Title: Why do we have to do this for money?
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00103072/00001
 Material Information
Title: Why do we have to do this for money? an ecological investigation of rewards and learning
Physical Description: xiv, 189 leaves : ill. ; 28 cm.
Language: English
Creator: Graham, Rita L., 1949-
Publication Date: 1982
Subjects / Keywords: Reward (Psychology)   ( lcsh )
Motivation (Psychology)   ( lcsh )
Learning   ( lcsh )
Genre: bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis (Ph. D.)--University of Florida, 1982.
Bibliography: Includes bibliographical references (leaves 183-187).
Statement of Responsibility: by Rita L. Graham.
General Note: Typescript.
General Note: Vita.
 Record Information
Bibliographic ID: UF00103072
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000352645
notis - ABZ0621
oclc - 09806851

Full Text







For my mother, Louise, who was my first and best teacher;

my husband, Jim, who has lovingly supported my education;

and my son, Jay, who gives meaning to all that I have learned.


The author wishes to acknowledge the following people,

without whose help the present research might never have

been completed:

Dr. Lee J. Mullally, chairman of the doctoral committee,

for his patience and encouragement. It was Dr. Mullally who

first gave the author the courage to go beyond her field in

search of answers to difficult educational questions.

Dr. John K. Bengston, co-chairman of the doctoral

committee, for his untiring dedication to this project.

Dr. Bengston has been more than an inspiring teacher and

advisor. He has been a friend.

Dr. Laurel Dickerson, Dr. James J. Algina, Dr. William

D. Hedges, and Dr. Kenneth A. Christiansen, members of the

doctoral committee, for their helpful ideas and careful

analyses of the manuscript.

Mr. Peter Reynolds, President of the Brio Scanditoy

Corporation, for graciously donating the Labyrinth games

which the author needed for the project.

Dr. John C. Condry, Professor of Human Development,

Cornell University, for his thoughtful critique of the

author's original prospectus, which helped immensely in

the redesign and implementation of the experiment.


Mr. Roy Thayer and Mr. Jarvice Caudron, staff members

of the Office of Instructional Resources, University of

Florida, for the technical expertise and equipment which

they contributed to the experiment.

Ms. Kathy Clower and Ms. Le Anne Brown, fellow doctoral

candidates, for taking time away from their own dissertations

to help the author.

The faculty, staff, and students of the P. K. Yonge

Laboratory School, University of Florida, for all their

kindness and cooperation during the weeks of the experiment.

Behind every achievement exists the motivation which is

at the foundation of it and which in turn is strengthened and

nourished by the accomplishment of the undertaking. Here

there is the greatest difference and it is of greatest impor-

tance to the educational value of the school.

-Albert Einstein

Out of My Later Years, 1950
The Philosophical Library, Inc.,
New York, N. Y.



ACKNOWLEDGEMENTS ........ ............................... iii

LIST OF TABLES .......................................... x

ABSTRACT .............................................. xiii


FRAMEWORK FOR THE STUDY ...................... 1

Problem Statement .............................. 1
Justification for the Study .................. 1
An Ecological Framework for the Study
of Motivation ............................. 6
The Ecological Tradition .................. 6
An Ecological Definition of
Motivation .............................. 9
The Concept of Competence and Some Other
Assumptions About Learning ................ 12
Process Versus Performance ................ 13
Learning Environments ..................... 14
Learning Cycles .......................... 18
Outline of the Present Research .............. 23
General Hypotheses of the Study ........... 23
Experimental Conditions .................. 24
Research Questions and Specific
Hypotheses ............................. 26
Summary ..................................... 31

II. A REVIEW OF RELATED LITERATURE .................. 32

Introduction .................................. 32
The Effect of Rewards on Interest ............ 32
The Effects of Rewards on Performance
Quality ................................... 35
The Effect of Rewards on Process
Variables ................................. 38
Incentive Characteristics .................... 50
Task Characteristics .......................... 52
Experimental Designs ......................... 54
Summary ...................................... 58



Chapter Page

III. THE DESIGN OF THE EXPERIMENT .................... 59

Introduction ................................ 59
The Sample ................................... 61
The Learning Activity ........................ 62
Experimental Setting ......................... 63
Contrived Reward ............................ 66
Procedure .................................... 68
First Experimental Session ................ 68
The Second Experimental Session ........... 73
The Postexperimental Session .............. 73
Data Collection for Within-Session
Variables ................................. 75
Qualitative Measures ..................... 75
Quantitative Measures .................... 78
The Questionnaire ......................... 79
Data Collection for Postexperimental
Variables ............................. 80
Data Analysis ................................ 81

IV. RESULTS ........................................ 84

Introduction ................................. 84
Learning Process .......... .. ............... 86
Percentage of Intervals Containing
Match Play (Hypothesis 1) .............. 86
Practice Time Preceding the First
Match of a Session (Hypothesis 2) ...... 89
Engaged Time (Hypothesis 3) .............. 92
The Mean Difficulty Level of Boards
Used During Play (Hypothesis 4) ........ 92
Instances of Employing Minimax
Strategies (Hypothesis 5) .............. 98
Social Interactions .......................... 100
Number of Instructional Events
(Hypothesis 6) ......................... 100
Instances of Interference
(Hypothesis 7) ......................... 100
Number of "Look-at-Me" Statements
(Hypothesis 8) ........................ 102
Number of Arguments (Hypothesis 9) ........ 106
Instances of Praise (Hypothesis 10) ....... 106
Instances of Encouragement
(Hypothesis 11) ....... ................ 106
Feelings of Frustration (Hypothesis 12) ... 109



Chapter Page

Number of References to Cheating
(Hypothesis 13) ........................ 112
Instances of Cheating a Partner
(Hypothesis 14) ......... ............... 112
Skill Development ........................... 115
Mean Scores on the Skill Test
(Hypothesis 15) ......................... 117
Accuracy of Predictions on the
Skill Test (Hypothesis 16) ............. 117
Ratings of Game Difficulty
(Hypothesis 17) ............... ..... .... 117
Expressions of Skill Requirements of
the Game (Hypothesis 18) ............... 120
Internality of Attributions
(Hypothesis 19) ............. .. ........ 123
Interest ..................................... 125
Expressed Liking of the Game
(Hypothesis 20) ........ ............... 125
Importance Placed on Competition
(Hypothesis 21) ... ................... 127
Reengagement (Hypothesis 22) .............. 127
Summary of Results ................ .......... 132


Introduction .................................. 134
Major Results ................................. 134
Interpretation of Results ................... 136
Learning Process .............................. 138
Social Interactions ........... ............... 145
Skill Development .......... ................. 151
Interest ................. ........................ 153
Implications ..... ........................... 157
Summary ...........................*.......... 164


A OBSERVATION TOOL ............................... 168

B QUESTIONNAIRE ....... .......................... 172

C POSTEXPERIMENTAL INSTRUMENT ..................... 175

D CORRELATION MATRICES ............................ 177

E "ARGUMENTS ABOUT CHEATING" ...................... 181




REFERENCES ............................................ 183

BIOGRAPHICAL SKETCH ................................... 188


Table Page

1 Summary Table of Variables, Predictions,
and Analysis of Variance Procedures
for the Hypotheses of the Study ............ 85

2 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Percentage of Intervals Containing
Match Play ................................... 87

3 Specific Comparison for Percentage of
Intervals Containing Match Play ............ 88

4 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Amount of Practice Time Preceding
the First Match ............................. 90

5 Specific Comparisons for Amount of
Practice Time Preceding the First
Match ........................................ 91

6 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Percentage of Engaged Time ................. 93

7 Specific Comparisons for Engaged Time ......... 94

8 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Mean Difficulty of Boards Used
During Play ................................ 96

9 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Mean Variability of Boards Used
During Play ................................ 97

10 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Instances of Employing "Minimax"
Strategies ................................. 99

LIST OF TABLES--(Continued)

Table Page

11 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Number of Instructional Events ............. 101

12 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Number of Instances of Interference ........ 103

13 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Number of "Look-at-Me" Statements .......... 104

14 Specific Comparisions for Number of
"Look-at-Me" Statements ................... 105

15 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Number of Arguments ........................ 107

16 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Number of Instances of Praise .............. 108

17 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Number of Instances of Encouragement ....... 110

18 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Feelings of Frustration .................... 111

19 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Number of References to Cheating ........... 113

20 Specific Comparisons for Number of
References to Cheating ..................... 114

21 Means, Standard Deviations, and 2 X 4
Split Plot ANOVA Summary Table for
Number of Instances of Cheating a
Partner .................................... 116

22 Means, Standard Deviations and ANOVA
Summary Table for Mean Scores on
Skill Test ................................. 118

LIST OF TABLES--(Continued)

Table Page

23 Means, Standard Deviations and ANOVA
Summary Table for Accuracy of
Predictions on the Skill Test .............. 119

24 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Ratings of Game Difficulty ................. 121

25 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Expression of Skill Requirements of
the Game ................................... 122

26 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Internality of Attributions for
Winning or Losing ......................... 124

27 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Expressed Liking of the Game ............... 126

28 Means, Standard Deviations, and 2 x 4
Split Plot ANOVA Summary Table for
Importance Placed on Competition ........... 128

29 Specific Comparisons for the Importance
Placed on Competition: Main Effect
of Condition ............................... 129

30 Means, Standard Deviations, and ANOVA
Summary Table for Reengagment of
the Game ................................... 130

31 Specific Comparisons for Reengagement
of the Game ................................ 131


Abstract of Dissertation Presented to the Graduate
Council of the University of Florida in Partial Fulfillment
of the Requirements for the Degree of Doctor of Philosophy



Rita L. Graham

December 1982

Chairman: Dr. Lee J. Mullally
Major Department: Curriculum and Instruction

This study was based on an ecological approach to

motivational research. Like other research in its tradi-

tion it focused on the influence of "extrinsic" rewards on

"intrinsic" motivation and learning. The terms intrinsic

and extrinsic are examined and redefined, however, in light

of an ecological definition for motivation.

The study examined the influence of a contingent

monetary reward on the learning process, skill development,

social interactions, and attitudes of 40 seven- to ten-year-

old boys learning to play a game of skill. The experiment

employed an ecological laboratory design in which subjects

were allowed a minimum of free choice in engaging, processing,

and terminating the activity. Subjects were randomly paired

and pairs were randomly assigned to four treatment conditions.


The treatment covered two sessions. During the first session,

two groups (reward) were offered money for playing the

game; two other groups (no-reward) were allowed to compete

if they so desired, but were not offered money. During the

second session, one of the reward groups had the offer of

money withdrawn, while one of the no-reward groups had the

offer of money introduced. The data were analyzed by means

of a 2 x 4 split plot ANOVA to determine whether reinforcement

for competition produced differences among the groups on

twenty-two dependent measures and whether the degree and

direction of the differences changed when the reward was

introduced or withdrawn during the second session.

Results indicated that reinforcement for competition

significantly increased the amount of competitive play,

reduced the amount of time subjects spent practicing, and

reduced overall game-engaged time. Subjects who had been

rewarded at any time for competition also spent significantly

less time engaged in playing the game during a postexperi-

mental free-choice period a week later. There were no

significant differences among the groups on measures of

skill development. Two measures of social interactions

were significant, but differences ran contrary to predic-

tions: No-reward subjects were more boastful and made more

references to cheating while playing the game. Descriptive

data are presented in the author's discussion and interpre-

tation of these results.




Problem Statement

The present study was designed to investigate how an

externally controlled source of reward would affect learning

process, skill development, social interactions, and attitudes

of children toward a learning activity. The effect of an

extrinsic reinforcer was examined in four different, but

related, learning conditions. All of the conditions were

designed to insure high intrinsic motivation for the activity

and to be representative of real world learning environments.

In three of the situations subjects were allowed to compete,

at various time, for an extrinsic reinforcer. The question

of interest was whether this reward, and the ensuing competi-

tion for it, would alter subjects' learning processes, skill

development, cooperative behavior, and attitudes toward the


Justification for the Study

Most children come to their first school experience

with enthusiasm. They are eager for exploration and they


approach learning situations with determination and diligence.

