REFOR T RESUMESED 010 137 24THE EFFECTS OF COUNSELOR REINFORCEMENT ON A STUDENT'SNEGATIVE RESPONSES.BY- SNELLGROVE, CHARLES E., JR.FLORIDA STATE UNIV., TALLAHASSEEREPORT NUMBER BR-6-8021 PUB DATECONTRACT OEC-2-6-06802?-0927EDRS PRICE MF-$0.09 HC-$2.24 56P.

SEP 66

DESCRIPTORS- *REINFORCEMENT, COUNSELING, GUIDANCE COUNSELING,COUNSELING SERVICES, *INDIVIDUAL COUNSELING, *COUNSELINGPROGRAMS, STUDENT PARTICIPATION, *STUDENT BEHAVIOR, RESEARCHPROJECTS, EXPERIMENTAL PROGRAMS, SIMULATION, *SELF CONCEPT,TALLAHASSEE, FLORIDA

THE SIMULATION OF A DESIRABLE COUNSELING PROCESS WITHEMPHASIS ON SYSTEMATIC COUNSELOR REINFORCEMENT OF NEGATIVESELF-REFERENCE STATEMENTS BY STUDENTS WAS REPORTED. THEOBJECTIVE WAS TO DEMONSTRATE THAT DEFINITE MEASURABLECOUNSELING PROCEDURES COULD BE EXPERIMENTALLY TESTED AND HAVEPRACTICAL CONSEQUENCES. THE VERBAL OPERANT CONDITIONINGPARADIGM WAS EMPLOYED IN AN ATTEMPT TO ANSWER SEVERALQUESTIONS--(I) CAN THE FREQUENCY OF NEGATIVE SELF-REFERENCESTATEMENTS BE MODIFIED BY VERBAL REINFORCEMENT IN AQUASI-COUNSELING SITUATION, (2) WILL SCHEDULES OF -

REINFORCEMENT PRODUCE DIFFERENT EFFECTS, AND (3) WOULDSUBJECTS RESPOND DIFFERENTLY TO THE SAME SCHEDULE OFREINFORCEMENT. A FUNCTIONAL RESEARCH DESIGN PERMITTED A FOCUSON INDIVIDUAL BEHAVIOR OF SUBJECTS. RESULTS AFTER 3 MONTHSINDICATED THAT REFLECTION (PARAPHRASING) WAS NOT AREINFORCING STIMULUS WHEN MADE CONTINGENT UPON A NEGATIVESELF-REFERENCE. IT WAS SHOWN THAT COUNSELOR REFLECTION OFNEGATIVE STATEMENTS HAD NO UNDESIRABLE EFFECT AND SEEMED TOCONTRIBUTE TO A DECLINE OF SUCH STATEMENTS. (0C)

1`,....

Tr'tII S. DEPARTMENT OF HEALTH, EDUCATION AND WELFA;--4'

Office of EducationCZ) This document has been reproduced exactly as received from fivperson or organ,zation ongrr,ating it. Points of view or opini;m3r --1

C.,stated do not necessarily represent official Office of Educabonposition or policy.

C":2

1.0

FINAL REPORT

Project No. 6-8027Contract No. OEC-2-6-066 )27-0927

THE EFFECTS

OF COUNSELOR REINFORCEMENT

ON A STUDENT'S NEGATIVE RESPONSES

September 1966

U. S. DEPARTMENT OF

HEALTH, EDUCATION, AND WELFARE

Office of EducationBureau of Research

THE EFFECTS OF COUNSELOR REINFORCEWEATON A STUDENT'S NEGATIVE RESPONSES

Project No. 6-8027Contract No. OEC-2-6-068027-0927

Charles E. Snellgrove, Jr.

September 1966

The research reported herein was performed pursuant to acontract with the Office of Education, U. S, DepartmentBrTggrfh, Education, and Ilelfare. Contractors undertakingsuch projects Lnder Government sponsorship are encouragedto express freely their professional judgment in the conductof the project. Points of view or opinions stated do nottherefore, necessarily represent official Office ofEducation position or policy.

Florida State University

Tallahassee, Florida

CONTENTS

PageA. 4.

ACKNOdLEDGEWENTS OOOOOOOOOOOOOO iv

INTRODUCTION. OOOOOOOOOO OOOOOOOO 1

ProblemTheoretical FoundationsRelated ResearchObjectives

l.IETHOD 7

SubjectsDesignData and InstrumentationAnalysis

RESULTSOOOOOOOOO . . . . . . OOOOOOOODISCUSSION. . O ............ 28

Operant PeriodReinforcement PeriodExtinctionGeneral Observations

CONCLUSIONS AND II4PLICATIONS 37

SUMARY 39

REFERENCES. . ............. . . . . . 41

GLOSSARY. . . . . 0 . .............. . 47

FIGURES

1-8 Total number of responses per experimentalsession for all subjects, one through eightrespectively.

ii

13

FIGURES

Page

9-11 Relative position of response rate for all

subjects during the last sessions of each

experimental condition. Grouping is by

sequence of conditions.

