
that children from cognitive group 1 spent
color, number, and material Other unrelated
more time in center 1 than they averaged in
materials (puzzles and books) were also pres-
centers 2 and 3, Z = 6 79, p < 01 Similarly,
ent, and the children were told they could
children from cognitive group 2 spent more
play with them if they wished
time m center 2 than they averaged m centers
After the tasks had been explained and 1 and 3, Z = 7 0, p < 01 And finally, cog-
demonstrated, the child was asked to repeat nitive group 3 children spent more time m cen-
the instructions for each center and then was ter 3 than they averaged in centers 1 and 2,
given 10 mm to play at any center (or centers) Z = 6 08, p < 01
of his or her choice Ghildren were told that
There was also a very close correspon-
they might find the tasks at any particular
dence between the children's ratings of the
center too hard or too easy and that they
T-VBLE 1
FREE-CHOICE TIKES AND RATINGS OF CENTERS AS A FUNCTION OF COGNITIVE GROUP
Center 3
Center 2
Center 1
Cognitive group 1
13 42 (31 90)
63 41 (27 62)
Mean time* 504 28 (81 52)
4 78 ( 71)
3 00 (92)
1 26 ( 38)
Interest''
4 84 (43)
3 82 (68)
2 00 (67)
Difficult)'quot;
Cognitive group 2
35 00 (38 05)
460 20 (66 00)
Mean time* 86 53 (51 48)
3 65 ( 82)
1 20 ( 38)
Interest*quot; 3 00 (96)
4 56 ( 82)
2 90 (90)
Difficult* 1 33 (64)
Cognitive group 3
344 96(89 30)
185 21 (71 23)
Mean time* 59 47 (71 90)
1 40 (56)
2 30 ( 81)
3 80 (1 10)
Interest*quot; 3 80 ( 61)
2 00 (61)
1 10 ( 31)
Difficultvquot;
NOTE —f = 30 for each cognitive group SDs in parentheses
I
» Out of total 600-sec free-choice period
> 1 quot; most interesting, to 5 = least interesting
>
• 1 = easiest, to 5 = hardest
4. 1046 Child Development
centers and the actual time they spent m each rewards and mtrmsic mobvabon, experimenters
of them Kendall's r stabstic was used to de- chose tasks which they thought their subjects
termme the degree of association between m- would find interestmg, and they were sabsfied
terest ratings and time spent at the centers to demonstrate that their subjects were m fact
Mean T values for ability groups 1, 2, and 3 wilhng to perform the tasks for no apparent
were 91, 88, and 82, respectively In deter- reward This approach has served as a useful
mining the significance of these mean T values, way to define mtnnsic mobvabon operabonally
each child's T was treated as an individual pnor to the mtroducbon of reward manipula-
score from a populabon of normally distributed bons It has also resulted in groups of subjects
scores with a mean T equal to zero The sig- who are, by this definition, already high m in-
nificance of a deviation from zero was tested trinsic mobvation before extrinsic rewards are
via a one sample t test With sample size of mtroduced (Gondry 1977) Only a few investi-
30 children for each cognitive group, the as- gators have attempted to establish a range of
sumption of normahty was not considered a lntnnsic interest pnor to assessing the effects of
problem (Hays 1973) This procedure indi- rewards (Loveland & Olley 1979, McLoyd
cated that the mean T values for each abihty 1979), and they report that the effects of re-
group greatly exceeded zero ( p < 001), re- wards mteract with initial interest levels
flectmg the fact that children rated most highly
In expenment 2, both the approach and
those centers in which they spent the most
the results of the first expenment were used
time
to define a range of uiibal levels of mtnnsic
In addition, there was an lnverted-U re- mobvation More specifically, all of the children
labonship between interest rabngs and diffi- from the first expenment worked on class-in
culty rabngs That is, tasks rated as either too elusion tasks like those used m center 2 Based
easy or too hard were not rated as interestmg on the predictions supported m expenment 1,
as tasks which were rated as moderately diffi- these class-inclusion tasks should be just beyond
cult This supports Deci's (1975) nobon that the lnibal level of children m cognibve ability
interest is a funcbon of optimal challenge group 2 and therefore highly mtnnsically moti-
With interest and difficulty as ordered vari- vabng On the other hand, these tasks are too
ables ( l e , Likert-type scales), a Kniskal- hard for the children in group 1 and too easy