It is not unusual to hear them say they "love" school. More-

over, children have already mastered many psychomotor and

cognitive skills. They have learned to walk, ride bicycles,

play a variety of games, and communicate rather skillfully in

one or more languages. And they have developed these profi-

ciencies often without the benefit of any formal instruction.

How to maintain their enthusiasm and motivate children to

learn at increasingly complex levels has been a perennial

educational concern. The present study addressed this

fundamental educational question: How can schools best

take advantage of children's natural enthusiasm and moti-

vation to learn while helping them develop the skills they

need to deal with the demands of their world?

Various approaches for enhancing student motivation

in schools have been tried. Prevalent among these has been

the use of incentives--in the form of course grades, points

on examinations, or tokens. Such incentives have, indeed,

proven effective for inducing children to do much of what

is required of them in schools. Teachers have successfully

used incentives to encourage children to read more books,

write longer compositions, solve more math problems, and

expand their vocabularies. Behavioral problems and

absenteeism have also been remedied by the careful applica-

tion of reward systems. Because of their success, reward

systems have become so commonplace in schools that it is

difficult to imagine a school functioning, or children

learning, without them.

But the truth of the matter is that children do learn

without the aid of externally controlled reward systems.

Outside the school setting, children learn efficiently and

rapidly those things which are important to their growth

and development. They do so without any formal curricular

or instructional plan, without the help of trained teachers,

and without any artificial inducements. For example,

children acquire language at a very early age. Incoherent

babblings gradually become shaped, through listening and

experimenting, into articulate sounds. Those sounds grow

into words and words grow into sentences. By the time

children go to school, they are able to use language to

tell stories, recount experiences, question, explain, and

build convincing arguments. Of all the skills which child-

ren develop in their first five years, perhaps in their

entire lifetimes, language is probably the most complex.

Yet it is universally mastered.

Why is this? Why do children seem to have less trouble

mastering the complexities of language than they have

learning the names of the fifty states, or the multiplica-

tion tables, or the anatomical parts of a frog? Why does

the mother feel no need to give grades or promise token

rewards while the school teacher does? It is not because

the skills developed outside schools are easier to master

or characteristically different from the skills taught in

schools. It is because motivational aspects of the environ-

ments in which the learning takes place are different.

In the world, babies learn to talk as a result of

environmental demands. Talking allows them to tell people

how they feel, to ask for things they want, to understand

and communicate with other people. Talking allows or

"affords" something which being silent, or even babbling,

does not. The act of talking and the consequences) of

talking are structurally linked. Hence, the meaning of

talking in the world is clear to the child. In the school

environment, however, the meaning of what is to be learned

is not so clear. Children learn the names of the fifty

states, for example, because they are told it is important,

not because its importance is readily apparent to them.

This is not to say that learning the names of the states

does not afford something. Certainly, these names afford

children a better understanding of their nation and an

improved ability to communicate and learn about that

nation. But this affordance is not readily perceived.

Children do not see how learning these names will allow

them to do things which they want to do and cannot do

already. If they fail to learn these names, children


feel no personal discomfort, no sense of loss, no frustra-

tion. The school environment provides no structural link

between the act of learning the names of the states and

the consequences) of that act. Therefore, schools have

found it useful to impose artificial consequences in the

form of rewards and punishments.

Questions about the utility of such rewards have been

raised by certain motivational research which indicates

that the use of artificial inducements may have undesirable

effects. This research is based upon the theory that human

motivation has, at least, two dimensions. Along one

dimension, our behavior can be motivated by external

factors such as environmental rewards and punishments.

Along the other dimension, it can be motivated by factors

which inhere in the individual and/or his behavior. These

dimensions have been called, respectively, extrinsic and

intrinsic motivation. Studies in this field have shown

that extrinsics, while increasing the frequency of reinforced

responses, can have a negative effect when used to motivate

behavior that would otherwise be freely chosen. A popular

way of stating this relationship is to say that extrinsic

rewards tend to undermine intrinsic motivation. The effect

appears to be most pronounced when receipt of a material

reward is made contingent upon performance of an already

interesting activity. Findings have been relatively

consistent across age levels, types and magnitudes of

reward, task requirements, and procedures. Even more

important from an educational perspective, extrinsic

rewards have been found to interfere with problem solving,

creativity, insight, and other processes considered criti-

cal to learning. This research is reviewed in detail in

Chapter II.

The present study represents a further attempt to

document the influence of extrinsic rewards on intrinsic

motivation and learning. The primary aim of this study

was to test the influence of an extrinsic reward in a

setting where learning could be expected to occur in

response to other demands. These other demands will be

called ecological rewards and their importance will be

made clear as the theoretical framework for this research

is described.

An Ecological Framework for the
Study of Motivation

The Ecological Tradition

The impetus for the present study owes greatly to the

pioneering efforts of James Gibson (1979), who established

an ecological foundation for the study of visual perception.

Gibson's work was committed to the principle of direct,

rather than mediated experience. The organism and the

environment constitute a system, he said, in which the

smallest unit of ecological analysis must be a dynamic

event, rather than a static stimulus, and in which the

behavior of the organism is viewed as a process of adapta-

tion to his experience of these events. How the organism

adapts will depend upon the "ecological information"

available in the ambient flux of stimulation, and upon

the pick-up of this information by the organism. Informa-

tion is considered "ecological" if it specifies nontrivial

properties which facilitate achievement or maintenance of

an adaptive relationship with the environment (Shaw et al.,

1974). Another way of stating this is to say that ecologi-

cal inforamtion specifies what things "afford" the organism

as he interacts with his surroundings. Thus, what one

perceives are not meaningless characteristics of things,

but, rather, what things afford in a particular behavioral/

environmental context. The ecological study of human

perception, then, is the study of a particular aspect of

"organism/environment mutuality" (Alley and Shaw, 1981,

p. 140).

The application of Gibson's ideas to the field of

motivational psychology awaited the discernment of John

Condry (1977), who first characterized task-extrinsic

incentives as bits of ecological information which

specify a particular motivational context. This

conceptualization has guided a research program at Cornell

University, and is summarized in an article by Condry and

Chambers (1978) as follows:

We do not think of rewards as "undermining"
interest or intrinsic motivation, although we
recognize the popular usage of this term. Talking
in this way, however, directs our thinking and
theorizing; it suggests that there is an ongoing
stream of intrinsically motivated behavior that
is somehow "undermined" by the offer of a task-
extrinsic incentive. In our research, we have
conceptualized these facts differently. We have
assumed that rather than undermine anything, the
offer of a reward acted as a "signal" to the
individual to call forth a given sequence of
activity. Thus, task-extrinsic rewards are part
of an informational array specifying the appro-
priate actions to take and perhaps the appropriate
way to conceptualize the "causes"of one's
behavior. The absence of a reward as well as
other aspects of the situation are part of an
informational array we call a "context," which
elicits a different pattern of motivated
behavior. . Thinking about it in this way
has some advantages:

1. It takes our attention away from the
reward "per se" and focuses it on the entire
complex of information facing the individual.

2. It avoids the conception of motivation
as being either "intrinsic" or "extrinsic."
Rather, we think the information specifying these
different contexts and what they "afford," what
they offer, exists and is discriminated by
individuals. This discrimination is reflected
in both behavior and conceptualization. . .
Thus, we avoid a false dichotomy (not a false
distinction) and leave open the question of
whether there are other informational arrays,
other "contexts" that give rise to other
patterns of motivated behavior. . (1978,
p. 62)

What Condry and Chambers offer here is an orientation

to motivational theory which centers on the fact of

organism/environment mutuality. This orientation avoids

the implication that extrinsics are either good or bad.

It also forces one to consider the effect of rewards

relative to the context in which they occur, rather than

as dichotomous catalysts with opposite and absolute powers

to influence behavior.

An Ecological Definition of Motivation

When motivation is viewed ecologically, it becomes

difficult to reduce the influence of rewards to a specific

effect. This view also makes the terms intrinsic and

extrinsic difficult to interpret. Just as the presence

or absence of extrinsic rewards changes the motivational

context, so other information in the context may alter

the significance of such rewards. It is likely that

certain activities, especially those which require

complex and prolonged learning, depend for their mastery

upon incentives which are extrinsic to actual task

demands but are intrinsic in the sense that they provide

important feedback regarding an individual's progress.

Such "extrinsic" incentives may actually increase

"intrinsic" motivation. Take, for example, the case of

a dancer who works harder in order to hear louder applause

from the audience, or demands a high salary, or strives

for recognition as the premier dancer of the company.

Such incentives are clearly extrinsic, in that they do

not inhere in the act of dancing. It is unlikely, however,

that any of these rewards would have the effect of under-

mining intrinsic motivation for dancing, i.e., they would

not diminish interest or reduce willingness to dance. In

fact, we would expect them to have just the opposite


As Condry and Chambers observe, it is at this point

that the dichotomous terms "intrinsic" and "extrinsic"

become troublesome. Typically, extrinsically motivated

behavior refers to behavior elicited by reinforcement

outside or external to actual task demands. Intrinsically

motivated behavior is that which the individual selects

due to inherently reinforcing properties of the behavior.

As used here, intrinsic seems to signify that the behavior

and the consequence of the behavior are inseparable, i.e.,

that the behavior is its own reward. This cannot be the

case. For any activity, even an apparent nonactivity

like resting, there is a consequence; and it is the

consequence of the act, not the act itself, which is the

"intrinsic" reinforcer.

A more precise terminology has been put forth by

Bengston (1981). He suggests that we characterize rewards

as either "ecological," in the sense that they issue from

and are structurally linked to activity, or as "contrived"

in the sense of being imposed as a function of forces

independent of the activity. An ecological reward, then,

is part of the fabric of experience. Its affordance issues

from the context of behavior and "its value or worth does

not depend upon a controller embedding it within events

that would not otherwise obtain" (Bengston, p. 2). The

ecological reward, being a consequence of an act, specifies

an event. The contrived reward, on the other hand, is

only arbitrarily related to an act. Therefore, its afford-

ance is unrelated to the demands of activity. Because it

is dispensed by a controller outside the activity, it

specifies only that the rule of dispensation has been met.

Whatever value it may have, or whatever power it may demon-

strate for controlling behavior, there is always the danger

that the contrived reward will divert attention from eco-

logical consequences of behavior.

For educators, the uncritical application of the term

intrinsic motivation is dangerous, because it allows us to

isolate motivation for learning within the learner. We

may too easily excuse poor instruction and failing student

work by claiming that the learner just was not interested

or motivated. We may resort to "making people feel good

without regard for the quality of the relationships between

learners and their natural and cultural worlds" (Bengston,

p. 3). We may become so preoccupied with making learning

fun that we neglect to consider, much less reveal for the

learner, the larger importance of what is learned. Such a

situation would scarcely improve upon the view that all

learning is the result of external forces which can be

(indeed, must be) controlled by the assiduous structuring

of contingent reinforcement schedules.

In the interest of precision, the present investiga-

tion will employ the terms "ecological" and "contrived"

to replace their more popular counterparts. The terms

"intrinsic" and "extrinsic" will be used only in discussing

previous research.

The Concept of Competence and Some
Other Assumptions About Learning

In 1959, Robert W. White introduced into motivational

theory what he called "the concept of competence." The

purpose of this paper was to reconceptualize motivation to

account for certain phenomena which drive theory had failed

to explain. In developing his thesis, White defined

"competence" as "an organism's capacity to interact

effectively with its environment," (p. 43) and he argued

compellingly for the idea that all animals strive con-

tinuously toward this goal. In human beings, competence

develops through prolonged and extensive cycles of learning.

The cumulative learning which begins at birth and increases

in complexity and efficiency throughout life is motivated

by the adaptive utility of becoming more skilled, more

capable, i.e., more competent in dealing with one's world.

The concept of competence entails three basic

assumptions. First, learning differs in important

ways from skilled performance. Therefore, what may

be appropriate means for motivating the one, may be

inappropriate for the other. Second, all learning is

environmental and adaptive with environmental demands.

Finally, learning is cyclical. The learning cycle

represents a process of learner/environment adaptations,

and, therefore, the nature of the cycle will depend upon

the nature of the learner/environment relationship. Each

of these assumptions is described in detail.

Process Versus Performance

The first assumption of the present study is that the

process of acquiring skill differs in important ways from

skilled responding. Research has established rather

clearly that the offer of a reward for the performance of

well-learned skills actually facilitates that performance.