12 Wean number of responses per experimental

session for all subjects. Grouping is by

sequence of experimental conditions.

22

26

13 Comparison of all subjects for the operant,

first CRF, and first extinction periods. 27

iii

ACKNOWLEDGEMENTS110

The author wishes to express his appreciation toDr. Harold F. Cottingham and Dr. Willard H. Nelson fortheir assistance in formulating the original researchplan and for their continued advice.

A note of thanks is also extended to ',tendon Hentonfor his assistance in preparing the electrical apparatusemployed in this research.

To my wife, goes a special debt of gratitudefor her many contributions in both time and effort.

iv

. INTRODUCTION

Oroblem

The problem was the specification and measurementof certain aspects of the counseling relationship. Theprimary concern was an attempt to determine which of severaltypes of action by the counselor produced expected(undesirable) behavior in the client. 3y manipulatingthe counselor's behavior in a controlled situation thenet effect on the client was determined.

Specifically, the client behavior investigatedwas the verbal response class of negative self-referencestatements. The independent variable was positive rein-forcement of negative client self-references, chosen tosimulate adventitious reinforcement observed to occur inthe counseling process.

by investigating the tendency of some counselorsto reinforce negative client statements, the possibledeleterious effects of such methods are established.The problem is highly significant for counselors at alleducational levels since the application of proven tech-niques or the avoidance of harmful procedures can beof value to those seeking to improve their methods.This study was an investigation of a negative variationof one possible technique a counselor may use.

Theoretical Foundations

Traditional counseling methods have been basedon the idea that the presenting problem was merely a symptomof some underlying disorder. Treatment focused on dis-covering the "cause" of the symptoms. This is essentiallyanalogous to physical medicine and is appropriately calledthe medical or disease model. This conception has ledto the postulation of numerous psychic entities whichsupposedly determine overt behavior.

The alternative to this approach is the psycholo-gical or behavioral model (Ullmann and Krasner, 1965).waladaptive behavior is believed to be learned and

I

therefore, subject to modification via learning principles.This eschews the idea of psychic determinants. :that

brings about the need for counseling is the learning ofinappropriate behaviors. This approach has come about forprimarily two reasons: first, as a reaction to the mentalismof the medical model; and second, as a form of counseling-therapy to replace it.

A number of criticisms have been leveled at themedical model. Probably one of the more well known isthat of Eysenck (1961) who says, "all methods of psycho-therapy fail to improve on the recovery rate obtainedthrough ordinary life experiences and non-specifictreatment." Krasner (1964) criticizes the uncertaintyand inefficiency of the psychotherapy "log jam." Counselinghas also bean criticized because of its close relationshipto psychotherapy, Strong (1964) suggests that becausethe counseling interview is essentially "reciprocal verbalbehavior between two people," perhaps attention shouldbe focused on verbal behavior instead of traditionaldynamic factors.

These criticisms point up the need for accuratedescription and measurement of the counseling relationship.Butler (1953) states that "the behavior territory ofcounseling must be mapped as it is done in the bestpsychological investigations." He further points out theneed to remedy the "basic lack of an organized set ofobservations of what happens during psychotherapeuticsessions."

Numerous writers have recently begun to explorethe possibility of using a behavioral approach in helpingpeople to deal with their problems. Grammer and Shostrom(196J) discuss its implications for counseling concludingthat it has promise for the future. Pichael and Meyerson(1962) discuss "A Behavioral Approach to Counseling andGuidance" in a special issue of the Harvard EducationalReview devoted to an examination of guidance. Neitz(1961) stresses its importance, suggesting a redirectionof behavior into more productive and satisfying channelsas a basic guidance function. Strong (1964) reviews theapplication of verbal conditioning techniques to counseling.Patterson (1963) says "The counselor is attempting toinfluence the behavior of the client--this is obvious;otherwise he wouldn't be engaged in counseling." Shoben(1963) identifies psychotherapy as an attempt "to direct ,

changes in human behavior" emphasizing the therapist's roleas a behavior model. If guidance is to take full advantageof the behavioristic model, it must have a closer

2

relationship with the behavioral sciences (Shoben, 1961).Rogers (Rogers and Skinner, 1956) agrees that "in client-centered therapy, we are deeply engaged in the predictionand influencing of behavior, or even the control of behavior."He later notes (Rogers, 1961) that in counseling, conditionsare established by external control which are followedby internal control by the individual. Krumboltz (1964)contrasts traditional and behavioral approaches to counsel-ing making a strong case for the latter.

