Matching-to-sample in young children

MISSING IMAGE

Material Information

Title:
Matching-to-sample in young children effects of the number of choices on microcomputer picture-word matching
Physical Description:
xii, 228 leaves : ill. ; 28 cm.
Language:
English
Creator:
Campbell, Pamela
Publication Date:

Subjects

Subjects / Keywords:
Word recognition   ( lcsh )
Computer-assisted instruction   ( lcsh )
Microcomputers   ( lcsh )
Education, Preschool   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1987.
Bibliography:
Includes bibliographical references (leaves 186-197).
Statement of Responsibility:
by Pamela Campbell.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001043395
notis - AFC6204
oclc - 18420385
System ID:
AA00002153:00001

Full Text














MATCHING-TO-SAMPLE IN YOUNG CHILDREN:
EFFECTS OF THE NUMBER OF CHOICES ON
MICROCOMPUTER PICTURE-WORD MATCHING





BY


PAMELA


CAMPBELL


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA
1987





























































Copyright


1987


n, ,, ,r,












ACKNOWLEDGMENTS

To the amazement of some, and the relief of others, this deed is


done.


The pages that follow represent yearslong demonstrations of


patience, trust, humor, and love from those who have shared this


experience with me.


They deserve more than the following; yet, let these


words serve as a beginning.


I am indebted to all members of my doctoral committee.


William


Wolking, Chair, demonstrated and demanded excellence and scientific


rigor and did so with thoughtful kindly guidance.


Drs. Cecil Mercer, Lee


Mullally, and Roy Bolduc provided models of effective teaching and


moments of clarity and did so as sensitive gentle persons.


Drs.


Bill Reid


and Stuart Schwartz critiqued and suggested and instructed in their


determination to make a competent writer out of me.


To all, I extend my


gratitude.
Others provided support in less formal, yet no less meaningful,


ways.


Drs. Larry O'Shea,


Vivian Correa, Bob Algozzine, Jo Hendrickson,


Cathy Morsink, and Sandra Fradd gave me opportunities to reach and


attempt the difficult task.


Those who passed through Norman Hall before


me--Drs. Lori Korinek, Bee Crews, Lee Clark, Charlie Hughes, Greg
Valcante, and Craig Smith--were friends who provided models of
determination and footwork, as well visions of life after the dissertation.








King, Karen Sealander, Kim Stoddard, and Cindy McCrery were good and


kind friends.


To all, I extend my gratitude.


There were also those who were not as near, and for whom these


years have been the longest and most difficult.


To Mark and Debbie


(whose love and enthusiasm were unquestioned) and to my Mom and Dad
(whose support and belief were always near), I extend my unceasing love


and admiration.


To Bill Baker (who shared the original vision), I offer my


love, my respect, and my gratitude.


His insight and ability to suggest the


alternative that made ultimate sense, his patience and steadfastness, and
the generosity of his spirit have sustained me.
These respected persons have given me an honored gift--the tools


with which to teach teachers.


I am now dutybound to share my skills with


those who follow--a commitment I take not lightly.














TABLE OF CONTENTS


ACKNOWI EM


LIST OF TABLES


LIST OF FIGURES


ABSTRACT


CHAPTERS


/INIRODUCION .


Background and Need for Study . .
'Statement ottie mIll. .. ..
Importance of the Problem. . .
Nature of the Investigation . .
Questions Under Investigation . .
Delimitations of the Study . .
Limitations of the Study . .
Summary and Overview of Remaining Chapters


REVIEW OF THE LITERATURE


Selection of Relevant Literature .
Overview of Matching-to-Sample. .
Learner Activity in Matching-to-Sample
Number of Matching-to-Sample Choices
Summary . .. .


. 16
. 18
. 22
. 41
.81


. 82


Subject Lsrpkn .
Setting . .
Variables Under Investigation
Measurement. . .
Experimental Design .
Experimental Procedures..
Materials . .
Data Collection and Analysis


. 85
. 86
. 87
. 90
. 91
. 96
. 99














RESULTS.


.103


Accuracy of Responding
Speed of Responding .
Learning . .
Pn o Resp Choices
Maintenance .
Summary of Results .


.103
.112
.116
.125
.133
.152


DISCUSSION.


. 154


Research Findings
Research Implimcatimns
Summary .


.154
.170
.183


GLOSSARY


.184


.186


REFERENCES


APPENDICES


PERMISSION AND INFORMED CONSENT FORMS


WORD RECOGNITION PROTESTS


. 199


.203


ENGLISH/SPANISH WORD LISTS


PICTURE NAMING AND RECOGNITION PROTESTS


MATCHING-TO-SAMPLE PROTESTS

DATA COLLECTION PROCEDURES


BIOGRAPHICAL SKETCH


.206
.212


.216


.224


.228













LIST OF TABLES


Table


Summary of Two-Choice Matching-to-Sample Studies
Summary of Three-Choice Matching-to-Sample Studies
Summary of Four-Choice Matching-to-Sample Studies
Summary of Five-Choice Matching-to-Sample Studies
Summary of Six-Choice Matching-to-Sample Studies
Summary of Eight-Choice Matching-to-Sample Studies


Page
44


Subject Data


Differences in Accuracy of Performance Between
Three- and Six-Choice Conditions


104


Differences in Speed of Performance Between
Three- and Six-Choice Conditions


Differences in Movements per Minute Between
Three-and Six-Choice Conditions


Celerations for Three-and Six-Choice Conditions


126


Percentages of Response Choice Positions
Differences of Response Choice Positions from
Computer Standard
Summary of Experimental Phase Performances:
Comparisons Between Conditions
Differences in Accuracy Between Three- and
Six-Choice Conditions-Maintenance


130


134










BagEe


Differences in Speed of Performance Between
Three- and Six-Choice Conditions- Maintenance


4-10


4-11


4-12


Differences in Correct Movements per Minute
Between Three- and Six-Choice Conditions-
Maintenance


Differences in Proportion of Responses by Response
Choice Position Between Three- and Six-Choice
Conditions-Maintenance


Differences in Proportion of Responses by Position
Between Phases


4-13


4-14


4-15


Response Choice Position Differences from
Computer Standard-Maintenance
Combined Response Choice Position Differences
from Computer Standard Across Experimental and
Maintenance Phases


Differences Between Experimental and Maintenance
Phase Performances


144













LIST OF FIGURES


Page


Linear two-choice pattern


2-2
2-3
2-4
2-5
2-6
2-7
2-8
2-9
2-10
2-11
2-12
2-13
2-14
2-15
2-16


2-18


2-19


Triangular two-choice pattern
Triangular/linear two-choice pattern
Rectangular two-choice pattern
Triangular two-choice pattern
Linear/random two-choice pattern
Linear/matrix two-choice pattern
Triangular three-choice pattern
Triangular/linear three-choice pattern
Circular/random three-choice pattern
Circular/lower three-choice pattern
Circular/pentagon four-choice pattern
Triangular/linear five-choice pattern
Circular five-choice pattern
Circular five-choice pattern
Rectangular six-choice pattern
Matrix eight-choice pattern
Circular eight-choice pattern
Frequency of matching-to-sample studies


Fi~ure










Page


4-5

4-6

4-7

4-8

4-9

4-10

4-11

4-12


Subject 1: Accuracy of responding in the three-
and six-choice accuracy conditions by sessions
Subject 2: Accuracy of responding in the three-
and six-choice accuracy conditions by sessions
Subject 3: Accuracy of responding in the three-
and six-choice accuracy conditions by sessions
Subject 4: Accuracy of responding in the three-
and six-choice accuracy conditions by sessions
Subject 5: Accuracy of responding in the three-
and six-choice accuracy conditions by sessions
Subject 6: Accuracy of responding in the three-
and six-choice accuracy conditions by sessions
Subject 1: Differences in learning between three-
and six-choice conditions


Subject 2: Differences in learning between three-
and six-choice conditions


Subject 3: Differences in learning between three-
and six-choice conditions


Subject 4: Differences in learning between three-
and six-choice conditions


Subject 5: Differences in learning between three-
and six-choice conditions


Subject 6: Differences in learning between three-
and six-choice conditions


Hypothetical comparisons of speed and accuracy


F1~1~








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


MATCHING-TO-SAMPLE IN YOUNG CHILDREN:
EFFECTS OF THE NUMBER OF CHOICES ON
MICROCOMPUTER PICTURE-WORD MATCHING
By
Pamela Campbell


December, 1987


Chairman: William D. Wolking
Major Department: Special Education


Matching-to-sample is an instructional procedure that has been
used successfully across a variety of learners, settings, and instructional


tasks.


Despite the proven success of this procedure with the learning of


new and abstract information, the effect of the number of choices on
learner performance has not been systematically investigated.
In this study, three-choice and six-choice response conditions were
manipulated to determine the effects upon accuracy and speed of
responding, learning, and response choice position in a word to picture


matching task.


The investigation was replicated within six subjects and


across two experiments.


Comparisons were also made to determine


whether observed effects on accuracy, speed of responding, and response
choice position were maintained after an interval of noninstruction.
Given three choices, subjects demonstrated higher percentages of
correct responses (accuracy) and greater numbers of correct responses








noted


, subjects generally maintained more efficient performances across


measures when learning and maintenance phases were compared.
Three choices appear to be more appropriate than six choices for
preschool-aged learners using this matching-to-sample procedure.


Implications for researchers,


instructional designers,


and consumers are


delineated with particular emphasis on the skills of young learners using
the matching-to-sample procedure.

















CHAPTER I
INTRODUCTION
Matching-to-sample is one instructional technology that has
proven effective across many types of learners and has particularly
interesting applications for handicapped learners (Etzel & LeBlanc,


1979; Mackay, 1985).


In this procedure, a sample stimulus is


presented and the learner is then able to select and match a response
from alternative choices surrounding the stimulus.
As microcomputers have become an integral component in the
educational process, matching-to-sample procedures formerly
reserved for the research laboratory can now be immediately
implemented in the classroom (LeBlanc, Hoko, Aangeenbrug, & Etzel,


1985).


Microcomputers that use touchscreens for responding make it


possible to implement precise and efficient matching-to-sample
programs that are particularly appropriate for very young or


handicapped learners (Rettig,


1986b).