In an excellent review of studies which have shown a facili-

tative effect of rewards, McGraw (1978) observes that such

studies may, in reality, have been measuring skilled

responding, and not learning. These studies have typically

measured rate and accuracy of responses on tasks such as

mental multiplication, serial learning, pair-association

learning and perceptual recognition. McGraw notes that

facilitation seems to occur only on problems for which

the person has a solution prior to beginning the task.

Therefore, what is being measured in these studies may

not be learning, but the speed and accuracy with which

one is able to apply a familiar algorithm. Bear in

mind that the question being investigated in this study

pertains to the influence of rewards on the acquisition

of new skills as the individual strives to develop compe-

tence with a learning activity.

Learning Environments

The second assumption of this study is that learning

occurs within an environment and is, therefore, shaped by

environmental demands. The term "environment" refers to

those aspects of the world with which a person interacts.

While two people may reside in the same area of the world,

they may interact with different aspects of the world,

depending upon the environment or niche which they inhabit.

The Aussie and the Aborigine both reside in Australia.

But the former occupies the civilized shoreline of the

continent, while the latter lives in the wild and barren

interior. Therefore, to say that one is an Aborigine or

an Aussie is to say something about the environment that

person inhabits and the demands with which he has adapted.

All environments are "learning environments," in the sense

that they require some form of individual adaptation. In

this paper, however, the term "learning environment" is

used specifically to refer to a context of events designed

to facilitate psychological adaptation, i.e., learning.

In the world, learning often occurs rather efficiently

in the absence of any formal instruction. There seems to

be no motivational problem. Skills which are important are

learned within the context to which they are relevant and,

moreover, they are learned to mastery. In ecological terms,

these things are so because, in the world, the act of learn-

ing and the consequence of learning are structurally linked.

This linkage is maintained by the influence of three prin-

ciples. First, there is information in the environment to

"engage" and direct the learner's attention toward important

aspects of the total informational array. Second, there is

"coherence" within events, and one can perceive that coher-

ence. Finally, there is "responsiveness" of environmental

information. That is, feedback is provided within events

to specify progress and completion of activity, so that the

learner knows how well he or she is doing. Consider how

engagement, coherence, and responsiveness control learning

in two real world learning environments.

First, consider the case of a boy in a tribal culture

as he learns to hunt. As the boy grows to manhood, he is

increasingly aware of the importance of skilled hunting to

the survival of his tribe. He watches and models the

behavior of the men as they prepare for and celebrate the

hunt. They reveal its importance by taking pride in their

own skill. Thus, the boy's attention is engaged toward

developing his own skill as a hunter. When he goes on

his first hunt, his father and the bush are his teachers,

responding promptly and informatively to his mistakes, as

well as to his successes. From his mistakes, he learns

the importance of precision and concentration. Each

success increases his confidence and paves the way for

more challenging adventure. Events and his perception of

those events cohere perfectly because his acts are struc-

turally linked to consequences. If they were not, he

would not master the skill of hunting. If he did not

become skillful, he might be killed by his prey, or his

family might starve.

Next consider the example of a boy in urban society

learning to drive a car. This task has engaged his

attention since childhood. He knows that being able to

drive gives adults independence which he does not have.

As he sits behind the wheel of a parked car and models

his father's behavior, he feels very grown-up and power-

ful. When the time comes for him to take the car out on

the road, his parents and the city streets teach him. His

mistakes are embarrassing, sometimes costly. But he learns

from them, because the coherence between his behavior and

the consequences of his behavior is clear. As the basics

of maneuvering the car and avoiding collision become

routine, his skill is challenged by new highways and

longer distances. His world grows larger and, as it

does, his skill as a driver steadily improves through

practice and experience. His life and the lives of

others depend upon this.

Motivation is not a problem in either of these

learning situations because learning occurs within the

context to which it is relevant (Condry, 1978). The

principles of coherence, responsiveness, and engagement

control the learning because of the structural linkage

between acts and their consequences. The desire, the

need, the willingness to learn grows out of the indi-

vidual's striving to adapt with the demands of the

environment he occupies. When learning is decontextual-

ized, however, environmental coherence, responsiveness,

and engagement may be missing or concealed, and stimu-

lating the willingness to learn becomes problematic. If

one doubts this, he might consider the motivational

problem inherent in teaching the urban dweller to snare

the tiger, or the primitive to master a driver's manual.

One way to solve this problem is to resort to the use of

contrived incentives.

Contrived incentives used in place of environmental

or ecological incentives may succeed in stimulating

activity. They may engage attention toward their attain-

ment. They may induce an artificial coherence in which

the consequence of acting becomes the receipt of a valued

reward. As long as activity leads to the reinforcer,

activity will continue. Contrived rewards may also take

the place of environmental responsiveness, providing

feedback as to the correctness of activity. But, when

contrived incentives control learning, there may also be

other outcomes. Because the contrived reinforcer has been

the purpose for activity, once it is removed, the rate of

the reinforced response is likely to drop; perhaps below

the baseline level. Moreover, the learning which results

from the control of contrived reinforcers will depend

almost entirely upon the performance contingencies which

are established. There will be no press to learn unre-

warded responses; therefore, incidental learning or

integration of learning is unlikely (unless contingencies

are also established for these). So it is that contrived

rewards may control the learning environment, but not

without undesirable effects.

Learning Cycles

The final assumption of the present study is that

learning is cyclical. Even the simplest act, like

lifting food to the mouth, depends for its mastery upon

prior learning and, in turn, opens new vistas to higher

learning. The process entails a cycle of perceptions,

adaptations, and renewed perceptions. Reduced to its

essential components, the learning cycle has four stages

which have been labeled by Condry and Chambers (1978)

engagement, process, disengagement, and reengagement.

Engagement refers to an individual's approach to a task;

the why, where, and how of his initial undertaking of a

learning activity. Process refers to the methods or

strategies used by an individual in acquiring a new skill

or concept; his patterns of attention and integration of

information, his use of time, and so forth. Disengagement

refers to an individual's termination of activity; the

when, why, and how of his willingness to leave the task.

Reengagement refers to an individual's return to a task

once it has been terminated; his willingness to reapporach

a completed activity (as in the case of puzzles which allow

varied solutions), or his willingness to persist in an

uncompleted activity (as in the case of skills which

require a number of cycles or repeated practice for


How the person moves through this cycle will depend

upon features of the learning environment; the coherence

of events, the responsiveness of feedback mechanisms, and

to what extent (and in what direction) attention is engaged.

The concern of this investigation is how the cycle will

differ, depending upon whether ecological rewards or

contrived rewards are in control. The more salient a

reward, the more control it is able to exercise. Consider

how the cycle of learning mathematics might differ depend-

ing upon whether ecological rewards or contrived rewards

were the more salient features of the learning context.

The assumption of this study is that when ecological

rewards are in control of the learning cycle, attention

is focused upon those events which seem most likely to

afford growth toward competence, i.e., mastery. If one's

attention were engaged, perhaps by a desire to understand

Einstein's theories, toward the task of mastering higher

mathematics, it would profit one to work mathematical

problems. The difficulty level of the problems engaged

would be determined by one's knowledge of his present

level of competence and the development afforded by the

problems. Working only problems that could be solved

easily would not serve one's purpose of mastery, that is,

it would not get him closer to his goal of understanding

higher mathematics.

Now, suppose a person were paid to learn math. In

this case, learning would be motivated by profit. More-

over, learning would be controlled by whatever performance

contingencies were established. These contingencies would

have to be observable or measurable, so it is not unlikely

that the receipt of payment might be made contingent upon

a number of math problems completed in a specified time.

A learning environment would then be created in which the

consequence of doing math problems was no longer the

learning of mathematics, but the receipt of payment. Math

problems would be engaged, not for what they afforded to

the task of doing higher mathematics, but for what they

afforded to receiving the payoff. Reason would suggest

that the difficulty level of math problems attempted under

this condition would be dictated by the ease with which

they could be executed. To do time consuming problems,

to explore various solutions to the same problem, or to

attempt problems which were obviously too difficult (but

which looked interesting) would all be contrary to the

profit motive. Research which supports this logic is

reviewed in Chapter II.

The actual process of solving problems, under the

control of a contrived reward, would probably become

solution, rather than development, oriented. There is

evidence to suggest, also, that solutions would become

stereotyped; that is, solutions which had worked in the

past would be tried over and over again (Schwartz, 1982).

Because of the narrowed attention of the goal-directed

learner, he or she would be prone to miss or ignore

important information. Thus efficiency would be reduced

and fewer problems might be solved. Frustration would be

likely to increase. Depending upon how highly the learner

valued the goal and upon the means available, shortcuts,

deceptions, or fabrication of solutions might be tried.

Of course, the learner might choose to ignore the contrived

reward; to make learning mathematics more important than

getting the payoff. In this case, however, the learner

would be operating under a different set of perceptions

than the one dictated by the context. Control would have

been returned to ecological rewards. If the contrived

reward were salient and highly valued, however, it is

more likely that the learner would adapt to its demands.

In this case, failing to use any means available to

complete problems would be counteradaptive.

Disengagement of the learning activity controlled

by contrived rewards would occur, not because curiosity

had been satisfied, or because a sense of mastery had

been achieved, but because the reward had been secured.

Moreover, the willingness to reengage the activity in the

absence of the contrived reward would be reduced. There

is substantial research, detailed in Chapter II, that

supports this claim. Why this is the case is still a

matter for speculation, however. It may be that one who

engages a task for the purpose of securing a contrived

reward comes to believe that the task has no inherent value.

That is, the mere offer of a reward may signal to the person

that the task is not worth doing in its own right. A more

tenable explanation is that the contrived reward reduces

willingness to reengage an activity because it focuses

attention upon itself, rather than upon ecological rewards

to be derived from the activity.

Outline of the Present Research

General Hypotheses of the Study

Two general hypotheses guided the study. The first

of these related to the presence or absence of a contrived

reward in a learning environment where ecological rewards

were already in place. This hypothesis might be formally

stated as follows: The introduction of a contrived reward

into a learning context where ecological rewards are salient

will negatively affect learning process, skill development,

social interactions, attitudes, and subsequent motivation

for the activity.

The second general hypothesis grew out of a desire to

know whether the timing of a contrived reward might alter

its influence. Typically, studies in this field have

introduced the contingent reinforcer as a condition of

initial engagement, allowing it to stay in place throughout

the learning cycle. This study attempted to determine

whether the effect of a contrived reward might change if

the reward were introduced after the activity had been

engaged under ecological conditions. Likewise, it seemed

important to test what effect withdrawing the reward would

have on the behavior of subjects who had engaged the

activity under contrived conditions. The second general

hypothesis might be formally stated as follows: The degree

and direction of a contrived reward's influence will change

depending upon the time of its introduction into the learn-

ing cycle.

Experimental Conditions

The ecological thrust of the present study dictated

that the experimental setting incorporate demand character-

istics of real world learning. To this end, an ecological

learning context was created, i.e., a context in which

learning could be expected to occur in response to

ecological demands. This basic context was then altered

in three ways, creating four experimental conditions. The

distinguishing characteristics of these conditions were the

presence or absence of a contrived reward and the time at

which this reward was introduced into the learning cycle.

In three of the conditions, receipt of a monetary reward

was made contingent upon competition with a partner.

While all subjects participated in two experimental

sessions, subjects in two of the three reward conditions

were offered the reward at only one of these sessions.

In all four conditions, pairs of children played a

maze game called Labyrinth, which has three playing boards

that require different levels of psychomotor skill. In

order to disguise the fact that their learning was the

focus of the study, children were told that they were

helping the experimenter conduct an evaluation of the

game. A general description of the four experimental

conditions follows:

No-reward/No-reward (NN). Subjects in this condition

received no monetary reward at either of the two experi-

mental sessions. They played the game in the same room

and were advised that they could play competitively. They

were also supplied with a score sheet on which they could

tally their wins. This group engaged, processed, and

disengaged the learning activity under ecological condi-


No-reward/Reward (NR). Subjects in this group were

offered a monetary reward for winning matches on the

second day, but not on the first day. Thus the learning

activity was engaged under ecological conditions, but was

reengaged under contrived conditions. This experimental

condition simulates the real world practice of rewarding

achievement after some skill has developed, in the hope

that this will also inspire greater achievement. The

degree/diploma system in our schools operates in this way.