The foregoing literature makes explicit the needfor an investigation of the counseling relationship interms of behavior modification. Since the most prominentovert behavior in a counseling session is verbal, thisarea of investigation has received the greatest attention.Skinner (195 ?) proposes that the same principles whichapply to animal behavior might also apply to verbal behavior.That is, the verbal behavior might be divided into responseclasses and studied in a manner similar to that of animalresponses. This has been demonstrated by Hildum andBrown (1956), Spivak and Parajohn (1957), Hartman (1955),Daily (1953), Cohen, Kalish, Thurston and Cohen (1955).Krumtboltz and Thoresen (1964) and Ryan and Krumboltz(1964) have reported successful investigations usingreinforcement techniques in a counseling situation tomodify client decision-making and information-seekingbehavior.

Related Research

In recent years, there has been increasing appli-cation of conditioning procedures to behavior disorders(Grossberg, 1964). One such technique is verbal conditioning.

Studies of verbal conditioning began with thework of Greenspoon (1951, 1954, 1955) in the early 1950's.A great deal of research followed as shown in reviewsby Krasner (1958), Salzinger (1959) and Greenspoon (1962).Krasner (1965) discusses the clinical implications of

verbal conditioning and its relationship to traditionalpsychotherapy. A number of studies have investigatedthe various verbal components of the therapy-counselingsituation.

Recent investigators have been using schedulesof reinforcement in the modification of verbal behavior.Kanfer (1954) found frequency of reinforcement to be acritical variable in the acquisition and extinction of averbal response class. McNair (1957), Salley and Santos (1958),

3

and Kanfer (1958) report successful investigations usingschedules of reinforcement.

Verplanck (1955) found that statements of opinionsincreased as a consequence of verbal reinforcement.Verbal agreement responses were used to modify attitudesin a survey by Hildum and Grown (1956). Insko (1965)also modified attitudes through verbal reinforcement.Waskow (1962) examined the extent to which selectiveresponding by a therapist would function as a reinforcerof content or feeling aspects of a subject's communication.Findings indicate some effect on content aspects. Fol-lowing Uerplanck's conversational procedures, Centers (1963)obtained an increase in opinion and information statements.In a semi-structured counseling session, decision-makingbehavior was successfully reinforced and it was found'that this behavior generalized to a non-counseling situa-tion (Ryan and Krumboltz, 1964).

The effect of reinforcement on negative responseshas also been the subject of investigation. Southwell(1962) conditioned hostile verbs in neurotics and normals.His results indicate that conditioning occurred as afunction of experimenter reinforcement and that verbalreinforcement was generally more effective than non-verbal reinforcement. Meyer, Swanson, and Kauchack (1964)obtained an increase in hostile verbalizations at gradelevels 4 and 6 using verbal approval as a reinforcer.

One of the more relevant response classes to beinvestigated is that of self-reference statements. (Many

counselors feel that the verbal aspects of the intervieware mainly concerned with the client's discussion ofhimself and that succ3ssful counseling involves a changein attitude as expressed by an increase in positive self-reference statements (Raimy, 1948). To a large extent,thin position is maintained by "client-centered" counselors.

Salzinger and Pisoni (1958) examined the effectof verbal agreement on the output of self-referred affectstatements during a clinical interview with schizophrenics.It was found that self - referred affect statements increasedas a consequence of verbal reinforcement and that dif-ferences in interviewers or sources of reinforcement hadlittle or no effect. A similar study was carried out bySalzinger, Portnoy, and Feldman (1964) in which self-referred affect statements were again conditioned inschizophrenics. Verbal assent was found to be an effec-tive reinforcer for self-reference statements in a studyof Adams and Hoffman (1960). Babbitt (1962), in an

4

attempt to produce attitudinal changes toward the selfby means of verbal conditioning, found conditioning effectswith one .group reinforced for positive self-referencesusing paraphrasing and verbal agreement as reinforcers.Nerbaum (1963) examined the effectiveness of three classesof verbal reinforcers; non-commital expressions, mildpositive expressions, and reflection of feelings. Ina series of clinical interviews he found reflection offeelings to-be significantly more effective in conditioningpositive and negative self-references. A later studyby Werbaum and Southwell (1965) showed paraphrasing ofcontent to. have a stronger reinforcing effect than merelyechoing the subject's affective self-reference. Anexperimental investigation of the effects of four psycho-therapy techniques was conducted by Adams and Frye (1964).It was found that self-reference statements were increasedby interpretive statements and minimal social reinforce-ment. Reflection and hostile statements decreasedpersonal references. Results indicate that differentphenomena of verbal behavior observed by various schoolsof psychotherapy, may be a function of different techni-ques employed. Rogers (1962) produced a significantincrease in negative self-reference statements by reinforc-ing with simple verbal agreement. Negative self-referencestatements were also reinforced by Gonzer and Sarason(1964) using several verbal agreement cues at random.They obtained a slight increase in the emission of suchresponses. Powell (1965) found reflection to be aneffective reinforcer for negative self-references.