Walhs ANOVA procedure involving direcbonal for those m group 3 and therefore less lntnnsi-
planned quadrabc trend comparisons was per- cally motivatmg The experiment was a test of
formed (Marascuilo & McSweeney 1977) For the effects of rewards and praise on the subse-
all three centers, the quadratic trends for the quent intrinsic mobvation of these three groups
interest X difficulty ratmgs were significant (p of children
< 05), supporting the inverted-U optimal chal-
One additional feature of the study should
lenge hypothesis
be mentioned here—a test of the effects of
These results indicate that the children praise Deci's (1975) cognitive evaluation the
from each ability group chose to work on, and ory states that receipt of positive feedback for
rated as most lnteresbng, those tasks which engaging m mtnnsically motivated acbvity
they (correctly) perceived to be of moderate should increase the intrinsic mobvation for that
difficulty In other words, they appeared to be activity Verbal praise, therefore, should in-
mtrmsically mobvated to engage in those tasks crease lntnnsic mobvation However, a recent
which were withm their reach but develop- review of the literature indicates that it does
mentally just beyond their current level These not always do so (Deci & Ryan 1980) For
results also supjjort Piaget's contenbon that in- some children, praise seems to act as an ex-
trinsic mobvation is a function of the match tnnsic reward and decreases subsequent intrin-
between a child's cognibve level and the cog-
sic mobvation, while for other children it seems
mbve demands of tasks
to increase feehngs of competence and there-
fore increases mtnnsic mobvabon This differ
Experiment 2 enbal reaction to praise may be related to the
child's concepbon of himseU as an active self-
The results of experiment 1 were so clear
determmmg agent For example, children who
that they provided a umque opportunity to
have an internal locus of control may view
assess the effects of rewards and praise on the
praise as a confirmation of their competence
motivation of children whose initial levels of
and therefore experience an mcrease m moti
motivabon varied both theorebcally and em-
vabon followmg praise On the other hand,
pirically In most previous studies of extnnsic
5. Danner and Lonky 1047
children who have an external locus of control quot;Very fine work quot; The experimenter used two
may be more hkely to view praise as an ex- hand-held counters to insure that each subject
ternal reward which subtly detracts from the received the same number (five) of individual
intrinsic value of a task quot;This hypothesis was positive verbal-feedback statements per session
tested by determining whether any differenbal After completing the activities, each child was
effects of praise on motivabon were related to told that not very many other children of their
individual differences m perceived locus of con- age had done as well No-reward subjects were
trol simply asked to work during these sessions, and
no mention was made of rewards, nor did they
Method
receive any positive verbal feedback
Subjects —The subjects were the same 90
children who served m expenment 1 On the day following the second 30-min
work session, each child returned individually
Procedure—The Bialer (1961) Ghildren's
to the experimental classroom which had re-
Locus of Gontrol Questionnaire was individ-
sumed its expenment 1 appearance The child
ually administered to each child Then 10 sub-
was posttested on the three classificabon tasks
jects (five male and five female) from each
and then told that since there was about 10
cognitive ability group were randomly assigned
mm left before returning to class, he or she was
to a reward condition, 10 to a verbal-praise
free to play m the room The experimenter left
condition, and 10 to a no-reward condition
for the observabon room where time as well as
Each of these nine treatment/ability groups
chosen activibes were unobtrusively recorded
spent two 30-min sessions on consecubve days
At the conclusion of the lO-mm free-choice
working together on class-mclusion problems
session, the child was again asked to rate the
with the second author Pnor to each day's
interest and difficulty level of each center
activities, two introductory lessons were pre-
sented by the experimenter These lessons Results and Discussion