Reward/Reward (RR). These subjects were offered the

monetary reward at both sessions. Therefore, their engage-

ment and processing of the activity was totally under the

influence of the contrived reinforcer.

Reward/No-reward (RN). Subjects in this group were

offered monetary reward on the first day, but not on the

second day. Here, initial engagement took place under

contrived conditions, but the activity was reengaged

under ecological conditions. This simulates the common

practice of offering people a contrived reward and then

withdrawing it after a certain time, hoping that the

rewarded response will maintain itself. Reading is

frequently taught in this way; particularly in the

primary grades, where children are awarded tokens or

special privileges for completing reading assignments.

Research Questions and Specific Hypotheses

The specific hypotheses are now presented according

to major areas of interest. For each area of interest,

the general research question is presented, followed by

the specific hypotheses which addressed that question.

The hypotheses are expressed in terms of no-reward (N)

groups, and the reader is asked to bear in mind that

interactions were anticipated among conditions when the

reward was introduced or withdrawn.

Research question 1--Learning process. The research

question was whether or not the learning process of sub-

jects in the ecological conditions would differ signifi-

cantly from that of subjects in the contrived conditions.

It was anticipated that subjects who were not offered the

contrived reward would be more oriented toward the process

of learning, while subjects who were offered the reward

would be more oriented toward product, i.e., getting the

payoff. The following hypotheses describe how these

different orientations were expected to manifest themselves:

Hypothesis 1. N subjects will spend less time playing

matches, i.e., competing.

Hypothesis 2. N subjects will spend more time

practicing before playing the first match of the session.

Hypothesis 3. N subjects will spend more of their

total session time engaged in playing their own game.

Hypothesis 4. N subjects will use more difficult

boards during play.

Hypothesis 5. N subjects will be less likely to

employ "minimax" strategies, i.e., strategies designed

to minimize effort or competition in order to maximize

scores (e.g., shortcutting through the game mazes,

falsifying scores, or making rules which negate competition).

Research question 2--Social interactions. The research

question was whether the presence or absence of a contrived

reinforcer would change patterns of cooperative and


competitive behavior. It was anticipated that subjects who

were not rewarded for competing would be more cooperative,

or, at least, more likely to display an attitude of

"friendly" competition than subjects who were given

money for each match they won. The following hypotheses

describe how cooperative versus competitive orientations

were expected to manifest themselves in the behavior and

attitudes of the subjects:

Hypothesis 6. N subjects will be more likely to

share information, give advice or offer instruction to

their partners with the intention of improving the

partner's ability to play the game.

Hypothesis 7. N subjects will be more likely to

praise their partners during play.

Hypothesis 8. N subjects will be more likely to

offer encouragement to their partners during play.

Hypothesis 9. N subjects will be less likely to

interfere, annoy, or distract their partners with the

intention of disrupting their play.

Hypothesis 10. N subjects will be less boastful

during play.

Hypothesis 11. N subjects will have fewer arguments.

Hypothesis 12. N subjects will be less likely to

talk about cheating.

Hypothesis 13. N subjects will be less likely to

cheat their partners.

Hypothesis 14. N subjects will rate their experience

with the game as less frustrating.

Research question 3--Skill development. The research

question was whether subjects who had worked through all

or part of the learning cycle under ecological conditions

would learn as much about playing the game as subjects who

had worked for the contrived reward. It was anticipated

that subjects who played the game under ecological condi-

tions would develop more skill and a better knowledge of

their own competence, because their attention during the

learning cycle would be focused primarily upon features of

the game which afforded mastery, rather than upon features

which afforded getting the payoff. These predictions

followed from those set forth in reference to the learning

process and were expected to manifest themselves in the

following ways:

Hypothesis 15. N subjects will develop more skill

with the game.

Hypothesis 16. N subjects will develop better knowl-

edge of their own competence with the game.

Hypothesis 17. N subjects will rate the game as being

less difficult.

Hypothesis 18. N subjects will be more likely to

express success at playing the game in terms of skill


Hypothesis 19. N subjects will be more likely to

make internal attributions for winning or losing matches;

i.e., they will be more likely to attribute wins or losses

to their own effort or practice than to luck or game


Research question 4--Interest. The research question

was whether subjects in the ecological conditions would

like the game as much as subjects in the contrived condi-

tions. It was anticipated that subjects who had not been

given any artificial inducement for playing the game

would say they liked it better and would be more likely

to reengage the game during a free-choice period a week

later. It was further anticipated that subjects who had

elected to compete in the ecological conditions would

place less emphasis on the competitive affordance of the

game than subjects who had competed for the contrived

reward. The following hypotheses describe how these

differing interest levels were expected to manifest


Hypothesis 20. N subjects will say that they like

the game better.

Hypothesis 21. N subjects will place less emphasis

on competition and winning as measures of game enjoyment.

Hypothesis 22. N subjects will be more likely to

reengage the game during a postexperimental free-choice

period; i.e., when there is no external pressure to do so.


This first chapter has described the nature of

ecological research in the field of motivation. It has

also outlined some assumptions about learning which are

basic to an ecological perspective. Specifically, it has

been suggested that learning outside the formal school

setting does not depend upon externally controlled

reinforcers, but upon rewards which inhere within the

context of activity. The researcher has chosen to call

these inherent rewards "ecological" rewards and has shown

that, when they are in place, motivation for learning is

not a problem. The externally controlled reward has been

renamed "contrived"; and it has been suggested that the

contrived reward alters environmental coherence, respon-

siveness, and engagement, thus changing the learning

process. A review of research which lends support to

this conceptualization of rewards is presented in

Chapter II.




In 1974, Levine and Fasnacht review a considerable body

of research on the use of extrinsic rewards for motivating

learning. The title of their review, "Token Rewards May Lead

to Token Learning," suggests what they found; specifically,

that short-term gains from token economies are offset by

deleterious long-term consequences. Since the appearance

of this article, evidence has continued to support this claim.

In general, the evidence suggests that rewards which are

contrived or extrinsic to task demands, anticipated by the

learner before engaging a task, and delivered contingent

upon performance of the task have a detrimental effect on

the learner's interest and the quality of his performance.

In addition, a recent line of research indicates that

rewards may actually alter patterns of learner/task inter-

actions during the learning process. What follows is a

selective review of the literature bearing upon these points.

The Effect of Rewards on Interest

Experiments conducted by Harolw (1950; Harlow, Harlow,

and Meyer, 1950) provided early insight into the effect of

extrinsic rewards on intrinsic motivation. Harlow et al.

allowed rhesus monkeys to play with a latch-puzzle that

could be opened by following a set of prescribed steps.

The monkeys found the puzzle interesting and soon learned

its solution. Some of these monkeys were then food-deprived

for 22 hours, and the puzzle, baited with food, was

returned to them. These monkeys displayed erratic behavior--

"attacking" the puzzle, rather than confidently applying

the manipulations that they had previously mastered. More-

over, once the puzzle was opened and the food eaten, the

monkeys showed no further interest in it.

Like much infrahuman research, Harlow's experiment was

regarded with skepticism. Some doubted that the findings

would generalize to human behavior. Moreover, there were

several competing explanations for the monkey's poor

performance and reduced interest, such as fatigue, satiation,

increased arousal, competing drives, and shifts in causal

attributions (Bates, 1979). Harlow's research had, none-

theless, raised significant questions about the effect of

extrinsic incentives upon performance and interest.

Deci (1971, 1972b) set out to document this same

behavior pattern in human subjects. Looking at the problem

from the perspective of self-perceptions theory (Bem, 1970),

Deci predicted that subjects paid to perform an intrinsically

rewarding task would experience a subsequent decline of

intrinsic motivation due to a shift in their perceived

"locus of causality" for doing the task. That is, they would

attribute the cause of their participation to the reward,

rather than to their own interest or desire. Deci tested

his hypothesis in two experiments using the SOMA puzzle, a

commercial puzzle composed of pieces which fit together to

form a variety of configurations. Subjects were asked to

reproduce a number of patterns using the puzzle and received

either (1) monetary reward, (2) verbal reward, (3) no reward,

or (4) various combinations of these conditions. In

both experiments, Deci's prediction was confirmed. The

offer of money, a reward extrinsic to the demands of the

task, tended to decrease intrinsic motivation (as measured

by the amount of time subjects spent playing with the

puzzle during a postexperimental free-choice period). On

the other hand, Deci found verbal reinforcement, which is

also considered an extrinsic reward, to have the effect of

increasing intrinsic motivation in varying degrees. More

will be said later regarding research on social rewards.

While Deci's research confirmed the possible deleterious

effects of rewards on intrinsic motivation, it failed to

address the issue of skill development. One might argue

that a little interest is expendable in light of the tre-

mendous power rewards have demonstrated for motivating

performance, especially on boring or repetitive activities

or in cases where individuals are decidedly "unmotivated."

Indeed, this line of reasoning led early researchers to urge

cautious application of their findings.

The Effects of Rewards on Performance Quality

A study by Kruglanski, Friedman, and Zeevi (1971)

opened an avenue of research which offered rebuttal to the

above argument. Reasoning that "liking" or interest was

only one aspect of general behavioral orientation toward

a task, Kruglanski et al. examined the effect of incentives

on qualitative features of performance. Their subjects

were 32 high school aged students who were asked to volun-

teer for "a survey on youth movements." Subjects were

randomly assigned to a reward condition or a no-reward

condition. Reward subjects received the promise of a tour

of the psychology department at Tel-Aviv University as a

"thank you" for their participation. The no-reward group

neither expected nor received any reward. All subjects

were then asked to complete five tasks, two of which

measured short-term recall, two creativity, and one the

Ziegarnik effect (i.e., the tendency to display better

recall for uncompleted as compared to completed tasks).

Following their completion of the tasks, subjects responded

to a questionnaire which asked them to rate their enjoyment

of the experiment and their willingness to participate in

similar projects in the future. The no-reward subjects

received higher scores than the reward subjects on measures

for creativity, recall and susceptibility to the Ziegarnik

effect. However, and somehwat surprisingly, differences

on the measures of interest were marginal.

In discussing the results of their study, Kruglanski

et al. noted that neither the self-perception theory pos-

tulated by Bem, nor the theory of cognitive dissonance

(Festinger, 1957), which had formed the basis of much

research, could adequately account for several nuances

of their own findings. They, therefore, offered two rival

hypotheses which they referred to as (1) the distraction

hypothesis and (2) the affect hypothesis. The former

assumed that extrinsic incentives might serve to distract

or divert one's attention away from the task toward the

attainment of the reward. The affect hypothesis suggested

that the presence of important incentives might create

"emotional interference" of a kind that disrupts skill

acquisition. The combined effect of these mechanisms could

account for the observed decline in the quality of per-

formance and subsequent "liking" for the task. Moreover,

the recognition of distracting and arousing consequences

of incentives implied that the very act of offering a

reward may have information value for the subject. That

is, it may act as a signal to the person to call forth

a particular pattern of attention and behavior.

A study by Lepper, Greene, and Nisbett (1973) sought

to confirm what the authors call the "overjustification

hypothesis" and lent support to the idea that rewards have

a distracting influence. The overjustification hypothesis,

combining elements of the self-perception and cognitive

dissonance theories, postulates that contracting explicitly

to engage in an activity for a reward should undermine

interest in the activity, while receiving an unanticipated

reward should have no such effect. Lepper et al. conducted

their research in a school setting. Children who demon-

strated initial interest in a drawing activity were randomly

assigned to either an expected-reward, an unexpected-reward,

or a no-reward condition. In the expected-reward group,

children were asked to draw pictures in order to obtain a

"Good Player Award." In the unexpected-reward condition,

no mention of an award was made until after the children

had completed their drawings. The no-reward group neither

expected nor received any award. Results of the study

showed that children who expected a reward displayed sig-

nificantly less interest in the drawing activity during a

postexperimental observation session. Moreover, the drawings

of these children were judged (by three judges blind to the

purpose of the experiment) as significantly lower in quality

than the pictures drawn by children in the other two groups.