Although a wide variety of response classes havebeen successfully manipulated, there have been occassionalfailures to achieve conditioning. Krasner's (1958)review of verbal conditioning ident).fies 12 studieswhere the reinforced behavior either did not increasesignificantly or was no more than in a control group.Kanfer (1962) suggests that such instability may be dueto a variety of concurrent controlling stimuli whichreduce the relationship between the reinforcer and theresponse class. Lubin (1965) takes the position thatverbal conditioning is a more elusive phenomena than wasfirst supposed. In his investigation of differences inconditioning effects, he concluded that failure to conditionoccurs through a failure to optimize the parameters ofthe reinforcement situation.

Friedman (1964) was unable to verbally conditionpositive attitudes to a questionaire of color preferencefollowing avoidance conditioning of neutral color stimuli.Cole (1965) reinforced negative statements in a counseling

5

analogue. The group reinforced for negative statementsabout a neutral concept conditioned while the groupreinforced for negative staterr,ents about the self-conceptactually showed a decrement in response rate. Ullmann,Krasner, and Knowles (1965) found that favorable attitudestoward medicine could be enhanced by verbal reinforcementbut under similar reinforcement conditions unfavorableattitudes were unaffected. Cahoon and denrich (1965)suggest that verbal conditioning is our most powerfulresearch and conceptual tool for a better understandingof complex interview interactions. They make a plea forfurther research, present controversies notwithstanding.

The Ince (1965) study, on which the present researchis based, is the only investigation to date employing afunctional research design. All of the studies reportedin the literature have employed a factorial or actuarialdesign. The functional design has the advantage ofpermitting a study of changes in the verbal behavior ofthe subject as a consequence of systematic reinforcementand the measurement of such changes over time. Thisresearch is a logical extension of a former study and anattempt to relate behavioral research to the counselingsituation.

Objectives

The objective was to demonstrate the definitemeasurable counseling procedures could be experimentallytested and have practical. consequences. Of primaryconcern was the extent to which certain types of counselorbehavior might produce undesirable behavior in the client.Specifically, an attempt was made to simulate a desirablecounseling process with the exception of the systematicreinforcement of negative self-reference statements.

As with most operant conditioning studies therewere no specific hypotheses to be tested. Instead, severalbasic questions were asked:

1. Is positive verbal reinforcement powerfulenough to modify the verbal behavior (negativeself-reference statements) of individualsin an interview setting, simulating acounseling relationship?

2. To what extent will the different reinforce-ment schedules produce different effectson the verbal behavior of the subjects?

6

3. To what extent will there be differencesin responding between subjects on thesame schedule of reinforcement?

METHOD

Subjects

Eight undergraduates at Florida State Universitywere selected for participation in the experiment. Theseincluded one male and seven females. Subjects werevolunteers who were aware of the time involved and agreedto daily attendance until completion of the experiment.A stipend was given for participation at the rate of$1.00 per session. Subjects' age range was 19 to 21 yearsrepresenting all college levels except the freshman.Selection was on the basis of interest and availability.These people were selected as subjects because it washoped that the results would be applicable to a collegepopulation.

Each subject was seen individually for one-quarterhour five days a week until the experiment was completed.This time period seemed feasible since previous research(Ince, 1965) indicates that the highest frequency ofresponse occurs during the first 15 minutes of each ses-sion. In addition, a pilot study has shown that one-quarter hour sessions are sufficient to demonstrate theexperimental effects and yet short enough to avoid suchconfounding variables as boredom, monotony, and awarenessof the artificiality of being a volunteer counselee.

Subjects were told that a study was being made tofind out how college student's views change from day today as a result of being in a college setting. They weretold that it was the day to day changes that were important.These instructions provided the subjects with a rationalefor talking and daily attendance. Questions about thenature of the study were answered by saying that theresults would be useful in learning more about counseling.Each experimental session was a free-operant situationin which students could talk about whatever they chosewithin the limits of the general instructions. Allsubjects received essentially the same instructions.

Subjects were aware that their responses werebeing recorded as a means of collecting data and that suchrecordings were reviewed only by the investigator. Theywere told that each interview was considered confidential

7

t

to insure the integrity of volunteered informations. Itwas stated that general information about the interviewscollectively would be made known but not specific informa-tion supplied by each participant, In addition, assurancewas given that there would be no invasion of privacy andthat interviewees would remain anonymous.