stressed class membership and the notions of Since the target of our reward and praise
subordinate and superordmate classes manipulations was center 2, the results and
discussion focus on the amount of free-choice
During the introductory lessons, reward-
time spent m center 2 after treatments com-
group subjects were shown the award they
pared with that before treatments In other
would receive for working on the tasks—a
words, the free-choice center 2 times from ex-
large quot;good workquot; certificate, which was promi-
periment 1 provided our mam index of initial
nently placed in the experimental room to in-
intrinsic motivabon which we compared with
sure salience (Ross 1975) Rewards were ad-
ministered to all subjects in the reward treat- the posttest center 2 times Analyses of the
ment condition on day 2 of their work sessions data from centers 1 and 3 are not reported here
Praise-group subjects individually received but may be obtained from the authors Table 2
positive verbal feedback statements such as presents the mean time spent at center 2 and
quot;That's the best work you've done so farquot; and the ratings of center 2 as a function of treat-
TBLE 2
CENTER 2 (Class Inclusion) POSTTEST TREATMENT MEANS AND STANDARD DEVIATIONS
Extnnsic Reward
No Reward Verbal Praise
Cogmtive group 1
151 80(77 46)
191 40(141 12)
236 42(174 01)
Mean time*
2 00 (66)
2 00 (74) 1 60 (51)
Interestquot;quot;
3 12 ( 32)
2 93 (1 10) 2 20 (63)
Difficulty'
Cognitive group 2
225 64(79 21)
405 88 (144 00)
Mean time* 441 40(119 25)
3 44 (70)
1 90 (87)
Interest*quot; 1 00 (00)
1 50 (70)
1 90 (56)
Difficultyquot; 2 32 (48)
Cognitive group 3
70 28 (58 85)
251 42 (58 61)
203 40(123 43)
Mean time*
1 90 (73)
1 56 (1 00)
1 90 (73)
Interestquot;quot;
1 90 (56)
1 66 (51)
2 10 ( 73)
Difficultyquot;
NOTE—jV =• lOfor each treatment/ability group SDs 10 parentheses
• Out of total 600-sec fiee-choice period
> 1 — most interesting, to 5 » least interesting
•
« 1 — easiest, to 5 - hardest
6. 1048 Child Development
Significant differences were in cognibve group
ment and cognibve group As in expenment 1,
2, where extnnsic-reward subjects differed sig-
interest ratmgs were highly associated with the
nificantly from no-reward subjects, F = 27 39,
amount of time spent in center 2, mean T
p < 01, and from verbal-praise subjects, F —
values of 82, 71, and 85 for abihty groups
13 63, p < 05 Thus cognibve group 2 was the
1, 2, and 3, respecbvely, and there was a sig-
only group for whom rewards significantly de-
nificant mverted-U relabonship between ratings
creaseid intrinsic motivation It is interesting
of interest and difficulty, p < 05 for quadrabc
to note that the children in this group were also
trends analysis
those who had been predicted and found to be
Analysis of the change m center 2 times high m lntrmsic mobvation prior to the intro-
as a funcbon of treatment and ability group duction of rewards
was done by means of a 3 (treatment) X 3
The treatment X pre-post time at center 2
(cognibve group) X 2 (pre- and posttest time
mteracbon is illustrated m figure 2 Scheffe post
m center 2) ANOVA The mam effect for cog-
hoc compansons were performed on the pre- and
nibve group was significant, F = 115 81, p <
posttest differences between cognibve groups
001, as was the mam effect for treatment, F
for each treatment There were no significant
= 9 16, p < 001 There were two significant
differences between cognibve groups m the no-
lnteracbons—cognibve group X pre-post center
reward condition In the verbal-praise condi-
2 tunes, F = 33 80, p < 001, and treatment X
tion, cognibve group 1 differed from cognitive
pre-post center 2 times, F = 14 63, p < 001
group 2, F = 21 20, p < 01 In the extrinsic-
reward condition, cognibve group 1 was sig-
The cogmtive group X pre-post center 2
nificantly different from both group 2, F =
bmes interaction is illustrated in figure 1
47 29, p < 01, and group 3, F = 13 93, p <
Scheffe post hoc comparisons were performed
05, and cognitive group 2 also differed from
on the pre- and posttest differences between
group 3, F = 9 89, p < 05
treatments at eacli cognitive level The only
No Raward
Ce9nttl«« group 1
too
•00
SOO
too -Cognltlva group 2
E 400
I 400
P
ei 300
M 300
Cognltiva group 1
o Raward
Cognltlva group 3
2 200
S 200 ,V*rbal PralM