In a replication of this study, Lepper and Greene (1975)

found that children who worked for an extrinsic reward

actually produced more drawings than children in the control

group, but that their drawings were, again, of lower quality.

Thus, the Lepper et al. studies support the arousal and

distraction hypotheses put forth by Kruglanski et al., and

have been referred to by Condry and Chambers (1978) as

evidence for the "signal" effect of task-extrinsic incentives.

That is, the anticipated reward may signal the individual to

focus attention on the attainment of the goal, rather than

on the demands of the task itself. The unanticipated reward

has no such effect because it is not part of the context in

which the task is undertaken.

With attention focused upon the receipt of a contrived

reward, the learner apparently loses sight of ecological

rewards of the task. Thus, the task seems less interesting--

less rewarding. It appears that working for a contrived

reward diverts attention away from process and exploration,

towards output or production. Thus, quality is sacrificed

for the sake of quantity. These outcomes are more readily

understood in light of research which has examined changes

that take place in the learning process when contrived

rewards are in control.

The Effect of Rewards on Process Variables

While most motivational studies have focused only on

the products of learning, there have been some attempts to

examine the learning process as it is influenced by

extrinsic rewards. These studies, though few in number,

argue compellingly for the thesis that anticipated rewards

alter patterns of attention and behavior, often to the

detriment of learning. Neimark and Lewis (1967) examined

learners' "information gathering" activities during a

problem-solving task. Children (10-12 years) were asked to

detect a pattern on a problem board in which the pattern

was concealed by a number of moveable shutters. Each

shutter opening or "move" revealed certain information

about the pattern. Subjects were asked to identify the

pattern on the problem board as being one of several pre-

sented to them on an answer sheet containing all pattern

possibilities, and to do so in as few moves as possible.

Subjects were also asked to keep a record of each move and

the pattern possibilities eliminated by it. Children in

the control group received no further instructions. Those

in the reward group were paid for successfully completing

problems; they also had money taken away for each move they

made. Neimark and Lewis examined several aspects of infor-

mation gathering, including strategy, total errors, redundant

information and ability to explain the correct rule for

solving the problem. While methodological problems kept

differences between the groups from yielding statistical

significance, the observed differences were clearly in the

direction of the authors' predictions. Rewards did tend

to have a detrimental effect on information gathering. The

problem of methodology will be discussed more fully in the

"Experimental Designs" section of this chapter.

In an attempt to understand more of process behavior,

Maehr and Stallings (1972) examined the level of difficulty

of tasks preferred by children who engaged an activity

expecting to be evaluated, as compared with that of children

who did the same activity for "fun." In this study, 8th

grade children were randomly assigned to four groups. Two

groups (external evaluation) were told that the task was a

test of their ability and that the results would be given

to their teacher. Two other groups (internal evaluation)

were told that the task was given for fun and that they would

be the only ones to judge how well they performed. Within each

of these conditions, one group was given a set of difficult

problems, and the other group a set of easy problems. All

groups then answered a questionnaire which measured their

interest in doing similar tasks in the future. Answers to

the questionnaire indicated that the internal evaluation

groups would be more willing to attempt a set of difficult

problems in the future, while the external evaluation groups

would prefer easier problems. Differences were strongest

for children identified as being high in need for achieve-

ment. Thus, even the "threat" of evaluation may alter one

of the key factors in learning--the willingness to challenge

oneself with increasingly complex activities.

In a study of "social interaction" Garbarino (1975)

extended the boundaries of process effects. Reasoning

that children who were paid to teach might focus more on

getting paid than on teaching, Garbarino offered 6th grade

children theater tickets for "successfully teaching" a

sorting game to a 1st grader. Process variables which

Garbarino investigated included emotional tone of teacher/

student interactions, teaching style, and use of time. In

addition, he took as performance measures the amount of

learning and the number of errors made by students. Garbarino

predicted that children who were promised a reward would

adopt an "instrumental" orientation toward tutoring that

would manifest itself in a more negative emotional tone,

intrusive teaching style, and less efficient use of time.

In addition, he predicted inferior learning and more errors

by the pupils of these teachers. Three of the four hypoth-

eses were confirmed. Differences in teaching style failed

to reach significance, though the differences were in the

predicted direction.

Rewarded tutors in the Garbarino study tended to

display the "quantity over quality" orientation discussed

earlier, engaging in more overall teaching behaviors but

fewer "high level specific" behaviors. Moreover, the

correlation between negative evaluations of task behavior

and negative evaluations of person (pupil) was significant

for the reward group, while it was nonsignificant for the

no-reward group. Garbarino interprets this finding as indi-

cating that, for the rewarded tutors, "when the child made

many errors, the tutor responded negatively to the child"

(1975, p. 426). This would logically follow from the

concept of rewards as having a distracting influence. When

one's attention is focused upon achieving a goal, anything

in the way becomes a means to an end, and it is likely to

be regarded negatively if it threatens to become an obstacle.

One of the most fruitful studies to examine process

was conducted by Chambers (1979). Sixty male undergraduates

were asked to solve concept attainment problems consisting

of an array of cards containing information, a worksheet

for notetaking and reference, and a wheel which could be

manipulated by the subject to obtain feedback. Subjects

were randomly assigned to four groups. Two groups (reward

groups) were paid $.75 for each solution, while two others

(no-reward groups) were paid $3.00 prior to their partici-

pation in the experiment. Within each of these conditions,

one group (specified) was told to do four problems, while

the other group (not-specified) was told only to work until

they were told to stop. All subjects were allowed to

complete four problems. They were then given a posttest

problem to see how much they had learned, and they were

asked to complete a short questionnaire designed to measure

attitudes toward the task, the learning context, and

future participation in a similar experiment.

Results indicated that subjects who were paid for each

solution adopted a different approach to the problem solving

process than subjects who were not paid contingent upon

their performance. Differences between the reward and no-

reward groups were strongest when the amount of work to

be done was not specified. Responses to the questionnaire

indicated that subjects who were paid for their solutions

were less willing to volunteer their time to similar projects

in the future and that they felt more pressure of time

during the experiment. In general, paid subjects began

guessing earlier and made proportionately more guesses

before solving problems. They worked more rapidly but

chose to work easier problems. In addition, they tended

to use a less efficient information gathering strategy.

This was manifested in their failure to make use of their

worksheets, and in their overreliance on a gambling

strategy to solve problems. The gambling strategy yielded

more information than a conservative focus only when the

response was positive. If the response was negative, no

information was obtained. Chambers points to this finding

as evidence of an "answer orientation" among rewarded

subjects. Indeed, one who viewed "getting" a solution as

paramount to "learning" a solution, would be more willing

to risk bypassing information in order to get the solution


The interactions between the specified/not-specified

conditions in the Chambers study, are also important.

Their net effect was to diminish the difference between

the reward and no-reward conditions. When the number of

problems to be done was specified, the reward group did

tend to work harder problems, work problems more completely,

use more efficient information gathering strategies, and

make better use of their worksheets. In his interpretation

of these interactions, Chambers notes that there is more to

be considered than the mere presence or absence of rewards.

The conditions under which incentives are offered may help

to determine the extent of their influence. When the amount

of work to be done was unspecified for the reward group,

subjects clearly had a choice between maximizing learning

or maximizing payoff. When the number of problems to be

done was specified, an element of choice was removed. There

was no longer any advantage to working faster, so subjects

worked more carefully. We will discuss the importance of

free choice more fully in the "Experimental Designs" section

of this chapter.

A fascinating study of the effect of contingent rein-

forcement on complex response sequences was conducted by

Schwartz (1982). In this investigation, a classic

behaviorist procedure was used to show that the negative

effects of reinforcement are not restricted to highly

probable or interesting activities, and that while

increasing the frequency of responses, reinforcement may

alter the form of those responses. Schwartz conducted

seven related experiments in which the subjects' task

involved pressing keys to illuminate a panel and earn money.

The reinforcement contingency required that subjects press

two keys, four times each, in any order. A red light came

on for three seconds following the completion of a

successful sequence, signaling to the subjects that they

had won the amount of money promised (either $.02 or $.05).

Seventy different successful sequences were possible. In

all seven variations of the expermient, the results were

similar: Reinforcement of successful response sequences

led to stereotypy; subjects tended to develop a dominant

response sequence, repeating over and over again precisely

what had worked in the past. When subjects were told that

their task was to discover the rule of reinforcement,

stereotypy did not develop. However, prior exposure to

contingent reinforcement (i.e., pretraining on successful

sequences) kept subjects from being able to discover the

rule. When subjects were reinforced for response vari-

ability, they merely developed higher order stereotypes,

producing just enough variability to satisfy the rein-

forcement contingency.

Schwartz observes that the tendency to develop

stereotyped patterns of response is not confined to

situations where contingent reinforcement is used. In the

world, "behavioral units" are common. People frequently

develop stereotyped behavioral patterns as part of their

daily routine. Moreover, successful execution of skilled

activities depends on repeating again and again precisely

those responses which work more efficiently and effectively

than others. The housewife, for example, develops a

routine for cleaning the house. Through practice and

experience, she learns that certain sequences of behavior

save time and energy. The tennis player, likewise, adopts

a stereotyped sequence of responses (at the muscular level)

for serving the ball, because that serve has proved more

effective than others.

Thus, the fact that stereotypy develops is not

surprising. What is surprising is that stereotypy can be

induced by contingent reinforcement in situations where it

is not required. In Schwartz' first experiment, not only

did every subject develop a dominant response sequence (when

the reinforcement contingency did not require it), but

half of the subjects said later that they believed the

response sequence they used was necessary, not merely

sufficient, to produce reinforcement. Even more disturbing,

from an educational perspective, is the fact that rein-

forcement induced stereotypy may transfer to related

activities. In three of his experiments, Schwartz found

that a history of reinforcement for successful sequences

interfered with subjects' problem solving ability on a later

task. The practical implication for classrooms is that

rewarding even low probability activities, such as memo-

rizing the multiplication tables, in the hope that they will

become high probability, may have a negative influence on

later activities, such as developing mathematical reasoning


The Schwartz study demonstrates that contrived rein-

forcement can change the nature of activity, inducing

stereotypy when it is not required and interfering with

performance on related activities. However, it leaves

open the question of why contrived rewards have negative

consequences, when conditions that produce stereotypy in

the world do not. Just because the housewife has developed

a routine for preparing meals, this routine does not

interfere with her ability to be creative in the kitchen.

In fact, one frequently finds that the most ingenious home-

makers are also the most organized in their daily routine.

Likewise, the most imaginative artists and craftsmen in

any field are the very people who have routinizedd" the

most features of their trade. It would be difficult to

imagine a pianist, who, having mastered technique (i.e.,

have developed stereotyped behavioral units for playing the

piano), would then find it difficult to compose. On the

contrary, the ability for composition should be enhanced,

because automatic responses have freed energy and attention

for creative use.

When contrived rewards control the nature of activity,

however, one finds just the opposite occurring. In the

learning process studies reported here, contrived rein-

forcement contingencies have been shown to have negative

influence on information gathering, problem solving, use

of time, and eagerness for challenge. In interpreting

the significance of his own study, Schwartz observes that

contingent reinforcement seems to take "control away from

something else" (1982, P. 57). That "something else" is

the structural coupling of act and consequence. When

ecological rewards control learning, the process and out-

comes of learning are virtually inseparable. To say that

one practices the scales daily in order to play better

music is to say something about the quality of that practice.

The outcome (better music) is structurally linked to the

process of creating it (practice of a certain kind). The

quality of one's practice will determine the quality of

the music one plays. In turn, the quality of the music

one wants to play will dictate the quality of the practice

in which one must engage. Thus, to say that a person plays

beautiful music is also to describe what that person has


However, to say that a person practices the scales in

order to earn a contrived reinforcer says nothing about the

quality of the practice. One must go beyond the musical

context to find out what kind of practice is required by

the reinforcement contingency. This is because the outcome

(securing reinforcement) is linked only indirectly to the

process (practicing the scales). The link is an event

external to the act of making music. Thus, the consequence

of the act tells the actor (or learner) nothing about the

quality of the action. How the learner gathers information,

what stereotyped units of behavior are developed, how time

is used, and what is learned from behavior will be dictated,

not by the demands of learning to make music, but by the

demands of reinforcement. In fact, if one wanted to

describe the behavior of practicing the scales in order

to earn a reinforcer, one could do so merely by describing

the reinforcement contingencies, without having to refer

to learning at all. One could simply say that the player

was able to execute a given pattern of muscular responses,

a certain number of times, and at a given rate. Of course,

accuracy would be important to the full description, and

so would consistency of tone. But there would be no reason

to refer to the player's internal states, to the awareness

of the music being played, or to the understanding of how

a given muscular response determined the quality of the

sound that issued from the instrument.