Design

For each subject an initial base rate of respondingwas obtained. During this time no reinforcement wasgiven for anything the subject said. Silence by theexperimenter was explained by saying that in the past ithad been found best when the student did all of the talkingat first. This operant period lasted until a stable rateof responding had occurred. While there is no rule tofollow in selecting a stability criterion, Sidman (1961)suggests that the criterion be based on three things;the phenomenon being investigated, the level of experimentalcontrol that can be maintained, and long term study of"steady state" behavior. With these considerations inmind, stability was judged to have occurred when subjects'responses were within a range of five responses over a

four day period. As in all operant studies, each indivi-dual was his own control.

Following the operant period, the method ofsuccessive approximation was used to bring the subjectsto the point where schedules of reinforcement might be

introduced. First, continuous reinforcement (CRF) wasused for all statements beginning with "I"; then forstatements beginning with "I" and containing negation(no, none, not, nothing, etc.). This was the final

--response class to be investigated--negative self-referencestatements. When a stable rate of responding had beenreached on CRF, two subjects were placed on a fixed ratio-2schedule of reinforcement. When the response rate beganto decline they were again placed on CRF in an attemptto recovef the higher rate. This was unsuccassful andboth subjects were placed on extinction during which timeno- "`reinforcement was given. The remaining five subjectstended to reach a level of response stability which wasessentially the same as the stability level of the operantperiod. In order to verify the apparent ineffectiveness

the reinforcer, an extinction period was begun. Withthree of these subjects, a second attempt was made toreinforce (CRF) the response class. This was followedby a second extinction period. One of the original

8

eight subjects withdrew from school during the first CRFperiod of successive approximation.

The reinforcement was reflection (paraphrasing)which has been shown to be reinforcing by Verplanck (1950,fierbaum (1963), Adams and Frye (1964), Werbaum andSouthwell (1965), and Powell (1965). At times when itwould have been necessary to interrupt by breaking into'subject's speech in order to giye reinforcement, simpleagreement cues (uh..huh, good, head nod, etc.) were used.These have been shown to be reinforcing by Rogers (1969),Southwell (1962), Portnoy and Salzinger (1964), Gonzer andSarason (1964), and Ince (1965). There are times when itwas clearly inappropriate to interrupt subject's speechin order to paraphrase. The pilot study showed this tobe awkward and detrimental.

Data and Instrumentation

The experimental room contained a desk and twochairs. A microswitch (80511-CE, Electrical Specialties)mounted on a wooden board was located under the desk.The experimenter operated this microswitch by foot. Itwas connected to a 6-digit electric counter (99 R 9011Lafayette) and a midget stepping relay (WER-115 ACGuardian). Each foot tap by the experimenter registeredone digit on the counter and at the same time advancedthe stepper on the relay. The counter was housed in asound insulated wooden box. The relay was wired to aterminal strip and both fastened to an aluminum chassis.This was mounted inside a steel box insulated with acousticfiberglass.

The stepping relay was wired to a small red lightbulb so that a given number of taps on the microswitchturned on the light for about 1 second. This was locatedso that it was visible only to the experimenter. Whenusing a fixed ratio schedule of reinforcement. the lightacted as a signal to the experimenter that the requirednumber of responses had occurred. The light actuallyflashed one response before the required number so thatthe experimenter could reinforce immediately upon the nextemission of the critical response. A tape programmer(LVE 1319FC) was also wired to the light bulb in orderto program the desired fixed interval and variable intervalschedules.

A tape recorder CT-1500 Wollensak) was used torecord the subjects verbalizations and the experimenter'sreinforcements. This acted as a check on the counter.

9

To all appearances, the experimental room wasas any other, counselor's office. The foot switch andsignal light were not visible from the subject's chair.The tape programmer, relay, and counter were located inan adjoining room. The tape recorder was located in acentral recorder bank which served all of the counselingoffices.

Analysis

Analysis consisted of identification and evaluationof modifications in the frequency of the response classas a consequence of programmed reinforcement. A simplefrequency count was made of the emission of the criticalresponse class for each subject. The total frequencyfor each session was plotted graphically. This procedureshowed the total number of critical responses per subject,per day. By comparing each session with the preceedingsession, the effects of reinforcement were determined.Intrasubject and intersubject comparisons were made.

This method of identifying behavior change hascome to be regarded as standard procedure in operantconditioning research (Skinner, 1953; Ferster and Skinner,1957; Sidman, 1961).

RESULTS

Figures 1 through 8 show the total numbar. ofresponses made by each subject (S) during each experimentalsession. This method of organizing data illustrated thechange in relative rate of response for each experimentalcondition. With the exception of Ss 2 and 8, all sub-jects showed a decrease in response rate over the courseof the experiment. This points up the fact that condition-ing did not occur. The fact that the response rate con-tinued to decline during periods of continuous reinforcement(CRF) indicated that the subjects' verbal behavior wasnot under the control of the reinforcing stimulus.