Although various investigators have demonstrated the
importance of the positioning of stimulus and response options and
the degree of learner involvement, unanswered questions remain








provide empirical answers for these questions before
microcomputer programs are written with technical features that
entrench less than optimal practices.
Whether learner achievement is a function of the number of
response choices in a matching-to-sample procedure warrants


investigation.


In this study, the well-established teaching procedure of


matching-to-sample was used to teach readiness skills to preschool


students via the microcomputer.


The effect that the number of choices


presented during matching-to-sample tasks has on the acquisition of
stimulus control was investigated.
Background and Need for Study

Learning Behavior
Skinner (1968) has defined teaching as the arrangement of


reinforcement


contingencies for tasks that are divided and


reinforced in small steps. As a result, behavior can be maintained
at given states of strength. Although researchers have been able to
demonstrate the effectiveness of these principles in the operant
laboratory, many teachers continue to provide instructional
delays, inconsistent program sequences, a lack of progressive
approximations to final behavior, and poor schedules of


reinforcement (Skinner & Epstein,


1982).


Reading, grammar,


reading comprehension, and mathematics skills are all examples
of important behaviors to which behavior technology has not been


applied in a systematic and thorough manner.


With










The Microcomputer
In less than 40 years, computer technology has progressed
from large and inaccessible vacuum tube structures to small and


accessible microcomputers,


with learner-friendly software (Bramble


& Mason,


1985).


This increased accessibility is reflected in larger


numbers of individual units in classrooms and concomitant growth
in federally supported funding during the past 10 years (Moskowitz


& Birman, 1985).


Widespread computer availability has been the


direct result of pressure from parents, administrators, and
"grassroots" classroom experts to infuse computers into educational


programming (Blaschke,


1985).


With advances pending in the areas of intelligent computing and
artificial intelligence, educators are facing further instructional


intrusions from electronic technologies.


Teachers have had little time


in which to evaluate the advantages and/or disadvantages of computer


use as either an instructional or non-instructional tool.


Were this


trend to continue, the distinct possibility exists that instructional
potential for computer instruction may go be unrealized by the
classroom teacher.
Computers have been employed in the instructional setting both
as an instructional vehicle and as a noninstructional tool for the


management of instruction.


Computer-Assisted Instruction (CAI)


refers to the use of the computer to achieve a direct instructional








goal (Behrmann, 1984; Hagan, 1985).


In accordance with the


procedures for implementing effective teaching procedures (Englert,
1984), microcomputers can be used to individualize across ability levels
and subjects through the systematic presentation of content,
provision of immediate reinforcement, and recording of subject
responses (Kameenui, Camine, Darch, & Stein, 1986; Torgesen, 1986).
Further, the use of the microcomputer as a delivery system and
research tool eliminates the need for "experimenter-blind" conditions
due to possible experimenter cueing effects.
Various types of CAI software programs have been developed
and are generally categorized as simulations, drill and practice, games,


or tutorial.


Tutorial programs provide learners with opportunities to


learn new concepts under the guidance and control of the computer


program.


Learner responses provide the basis for reinforcement and


remediation (Kelly, Carnine, Gersten, & Grossen, 1986).


In a similar


way, matching-to-sample provides for the presentation of a stimulus
and an opportunity for the learner to respond by selecting one of


several response choices.


Contingent upon correct responding,


reinforcement or remediation is provided,


followed by the presentation


of the next stimulus.
Matching-to-Sample
Matching-to-sample provides the means for developing
stimulus control for a variety of behaviors and is especially effective for


complex and abstract behaviors (Ferster & Perrott,


1968).


This








procedure has a lengthy history of proven success.


In the 16th


century, matching-to-sample was used by Itard to teach Victor, the
Wild Boy of Aveyron, to match spoken words to objects and actions.
Using auditory and visual discrimination embedded within chains of
behavior, Victor learned to wait, look, and listen carefully in order to
obtain food. The matching-to-sample procedure was used to extend
communication skills to include his spoken name and finally the


association of spoken words with pictures.


Although Itard is often


cited as being disappointed with the overall effects of the instructional


program,


Victor was finally able to name objects after matching-to-


sample training (Stoddard,


1986).


More recently, both animals and handicapped learners
have been used in laboratory investigations of behavior in which
the matching-to-sample procedure has been utilized (Ferster &


Perrott,


1968).


Researchers have been successful in teaching


pigeons to discriminate and reject defective drug capsules


(Verhave,


1966) or to match playing cards to their


corresponding suit name (Skinner & Epstein, 1982).


Five- and


8-year-old autistic and retarded subjects have been trained via
matching-to-sample to discriminate circles and ellipses with
greater precision than Harvard graduate students (Sidman &


Stoddard,


1967).


Finally,


when matching-to-sample


techniques are used, learning can occur rapidly as learners








willingly persist at difficult tasks for long periods of time (W


Wolking, personal communication, April,


1987).


Because instruction based upon behavioral principles
requires incremental sequential instructional steps and
immediate reinforcement, elaborate equipment has been
developed for use in laboratory settings. This equipment includes
slide projectors, photocell response keys, and reinforcement


dispensers.


Implementation of effective technologies has


subsequently been difficult for the individual classroom teacher.
Based on the earlier work of Pressey. Skinner and Holland
developed rudimentary teaching machines that presented
information and provided opportunities for learner response and


large numbers of very small reinforcers.


Many educators, not


understanding the purpose or potential role of such machines,


resisted their placement in the classroom (Skinner,


1986).


Although the effectiveness of the matching-to-sample procedure,
both an instructional and as a record-keeping tool, has been
documented, many researchers continue to use crude equipment


of questionable reliability (Devaney, Hayes, & Nelson,


Spradlin & Saunders,


procedure.


1986;


1986) with which to implement the


Those who have implemented electronic technologies


have done so only in a peripheral manner and not for the
purposes of providing instruction in a manner that is systematic
and based upon effective instructional design principles.








skills have often been used as stimulus and response choices
have not been investigated in a sequential and hierarchical


manner (Spradlin & Saunders,


1986).


Therefore, despite the


proven success of matching-to-sample, relatively few studies exist
with learners in actual classrooms.
The microcomputer can function as a tool to not only implement
the matching-to-sample procedure as an effective teaching procedure,
but also to examine procedural variables within the procedure that are


manipulable.


Simultaneous or delayed responding, the degree of


learner activity, the arrangement of sample and response stimuli, or
number of response choices are procedural variables that can be
manipulated in order to systematically vary the matching-to-sample
procedure.
In delayed responding, the sample stimulus is removed prior
to the presentation of the surrounding matching choices and the


learner is given the opportunity to remember.


This type of format


appears most suited for maintenance and retention of previously


presented information.


Delayed responding may not be most


appropriate for the acquisition of new skills.
Learner activity can be increased by requiring the learner to


point to the sample stimulus.


In early studies with animals, birds


did not demonstrate the strong behavior of looking at the sample.


separate reinforcer is required.


By requiring the subject to touch


the sample stimulus, attention is fostered and the likelihood








of control by the sample stimulus is increased (Skinner & Epstein,


1982).


Skinner has suggested that humans, monkeys, and apes


are reinforced simply by the ultimate success potential provided by


choosing (Skinner & Epstein, 1982).


However, preattention skills


are often the concern of educators of young children or those with


immature attending skills.


The use of the matching-to-sample


procedure of touching the sample stimulus, together with the
technology of the touchscreen and the microcomputer, has been
suggested by Rettig (1986a) as the most efficient teaching strategy
for learners with young mental ages.
Sample and matching choices can be arranged in a variety


of patterns.


In many of the early studies, pigeons were required to


peck keys on the left or right of the sample (Ferster & Perrott,


1968).


In early studies with humans, a sample stimulus was


presented in the center of either a 3x3 array or circular pattern


(Sidman,


1971; Sidman & Stoddard.


1967).


Patterns and stimuli


arrangements of stimuli can also be varied to provide the
opportunity for alternative numbers of response choices.


Numbers of choices have ranged from two to eight.


Sidman has


supported the notion that more than two choices are needed but
exactly how many remains unclear (personal communication,


October 29, 1986).


Standard descriptions of matching-to-sample


describe general procedures for the sequence of sample stimuli
and reinforcement schedules yet fail to address the issue of








(cited in Becker, Engelmann, & Thomas,


1975) has demonstrated


a precise structure for the systematic and cumulative variation of
alternatives in which all relevant and irrelevant characteristics of
the positive and negative stimuli are presented in two-choice


discrimination.


Engelmann and Carnine have also stated that


learners can process up to 11 items on a page (Engelmann &


Carnine,
than 10.


1982), while Woodward (1987) has suggested no more
Miller (1968) has affirmed that learners can recall or


efficiently use up to 7 individual pieces or clusters of information.


Thus


, although effective learning has been demonstrated using


each of the above formats, some question remains as to the most
efficient and effective number of choices for learners at specific
ability levels or learning phases.
Summary


The matching-to-sample procedure is an instructional
format that is well-suited to the presentation of new and complex


concepts.


With the microcomputer, instructional procedures can


be implemented in a consistent, reliable, and efficient manner


according to the needs of the individual learner.


The combination


of matching-to-sample and the microcomputer in classroom
settings can provide rapid access, consistent presentation, and
reliable recording of the accuracy, speed and position of learner
responses to an efficient instructional procedure.








Statement of the Problem


Although matching-to-sample has been demonstrated over

time to be an effective and efficient instructional technique, the

application of this procedure to the classroom has been limited.

Concurrently, a number of procedural variables in the matching-to-


sample procedure have not been systematically investigated.


The


rapidly expanding capability of the microcomputer has occurred

simultaneously with the placement of increasing numbers of


microcomputers in classrooms.