Now, if one can circumvent references to learning in

describing the act of playing the piano for contingent

reinforcement, is it not possible that the player can

circumvent learning while, at the same time, accomplishing

that act? If so, this might explain the person's inability

to apply the meaning of what should have been learned in

one context to the requirements of a related context, i.e.,

the negative transfer of reinforcement to later problem

solving ability. It would also help to account for the

person's apparent lack of knowledge about what was done, as

in the case of Schwartz's subjects who were unable to

explain the rule governing their reinforcement. In any

event, it would clarify the difference between reinforcement

induced stereotypy and stereotypy which develops in the

process of task mastery: The one does not require learning,

in the richer sense of the word, the other does. Of course,

to say that a person does not learn anything while working

for reinforcement would be inaccurate. A more accurate

statement would be that what appears to be learned are only

those aspects of activity which are made salient by the

conditions of reinforcement.

Incentive Characteristics

We have already seen that many types of incentives

(e.g., money, awards, evaluations) can affect interest,

performance, and learning. Nevertheless, certain attempts

to manipulate incentive characteristics are noteworthy.

One series of experiments (Deci, 1971, 1972a; Deci, Cascio,

and Krusell, 1975) found social rewards (praise) to increase

subsequent interest for males, but to have the opposite

effect for females. Deci et al. used only an unanticipated

reward condition; however, when Smith (1975) attempted to

replicate this finding using monetary and social incentives

as anticipated rewards, he found both to have an undermining

influence, and no interaction with sex.

A study by Anderson, Manoogian, and Resnik (1796) examined

the effect of various incentives--social, monetary, and

symbolic--on the interest of four and five year-olds. Here,

subsequent interest was decreased by money and symbolic

rewards but was increased by praise. Chambers (1979) has

offered an interesting analysis of social rewards which may

explain contradictions in the literature. He points out

that a conceptualization of any reward must take into

account not only the nature of the reward (verbal, physical,

etc.) but also the relationship between the recipient and

the provider of the reward. If one receives a social

reward from a person with whom he identifies strongly (i.e.,

whose role he has "internalized"), it may seem equivalent to

giving oneself the reward, and, thus, have more intrinsic

relevance for the person. This possibility is most credible

with young children, who readily adopt adult role models.

The point to bear in mind is that rewards, like anything

else, are not all good or all bad. Their benefit or detriment

will always be tempered by factors specific to the situation

in which they are used.

Related to social incentives is the use of surveillance.

Zajonc (1965) reviewed research on the effect of the presence

of others on learning and performance. He concluded that,

while surveillance facilitates the performance of well-

learned skills, it tends to inhibit the development of

poorly learned skills. Indeed, in the Anderson et al.

study just described, the greatest motivational decline

was found in a "control" condition where the experimenter

was in the room with children but did not interact with

them in any way. Likewise, Lepper and Greene (1975) found

that being "watched" by a TV camera decreased subsequent

interest in a task among nursery school children. Even

the imposition of a deadline has been found to depress

interest (Amabile, DeJong, and Lepper, 1976). It would

seem that almost any external constraint can have the power

to change the learning context and the potential for under-

mining intrinsic factors.

Task Characteristics

The detrimental effect of rewards seems to generalize

across a wide variety of tasks. McGraw (1978) has presented

an excellent review of some 50 studies in which the effect


has been documented. McGraw points out that the bulk of

research, spurred by the pioneering efforts of Miller and

Estes (1961), has focused on children's discrimination

learning. The Miller and Estes study set out to investigate

the effects of different levels of monetary incentives on

the learning of a perceptual discrimination rule by 3rd

grade children. The investigators found no differences in

the number of errors made by subjects who were paid $.01

for each correct discrimination and those who were paid $.50.

Both paid groups, however, made significantly more errors

than a "knowledge of results only" group.

Since the Miller and Estes study, others have found

similar decreases in performance with tasks involving verbal

discrimination (Spence, 1970), picture discrimination (Spence

and Dunton, 1967) and pattern probability learning (McGraw

and McCullers, 1974). Concept discrimination (McCullers

and Martin, 1971) and concept attainment (Condry, 1975;

Chambers, 1979) tasks have also been used in similar studies,

with the same results. Other tasks which have been shown

susceptible to the detrimental influence of rewards include

choice preference (Masters and Mokros, 1973), functional

fixedness problem-solving (Glucksberg, 1962, 1964), insight-

ful learning (Vesti, 1971) and incidental learning (Bahrick,

Fitts, and Rankin, 1952; Johnson and Thomson, 1962; Staat

and McCullers, 1974).

The research on incidental learning is particularly

important to the present study, because it supports the

notion that rewards act as an attention focusing signal.

In the Staat and McCullers (1974) study, subjects were

asked to learn nine paired associations, with reward

subjects earning $.05 for each correct response. When the

subjects had reached a criterion of six correct responses,

the "training" period ended and they were given a recall

test designed to check their incidental learning: Response

items were supplied and subjects were asked to provide the

stimulus item. Unpaid subjects recalled nearly twice as

many of the stimulus items as the paid subjects. Thus, the

attention focusing properties of extrinsic rewards may

cause the learner to ignore other information in the

learning context. Granted, the stimulus items in an experi-

mental association task are not crucial information. But

what of the questions on a classroom test? Surely they are

as important as the answers which students are expected to


Experimental Designs

Experiments in this field of research have relied

heavily on traditional designs. Frequently, the result

has been a lack of differences among groups, or findings

that run contrary to intuition. Recall that in the Neimark

and Lewis study (1967), children were rewarded for detecting

patterns, and their information gathering activities were

observed. Chambers (1979) notes two important factors

which may have contributed to lack of statistical signifi-

cance in this study. First, the way rewards were admin-

istered may have offered beneficial information to rewarded

subjects, confounding the results. Penalties exacted for

each move, and the fact that no strategy but the correct

one ever paid off may have signaled that careful work was

required. In essence, the reward contingency shaped the

behavior of rewarded subjects in ways that made them look

like nonrewarded subjects. Second, constraints placed on

the control group, such as self-monitoring of moves and

instructions to keep the number of moves to a minimum, may

have served to "reduce the freedom of an intrinsic context"

(1979, p. 8). Both criticisms point to "choice" as being

a critical factor in motivation research. If we are to

discover what personal and intellectual factors are

influenced by the presence of absence of rewards, then we

must try to create situations in which the individual has

the freedom to determine for himself the conditions under

which a task will be engaged, processed, and terminated.

When attempts are made to preserve freedom of choice,

results can be surprisingly realistic. Winston, Torney,

and Labbee (1978), for example, investigated children's

self-reinforcement and found, in contrast to most similar

studies, a tendency toward maximization of payoff and minimi-

zation of effort. The authors argue that earlier experiments

had exerted too much control on children's behavior.

Experimental instructions, training procedures, modeling

cues given by adults, and little freedom to vary performance

standards may all have contributed to the traditional finding

that children's self-reinforcement schedules are self-

demanding, and relatively stable across time. The effect

may have been further enhanced, the authors note, by the

fact that these studies typically lasted only one session.

Winston et al. attempted a study in which external cues and

controls were minimized. They asked 5th grade children to

work at a wheel-turning task for four consecutive sessions.

The children could select on each trial how many wheel turns

would earn one token. Tokens were redeemable for a certain

amount of cartoon-viewing time. Children were neither

rewarded for maintaining high standards nor punished for

lowering their standards. Reinforcement was simply con-

tingent upon their performance, and the standard of

performance was up to the individual. Thus, children were

free to adopt one of two strategies for earning enough

tokens to view a whole cartoon: They could either lower

their standards, so that fewer wheel turns were required, or

they could maintain high standards, which necessitated more

(or faster) wheel turning.

Winston et al. found a significant drop in performance

standards across sessions. Not only did children reduce

their standards, they also lowered their overall rate of

wheel turning. Given that the wheel-turning task offered

few, if any, "intrinsic" rewards, and that it was presented

strictly as a means to an end, and that cartoon viewing is

probably a very desirable end for a 5th grader, these

results are realistic. In the absence of external controls

to guide them, and allowed to experience the fruits of a

truly self-determined reward system over a period of time,

children responded in a very predictable way: They tried

to get the most reward for the least effort. Under normal

(i.e., ecological) circumstances, would not any other

response from a normal human being be considered impractical,

if not altogether irrational?

While the Winston et al. study is intriguing for several

reasons, it is presented here mainly to point out that

changes in experimental conditions may have drastic influence

on experimental results. To understand a human behavior,

like learning, the first consideration should be the optimal

circumstances under which that behavior is found to occur.

Both research and experience suggests that learning can

occur most efficiently when external (contrived) constraints

and demands are minimized, and when the ecological con-

sequences of learning are salient features of the learning

context. Therefore, learning studies should attempt to

incorporate these features into research designs. Such

an approach has been encouraged by researchers who are

working to establish an ecological basis for the study

of learning (e.g., Johnston, 1981; Bronfenbrenner, 1976),

and its value to scientific inquiry is aptly summarized in

W. I. Thomas' observation: "If men define situations as

real, they are real in their consequences" (Thomas and

Thomas, 1928, p. 572).


This chapter has presented research which indicates

that many types of extrinsic (i.e., contrived) reinforcement

can have undesirable consequences for learning. First,

contrived incentives appear to decrease motivation or

interest for learning activities. Second, they encourage

an attentional and behavioral orientation which places

quantity before quality and product before process. Third,

they inhibit certain processes, including creativity,

insight, and problem solving, which are generally considered

important to learning. The present study sought to examine

the occurrence of such effects in a setting where eco-

logical rewards were salient and where the individual was

given a maximum of free choice in engaging, processing, and

terminating the learning activity.




In Chapter I, some assupmtions about learning were

adopted. One of these was that motivation to learn grows

out of the human being's striving to become more competent

in dealing with the demands of the environment he or she

occupies. Another was that the use of extrinsic incentives

to motivate learning becomes unnecessary when the context

for learning provides coherence, responsiveness and engage-

ment. These assupmtions guided the design of the present

research along ecologically valid lines.

A primary concern was to create an environment, or

context, where learning could occur but was not demanded

or controlled from without. In Chapter II it was pointed

out that certain laboratory experiments investigating

intrinsic motivation may have placed demands on subjects

which diminished the differences between rewarded and

nonrewarded groups. Being asked to do a specified amount

of work, anticipating a time limit, or just being told that

one's task was to learn something may have signaled rewarded

subjects to work more carefully than they would have

otherwise, or may have actually interfered with the

standards for practice and progress which nonrewarded

subjects would have set for themselves. Therefore, care

was taken in the present study to minimize procedural

constraints and experimenter influence.

Another important design consideration was creating

a learning environment which would be coherent, responsive

and engaging. In the world, motivation to learn only

becomes a problem when one or more of these environmental

aspects is missing or concealed. If actions do not seem

to be logically or consistently connected with events, if

there is not information to signal progress or completion

of activities, if nothing attracts or requires attention,

then the environment ceases to have meaning, and learning

or not learning has no consequence. Admittedly, when

these conditions obtain, contrived rewards have some

value. They may induce artificial coherence, provide an

externally controlled source or response, and engage

attention, if only toward their own attainment. In this

way, contrived reinforcers can increase the frequency of

responses which otherwise might not occur at all. The

aim of this study was to investigate the influence of a

contrived reward in an ecological setting, i.e., where

learning could be expected to occur in response to eco-

logical rewards. Such a setting had to provide a logical

framework for activity. All instructive feedback necessary

for progress had to inhere within the context of the

activity. Finally, the activity and its context had to

engage interest and attention. These requirements deter-

mined the choice of the learning activity and the

arrangement of the experimental setting.