Subject 2 showed some increase in response rateduring the shaping period (CRF) and the fixed ratio (FR)reinforcement period. The FR period, however, appearedto be unstable and when the response rate began to decline,CRF was again employed. The response rate stabilized atabout the same level as that of the first CRF shaping

period. The increase for this subject was so slight that

conditioning could hardly be said to have occurred.

10

Figure 8 was the only subject to show a consistentincrease in response rate over a period of 22 experimentalsessions. The data suggested that conditioning mighthave occurred during the CRF shaping period had thissubject continued to participate.

The sequence of experimental conditions waslargely a function of the extent to which each subjectshowed an increase in total responses per session as aconsequence of programmed reinforcement. As a result,three sequences were employed after the first period ofcontinuous reinforcement (CRF).

Subjects on the first sequence (Fig. 1 & 2)showed a slight increase in total response during the CRFperiod. A fixed ratio schedule of reinforcement was begunwhere reinforcement was delivered for every 2 responses.The response rate showed a gradual decline and continuousreinforcement was again employed. The response raterecovered slightly, then began to decline once again.In Fig. 1, stability occurred at a level approximately50/* lower than that of the operant period. Figure 2shows less variability. Response rate over the entireexperiment appeared to be relatively stable. In both casesthe introduction of reinforcement effected only a slight,temporary increase in response rate. During extinctionthe rate of response increased for subject 1 (Fig. 1)and remained about the same for subject 2 (Fig. 2).

Figures 3, 4, and 5 illustrate the second sequenceof experimental conditions. Response stability on con-tinuous reinforcement occurred at about the same levelas during the operant period. At that point, reinforce-ment was withdrawn. This extinction period continueduntil the response rate again stabilized. The level ofstability was then below that of the two preceeding periods.Reinforcement was again introduced on a continuous schedule.Stability occurred rather rapidly with no change in responserate. Final extinction resulted in no change for 2 sub- -,

jects (Fig. 3 & 4) and a slight decrease for the remainingsubject (Fig. 5).

The final sequence (Fig. 6 & 7) was similar tothe preceeding one except that the experiment was terminatedafter the first extinction period. Like the three pre-ceeding subjects, the response rate stabilized at orbelow the level of the operant period. The reinforcementwas then withdrawn. Stability occurred below the rateof the CRF period. Time did not permit a second attemptto reinforce the critical response class.

11

(

/

rig. 1-8.

Totpl number of responses

per experimental

session

for all subjects, one

1-a

13through eight respectively.

I

RESPONSES t I 1

NI 5 o W )4 0 1

a 1

0

60 50

10

OP

ER

AN

TI

CR

FF

R2

IC

RF

IE

XT

510

1520

2530

3540

4550

SE

SS

ION

S

Fig

. 2

60 50

c

10

RA

IIL._

10I

CR

F__

LE'X

itrz.

B F

1 E

XT

____

__15

2025

3035

4045

SE

SS

ION

S

50

cn

60 50 40

L.L1 U) Z

30

U)0 1.LJ

20 I0

RF

1015

2025

SESS

ION

S

XT

CR

PI

EX

T30

3540

4550

60 50 40 10C

°

OPE

RA

NIL

CE

E L

____

_EX

L_

LC

RE

LE

XT

_5

1015

2025

3035

4045

50

SESS

ION

S

9'f

)T .1

SNO

ISSES

0(.1gi,

OV

c r4 ../09.

9303

9101

DC

-31

...]Y.3-1

INV

d-i-Id0Is.

.1.111.1.0.0e....

'''''......... .... . s ...*. ...........

.. .......m.w

.,.......e.......m..

..............,.......p....,..InlIIIIY.0

11

01

09

109

60 50 40

Z 3

0R cc

20 10

a

..fo.

..10

. .,

OPE

RA

NT

1G

RE

IF

>"1

-10

1520

2530

35

SESS

ION

S

aN

I44

it

40b0

RESPONSES

O N

0 0 Gn 0

o0 -o rn

0z

r--

- ' Cf) Fl

(.1)

ci4 0

A

The subject whose responses are shown in Figure 8would have been placed on a variable interval (VI)schedule of reinforcement had she continued to participate.This was the only subject to show an increase in responserate as a consequence of reinforcement.

These three sequences of experimental conditionsare illustrated in Fig. 9, 11, & 11. The data points ineach experimental period represent the total responsesduring the last four sessions. In most cases these werethe four response stability criteria which marked the endof a period. These figures tend to show the relativeposition of the response rate for each experimental condition.

Intersubject comparison showed that while thegeneral pattern of responding was similar for subjectsin each group, there was considerable individual varia-bility. In Fig. 9, 51 showed a wide range of variationwhile 52 remained relatively stable.