The potential therefore exists for


the implementation of instructional programming, including
matching-to-sample, that is neither well-designed nor appropriate

for the learner.


rT^
<4 Shf


Importance of the Problem


The problem of measuring and documenting the details of


~~~Vta


performance in stimulus discrimination via matching-to-sample


procedures is important for several reasons.


First, the success of


the matching-to-sample technique as an instructional technique
has been demonstrated and replicated with animals and human

learners; yet, investigators have failed to address the


effectiveness of the numbers of response choices.


Second, applied


research in the classroom setting has been limited due to the
need for costly equipment and time-consuming investigatory


procedures.


Descriptions of learning performances by the


average or mildly handicapped learner using matching-to-


A


,.pl








exists or is potentially available in most classrooms to implement
a strategy that is similar in design and purpose to that proposed


in this study.


Fourth, the abilities to attend, discriminate, and


match stimuli are considered important skills in the acquisition
of readiness skills and have been linked to student achievement


in both regular and special education.


Fifth, Engelmann and


Carnine (1982) have outlined the systematic presentation of
non-examples as well as examples during the acquisition stage of


learning.


Further information is needed as to both the role and


the numbers of non-examples in learning tasks with more than


two choices.


Sixth, many learners of elementary school age have


persistent and resistant problems with stimulus discrimination.
Stimulus discrimination requires that the learner respond
accurately even when another stimulus with similar physical
properties or previous stimulus association is presented.
According to Pryor (1984), stimulus control is demonstrated
when the learner responds appropriately only in the presence of
the specific stimulus, never in the presence of an alternative
stimulus, never in the absence of the specific stimulus, and never
with an alternative response in the presence of the stimulus.
Methods for developing stimulus discrimination have included
two-choice discrimination and matching-to-sample procedures.
Seventh, given the current emphasis on the identification and
development of programs for young learners with special needs,








matching-to-sample procedure can provide such


demonstrations.


This study may provide a foundation upon


which to build a feasible and reliable instructional strategy with
applicability to students of varied ages and discriminative ability.
The results of this investigation should add to the developing
body of knowledge on the technology of teaching.
Nature of the Investigation
The number of response choices in the matching-to-sample
procedure is a manipulable variable that has not been systematically


studied and is in the early phases of investigation.


This study was


therefore exploratory to determine whether the notion of learner
differences based upon the number of choices is sensible and, if so,
whether differences are large enough to recommend additional


research or alternative matching-to-sample teaching procedures.
single subject experiment by design permits and accounts for the
amount and nature of any performance differences.
Questions Under Investigation
This study was conducted to examine the effect that the
number of choices has on the ability of the student to learn and


develop stimulus discrimination.


The


Two conditions were


systematically manipulated in the context of matching words to


pictures by preschool-age learners.


The number of correct and


error responses, response time, and response position were


recorded under both conditions for each subject.


The following








performance in a three-choice as compared to a six-choice


condition?


In addition, five related questions were investigated.


Does the number of response alternatives affect learning accuracy


during the acquisition of stimulus control?


Does the number of


response alternatives affect speed of responding during the


acquisition of stimulus control?


Does the number of response


alternatives affect celebration (rate of learning or improvement)


during the acquisition of stimulus control?


Does the number of


response alternatives affect response choice position during the


acquisition of stimulus control?


Does the number of response


alternatives affect learning performance during the maintenance


of stimulus control?


Maintenance of stimulus control was


determined by a single measurement of learning accuracy,
learning speed, and response choice position.
Delimitations of the Study


This study was delimited in several ways.


The investigation was


conducted in Alachua County, a medium-sized county in north-


central Florida.


The subjects for this study were of preschool age, not


diagnosed as handicapped, and selected from a preschool in


Gainesville, Florida.


Additionally this study involved only sufects


whose parents who volunteered to permit their children to participate


in this research project.


There was no consideration given to the sex or


socioeconomic status of the subjects.


The tasks employed for


instruction in the individual learning trials for this investigation








Limitations of the Study

Several limitations were imposed that may restrict the


generalizibility of the findings.


Specific eligibility or entrance criteria


used at the Baby Gator Preschool may limit the generality of findings


to other preschool settings.


Since the learners in this study were


selected from a preschool program, generalization to grade levels in
which younger or older students are enrolled should be made with


caution.


Further, results may not be generalized to either older or


younger students or to those individuals diagnosed as handicapped,
without systematic replication with samples from those populations.
Summary and Overview of Remaining Chapters
The effects of three- and six-choice alternatives in a matching-
to-sample procedure were addressed in this study for the purpose of
identifying whether one condition is more effective in the acquisition
and maintenance of picture/word matching skills among preschool


learners.


Empirical evidence linking this variable via computer


instruction is lacking in both regular and special education.


This


investigation was designed to further define the components of the
matching-to-sample procedure via the microcomputer.


In Chapter II,


a review and analysis of pertinent matching-to-


sample studies are presented with emphasis on the following: learner
activity, the arrangement of sample and matching stimuli, and the


number of response choices.


The implications of previous research for


educators of preschool and special education students are discussed








investigation are described and, in Chapter IV,


described.


the obtained results are


Finally, a discussion of the results in terms of current


knowledge,


implications for future research,


instructional designers, is and


classroom implementation are presented in Chapter V


. Definitions of


terms used in this investigation are presented in the Glossary.















CHAPTER II
REVIEW OF THE LITERATURE

The purpose of this chapter is to synthesize existing
literature related to manipulable procedural variables of the


matching-to-sample procedure.


Examples of procedural variables


are the relative position of sample or choice stimuli, the size of the
choice stimuli, or the time between presentation of the stimulus


and the choice stimuli.


These variables may require definition and


description in order to attribute results to independent variables


or to the matching-to-sample procedure itself.


Criteria for


inclusion in this review are stated first, followed by a discussion of


learner activity as it relates to matching-to-sample.


This chapter


concludes with a review of matching-to-sample studies according
to the numbers of choices provided the learner.
Selection of Relevant Literature


The selection of relevant literature for inclusion in this study


required the delineation of selection criteria.


The goal of the review


was to examine matching-to-sample studies in which descriptions


of manipulable variables have been included.


Further, literature


related to the activity of the learner when participating in the






17

All relevant research published in the last 20 years (1967-1987)
as well as other important investigations or findings prior to that


period of time were reviewed.


Reference sections from related articles,


chapters, and books reviewed were sources for locating less recent


research.


Studies that used the matching-to-sample procedure had to


meet the following criteria for inclusion in this review:
1. Some aspect of learner behavior was included in the
dependent measure.
2. A description of the numbers of stimulus and response
choices was included.
3. A description of the arrangement of the stimulus and
response choices was included.
4. A description of learner activity was included.
Sources reviewed included Current Index of Journals in


Education (CIJE),


Education Index Knowledge Index. Educational


Resources Information Center (ERIC),


Library User Information


Service (LUIS), card catalogs at the University of Florida libraries,
and personal communication with professionals in the fields of


special education and matching-to-sample research.


Descriptors


used for the literature search included matching, sample,
matching-to-sample, attention, preattention, stimulus control,
effective learning, and learner responses.
Certain criteria were applied to the studies located in the initial
search to ensure that only methodologically sound studies would be








included in the literature


review.


These criteria were as follows:


. The subjects and setting for the study were thoroughly
described.


The dependent variables,


procedures, and materials were


thoroughly described.
The data display and analysis were presented without
significant losses of information.


The investigator's
results displayed.


conclusions were consistent with the


Forty-two matching-to-sample studies were reviewed.


Because


one study incorporated two different numbers of choices within the


same study, this review actually encompasses 43 studies.


Only one


study was located in which a matching-to-sample procedural variable
related to numbers of choices was investigated.
Overview of Matching-to-Samnle


The matching-to-sample procedure has been studied and
implemented successfully across a wide spectrum of learners and


learning activities.


Studies with nonhuman learners such as


pigeons (Cumming & Berryman,


1965


Verhave


,1966),and


monkeys and baboons (Sidman, Rauzin, Lazar, Cunningham,
Tailby, & Carrigan, 1982) have been and continue to be conducted in


laboratory settings.


Studies with human learners have been


conducted both in the laboratory and in classrooms with retarded


children and adults (Sidman,


1971


Sidman & Cresson, 1973;








Kotlarchyk, in press),


preschoolers (Lahey & McNees,


adolescents (Sidman, Cresson, & Willson-Morris,


1974


1975),
), college


students (Skinner, 1968),


and adults (Lazar,


1977).


Matching-to-sample has also been used successfully to