Eitrincic Reward
« 100
O 100
0
0
Varbal Pralaa
Cognltlva group 2
•00
•00
SOO
SOO
lo Raward m ^Cognitlva group 2
'Varbal Pralaa S 400
E 400
M 300
'Cognltlva group 3
xtrtnalc Raward 1200
I 200 Cognltlva group 1
100
o 100
0
0
Extrlnalc Raward
Cognltlva group 3
•00
•00
SOO
SOO
m
I 400
e 400
ti 300
H 300
-Varbal Pralaa lognlttva group 2
I 200
-No Raward
S 200 :ognltlva group 1
O 100
» 100 gnitlva group 3
•ExtrtiMle Reward
0
Pra
FIG 2 -Pre- and posttest center 2 bmes as a
FIG 1 —Pre- and posttest center 2 bmes as a
(unction of treatments
function of cognitive group
7. Danner and Lonky 1049
Effects of Rewards smaller increases in classification performance,
Consider first the effect of extrinsic re- when mtnnsic motivation for a difficult task was
wards The highly motivated group 2 children initially low, rewards had httle effect on moti-
showed a decrease in interest following re- vation and led to greater increases in classifi-
, wards, while the less highly motivated children cation performance than did no rewards
from groups 1 and 3 did not suffer these ill
Effects of Praise
' effects In fact, the group 1 children actually
The effects of verbal praise were very
* increased shghtly on our measure of intrinsic
similar to those of extrinsic rewards but not as
motivation when they were rewarded for work-
strong There were no significant effects of
ing on center 2 tasks—tasks which were more
praise on intrinsic motivation in any of the cog-
difficult than those they tended to choose in
nitive groups and only a slight suggestion that
the first experiment This finding could be in-
praise sometimes improved subsequent test per-
terpreted to mean that rewards might increase
formance Nevertheless, these nonsignificant
a child's interest in tasks which are more chal-
overall effects may have masked important in-
' Ienging than those he or she would ordinarily
dividual differences m response to praise Our
', choose However, a similar increase in the
hypothesis was that children with an internal
amount of tune spent m center 2 occurred
locus of control would view praise as a con-
among the group 1 children m the no-reward
firmation of their competence and would there-
condition Therefore, extnnsic rewards were no
fore show an increase m lntrmsic motivation
more effective m increasing the motivation of
following praise, while children with an ex-
children whose initial level of interest was low
ternal locus of control would view praise as an
than were simple requests to work on the tasks
external reward and would respond to it as if it
Another way to look at the effects of rewards were a reward Since cognitive group 2 was
IS to consider their impact on learning Al- most strongly affected by extrinsic rewards, any
though the experiment did not involve exten- differential reaction to praise as a function of
sive training, there was enough interaction with locus of control should be most evident in this
classification tasks that one might expect some group If group 2 children with an external
improvement in classification performance The locus of control really do view praise as a re-
children m cognitive group 1 had produced ward, then they should show a decrease m in-
only one complete sort on the dichotomous trinsic motivation following praise similar to the
sorting task during the pretest This corre- decrease produced by a reward
sponds to a score o^ 10 out of 30, according to
f
The tests of these hvpotheses were based
the scoring procedures described by Hoojjer,
on a comparison of the amount of time spent
Brainerd, and Sipple (Note 2) On the post-
in center 2 by children who were high or low
test, the average scores m each group 1 treat-
on a measure of internal locus of control A
ment condition were as follows extrinsic re-
median split of scores on the Bialer (1961)
ward = 24, verbal praise = 22, and no reward
Children's Locus of Control Questionnaire was
= 14 Dunn's procedure comparing all pairwise
used to create the high and low internal locus
means indicated that the extrmsic-reward chil-
of control groups This questionnaire yields
dren performed significantly better than the
scores ranging from 0 to 23, with higher scores
no-reward children, d= 92, p < 05 The chil-
indicating greater internal locus of control The
dren in cognitive group 2 had failed class in-