The Sample

Subjects were forty 2nd, 3rd, and 4th grade boys at

the P.K. Yonge Laboratory School. The decision to use

only boys in the study was based on pilot data in which

boys were generally found to be more attracted to

Labyrinth than girls. Parents of all boys in the 2nd, 3rd,

and 4th grades were mailed an informed consent letter,

explaining the purpose and procedures of the experiment,

and asking that they give written permission for their

son to participate. Fifty-five parents returned forms

giving their consent. These returns were stratified by

grade level, and 40 boys were randomly selected to par-

ticipate in the experiment. The other 15 names were

placed in a pool, and drawn out when subjects had to be

dropped. In all, seven subjects were replaced in the

course of the experiment: Four subjects were dropped

due to prior familiarity with the game; two were replaced

due to absence on the first day of their treatment; one

2nd grader elected to remain in his classroom because he

did not want to get behind in his work.

Subjects were randomly paired within grade levels.

Each pair, or team, was randomly assigned to one of four

treatment conditions: No-reward/No-reward (NN); No-reward/

Reward (NR); Reward/Reward (RR); Reward/No-reward (RN).

Each team participated in two experimental sessions. The

NN subjects were not offered the contrived reward at either

session. The NR subjects were offered the reward at the

second session, but not at the first session. The RR

subjects were offered the reward at both sessions. The RN

subjects were offered the reward at the first session, but

not the second session.

The Learning Activity

One criterion for choosing the experimental activity

was that it must have high ecological (intrinsic) appeal.

That is, the task had to look interesting and be able to

engage subjects' interest with a minimum of introduction

by the experimenter. Other criteria dealt with the

learning requirements of the task. It was essential that

the task entail a hierarchy of skills through which a

subject could progress rather rapidly with practice. More-

over, all the feedback necessary for skill development had

to be contained within the activity itself. Thus, skill

levels could be distinguished and experimenter influence

could be reduced. The final criterion for selection of

the task was that it should be relatively content free.

This eliminated the need to assess prior learning in a con-

tent area.

The activity chosen was a commercial board game called

Labyrinth, manufactured by the Brio Scanditoy Corporation.

The game, shown in Figure 1, consists of a box on which an

assortment of three trays can be mounted and tilted by

manipulating two knobs on adjacent sides of the box. The

trays contain mazes of varying difficulty through which the

player must move a steel ball. The ball can be made to

roll in directions parallel to the sides of the tray by

moving one knob at a time. To make the ball move diag-

onally, both knobs must be turned simultaneously. The

object of the game is to move the ball through the maze

without dropping in any of the holes placed strategi-

cally throughout the maze. For a more complete description

of the game, the reader may refer to the "Procedures"

section of this chapter.

Experimental Setting

The ideal setting for this type of study would probably

be in the home, where games are typically played. Placing

Labyrinth in the homes of children, however, would have

meant a loss of experimental control and would have

necessitated elaborate systems for monitoring or


OI O i ;0

-- LOw sO j a-

L' L0--
-J>,--. 1Qo- C-.a

0 00
_ _-*o 7 O,


'I O.



j ^ ^_
L~j^ Ir-^^
~ k0} '
____ s I
CD 7^

self-reporting of activity. Therefore, it was decided that

the experiment should be conducted under the guise of an

evaluation. Subjects were told that the experimenter was

evaluating Labyrinth for possible purchase by the local

school system, and that they had been chosen to help her

by playing the game and answering some questions. This

"cover" helped to account for the questionnaire administered

at the close of each session. It also concealed from

subjects the fact that their skill development with the

game was a focus of the experiment.

Subjects were teamed to avoid the uncomfortable

effects of isolation in a lab. This teaming also made it

possible for subjects to pit their skill against one another

if they so desired. Allowing this competitive affordance of

the game to be realized also increased the ecological

validity of the design.

All activity in the experimental room was recorded on

videotape by a concealed camera. (Only one subject detected

the presence of the camera, and this was after all data

regarding him had been collected.) One reason for this

unobtrusive observation is obvious: People often feel they

need to "perform" if there is a camera present. Moreover,

in the Lepper and Green study (1975) being monitored by a

TV camera worked to decrease intrinsic motivation. The

most important reason for videotaping, however, was that it

allowed the experimenter, who was busy monitoring the

sessions, to collect from the tapes a great deal of quali-

tative data which might have been lost otherwise.

Contrived Reward

Subjects in the reward conditions (NR, RR, RN) were

paid for winning matches. Recall that this offer was made

to the NR group only on the second day, and it was made to

the RN group only on the first day. Subjects were told

that each time they played a match the winner would be

paid a dime. A "payoff" was made every ten minutes during

the session when the experimenter went into the experimental

room to inquire about how the children were doing and

whether they wanted to leave yet.

Money was chosen as the reward for three reasons.

First, in experiments that have used various incentives,

subjects typically value money above other material rewards,

such as gum, candy, or toys (Haddad et al., 1976; McCullers

and Martin, 1971). Second, it was felt that the univer-

sality of monetary reward would improve the generaliza-

bility of the study. Finally, a purely practical concern:

Money was easier to handle and distribute, and did not

threaten to conflict with any school or parental rules.

What performance criterion to use as the reward

contingency was a crucial question. Bearing in mind that

rewards have clearly demonstrated their effectiveness for

increasing the frequency of reinforced responses, the

possibility of shaping the behavior of subjects had to be

avoided. Therefore, the possibilities of making the reward

contingent upon the number of trials to mastery, or the

amount of play, or the difficulty level of the boards

used were all eliminated. Another consideration was main-

taining the ecological validity of the reward context. In

the real world, there are rarely enough "goodies" to go

around, and people are forced to vie for salaries, goods

and services, or recognition. In most schools, the reward

structure is equally competitive. Only the "best" students

get the high grades, the scholarships, and the diplomas.

Furthermore, while the primary interest of this study was

learning, the fact that the experimental activity was a

game could not be overlooked. In the world, games have

both individual and social aspects. A player may practice

individually to increase his skill or competence at the

game. He plays or competes in order to test that skill.

Even so-called "individual sports" like golf, while practiced

and scored individually, are played competitively.

In order to simulate these aspects of real-world

reward structures, the decision was made to make the

reward contingent on winning matches. Thus, teams were

free to set their own standards for competition, in that

they could choose among three difficulty levels of boards

on which to play; but individuals were required to develop

superior skill in order to be rewarded. Subjects kept

their own record of wins on a tally sheet provided by the

experimenter. A win was clearly defined as "getting further

through the maze" than one's partner on a given match. All

subjects were told that they were not required to play

matches, but, if they did, the should score themselves by

this standard. In addition, rewarded subjects were told

that they would receive a dime for each match they won.

Details of the instructions and procedures in the experiment



The experiment took place during regular school hours,

over a period of six weeks. All subjects participated in

two experimental sessions, scheduled within 72 hours of

each other. Seven days after the second experimental

session, subjects were called back for a postexperimental

session in which Reengagement Skill and Knowledge of

Competence were assessed.

First Experimental Session

The children were brought to the experimental room by

the experimenter (E) and seated on either side of a large table.

E seated herself at the end of the table in front of a

Labyrinth game.

"This is a game called Labyrinth. Has either of you ever

seen a game like this before?" If either child said, "Yes,"

E probed to determine if the child had one at home or was

very skilled with the game already. If so, both subjects

were dismissed. E thanked them for their willingness to

participate in the evaluation, but explained that the

evaluation required the opinions of children who had never

played the game before.

"I have been asked to evaluate this game for the school

system. You have been chosen to help me. In a minute, I

will give each of you a game like this one. I would like

for you to play it and, then, answer some questions for me."

The RR and RN groups were then told, "I am going to

give you a chance to win some money. We will come back to

that in a minute."

E referred to boards #1 (the easiest board) and #2 (the

moderately difficult board) which were placed beside her

game. "These are the playing boards. As you can see, this

one (board #2) is a little more difficult than this one

(board #1). This one (board #3, the very difficult board)

is the advanced board."

E picked up board #1 and placed it on her game. "The

object of the game is to roll this little ball along the

path marked by this brown line without dropping it in any

holes. To move the ball, you must move the board using the


E demonstrated how the maze should be negotiated . .

"The front knob moves the board from side to side. The knob

on the right moves the board from front to back. If you

turn both knobs at the same time, the board will move diag-

onally. Now, you try it." E allowed each child one practice


"That is basically how the game is played by one person.

You might want to practice until you can get through the

maze every time. Then, you might work to improve your speed

on that maze, or you might go on to a more difficult board.

Do you have any questions about how the game is played?"

E waited, and answered any game-related questions, being

careful not to give informaiton which might prejudice the

child's attitudes or tip him off to efficient playing


"Let me ask you some questions to be sure you under-

stand. How do you make the ball move? . Which way does

the board move if you turn the front knob? . The side

knob? . Both knobs together?"

"you have seen how this game is played by one person.

It can also be played by two people in a match. If the

two of you want to play against each other, both of you

will move the ball through the maze until you reach the end

or drop in a hole. The player who gets further through the

maze wins the match. If both of you get to the end or drop

into the same hole, that match is tied--nobody wins."

E referred to the score sheet. "Here is a sheet for

you to keep score on if you decide to play matches. Just

make a mark under your name each time you win a match."

The RR and RN groups were then told, Each time one of

you wins a match, I will give you a dime."

"Let me ask you a couple of questions to be sure you

understand about scoring. If you decide to play a match,

what is the first thing you must do? . Who wins a

match? . (Referring to board #2, hole 9/21) If you

dropped in this hole, what would your score be? . How

do you know that? . Who wins a tie?" E explained

scoring and rules again if the children seemed to be con-


"Remember, you do not have to play matches. You may

have just as much fun playing by yourself. Do you have any

more questions before we begin?"

When there were no further questions, E gave each

child a game and made sure he was comfortably seated. She

watched for a moment to make sure the children understood

how to manipulate the game, then excused herself. "I have

reports to fill out, so I'm going to work in the next room.

If you need me, just knock on the door. You may leave any

time you wish, but remember, before you do, I need to ask

you some questions."

E left the room and observed the children from a

remote VTR in the next room. She returned in ten minutes

and said:

To the NN and NR groups, "How are you getting along?

. . Have you played any matches yet?"

To the RR and RN groups, "How are you getting along?

S. Have you won any money yet?" E then gave these

reward subjects a dime for each mark on the score sheet.

E stayed approximately one minute, then excused her-

self again, reminding the children that they were free to

leave whenever they liked, and that they could take a

break if they wanted. This process was repeated every ten

minutes until one or both subjects asked to be excused, or

until the one-hour time limit established for the sessions

(but never mentioned to the children) was up, or until

both subjects showed signs of losing interest in the game

by being off-task for four out of five minutes.

At the end of the session, E asked one child to wait

outside while she asked the other child some questions. E

administered the questionnaire to the first child, then

excused him for the day, then did the same with the second

child. Both children were told that they might be asked

to come back again the next day. RR and RN subjects were

requested not to mention the money to their classmates,

because the E's funds were limited and she might not be

able to give dimes to everyone.

The Second Experimental Session

The children arrived and were seated in front of their


"Yesterday you had a chance to get acquainted with

Labyrinth. Today, I would like you to work with it, again,

so I can ask you some more questions. Do you remember all

the things we talked about yesterday?. . What is the

object of the game?. . How do you keep score if you want

to play matches?"

The NR group was then told, "Everything is the same

today, except that, today, each time one of you wins a

match, I will give you a dime."

The RN group was told, "Everything is the same today,

except that, today, I will not be able to give you a dime

for each match you win. My money has run out."

E allowed the children to ask any questions they had

and then excused herself as in Session One. The same cycle

of observation, returning to the room, giving dimes to the

reward groups, and administering the questionnaire was

repeated. When the children were excused from this session,

E thanked them for their help in the evaluation and said

that she might ask to see them again in a few days.

The Postexperimental Session

Each child was called to the experimental room indi-

vidually for the followup. E greeted the child in the

hallway outside the room and asked him to wait inside for

a few minutes while she tried to locate a misplaced file

which she needed before they could proceed. When the

child entered the experimental room, he found a Labyrinth

game, assorted comic books, and a Rubik's Cube on the table.