teach a variety of tasks.


Simple discrimination tasks such as


identity matching (Carter & Eckerman, 1973


Sidman


, 1971),


well as complex concepts and symbolic tasks (Sidman, Rauzin,


Lazar, Cunningham,


Tailby, & Carrigan, 1982) have been taught


with high accuracy and efficiency.


The procedure has also been


used successfully to teach spelling (Mackay


comprehension (Sidman,


1985) and reading


1971).


As the matching-to-sample procedure has evolved in the


past 30 years,


techniques for implementing the procedure have


changed in ways not necessarily related to substantiated findings


of increased effectiveness.


Initial studies in which matching-to-


sample was used were conducted with animals in specially


constructed chambers.


As a result of these early studies, more


sophisticated laboratory equipment that could deliver greater
numbers of alternatives in a wider variety of formats to both


human and nonhuman learners was designed.


The teaching


machine was developed in an attempt to translate the matching-
to-sample technique into a teaching technology that could be


implemented in applied classroom settings.


Resistance from


educators resulted in the continued use of flashcards, semi-






20

implementing matching-to-sample procedures classroom
settings.
The ability of technologies to deliver more sophisticated
patterns has dictated the format of certain procedural variables.
In laboratory studies, the matching-to-sample procedure has been
continually refined electronically without certain procedural variables


being systematically challenged and investigated.


Because the


microcomputer is now capable of delivering sophisticated and reliable
matching-to-sample procedures in the classroom in an atmosphere


of ready acceptance and high motivation (Serna,


1987), a


determination of procedural variables is needed to reduce the
possibility of misinterpretation and inefficient learning.
Cumming and Berryman (1963) suggested three classes
of matching-to-sample procedures that can be manipulated


and therefore warrant investigation.


The first category,


reinforcement contingencies, that govern the effects of
responses to either the choice or comparison stimuli, has been
studied by Verhave (1966) and Sidman and Stoddard (1967).
The second variable, the temporal relations between the choice
and comparison stimuli, has been the subject of studies by


Sidman (1969) and Nelson and Wasserman (1978).


The effects


of simultaneous versus delayed presentation of choice stimuli


have been the focus of these studies.


The third procedural


variable, the interrelations of the choice and comparison stimuli








both in terms of the number of stimuli involved and their


specific stimulus properties, has not been investigated. A
fourth procedural variable deserving investigation is whether


the matching task is iconic or symbolic.


Iconic or identity


matching requires the learner to match visually identical


symbols, such as a circle to a circle.


Symbolic matching requires


the learner to match visually dissimilar symbols, such as a
lowercase letter to its corresponding uppercase match.
Cumming and Berryman stressed the importance of


investigating the numbers of choices in 1963


more recently, this


same need has been restated and reinforced by Sidman (1987).


The


number of choices offered to the learner has continued to vary in
matching-to-sample studies and appears contingent on the
investigator's decision as to the type of pattern and equipment to be


used.


Therefore the number of alternative choices is a direct


function of the arrangement of the stimuli.


Only one study in


which a procedural variable related to the interrelationship between
sample and choices was systematically manipulated and examined.
In this study, two rhesus monkeys learned hue and line


discrimination.


Iverson, Sidman, and Carrigan (1986)


manipulated the position of the sample stimulus in a two-choice


linear format across two conditions.


In the experimental condition,


the sample choice randomly appeared in any of three positions;
this randomization determined the subsequent position of the two-








appeared in the center key.
the location of the stimulus.


Learners made choices based on
Discriminations that had been


learned in the control condition then deteriorated under the


experimental condition.


The investigators concluded that an


accurate description of the controlling choice stimuli should
therefore also include spatial location.
The success of matching-to-sample is also dependent on


learner involvement during the procedure.


Learner activity,


which has been standardized as looking and touching behavior,


is therefore discussed next.


As no studies were located in which


numbers of choices were experimentally manipulated in a
systematic and direct manner, a descriptive review of major


matching-to-sample studies follows.


This review contains an


historical overview of matching-to-sample and is organized by
numbers of choices.
Learner Activity in Matching-to-Sample
As defined by Cumming and Berryman (1963), matching-
to-sample is a conditional discrimination procedure in which the
learner must first respond to a sample stimulus in order to


produce comparison stimuli.


This initial response by the learner


serves to set the occasion for the presentation of the matching


choices.


Traditionally, attention to the stimulus has been obtained


by direct learner contact with the sample stimulus either by


pecking, pointing, or touching.


A discussion of the importance of








this initial activity as a situational setup for initiating stimulus
presentation and choices follows in terms of the learner first
looking and then touching.
Looking as a Feature of Attention


The ability of learners to learn through looking is dependent
on both the ability of the learner and the presentation of the


stimulus.


Learner ability and stimulus content are therefore


discussed below.
Learner abilities


Dependent on several skills that are often taken for granted
in the learning situation, learners are able to attend to stimuli.
These skills include being able to see and process information, to


prepare for the presentation of stimuli via preattention skills,


or to


benefit from specific training that emphasizes the development of


preattention or orienting behaviors.


A thorough review of this body


of literature is beyond the scope of this study; therefore an overview
of the major studies is provided.


Visual Drocessineg.


The visual acquisition and processing


system is the human processing system that is most accurate and
at the same time most basic to school learning tasks (Getman,


1965).


Reaseachers have described infant visual systems that are


functionally near adult level at 4 months of age (Clark-Stewart &


Koch,


1983).


In addition, scanning (Salepatek, 1969; Wollf, 1965),


discrimination (Brinker & Lewis,


1982; Fantz, 1966),


and choice








have been described among infants and very young children.
Learners of preschool age can therefore perceive and process visual
information with high degrees of accuracy.


Preattention.


Learning is also dependent on the learner's


ability to look, listen, and sit quietly during instruction.


In the


classroom, behaviors such as a particular posture, observing, or
orienting towards a stimulus can be indicative of preattending


Students who are looking at materials, looking at the teacher


when the teacher is modeling materials, listening to oral
instruction, or sitting quietly for short periods of time are
demonstrating preattending or orienting behaviors (Etzel,


LeBlanc,


Schilmoeller, & Stella, 1981).
Preattention can also be detected through the careful
observation of eye and head movement or other physical


symptoms (Lewis,


1972; Neisser, 1967).


Stella and Etzel (1986)


have suggested that different stimulus manipulations can produce
reliable differences in the control of young learner's eye
orientations, that orientations might be predictors of pointing,
and that there are ways to gain control of visual orientations
through the manipulation of antecedent stimulus changes.


Training preattention.


Winslow and Etzel (1972)


investigated the preattending skills of 80 heterogeneous students


who were 4- to 5-years-old.


Manipulations of the learning


environment were implemented to alter preattending behavior.


skills.








inappropriate preattending.


Increases in preattending behavior or


reductions in inappropriate preattending were not necessarily
affected by changes in consequences, easier tasks, or errorless


tasks.


Rather the training of learner activities such as head and


upper body orientation, pointing on demand, and pointing with


index finger extended has increased preattending.


Children have


been taught preattending skills via modeling, prompting, and


shaping.


For example, one component of such training included a


changing-criterion reinforcement schedule that shifted from
preattending to task accuracy from the beginning to the end of the


training session (Winslow & Etzel, 1972).


Others have stated that


preattending may not necessarily be related to internal
motivation, but rather to the age or disability of the learner.
Preattention may therefore be amenable to training (Loper,
Hallahan, & Ianna, 1982; Miller & Bigi, 1979).
Stimulus .resentation


When considering the ability of the learner to attend, other
researchers have emphasized the content or topography of the


stimulus (Fleming & Levie, 1978; Vernon, 1952).


Cohen and


Gross (1979) have identified visual stimuli dimensions
(brightness, object movement, patterned and color presentation,
external versus internal features, size, number of elements,
attention to faces, functionality, and novelty) that follow a


developmental sequence.


Limitations in the ability to process any








or focus on an individual stimulus dimension.


For this study, the


amount of visual information that can be presented to and
successfully processed by the learner was considered in terms of
the size and organization of the stimuli, as well as the number of


learning opportunities.


To date, researchers have addressed the


number of learning opportunities in terms of the number of
problems, the number of trials per problem, and the number of
items per trial only.


Size and organization of the stimuli.


The size and


organization of stimuli are important features in the processing of


visual information.


Hershenson (1964); and Brennan, Ames, and


Moore (1966); as well as Fantz and Fagan (1975) have noted
learner preference for smaller elements except when the number of


elements has been controlled.


Hochberg (1966) and Murch (1973)


have suggested that greater accuracy in less time is possible when


learners process forms that are simple,


regular, closed,


continuous, smaller, and symmetrical with uninterrupted,


straight lines or smoothly curved contours.


The ability to process


forms in various organizational patterns has been described as a


developmental sequence by Fantz (1966)


Fantz and Nevis (1967);


and Fantz, Ordy, and Udelf (1973).
In this review of 43 matching-to-sample studies, sample and
choice stimuli were incorporated into closed four-sided figures in 35
of the 43 matching-to-sample studies reviewed for this








Of this number, 30 patterns were incorporated into horizontal