mean scores were 10 16, 11 36, and 11 60 for
clusion on pretest Their posttest averages on
cognitive groups 1, 2, and 3, respectively An
the class-mclusion task, on a scale of 0-10,
analysis of variance conducted on these scores
were as follows 6 for the no-reward treatment,
indicated no significant effects of cognitive
3 for the verbal-praise treatment, and 2 for
group, sex of subject, or the mtera9tion be-
the extrinsic-reward treatment The no-reward
tween the two
group performed significantly better than the
extnnsic-reward group, Z = 2 0, p < 05 None Figure 3 presents the mean times spent by
of the children m cognitive group 3 passed the verbal praise children m center 2 as a function
combinatorial reasoning task at either pre- or of cognitive group and high or low internal
posttest, thus no performance differences are locus of control Our primary interest was m
available for this group the effects of praise on high and Jow internal
locus of control children from group 2—the
In summary, rewards had the following
group which had been most strongly affected
effects When intrinsic motivation for an ap-
by extnnsic rewards As shown in figure 3, only
propriately challenging task was initially high,
group 2 children with an external locus of con-
rewards decreased motivation and produced
8. 1050 Child Development
•00
so
s
SOO
490
400
1 9S0
u
900
M
tso
200
£
1 ISO
100
SO
0
FIG 3 —Pre- and posttest center 2 times as a function of cogmtive group and locus of control
els of lntnnsic motivation before they develop
trol spent less time m center 2 following praise
an internalized self-reward system, even the 5-
This difference from pre- to posttest was sig-
year-olds in our study freely chose those tasks
nificant, t — 4 04, p < 01, and it indicates that
which provided an optimal cognitive challenge
these children had a negative reaction to praise
which was similar to the reaction which other Certainly there is more to motivation than
group 2 children had to rewards the match between cognitive ability and task
demands, but it is both interesting and unpor-
Implications tant that the children chose to work on those
tasks which theoretically would provide opti
The results of these two studies have sev-
mal stimulation for cognitive growth It is in-
eral implications for the understanding of in-
teresting because it supports Piaget's disequi
trinsic motivation and its relabonship to extrin-
libnum model of cognibve growth, and it is
sic motivation First, Piaget's (1977) disequi-
important because it suggests that at least part
hbnum model of cognitive growth provides a
of the teacher's difiBcult problem of matching
useful theoretical framework for defining m-
tasks to children can be solved by providing
tnnsic mobvation Its focus on the relabonship
children wath more choices than they are typi-
between children's cognitive level and the cog-
cally offered This is consistent with Duck-
nitive demands of tasks leads to clear predic-
worth's (1973, 1979) suggestion that if a cli-
bons about which children will be interested
mate of respect for children's ideas is created
m which acbvities These predictions were con-
in a classroom and a choice of tasks is provided
firmed at all three developmental levels sam-
children can be trusted to select those tasks
pled m our study Children spent the most
which will provide stimulation appropriate to
tune on and rated as most interesting those
their intellectual level The fact that we used
activities which were one step ahead of them
tasks involving important cognitive skills makes
m a developmental progression The children
the leap to such speculations about classroom
not only chose the predicted tasks and found
learning a little easier to justify than has been
them most mteresting, but they also considered
the case with previous studies of intrinsic moti
them to be of intermediate difiBculty These
vation which have used magic markers, drums,
results are consistent -with Harter's (1978b)
or puzzles as stimuli (Notz 1975)
finding of a curvihnear relationship between
task difficulty and intrinsic pleasure as indi- A second lmphcation of the studies con
cated by the amount of smiling while solving cerns the effects of rewards on intrinsic moti
problems While Harter (1978a) has suggested vation and task performance Put most simply.