The Fig. 13 sequence stowed 53 to be relativelystable while Ss 4 and c' showed pronounced decreases instability levels. This pattern was also found for Ss 6and 7 of Fig. 11. The conditioning effect was clearlyillustrated for 58.

In terms of the level at which the response ratestabilized, there appeared to be two distinct areas. Formost subjects the response rate was at or about the samelevel at the end of the operant and CRF shaping periods.In subsequent periods, stability occurred at a much lowerlevel. The exceptions were 52 (Fig. 9), S3 (Fig. lj),and 58 (Fig. 11). This further indicated a generaldecline in total responses as the experiment progressed.Regardless of the sequence of experimental conditions,reinforcement did not bring about a sustained increasein response rate.

Figure 12 shows the mean number of responses persubject for each experimental condition. The threegroupings are by sequence of experimental conditions asin Fig. 9-11. The advantage of graphing means was tofacillitate a comparison of intersubject variability.It can be seen that Ss 3, 4, and 5 generated similarpatterns of mean responses under the same experimentalconditions. This was also true of Ss 3-7. Subjects 2 and 8were atypical; S2 being relatively unaffected by the changeof experimental conditions and 58 showing evidence of aconditioning effect.

21

VSO

4,0.S.* .. ... aVI+

1..a... 11 ,/ ...... V

II 41,0. NW

...VV

..0..., N

.G.., .

.ue,%

.0.0.....raw

er a...(),°*...""

.)

CI)

0.0111

lo

(,)

cjO

0.

F W.I.

M. .444 . Y

ea.

.4.(I)

OM

.- .

YM

VIM

,co.

v.v....* Mb

II.

I). I

.11IAtr)

t:1-I

C1

3:

1

t1

I_.--1

n.

1i

01)

5)

1. A.

a

SI

S+

U)

r11

Vow

..ey:

wW

m:A

Lam

otor *44*. . ov

I.)to

(5); )

(I)V

)1:7

C1

Jr-

ft11

;1f IC

.()".);)

r.)I

,

))Z

T1

(Si

,..)U

J.1

. -1

(.1

(.1(

.1

1 t.

0;

.,

,.".

*"

)

rS

;

Fig, 12. Mean number. of responses perexperimental session for all subjects. Groupingis by sequence of experimental conditions.

25

(

.1

L.

RE

;1\

1) 00

0co

cD

1]

IN)

Ot

00

0

FT-

II

----

-.--

-1-

rt.

/0

/e.

13 1

>v

ocr

)N

Cl)

co-4

v)

C (b

3

i'( )1"(I)

1j':

sl

IL

.Ito';

;'

:,)t

II

(

et et...et

.....

Fla

0I0

The mean number of responses for the operant:first reinforcement, and first extinction periods areshown in Fig. 13. This permits a graphic comparison ofall subjects and amplifies two recurrent facts: the failureto achieve conditioning and the decrement in response rate.

DISCUSSION

This experiment was performed in order to provideinformation about counseling which might be used in answer-ing the questions stated earlier. A careful considerationof each question appears to be the logical point fromwhich to begin an extensive examination of the results.

s

Is positive verbal reinforcement powerful enoughto modify the verbal behavior (negative self-referencestatements) of individuals in an interview setting,simulating a counseling relationship?

The data indicated that verbal reinforcement ispowerful enough to modify client negative self-referencesbut that this modification is in the form of a decreasein responding. The effects of reinforcement were morethat of suppressing the critical response class thanconditioning. The fact that negative self-referencestatements did not increase during the reinforcement periodindicated that reflection did not function as a positivereinforcer.

If the assumption is made that the typical under-graduate student does not wish to become more negativein his appraisal of himself, then any method of confrontinghim with such behavior would result in a suppression ofthat behavior. ?Then a reflection is made on a negativeself-reference statement, the counselor is identifyingand calling to attention the client's negative verbalbehavior. It would not seem rewarding to be made awarethat one is speaking poorly about himself. The presumedeffect of reflection in counseling is to convey under-standing, recognize feelings, facillitate insight andhelp to clarify the client's behavior. If reflectionfunctioned as such in this experiment, the effect couldwell have been one of punishment rather than reinforcement.It is doubtful that the student likes to be consistentlyconfronted with his self-negativism.

Cole (1965) found results which tend to supportthis point of view. Two groups of 10 subjects each were

28

seen in a counseling interview. The group reinforced fornegative statements about a neutral concept showed aconditioning effect while the group reinforced for negativestatements about the self-concept showed a decrement inresponse rate. The point of particular relevance isthat in both the Cole study and the present research,conditioning did not occur when the object of the reinforcedbehavior was oneself.