rectangles, 1 pattern into a vertical rectangle (see Figure 2-16),.


and


5 patterns into a square (see Figures 2-4, 2-7


and 2-


In6


studies


, stimuli were incorporated into a circle pattern (see Figures


2-10, 2-


,2-14, 2-15, and 2-


One


study was reviewed in which


stimuli were incorporated into a pentagon pattern (see Figure 2-


In 6 of the 18 organizational choice patterns,


circles rather than


squares were used for individual stimuli (see Figures 2-


.2-14, 2-


10, 2-


15, and 2-


Berelson and Steiner (1964) have considered learner activity in
terms of responding to patterns that are similarly organized and


grouped4
and rate


In the matching-to-sample procedure, learning accuracy


.


, as well as response direction, are directly affected by sample


and choice organization (Rettig,


1986a).


Response direction is the


direction of learner movement from the first orienting response


(touching the sample stimulus) toward the response choice.


Rettig


has stated that learners naturally proceed in a left to right
progression and has designed matching-to-sample instructional

software to conform to that motion pattern (M. Rettig, personal


communication


, April 22,


1987).


Of 43 matching-to-sample studies


reviewed for this investigation,


response demands in 16 studies


restricted the learner to a left to right motion (see Figures 2-


and 2-7) and in 3 studies,


to a downward motion (see Figures 2-2,


3. 2-5,


.2-9, 2-11


2-13


,.2-15, and 2-


Learner responses in 3























Figure 2-1.


Figure 2-2.


Linear two-choice pattern.


Triangular two-choice pattern.


Figure 2-3.


Triangular/linear two-choice pattern.


Note: s denotes sample.
c dnote x ikr.


c -


S


I, ; ::- ;;
C** \\ .r. ,.*...*..* "**.*' t.'^'-''"''' '''


















Figure 2-4.


Rectangular two-choice pattern.


Figure 2-5.


Triangular two-choice pattern.


S


S


C'.-'.. -:. .,: : .:


C : -:' l*:.' *: .


:'C.- -. :*: :


L]


[7


17



























Figure 2-6.


Linear/random two-choice pattern.


FiEure 2-7.


Linear/matrix two-choice pattern.


























Figure 2-8.


Triangular three-choice pattern.


Figure 2-9.


Triangular/linear three-choice pattern.


S




























@C


C ..-


' '


C


Figure 2-10.


Circular/random three-choice pattern.


C


@=^Sa^


C


C



























Figure 2-


Circular/lower three-choice pattern.


Figure 2-12.


Circular/pentagon four-choice pattern
(4 choices in any of 5 rim circles).


S

* -.C .
,<-*:- ^ y..:'| ^-l*1* l'-* |* q :: _- .|























Figure 2-13. Triangular/linear five-choice pattern.













w^J-D
:[ pi!iii i
C CH


Figure


2-14.


Circular five-choice pattern.
(Choices in any of eight locations).


S

























Figure


2-15.


Circular five-choice pattern
(Choices in five lower locations).


Figure


2-16.


Rectangular six-choice pattern
(Choices in any of 11 locations).


oa
^^8\ /c\ fl^
pm^^] I 9 J ^i^i:i:J
^-;.;.;.:-^- .'.***>* ^ f ; '*^^^*M111'-!-'-;-^
^SS.iiiiiiii^^^^ ^A^a.^^:^:^
t:::::::y^^^:::-| ^^ |.-:::;aT|::::::^:|
Vitiii^^l^^
\|lBll||


S
..: .......*.;.*;
C ... *.


Cr


*II




































Figure


2-17.


Matrix eight-choice pattern.


Figure


2-18.


Circular eight-choice pattern.


ft.
ft...
'tee
ft.
rn
a-. NW.
ft.... ,, ft.
C 'a


ft-ft
ft .-..-
tjd


f'^ ^^: -'-%?;^^*:*:^^. .1 '- ^ -^'^^








studies were permitted in either upward or downward motion (see


Figures 2-4, 2-10,


and 2-12); and,


in 6 studies, in any direction (see


Figures 2-14, 2-17, and 2-18).


Number of learning opportunities.


The effect of the number


of opportunities to learn has been addressed in various ways.
Carnmine (1980) advocated large numbers of examples while
Harlow (1959) noted that rate, efficiency, and amount of learning


were directly related to the number of trials per problem.


The


smaller number of trials given per problem resulted in a larger


number of problems needed to reach criterion.


Harlow also


correlated the number of trials per problem with stages of learning


and the amount learned.


Larger number of problems during


initial trials (1-100) resulted in more efficient learning as well as


greater transfer of skills to successive problems.


Hayes,


Thompson, and Hayes (1953) corroborated these findings using


chimpanzees who,
learn a new task.


with 2 trials per problem, required 300 trials to
With 10 trials per problem, learning occurred


more rapidly.


Hershenson, (1964),


Brennan et al. (1966), and Fantz and


Fagan (1975) have stated that learners prefer large numbers of


items.


Engelmann and Carnine (1982) have noted that learners


are able to work successfully with up to 11 items on a page. In
contrast, Miller (1968) has stated that there is a clear and definite
limit to the accuracy with which we can interpret and remember








judgments, the span of absolute judgment is approximately 7


units.


The commonality of the number seven has been noted


across a wide range of disciplines, stimuli, and modalities (Miller,


1968).


Chunking or clustering individual units of information can


result in increased capacity for remembering; yet the numbers of
items on a page in relation to the encoding process can limit other
information processing.
Consideration of the units of size, number, as well as
organization contribute to the presentation of stimuli in a manner


that can facilitate learning.


This consideration, in combination with


the ability of the learner, sets the occasion for the learner to
respond and produce the discriminative stimulus in the matching-
to-sample procedure.
Touching as an Active Learner Response

Often, to facilitate the preattention of the learner,
topographical cues such highlighting, arrows, color, sound, and


movement are incorporated into the stimulus presentation.


alternative to this external control,


Asan


Lewis (1972) has referred to


responses by the learner that facilitate the reception of stimuli and
help direct action to those stimuli.
Touching in matching-to-sample


In discriminative learning experiments,


learners have been


required to take some action such as pressing a response key


prior to being exposed to the discriminative stimulus


(Ferster,








demonstrated increased key pressing (mediating response) to
produce a color change when that color change was reinforced
under a fixed interval (FI) 30-second reinforcement schedule.
Kelleher (1958) used monkeys to illustrate discriminative learning
with a similar task that required a mediating response to produce


a discriminative stimulus.


Lindsley (1962) noted that the operant


response of pushing a button at 60 presses per minute to
maintain audio or visual intensity eventually became equated with
a response that resembled looking, such as bending or turning


the head.


Findings have been expanded further to include


responses to both looking and listening tasks presented in
schedules designed according to the individual learner's baseline


performance.


The act of pointing or reaching orients the learner


toward the presentation of the sample stimulus in the matching-


to-sample procedure.


Pointing can also provide an accurate


measurement of active learner responding and has been
correlated with increases in achievement (Greenwood, Delquadri,
& Hall, 1984).
Touching with young learners and microcomputers

Rettig (1986a) has insisted that touch is the only logical
response mechanism for learners of preschool age or for those


with young mental ages.


This response option has been facilitated


further by the introduction of the microcomputer into the


classroom.


The ability to scan the contents of a computer screen








choices and determine preferences as described above.


The


microcomputer can provide a tool for examining the learning of


infants and handicapped learners (Brinker,


1984).


While Porter


et al. (1986) have stated that learners as young as 15 months can
use the computer as a functional tool, Brinkley and Watson
(1986) have suggested 3 years of age as optimum for the
successful introduction to the microcomputer as a learning tool.
Rettig (1987) has corroborated this view by stating that the years
from 2 to 5 are appropriate for introducing the microcomputer to
the young learner.
The existence of large numbers of microcomputers in
classrooms also requires that software developers become aware
of the developmental age and abilities of the intended users in
order to identify appropriate software content, as well as


presentation and response modes (Johnson,


1986).


Rettig


(1986b) has stated that inaccurate measurement of a learner's
ability may increase when inappropriate computer response


modes are used.


When determining the appropriate response


mode, one should consider that the learner should be able to
respond without taking time to think about how to respond.
Rettig (1986b) investigated alternate response modes on the
microcomputer with young mentally retarded and physically
handicapped children to determine possible differences in


response frequency or accuracy.


The color matching skills of 20








years of age were compared using four different response modes
(keyboard, keyboard adaptation, paddles, and light pen) on a
microcomputer. Keyboard responding was highest in frequency
and lowest in accuracy. The paddles response mode generated the


highest improvement in accuracy over time.


Because of the


distraction factor (focus on the keyboard and not on the stimuli),
the keyboard response mode was determined to be the least
appropriate for young learners.
Rettig (1986b) has stated further that almost any adaptation in
response mode can improve the accuracy of young learners when
using the microcomputer and has targeted the touchscreen as the
most appropriate response mechanism for preschool students and


learners with young mental ages.


With a touchscreen, learners touch


directly on the screen; attention is therefore focused on the screen


and this interactivity increases learner involvement.


Responses can


be fast, easy, direct, natural, and more appropriate than other
response modes such as paddles, keyboards, or Joysticks (Semrna,


1987).


Touchscreen and microcomputer technology offer effective,


efficient, appropriate, and reliable means by which to investigate the
effect of the numbers of choices in a matching-to-sample procedure
with young learners.
Number of Matching-to-Sample Choices
The number of choices offered the learner in matching-to-


sample tasks has continued to vary.


In recent years, new








sample.


Electronic equipment has facilitated the delivery of more


sophisticated patterns that have subsequently determined the


number of choices offered to the learner.


Investigators have


appeared to change the number of choices based on findings
related to other experimental variables or the need to save


experimental time (Sidman, 1987).


These changes have occurred


not only from experiment to experiment (Sidman,
within the same experiment (Sidman & Cresson,


1987) but also
1973: Sidman, et


,1974).


A review of the historical development of the matching-to-
sample procedure resulted in the location of 43 major studies in
which the number of choices and pattern arrangement were


clearly described.


Only one study was found in which a procedural


variable that indirectly affected the number of response choices