that young children may not exhibit high lev- rewards do not nave a uniform effect on chil
9. Danner and Lonky 1051
dren's motivation The impact of rewards in- locus of control children viewed praise as a
teracted quite strongly with the initial develop- reward designed to keep them on task, and
mental and motivational levels of the children they lost interest when this reward was with-
drawn
For highly motivated children, rewards de-
creased subsequent motivabon and also led to
However these findings are interpreted,
smaller improvements in task performance than
they suggest that the effects of rewards and
those achieved by highly motivated children
praise on intrinsic motivation are more compb-
who were not rewarded But for the younger
cated than they first appear These effects de-
children, whose intrinsic motivation for class-
pend on children's initial levels of intrinsic
inclusion tasks was initially low, rewards were
motivation which are, m turn, affected by their
not detnmental to subsequent motivation and
cognitive levels, and they are complicated fur-
led to greater increases m task performance
ther by differences in the ways m which chil-
than did no rewards Perhaps for the younger
dren interpret rewards and other forms of feed-
children the controlling aspects of an extnnsic
back Therefore, the routine use of powerful
reward (Deci 1975) served to focus their ac-
external reward systems with groups of children
tivity on a more difiBcult task than thev ordi-
—given the inevitable ability differences be-
narily would have chosen for themselves (Har-
tween them—IS bound to have mixed effects
ter 1978a)
(Lepper 1980) The irony is that the most det-
rimental effects seem to be reserved for those
It IS important to note, however, that the
highly motivated children for whom no reward
slightly positive effect of rewards on the in-
system is necessary m the first place
trinsic motivation of young children was no
greater than the effect evident m the no-reward
condition That is, exposure to relatively diffi- Reference Notes
cult tasks m an open settmg where there was
1 Hooper, F H , Swinton, S S , & Sipple, T S
time to manipulate and explore seemed to gen-
Logical reasoning m middle childhood a study
erate as much interest as did this experience
of the Piagetian concrete operations stage Re-
plus rewards While the effects of extrinsic re-
search summary presented at the Nineteenth
wards might have been different if a more
International Congress of Apphed Psychology,
powerful multiple-trial reward procedure had
Munich, July 30-August 5, 1978
been used (Vasta 1981), our findings support
2 Hooper, F H , Brainerd, C J , & Sipple, T S
Lepper's (1980) contention that rewards are
A representative series of Piagetian concrete
often either detrimental to subsequent intrinsic
operations tasks (Theoretical Paper no 57)
mobvation or superfluous and, m either case,
Madison University of Wisconsin, Research
should be used with caution
and Development Center for Cognitive Learn-
ing, 1975
A final lmphcation of the studies is based
on the effects of praise on intrinsic motivation
Praise had generally positive effects on moti- References
vation with one notable exception Highly
Appel, M H , & Goldberg, L S (Eds ) Topics in
motivated children (cognitive group 2) tended
cognitive development Vol 1. Equtkbratton
to decrease m motivation following praise But
theory research and application New York
this slight group decline in motivation masked
Plenum, 1977
important individual differences in response to
Bialer, I Conceptualization of success and failure
praise Those group 2 children with an inter-
m mentally retarded and normal children
nal locus of control increased m motivation fol-
Journal of Personality, 1961, 29, 303-320
lowing praise, while those with an external
Calder, B J , & Staw, B M Self-perception of
locus of control decreased in motivation m re-
intrinsic motivation Journal of Personality and
sponse to praise These findings suggest that,
Social Psychology, 1975, 31, 599-^5
to use Deci's (1975) terms, the informational
Condry, J Enemies of exploration self-initiated