E watched him on the remote monitor in the next room and

recorded his activity, as it occurred. Time spent engaged

in playing Labyrinth, reading comics, manipulating the Cube,

or just sitting quitely was noted.

After five minutes, E returned to the experimental

room, apologized for the delay and said that she had

found the papers she needed.

"Today, I would like to find out how well you have

learned to play Labyrinth. I would also like to see if you

know how well you play. Before we begin, would you like

to take a few minutes to warm up?"

When the child indicated that he was ready to begin,

E explained, "Now, I would like you to try to get all the

way through each of the three mazes. I'll give you as many

as five tries on each maze. Before you take each try,

think for a minute, and tell me, before you begin, if you

will make it all the way through the maze on that try. Do

you understand?".. After answering any questions the

child had, E asked him to put board #1 on the game.

If a child completed a maze on the first try, or if

he completed it twice in succession, E asked him to go on

to the next board. For board #3, the "end" of the maze

was set by the subject himself. To do this, E asked, "How

far have you gone on the advanced board before?" Whatever

hole was named by the child as his highest score on board

#3 was then taken as the end of that maze.

E recorded the child's score on each trial and his

prediction. When all the trials were completed, E thanked

the child for his help and excused him.

Data Collection for Within-Session Variables

Some within-session data were collected during the

experimental sessions. These included, the time the session

started, the time and reason it was terminated, and the

time of the first match. All other within-session data

were collected from videotaped recordings of the sessions.

The instrument used to code these data is shown in Appendix


Qualitative Measures

Instructional Events. Coaching or offering advice

with the intent of improving a partner's play.

Praise. Comment intended to acknowledge partner's

good play (included nonverbal behavior, such as applause).

Encouragement. Comment intended to console partner

for poor play or relieve partner's frustration.

Look-at-Me-Statements. Boasting or demanding partner's

attention or praise.

Arguments. Prolonged disagreements over turn, score,

or rules of play.

References to Cheating. Any statement containing the

word "cheat," e.g., "You cheated." "This is how I cheat,"


Interference. Statement or action intended to annoy,

distract, or disrupt partner's play.

Instructional Events, Praise, and Encouragement were

aimed at measuring the degree of cooperation in the subject's

social interactions. Look-at-Me-Statements, Arguments,

References to Cheating, and Interference attempted to

measure the negative influence of competition, or what

might be called "unfriendly competition."

These data were recorded for consecutive intervals of

a session. An interval was defined as 50 feet of videotape

(approximately 3 minutes). A behavior was coded only once

for a given interval, regardless of the number of times it

occurred during that interval. Any behavior which was

ongoing for more than one interval was recorded for each

interval during which it occurred. In addition, specific

observations were noted for each session. Subject's exact

words to each other during Arguments and Instructional

Events were recorded whenever possible. The nature and

degree of Interference was noted wherever it occurred.

Any instance of cheating, whether or not it was accompanied

by a verbal Reference to Cheating, was also described in


Descriptions of actual cheating were compiled to

determine whether patterns of cheating might exist and

whether these patterns were consistent among groups. Three

specific types of cheating were observed:

Cheating a Partner. Lying to one's partner or falsi-

fying the score card in order to win a match or improve

one's cumulative score.

Cheating the Experimenter. Lying to the experimenter

or falsifying the scorecard in order to win more money.

This form of cheating, of course, was unique to reward

groups, and therefore, was not included in the primary

data analysis. This type of cheating involved a cooperative

deception and the dialogue which transpired was recorded

verbatim by the experimenter.

Minimax Strategies. Attempts to minimize effort or

competition in order to maximize scores or profits. This

might also be called "cheating oneself" or "reducing

challenge." Usually, it was accomplished by making a rule

such as "finishing the maze equals winning the match," or

shortcuttingg through the maze is fair play," or "the first

person to drop in loses, regardless of the other person's

position at the time."

Quantitative Measures

Mean Difficulty of Boards. The board used by each

subject during each interval was noted. The number of the

board (either #1, #2, or #3) was recorded, and these numbers

were averaged across intervals for both subjects to yield

a team measure of the average difficulty of boards used

during that session.

Intervals Containing Match Play. Match Play was

recorded for every interval during which the game was

being played competitively, whether or not subjects were

tallying their scores. The percentage of total session

intervals during which Match Play occurred was calculated

as a measure of time spent in competition, rather than


Practice Time Preceeding the First Match. This measure

was calculated as the number of minutes which elapsed

between the time subjects first began playing the game to

the time of the first interval in which match play was


Engaged Time. This measure was defined as the per-

centage of total session time subjects spent actually

engaged in playing their games. A stopwatch was used to

measure the amount of time each subject was nonengaged

during each interval. Nonengagement was clocked whenever

a subject was not playing his own game. Thus, stopping

to watch one's partner play was included in nonengaged time,

along with resting, looking around the room, reading a

book, and so forth. Nonengaged time was subtracted from the

total session time to yield a raw measure of time engaged

in playing the game. Engaged Time was then calculated as

a percentage of total session time.

The Questionnaire

The questionnaire administered at the close of each

session contained 15 items. A copy of the questionnaire

appears in Appendix B. Questions were presented orally

and subjects' responses were marked on the questionnaire.

Responses were then coded numerically to yield a combined

score for each team in the following six categories:

Liking of the Game. This was measured by questions

1, 2a and 2d, 3, and 14.

Feelings of Frustration. This was assessed by ques-

tions 2b, 8, 9, and 10.

Ratings of Game Difficulty. This was assessed by

questions 2c and 2e.

Expressions of Skill Requirements of the Game. This

was measured by items 6, 7, and 15.

Internality of Attributions (for winning or losing

matches). This was measured by questions 12 and 13.

Importance Placed on Competition. This was assessed

by items 5 and 11.

(In the final data analysis, items 1 and 2a were

excluded because of ceiling effects. Probably due to the

wording of the questions, all subjects gave the same

answer. Item 4 was excluded from the analysis because it

had no direct bearing upon the experiment, but was used

in the questionnaire to introduce question 5.)

Data Collection for Postexperimental Variables

Data on Reengagment, Mean Sources on the Skill Test,

and Accuracy of Predictions were collected during the

postexperimental session. Appendix C shows the instrument

used to collect data on these measures:

Reengagement was defined as the amount of time during

a five-minute free-choice period a subject spent playing

Labyrinth. This time was recorded to the nearest half-


Mean Scores on the Skill Test were measured over a

maximum of 15 trials, five trials per board. On each trial,

a raw score was recorded as the number of the hole where the

ball fell in. These scores were then adjusted by assigning

consecutive numbers to each hole for boards #1 through #3.

Thus, a raw score of 12 (hole 12) on board #2 became an

adjusted score of 21. A subject's adjusted scores were

then averaged across trials to yield a mean score.

Accuracy of Prediction of success or failure over a

series of trials on the skill test served as a measure of

subjects' knowledge of their competence with the game.

Before each trial, a subject said "yes," he would get all

the way through the maze on that trial, or "no," he would

not. Predictions were compared to outcomes across trials

and the percentage of trials on which the subject predicted

the outcome correctly was then calculated.

Data Analysis

All within-session variables were analyzed by team.

Where data had been collected individually, the subjects'

scores were averaged into a team score. A 2 x 4 split-plot

ANOVA was used to determine whether there were differences

between teams in the four treatment conditions for each of

the within-session measures. Nine qualitative variables,

4 quantitative variables, and 6 variables defined by cate-

gories of the questionnaire entered into the primary

analysis. The ANOVA for each of these variables contained

two levels of Time and four levels of Condition. Every

team received two treatments, at Time 1 and Time 2, but

the treatment differed according to the Condition of the

subjects. Graphically depicted, the design for the analysis

of within-session variables was as follows:

Condition Time 1 Time 2

NN (N) (N)
NR (N) (R)
RN (R) (N)
RR (R) (R)

Reliability of the experimenter's observations of the

qualitative variables was established by estimating the

variance among ratings done by the experimenter and two

other raters. One tape from each of the four conditions

was chosen at random after the experimenter had coded all

data. Two persons blind to the hypotheses of the experi-

ment were instructed in using the observation instrument.

Instruction was based on tapes recorded during the pilot.

Following the instruction period, each observer watched

the tapes and coded the data privately. For each tape,

rating of the observers and the experimenter were analyzed

by means of the Variance Component Estimate procedure in the

SAS package. The percentage of total variance among the

three ratings which was actually due to differences among

conditions was then calculated for each behavior using the


variance associated with tape
var. assoc. with tape + var. assoc. with rater + error

The postexperimental variables were analyzed as indi-

vidual data. Subjects' scores on Reengagement, Mean Scores

on the Skill Test,and Accuracy of Predictions were sub-

jected to a one-way ANOVA to determine whether there were

reliable differences among conditions. For all variables,

the SAS package was used to conduct the Analysis of Variance.


Follow-up comparisons were made using either the Bonferroni

t statistic for complex comparisons or Tukey's procedure

for pairwise comparisons. Both statistics control the

error rate per family. The significance level for proba-

bilities was set at .05. The results of these analyses are

presented in Chapter IV.




To address the questions of how a contingent monetary

reward would influence learning process, social interactions,

and attitudes of children learning to play a game of skill,

data were subjected to a 2 x 4 split plot analysis of

variance. Questions pertaining to skill development,

knowledge of competence, and willingness to reengage the

game were addressed by means of a one-way analysis of

variance. Table 1 presents a summary table of the vari-

ables, the direction of predictions for hypotheses as

they were presented in Chapter I, and a description of

the analysis of variance procedures.

The extent of the relationship among certain variables

was evaluated by Pearson product moment correlations. Com-

plete correlation matrices are presented in Appendix D (see

Tables A-1, A-2, and A-3). Wherever these correlations are

pertinent to the results, they are reported as coefficients

of determination (R2), an expression of the amount of

variance two measures hold in common.


Summary Table of Variables, Predictions, and
Analysis of Variance Procedures for the
Hypotheses of the Study

2x4 Split Plot* One-Way
Variable Prediction ANOVA ANOVA

1. Intervals Containing Match N < R
2. Practice Time Preceeding N > R
First Match
3. Engaged Time N > R v
4. Mean Difficulty of Boards N > R /
5. Minimax Strategies N < R /
6. Instructional Events N > R /
7. Interference N < R /
8. Look-at-Me Statements N < R /
9. Arguments N < R /
10. Praise N > R /
11. Encouragement N > R /
12. Feelings of Frustration N < R /
13. References to Cheating N < R /
14. Cheating a Partner N < R /
15. Mean Scores on Skill Test N > R /
16. Accuracy of Predictions N > R /
on Skill Test
17. Ratings of Game Difficulty N < R /
18. Expressions of Skill Require- N > R /
ments of the Game
19. Internality of Attributions N > R /
20. Expressed Liking of the N > R /
21. Importance Placed on N < R /
22. Reengagement N > R

*The nature of the hypotheses presupposed a two-factor interaction.
Therefore, wherever the overall F test was nonsignificant for the
condition x time interaction, the null hypothesis of no differences
was not rejected. If a significant condition x time interaction was
found, the analysis proceeded to test whether the interaction was due
to specific N vs. R differences (i.e., Time 1 vs. Time 2) or to com-
bined NN, NR, RN, or RR differences (i.e., Condition). Specific
comparisons were then made to determine the location and direction
of the differences.

An estimate of the total variance among the ratings of

the experimenter and two observers blind to the hypotheses

of the experiment which was due to actual differences among

conditions is reported for qualitative variables. The

results of the data analysis are presented by major areas

of interest. For each variable, results of the analysis

are presented, followed by summary tables of the analysis

of variance and any follow-up comparisons that were performed.

Learning Process

Five variables were analyzed in an attempt to determine

whether the four groups differed in their approach to the

learning process. (The first three of these assessed the

influence of the contrived reward on the amount of time

subjects spent competing, practicing, and actually engaged

in playing their own game. The other two examined the

possible tendency of subjects to undermine their own skill


Percentage of Intervals Containing Match Play
(Hypothesis 1)

Table 2 summarizes the results of the split plot

analysis of variance. Because there was a significant

condition x time interaction, simple effects were analyzed

for time 1 and time 2. The results of these follow-up

comparisons of simple effects are summarized in Table 3.

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.9 - mvs