was systematically manipulated (Iverson et al.,


1986).


No studies


in which the number of choices in a matching-to-sample
procedure was directly and systematically varied were located. In
order to provide a review of important, well-designed, and
completely reported matching-to-sample studies, these
investigations have been organized into categories that reflect the


numbers of response choices.


'Ivo-


, three-, four-


, five-, six-, and


eight-choice studies are presented in Tables 2-1 through 2-6.
These studies will be discussed according to the numbers of
choices, primarily in terms of pattern arrangement, types of






43

been the manipulated variable in a matching-to-sample study,
only those findings of major significance for matching-to-sample
technology or those indirectly related to a subsequent change in
the number of choices will be discussed.
Two-Choice Matching-to-Sample Studies
Matching-to-sample tasks in which a two-choice format (see
Table 2-1) was used originated from attempts to extend simple yes-
no conditional discrimination tasks that had been used successfully


in the laboratory.


Terrace (1963a,


1963b) taught pigeons to


discriminate without making any "errors" and then transfer this


control to other stimuli.


Sidman and Stoddard (1966) demonstrated


the effects of subtle discrimination programming on the ability of a


severely retarded adult to discriminate circular forms.


Nine


translucent panels were arranged in a three by three matrix onto


which circles or ellipses were projected.


Through instruction and


reinforcement, the subject learned first to touch the panel and then


to discriminate circles from ellipses.


Electronic control of stimulus


presentation and reinforcement provided the tools with which to
implement these studies.
Two-choice matching-to-sample equipment
Catania (1979) has described a three-key pigeon chamber
in which the sample stimulus was located in the center and
























U


0
o3 5
6 ao
01
59SP2






















0,
-a.
I)
.0
I

'00)
-4-'
'-4
P4w
4)







.4-'
V
2
a





.4-'
4)
2

U
.4-'
.4-'
04i~
Cu





0)
-a.-


o V
.4-' 1)
V P4
0
4-' II --
V
'0
V
.04)
ecu


ijE1~
~o4.J


V
o ,0
be
j~L
~ 0

vct-~U

V
*jG) 1)0
Ut)
~4Iw4


ei *4


P4


b s?
pi ,
-^ **


a!4%


U)1


2
0&
^ 8





















U)
U.)
-4
,0

I
I
~
-4
45)0)






4-h
as)
a
a





1'
0)
I)
*.Jcu






U)


E &
as
4-'
R a50



a s o%

*w-- a^ 1
*fjtUI Ii'
w *03^


U
4-1

04
o
*~6
o 0)

Ow


.43


1-4~p


4-,

I..
0
5.


bUS


&01





















0)
4.)
.0
*1

0
4)
'00)
O .d~D
a02
6~)






V
0
4.)
2


r~4



0
4)
2
~ 4)
to
4)0
Cu






0)
NW-


U
0
U
'-4
0
I.'
4)
.0
2
z


U)


O0.4~J


~ui0


F


P~oS~


'0 4.)
6)
'U
'CE'

~QI4
0)0
B~
Cu'-

to
cuj~
.4..Jt)


1.4



.0









'-4A'

S








0
4) ~6%
04


6

6) )
UI

Ucul

00
gObo
So Co

^c'a n

a e -4
0 y 3 5

1iI 2;t
OJ-d^ C


0
a
0
~4)

'ECu








to
dl


3jt
















U)
4)

I

0.)
'CU,
V
~~U)






02
2'
r~4


0)


.~ 4.J





te-.


&
U
0
4)
~ 0.




A
'-I

a


~II
~ flU


*5


LI


















0
N-
4)
-I
I
.me
0
4)







4.'
0
4)
S
a
'4



4)
S
Es?
.me
041
0




0
N.


N
0
U
'4-.

I
a
N.
S


tb.c
'0

~


,40


07

osp


ta

.0
2
4)
Es?









*1
2


0
o0
Thu
*00

0 Cu
0

Uv~4


V
El)
tAO4) -
.0*1
tiIlt4E*1
6).0
~0
0~*'





544

S.


8 b.
Nm W^


















0)
4)


I
V
4)
002
-
~40)







4)
2
04

rz~



4)
2
~ 4)
to
I-
041
Cu




U)


.5
K


gliai
a l 4)
"12^-l^~a*M b

I^Ilt^


rI


*0 14


4.)
u&o~
4-h
.~Cu
14,0411


U,


bSU Si


.4-I



~~, cU


4)
~uJ4
a
"B";





214
0
El)


104
--'Cu
c~ Go





















cc
43
z
Cu
Ii

C
0)
V.,
-
*4~
4)4)







C
a







4)
a
~ I)
4.' CU
.4-.
CU





U)


8



S
44
0
V
.0

z




$4




CV
~


$4
@3
14-Mg .4-b
On @3
430 (/2
4)
CU 0.c U
cc:j!:;'~ :~



~U2$4~OS&CU


V


$4

I

4)
4
~ Q4


4.'


Cu
.EU~4


CU


I FV


nl$





















0)
6)

I

U
'00)
-
-
V








62
2
a-
0~
rz~




4)
2
~ 4)
to
4-JCU
'4-I
Cu






0)


0)
4)
U
C
I.
.3
0
U
'4-4
0
I-.

2
z




tab'
CU
0)
a-
'0
U)

.2

I"


'o
u0

Il k
am

c a 4

.'8
Ti .w *


i&~


'SJJ



'0


-4


~ its u





S

-4
0
top,


'0

.~Cu




U) ~4.J

-

>c:~




2
flu '0

O~-CJ
iijIfliiI'
ucu'~
CUG.)


~lii


i Or-b
C-

-4

ci


Ls






















U)
4)


I

4)
vy)
~ 4-'
-4
acn
4)4)








I)
2
a

I~4



.4...
V
2
~ 0)
to
0)
4-'
Q4~
Cu




U)
4)4-.


8
p.


0








te
U





O
1$4




I-




U 04


cM
$~ 0
04


pt 5



ags u*- = o .>
.. o u ,s .a u r

pa c>.a "a
fl2 "as~g'goa

1jihj~i4)II

g||g~e'4

5SjVg'S
^ a(


V

4) 4) CU
a,
CUv

M4 0 4)UP4CJ
~ 2.Y~ pM 2
(Z~


Cu
S
..piu



U)

e'14
St.-
04


'I)


4)


orJ~






















U)
4)


I

'00)
-

4)






4-,
V








V
~ 6)
to
4.60
Cu
Q4.~





U)
t


c



OgS g

M^ag8^
il~Lll
'0 0^I
C.',


I-u V
iso
40 4)
.9.0 0
U
1.4


h-i

0)

0
P4

jP'.
4%~
4444


0
1.4
Rb
Pee
~ rn
2.2~ ~
p


00-"


I.O
MU*^


~ ML


*1d



















02
N.

4,

I
Es)

4)0)








U)
a






I)
~ U)
*00
4)
Cu





0)
N.

I~)


> .=-g g 8

0-0
0 ^E? Ca

ir-^
"5 ^ 073


*0
IKe,

.~vu
i0)


iii


04)
isl-
oo
04 a


-4


S





0)
K

~Q4~


~4$4 -


"'U


V.


I.
513
A^

























@2
Gil



F

4.)
00

~4u)
Cc








.4-I
4)

04

I~4





4)



"-'Cu
CU







0)


0


*1
.0


O0





s 0"
Os*^^ p ^


I-
4)
4-h



hi

~2~& g r6







t~ 0'~CQ 0
'.4
CU
S *. S
U







U)

C 0

eq


0
~S
0




5P4~VCU
U)


CE'


04)

4-h
4)
o~ 6
- 04


'.4
4)v~
4-
CU


~bo4)
hi hi


I-

U,"'-' CU


0




o 'C .CS 3'0
g ca~u c


-CU
c~ C')






















0)
4)
-4
.0


F

Es)
'00)
-
*4~~
s)U)









a)
2


r~4



4.'
Es)
2
~ 6)
6)
A
Cu





0)


U)

C)
U
~4mD
0
V


8v~
'4-
o~i
U)



Cs)







4-J0)














i'I:ii
~ fri U




0)
*1


-4


-g -
5e o
t -a- 3 <
3 Ss^-St
*o^ Op
^* 5 U r -
ft L 4ai ^e


C




V
04
0 .~CU


4-'
.4-h
C
6) '0
-4
04
cc


U)
U

'.4
~U)E)
4J
04 CU


~,aI
~ W0tJ


U


0)


0

Cu
V

'.4
Cu




I
6)


V
Cu

Cd
'-4
0

Cd
,~
0


a
V -

'at








comparison stimuli on the left and right of the sample stimulus


(see Figure 2-1).


Pecking on the center key ensured looking and


also illuminated the two side (choice) keys.


Pecking on the


matching choice key produced a reinforcer, followed by the
presentation of a new sample and two new comparison or choice


stimuli.


A nonmatching side-key peck could be followed directly


by the presentation of a new stimulus without a reinforcer or an


intertrial interval without a reinforcer.


Sample stimuli were varied


from trial to trial as was the position of the matching choice.
These chambers were operated electronically and were used in the
early studies by Skinner (1950); Ferster (1960); Cumming and


Berryman (1961); Nevin,


Cumming, and Berryman (1963)


Nevin


(1967); Wright and Cumming (1971)


Maki and Levin (1972);


Carter and Eckerman (1975); Thomas (1973)


Farthing and


Opuda (1974); and Nelson and Wasserman (1978).


The success of


the procedure with pigeons was also demonstrated by Verhave
(1966) in a slightly different format (see Figure 2-2) in which
pigeons matched pills to a standard as a measure of quality
control.
The equipment used in two-choice studies has evolved from
the simple three-key pigeon chamber to sophisticated electronic
equipment that can deliver stimuli consistently and record


behavior reliably.


Reinforcement devices have been modified from


pellet dispensers (Verhave, 1966) to bells, chimes, and buzzers








(Spradlin et al., 1973).


Other modifications have included sound-


attenuated chambers (Maki & Levin,


1972); white noise (Thomas,


1973); tape-recorded presentation (Spradlin & Dixon,


1976);


electronically operated microswitches sensitive to the touch;
computerized presentation of stimuli (Maki & Levin, 1972);


punched-tape readers (Sidman et al., 1982)


electro-mechanical,


electronic 20-pen event recorders (Sidman et al., 1982); digital
counters (Stromer & Osborne, 1982); carousel slide projectors
(Wetherby, Karlin, & Spradlin, 1983); and computer and monitors
that have incorporated joysticks, keyboards, modified keyboards,


paddles, and touchscreens (Rettig, 1987; Serna,


1987).


Two-choice matching-to-sample pattern organization
Changes in equipment provided the opportunity to alter


the organizational patterns of stimuli.


Choice stimuli could be


presented either above the response choices (see Figures 2-2, 2-3,
2-4, and 2-5), below the response choices (see Figures 2-4 and 2-


8), or within a matrix that totally surrounded the sample


stimulus (see Figure 2-7).


The opportunity for moving stimuli to


different positions also provided the means for additional areas
in which no response choices were given (see Figure 2-3, 2-4, and


2-7).


Finally, overall patterns within which stimuli were


organized changed from the linear three-key pigeon format to an
overall triangular (see Figures 2-2, 2-3, and 2-5). rectangular (see
Figure 2-4), and square format (see Figure 2-7).









Two-choice matchine-to-samole findings


Laboratory experiments (often with other numbers of


choices) with retarded,


nonreading adolescents resulted in the


development of performances that had not been specifically trained.


After matching pictures to words,


learners were able to read words


(Sidman,


1971)


. These findings of untrained performances


(stimulus equivalence) led to subsequent studies in which the
modality of stimulus presentation and ultimate learner responses
(Dixon & Spradlin, 1976; Joyce & Wolking. in press) were


addressed.


Spradlin and Dixon (1976) demonstrated the


effectiveness of visual to visual (intramodal) matching while Sidman
and Cresson (1973) demonstrated the effectiveness of auditory to
visual training. These findings were later substantiated by
Wetherby et al. (1983). Attempts have been made to investigate


intraspecies (Wright & Cumming,


differences.


1971) as well as interspecies


Researchers have sought to determine whether


nonhumans (monkeys and baboons) are truly able to achieve


stimulus equivalence (D'Amato, Salmon, Loukas, & Tomie,


Sidman et al.


1985).


(1982) have attempted to determine if differences exist


in animal and human performance through tests of transitivity,


symmetry,


and reflexivity.


It remains unclear as to whether the


emergence of untrained matching-to-sample tasks is limited to
humans or may be possible in other primates.






61


In other studies in which two-response choices were made
available to the learner, researchers have investigated reinforcement


schedules.


Both human and nonhuman learners have been exposed


to schedules of intermittent reinforcement (Skinner,


1950) and


schedules of variable, interval, and fixed-ratio reinforcement (Ferster,
1960; Nevin et al., 1963).
In some studies, findings regarding matching-to-sample
procedures have simply been noted, without manipulation of these


procedural variables within the context of the study.


Cumming


and Berryman (1961) reaffirmed Sidman and Stoddard's


(1967)


earlier findings that learners tend to develop position habits and
respond to the same key if sample stimulus presentation is not


randomized.


Nevin (1967) noted that response latency appeared


to decrease when subjects were given larger amounts of time to
observe the sample; Makin and Levin (1972) related response
latency to the length of the intertrial interval or the complexity of


the stimuli.


Rapid response and learning have been substantiated


by Cumming and Berryman (1961) and Sidman and Stoddard