aspects of praise were most salient to the in-
versus other-initiated learmng Journal of Per-
ternal locus of control children, while the con-
sonality and Soctd Psychology, 1977, 35, 459-
trolling aspects of praise were more salient to
477
the external locus of control children That is,
Deci, E L Effects of externally mediated rewards
children with an internal locus of control con-
on lntruisic motivation Journal of Personality
sidered the praise positive information about
and Social Psychology, 1971, 18, 105-118
their competence, and they transferred some of
this positive feeling to the task, while external Deci, E L Effects of contingent and non-contm-
10. 1052 Child Development
tal psychology New York Holt, Rinehart &
gent rewards and controls on mtnnsic mobva-
Winston, 1969
bon Organizational Behaoior and Human Per-
formance, 1972, 8, 217-229 Langer, J International aspects of cogmbve orga-
nization Cognitwn, 1974, 3 , 9-28
Deci, E L Intnnstc mottvatton New York Ple-
Lepper, M R Intrinsic and extnnsic mobvabon
num, 1975
m children detnmental effects of superfluous
Deci, E L , & Ryan, R M The empirical explora-
social controk In W A Collins (Ed ), Min-
tion of intrinsic motivational processes In
nesota symposia on chdd psychology Vol 14
L Berkowitz (Ed ), Advances m experimental
Hillsdale, N J Erlbaum, 1980
soctd psychology Vol 13. New York Aca-
demic Press, 1980 Lepper, M R, & Creene, D (Eds ) The hidden
costs of reward Hillsdale, N J Erlbaum,
Duckworth, E The having of wonderful ideas In
1978
J Raph & M Schwebel (Eds ), Piaget m the
Lepper, M R , Creene, D , & Nisbett, R E Un-
classroom New York Basic, 1973
dermuung children's mtnnsic interest with ex-
Duckworth, E Either we're too early and they
tnnsic rewards a test of the quot;over-jusbfica-
can't learn it or we're too late and they know
tion' hypothesis Joumd of Persorudity and
it already the dilemma of quot;applying Piaget'
Social Psychology, 1973, 2 8 , 129-137
Harvard Educational Review, 1979, 4 9 , 297-
312 Loveland, K, & Olley, J The effect of external
reward on interest and quality of task per-
Flavell, J H An analysis of cognitive-developmen-
formance in children of high and low mtnnsic
tal sequences Genetic Psychology Monographs,
1972, 86, 279-350 mobvation ChM Development, 1979, 50,
1207-1210
Flavell, J H Cogntttve development Englewood
McLoyd, V The effects of extnnsic rewards of
Chffs, N J Prentice-Hall, 1977
differential value on high and low mtnnsic
Harter, S Effectance mobvabon reconsidered to-
interest Chdd Development, 1979, 50, 1010-
ward a developmental model Human Devel-
1019
opment, 1978, 2 1 , 34-64 (a)
Marascuilo, L A , & McSweeney, M Nonpara-
Harter, S Pleasure derived from challenge and the
metnc and distribution free methods for the
effects of receiving grades on children's diffi-
socud sciences Belmont, Calif Wadsworth,
culty level choices Chdd Development, 1978,
1977
49, 788-798 ( b )
Mischel, T Piaget cogmtive conflict and the moti-
Hays, W Statistics for the social sciences (2d ed )
vabon of thought In T Mischel (Ed ), Cog-
New York Holt, Rinehart & Winston, 1973
ntttve development and eptstemology New
Hunt, J McV Intrinsic mobvabon and its role in
York Academic Press, 1971
psychological development In D Levine
Notz, W M Work motivation and the negative
(Ed ), Nebraska Symposium on Motivation
effects of extnnsic rewards American Psy-
Lmcoln University of Nebraska Press, 1965
chologist, 1975, 3 0 , 884-891
Kessen, W , Haith, M , & Salapatek, P Infancy In
P H Mussen (Ed ), Carmtchael's manual of Piaget, J Problems of equihbrabon In M H Ap-
child psychology (3d ed ) Vol 1. New York pel & L S Coldberg (Eds ), Topics m cog-
Wiley, 1970 ntttve development Vol 1. Plenum, 1977
Ross, M Salience of reward and lntnnsic mobva-
Kofsky, E A scalogram study of classificatory de-
tion Journal of Personality and Soctal Psy-
velopment Chdd Development, 1966, 37,
chology, 1975, 32, 245-254
191-204
Vasta, R On token rewards and real dangers-a
Langer, J Disequilibrium as a source of develop-
look at the data Behavior Modification, 1981,
ment In P Mussen, J Langer, & M Coving-
5, 129-140
ton (Eds ), Trends and issues m developmen-