(1966) who have described learning within a period of 5 minutes


with a severely retarded subject.


Cumming and Berryman (1961)


also noted that the large number of correct responses appeared
related to Harlow's learning set theory in which high accuracy is
correlated with many practice opportunities (Harlow, 1959).








Although several investigators (Semrna. 1987; Sidman, 1987)
have suggested that greater attention be paid to the manipulation
and subsequent effect of procedural variables such as intertrial
interval, feedback, reinforcement, and numbers of choices, only
one study was located in which a procedural variable related to


numbers of choices was manipulated.


Iverson et al. (1986) found


that learner accuracy was directly related to the consistent


position of the sample and therefore the choice stimuli.


Iverson et


al. (1986) have suggested that failure to specify stimulus location
may generate unwarranted conclusions regarding performance
and that findings by Cumming and Berryman (1965),Farthing


and Opuda (1974),


and Sidman et al. (1982) may have been


confounded by the position or number of choices in the matching-
to-sample procedure.
Three-Choice Matching-to-Sample Studies
Matching-to-sample studies in which three-response
choices have been incorporated into the procedures (see Table 2-2)


were based on


teaching machines developed by Skinner (1968)


who modified the earlier testing machines of Pressey (1926).
Subsequent interpretations of the teaching machine used
nonmechanical, mechanical, and electronic devices.
Three-choice matching-to-sample equipment
Teaching machines were devised to provide necessary and
precise arrangements of contingencies of reinforcement to bring
















































'.4
Ii
Cu



In
"4


~I

























0)
V

-4

I
0
V
'00)
O
-4
aco
V








4-,
0
V









0
V
~ V
to
C
4-a
4-A
041
Cu






a,


ii


02
V
U


4-I
C.)


t~4
0
.4.J







'-I
4)
Cu
Cu
as)
&-Dg)

hi
a
0411
Cu


10

I.'




I4-. .9
0

~ U


0)
-4
0

-4
UVQ)
-
- CI)
V
ci) ~


'0





Q0)cjoUs'I


Vs


o 00




4 C-







1.


i C
4)
$
to
4')
~hi~0
cn~~oE-4


131
to
i0#


hip


hi
4.)
hi



U us0'


90


IOU


I-
0



















0
4)

II

0
I)
'0
ii








0
4.)

3





V
~ V
I.)
I.
041
Cu




U)


ii


'0

>s o
4) '3 -4

4)31

^0 -'~
Silt!

c o f


~Ii
4)
4)
V '0
4)
-OWE)
4-'
I ~~bo2~'~ 4) 0

~C~4 U

Ct,


4)


CO



rrcon


016'


I.I


0
A
Cu
'0


I

'S
U
0
4)

'0


0
A




-



II
-S








believed that subtle discrimination could be learned and that such


behavior could be maintained in strength for long periods of time.


was hoped that this type of teaching might make it


possible to


translate the successes of the operant conditioning laboratory into
opportunities for any learner.
Skinner (1968) noted many advantages of the teaching


machine.


These advantages included immediate reinforcement,


individualized presentation and modifications, eventual errorless
learning, the incorporation of supplemental reinforcers, alternative
uses of teacher time, sequenced instruction, and automatic recording


of learner performance.


Skinner viewed young children as


disadvantaged if such tools were not present in their learning
environment.
Other investigators have used three-choices format (see


Figures 2-9, 2-10, and 2-11).


Lahey and McNees (1975) used


three choices in a traditional format, but with a slightly altered


teaching format (see Figure 2-9).


While some investigators have


used 3 by 5 inch cards and human recorders to teach letter


discrimination to preschoolers (Lahey & McNees,


1975), others


have used electronic equipment when working with both normal
and retarded children and adults (Lazar. Davis-Lang, & Sanchez,


1984; Sidman et al., 1985; Sidman,


Willson-Morris, & Kirk, 1986).


This equipment included carousel slide projectors, solid-state
computerized programming equipment, impulse counters, a








investigators could modify the format of the sample and choice
stimuli.
Three-choice matching-to-sample pattern organization

Teaching machines were designed in many instructional


formats, one of which was matching-to-sample.


A standard model


used three response choices below the sample stimulus (see Figure 2
8). Skinner (1968) used this format to teach part of a course in
human behavior to undergraduate college students.
Sidman et al. (1985, 1986) used an eight-circle arrangement in
which only the bottom three circles contained response options.
Lazar et al. (1984) used a six-circle arrangement in which any of the
top or bottom circles might have contained the three response


options.


Findings from studies by Sidman et al.


(1985, 1986) and


Lazar et al. (1984) supported and extended previous findings with
regard to the emergence of untrained learning and expansion of
stimulus classes.
Four-Choice Matching-to-Sample Studies
Saunders and Sherman (1986) have conducted the only
study reported in this review in which a four-choice procedure was


utilized (see Table 2-3).


Their instructional device was constructed


from a white plexiglas panel that was mounted on a plywood wall.
Five holes were cut in a rough circle for responses; stimuli were
projected onto four of the five openings from the rear and
illuminated with 60-watt light bulbs (see Figure 2-12).























0)
4)

-4

I
4-a
V
O
-4
~
0)







.6-h
0
4)



r~d


U)


'3




sjfi aoi
it-s s



a,S r! C o<

ii 8pi6

agigy S a
fl~g 8 85 0
V co0.g
lia'S6legg


$4]


Cu-


o ** -3
ASP tJ


0)
4-h
0



VvV
~I4WC4~


'C
0

Cu
V










F
V





t~4
0

.0
0
~0L)




to
4.)
a


U,
V
U-
V

(/2
am)



(12
a
0
.4-a
I
0
U-
.0
U


4;
U
U-
0
.0
C)
I
'-4
0
(Z4
'4-








sample stimulus was centered within the response option.


This


study was conducted with two male and one female mentally
retarded adolescents and intertrial interval length was identified as
a potentially important variable in matching-to-sample and choice


behavior.


Position bias,


wherein the learner appears to make


choices based upon the location of the response key, was noted and
reaffirmed the previous findings of Cumming and Berryman
(1961) and Sidman and Stoddard (1967).
Five-Choice Matching-to-Sample Studies
The three five-choice matching-to-sample studies located for
this review used electronic equipment similar to that used in previous


studies (Sidman et al.,


1982, 1985, 1986) and these


studies are


summarized in Table 2-4.


Tawney (1972) conducted a group study


with 13 males and 7 females of preschool age in which a more typical


triangular/linear formation was used (see Figure 2-13).


Sidman and


Tailby (1982) confirmed the emergence of multiple numbers of
stimulus relations without directly training a common sample or


choice.


A circular format was used in which only the lower five circles


(out of a possible eight) were used (see Figure 2-15).
Sidman et al. (1974) conducted two separate experiments.
each, subjects were required to see/touch a picture and then


see/touch its corresponding word.


One subject completed the


experiment with a five-choice condition (see Figure 2-14) and the


other with an eight-choice condition (see Figure 2-18).


Both



















0)
4)
.0

.4-,
4)
'00)
~ 4-'
-4
~0)






4.'
4)
a


0)
4)
U
0
4-,
'0)
04-'
4)
0
1.4J14
0-..

z


V
'0

'-I 04
4)






~ 0)


0)
1-4 02
0
p4 (0
U 4.)
-

'8
(U Co
(o~
$8
00
it!'Z I 4)


1*4
0
t
4)
hi
V


'.4
0
-' -
cup'
Q4(I) 0)
04
0)
s-i


-4


0friU


S~ 3

I ir e S
< g-li i
&Rss S"
aflfae0)


i-i

" SB' :
II S 'S^
s '" -.(0U


se a,


'0
A
002
~4)
QkC~)
%~.4~J I-,
0


U)
(0
'.&
II





















U)
C
a
I

'0W)
Cm







4,
C
a
0






4)
a
~ C

.4-A CU
.4-,
cU






02


I0 0
So
a -


lMC C0


4)
,0 4)
o
'-I
4) ti~

Ok



k0
~m4-Jo





Co

0~

kO


Q4~


SM;
0
0
Cu
'0




V
8
V
0

V








a:
09
is


I
V








subjects demonstrated mediated transfer (oral naming) to


additional equivalence classes.


Intramodal training (visual to


visual) yielded crossmodal (see to say) results.


In discussing their


findings,


Sidman et al.


(1974) never mentioned the possible effect


that the different number of choices might have had on results.
Six-Choice Matching-to-Sample Studies
Mackay (1985) has conducted the only matching-to-sample
study in this review that used a six-choice response choice format


(see Table 2-5).


Two severely retarded adolescent males were required


to see/touch colors and words as well as hear color names/touch


color words.


Electronic equipment was not used by Mackay.


Rectangular pieces of cardboard,


pieces of colored paper, and praise


in the form of tokens or verbal remarks comprised the materials used
in the study. Mackay (1985) supported the use of these materials as
being inexpensive, practical, readily available, and easily prepared by
retarded students. The cardboard rectangle was divided into 12
sections and the sample stimulus was presented in the top lefthand


section.


Six response choices appeared randomly in any of the


remaining 11 sections.


The emergence of equivalence relations as a


result of the visual to visual training was reported.
Eight-Choice Matching-to-Sample Studies
Five studies were located in which an eight-choice comparison


format was used by investigators.


One study (described above and


in Table 2-4) by Sidman et al. (1974) included a component in




























*1
bM
U
I


r.3B








which one of the two subjects used a five-choice condition and the


other subject an eight-choice condition.


Studies by Sidman


(1969, 1971), Sidman and Cresson (1973), and Sidman and Kirk
(1974) used electronic equipment, including a three-by-three
matrix response board that had been used in an earlier (non-


matching-to-sample) study (Sidman & Stoddard,


1966).


Modified


for use as a matching-to-sample tool, the matrix provided the
learner with the sample choice in the center and later surrounded


by eight choices (see Figure 2-17).


Sidman used this format in


three studies and for part of a third (Sidman & Cresson, 1973).
For some reason that was not explained, the format was changed
in mid-experiment to a circular arrangement (see Figure 2-18)
without, according to the investigators, affecting the results. It
can be hypothesized that a possible reason for changing the
format might have been the proclivity of learners to demonstrate
position habits as previously described with two-choice conditions


(Cumming & Berryman, 1961
(Saunders & Sherman. 1986).


) and four-choice conditions
In the original three-by-three


matrix, the corner choices were farther away from the sample
stimulus than the other choices and also required learner
movement on the diagonal, rather than in a horizontal or vertical


direction.


Placing the response choices in a circle with choices


equidistant from each other and the sample stimulus eliminated


the potential for bias based upon distance or direction alone.


The


















0)
U
oSfl
U-.



to
Jig
a




I-.
o
1)


-ii 3M
~Uo C)








N
~QtJM


a)



ji~ii
1*44)1


*0


0
-.4
0)
V
V*4)
IL




~cCO



'-I
V
4)2
041.4
'-S a ~a
0.0 0
t
cm)

et U


N






'5
1~v
p4


U)
























0)
V
.0



C
Oh?
*00)
4-.
I"4 -

Ii)









C
as)

p4







I.
C
V
~ 4)
to

.4a
Cu







0)


0)


IJ -



014.4 C
0 so
U :5 -


V0WI


Cs


1.4 0)
o .4-a
C -
p4
tOm
V GM)
.~ I'.4

'0
#4- 0



U



4.)


k
0
.4-i


p4-lv


0 M
N^-


c/nj,


CUD)


eq


V
A


~ 1.4
c~


































I


4J~
4.?
t
'Co
Of-


0
Ir


4 I f i i t


I-
0








0-9


S






78

types of learners and were primarily used from 1969-1974 to
either corroborate or expand upon the work on stimulus
equivalence and mediated transfer.
Implications for Control of Numbers of Choices
A search for matching-to-sample studies conducted from
1950 into 1987 resulted in 43 major studies in which this procedure


was used.


When the studies are considered in terms of the numbers


of response choices given to the learner, it is obvious that an
overwhelming number of investigations have incorporated the two-


choice condition (see Figure 2-19).


This dependence on the two-


choice format has remained relatively constant throughout the last


four decades.


Although consideration of number of elements has


been a concern of researchers in various fields (Brennan et al.,


Fantz & Fagan, 1975; Hershenson, 1964; Miller,


1966;


1968; Miranda &


Fantz, 1971) the optimum number of choices has remained unclear.
Sidman (1987) has recently stated a more specific concern
with regard to the numbers of choices used in the matching-to-


sample procedure.


This concern arose initially from careful


analysis of potential errors of interpretation when only two


choices are given.


Most of Sidman's early matching-to-sample


studies incorporated a matrix or circular pattern with eight
response options surrounding a center sample choice (see Figures


2-17 and 2-18) in symbolic matching tasks.


Findings of stimulus


equivalence (untrained performances) resulted in the







79




















2 CHOICES
2 CHOICES
4 CHOICES
5 CHOICES
6 CHOICES
8 CHOICES


1950-59


1960-69


1970-79


1980-87


DECADE


Figure 2-


Frequency of matching-to-sample studies.








development of testing procedures that were needed to validate the


unexplained learning.


With eight response choices, protesting


became a time-consuming and unwieldy process.


Therefore, many


of the subsequent studies, reverted to the earlier two-choice


matching patterns.


Questions regarding possible


misinterpretation due to the presence of false negatives (failure to
learn the original discrimination), false positives (selection based
simply on the fact that the stimulus is the "other" one), and
exclusion (selection as a by-product of excluding the alternate
choice) led Sidman to question seriously the validity of using only


two choices.


Sidman has stated that as the size of the stimulus


class increases, the opportunity for false positives decreases and


the opportunity for selection by choice increases.


This concern


has increased validity when consideration is given to the fact that
two-choice conditions in the matching-to-sample remain the most
frequently-used format
The speed and dependability of learning across situations may
also be related to the number of nonexamples that are provided when


making new discrimination.


If greater numbers of choices are


presented in the matching-to-sample procedure, greater numbers of
controlled nonexamples are possible and the learner may have a
better chance to develop discrimination that generalize and


maintain over time (Engelmann & Carnine,


Goldiamond,


1982; Moore &


1964).






81



Summary
As Sidman (1987) has stated, attention to sound methodology
provides the foundation for valid research and effective application.
Accurate definitions of stimuli may reduce confounded conclusions


regarding learning.


This study, therefore, focused on one procedural


variable, the number of response choices, and explored whether
differences existed in the matching-to-sample performances
generated under three- and six-choice conditions with preschool


learners.


As no previous experimental work in this area existed, the


results of this investigation may have great significance, especially
with the anticipated increased application of the matching-to-sample
procedure via the microcomputer.















CHAPTER III
METHOD
A study of the effects of the number of choices on
performance in matching-to-sample instruction was implemented.
Picture-word matching acquisition skills were the focus of the
study to assist in subsequent development of instructional


materials and strategies that are effective across learners.


The


general question was as follows: Are there differences in learner
performance in a three-choice as compared to a six-choice


condition?


The following specific questions were addressed:


1. Does the number of response alternatives affect learning
accuracy (percentages of correct responses) during the acquisition of
stimulus control?
2. Does the number of response alternatives affect speed of
responding during the acquisition of stimulus control?
3. Does the number of response alternatives affect celebration
(rate of learning or improvement) during the acquisition of stimulus
control?
4. Does the number of response alternatives affect response
choice position during the acquisition of stimulus control?










Applied behavior analysis procedures were used according to


the research method described in this chapter.


setting,


The subjects,


variable under investigation, measurement procedures,


experimental design, experimental procedures, materials, data
collection, and data analysis are described.


Prior to subject selection,


permission to teach the subjects


prereading skills was obtained from the Baby Gator Research
Review Committee and the University of Florida Institutional Review
Board (Committee for the Protection of Human Subjects). Parental
permission for the subjects recommended by their preschool


teachers was obtained


, via letters of consent, prior to protesting


(See Appendix A).


Subject DescriDtion


3-1).


Six preschool subjects participated in the study (see Table
All subjects were enrolled in the Baby Gator Preschool


Program on the University of Florida campus.


Preschool-age


subjects were selected for participation in this study because
manipulations of the independent variables required that subjects


be nonreaders.


Further, nonhandicapped preschool-age children


demonstrate mental ages that compare with those of older
handicapped learners.
Subject I was a 4-year-old male Caucasian whose home


language was English.


He had attended preschool for 11.5
























4-h e


4)
I..'
0



0
0
Cs)
Co
a)
&





CL






I)
to
Cu
to


r









Subject 2 was a 4-year, 5-month-old Caucasian male,


whose home language was also English.


attended preschool for


age of 4 years,


This subject had


11 months and demonstrated a mental


1 month.


Subject 3 was a 4-year, 3-month-old Caucasian female who
spoke English in her home and had attended preschool for 15.5


months.


Her mental age was 4 years,


1 month.


Subject 4 was a 4-year, 5-month-old bilingual Hispanic male
whose home language was Spanish. He had attended preschool for
12 months and had a mental age of 3 years, 11 months.
Subject 5 was a 4-year, 3-month-old bilingual Asian female


whose home language was Korean.


She had attended preschool for


12.5 months and demonstrated a mental age of 4 years, 5 months.
Subject 6 was a 4-year, 9-month-old bilingual Asian female who


spoke Chinese in her home.


She had attended preschool for 13


months and had a mental age of 4 years.

Setting


The study was conducted in the Baby Gator Research
Center for Child Development of the University of Florida in


Alachua County.


The Baby Gator Research Center for Child


Development provided preschool activities for approximately


105 children.


This setting was familiar to the subjects as they


spent a portion of their normal day in the same environment.










All sessions took place in a quiet conference room (8' x 10') that
contained a large conference table (3' x 7') and several student


chairs.


Equipment was set up in a consistent manner for each


session with the subject and investigator sitting side-by-side at
the end of the conference table facing the monitor and


touchscreen.


A videotape camera was positioned as


unobtrusively as possible, behind the subject, during two


sessions.


Each session occurred during the regular preschool


day and consisted of individual subject interaction with a
computer programmed for matching-to-sample instruction.
Variables Under Investigation

Independent Variable
The number of matching/response alternatives was the
independent variable in this investigation.
Matching-to-sample procedures begin with the presentation of a
sample stimulus to the subject, followed by the opportunity to select


one of several comparison words.


In this study, a sample stimulus


in picture form was presented to the subject within a rectangle


located in the center of the computer screen.


When the subject


touched the stimulus rectangle, comparison words appeared in six
rectangles on the rim of the computer screen in one of two


experimental conditions: three choices or six choices.


Subject


performances were compared under each condition (three- or six-










Dependent Variables
The dependent variables in this investigation were the
accuracy and speed (response latencies) of the subjects'
performance, rate of learning or improvement over time


celebration) ,


and the position of response choices.


Accuracy,


speed, and response position data are described in percentages.
These measures permit quantitative evaluation of the


experimental questions.


Measurement of both the independent


and dependent variables is explained further in the following
sections.


Measurement


Independent Variable
Control of the independent variable in the form of
procedural reliability was maintained through the use of the


computer program.


The sequence of presentation of either the


three- or six-choice alternative conditions was determined by a


computer-directed randomization procedure.


The sequence,


timing, and reinforcement procedures were replicated exactly in


each of the 40 trials per session.


A pilot program with one


subject was conducted to demonstrate the consistency and
reliability of both the computer equipment and the software
program, as well as the appropriateness of the task for the
proposed subjects.












Dependent Variables
All dependent measures in this study were recorded


throughout all instructional sessions.


The computer program


recorded correct and error responses, response position, and


response latencies for each condition.


Total response time for


each


sess


ion of 20 trials was recorded by the investigator;


corrected performance time was calculated by a separate


computer program.


Corrected performance time reflected the


removal of reinforcement time for correct answers and
transition times between the end of a trial and the beginning of
presentation of the next picture sample.
Response accuracy, the matching of the written word


(comparison) to its corresponding picture (sample),


was


measured for each condition. The number of correct and incorrect
choices across 20 trials for each condition was recorded and
represented as percentages of correct responses.
Speed of responding was measured in terms of response


latencies across 20 trials.


These latencies were ranked and


median performances for each condition were determined.
Rates of learning over time were calculated by celebrations.
Celeration is a standard measure of precision teaching in which
the frequencies of correct or error responses for a week are