The response classes of positive and negativeself-reference statements have been the subject of consider-able research. Some investigators report success whileothers have been unsuccessful. In fact, this has beentrue of virtually all response classes. These discrepanciespreclude any definitive conclusion about the conditioningof verbal behavior in a counseling-therapy situation.It is one thing to achieve conditioning of hostile verbswith beginning psychology students or shape opinionsby phone, and quite another to alter meaningful verbalbehavior in a counseling situation.

Lubin (1965) recognizes the problem of partialsuccess and suggests that the parameters of the rein-forcement situation operate as do those of primary rein-forcement. His study was particularly concerned withspeed and consistency of reinforcement as related todifferences in conditioning effects.

In the present research, reinforcement was givenimmediately after the emission of a critical response.With very few exceptions, reinforcement was always reflec-tion. Simple agreement cues were given only on occassionswhere it was felt that reflection would have a detrimentaleffect on the flow and content of the subject's speech.As for the consistency of reinforcement, very little canbe said. It seems that a true reflection, like rein-forcement, can be identified only by its consequences.There is no question that some variability occurred betweenattempts by the experimenter to reflect negative self-references. On the other hand, it is relatively certainthat such statements were paraphrased. The extent towhich the client perceived this as a reflection may havebeen another variable. Reflection (paraphrasing) can beconsidered a reinforcer only if it is a true and appro-priate restatement with all essential qualities. Thismakes meaning implicit. In this attempt to modify verbalbehavior, there is little doubt that such private experiencesas the perception of meaning functioned as uncontrolledvariables, One obstacle to the investigation of verbalbehavior is that the consequences which maintain the verbaloperants are usually difficult to specify.

29

Recently, numerous investigations have beencarried out in an attempt to determine the necessaryconditions for verbal conditioning to occur. Merbaum (1964),Werbaum and Southwell (1965) and Powell (1965) have shownthat different types of experimenter responses have variedeffects on subjects' verbal behavior. Reflection andparaphrasing have been found to be highly effectiveexperimenter responses although, one study reports successwith only negative self-references. The discrepancybetween these findings and the results of this experimentmight stem from a major difference in design. Thisresearch attempted, rather successfully, to simulate adesireable counseling process with daily sessions over aperiod of several months. Interview sessions over ashorter time span could understandably have a differenteffect.

The issue of subjects' awareness of the response-rainforcement contingency is another possible explanationfor the failure to achieve conditioning. Speilbergerand DeNike (1966) present evidence that in verbal condi-tioning, it is this awareness that is learned. Maltzmann(1966) recognizes the inadequacy with which cognitivepsychology interprets reports of awareness in verbalconditioning. Numerous other writers have discussed theissue of awareness with evidence being presented in supportof its necessity or irrelevance. In this present research,there was no attempt to assess awareness, however, allsubjects reported complete naivete in regard to anypurpose beyond the general instructions. With one excep-tion, all subjects reported that within a week or so thesessions actually became a counseling or "catharsis"session in which they felt free to express themselves.All subjects reported that it had been a pleasant andworthwhile experience.

Recent investigations have also identified setas a major determinant of conditioning effects. Thesubjects reported that they had no idea what to expectbeyond the general instructions. This absence of any setmight account for their failure to become aware of theresponse-reinforcement contingency, as suggested earlier.Simkins (1963) studied the effects of preconditioninginstructions, concluding that they affect awareness andlearning. In a more recent study, Farley and Hokanson(1966) found that informational set had a significanteffect on the level and rate of verbal conditioning. Thedifferences in informational set might account for manyof the discrepancies found in the literature and thefailure of subjects in the present study to acquire thedesired behavior.

3.0

A new direction in verbal conditioning researchis the investigation of verbal habits. Clance and Dixon(1965) found that the type of verb used in a sentencehad a systematic effect on the frequency of first personpronouns. This effect is presumably a verbal habitestablished prior to the experiment. They add that theeffect is greater during acquisition than during theoperant period. This could account for the negativedifference between the response rate of the operant andfirst CRF shaping periods.

Dhatarazzo, Saslow and Pareis (1960) and Simkins(1963) have identified the response class as an importantvariable. In the case of self-reference statements, itis not an easy task to operationally define those thatare positive or negative. The terms positive and negativeappear to be relative to some arbitrary standard. Whenused to dichotomize statements about oneself, they implyemotional feelings which, like other private experiences,are not accessable to the experimenter. Positive self-references are usually defined as self-references whichclearly contain no expression of the negative. This wouldinclude most self-reference statements. Suppose 20% wasa figure arbitrarily selected to represent the negativeportion of all self-references. To reinforce positiveself-references would be to reinforce OD% of all self-references, according to the definition above. Thispercentage would be sufficient to show a conditioningeffect. On the other hand, to reinforce negative self-references would involve only 20% of all self-referencesand the response class would extinguish. A self-referencestatement as a group of words has no inherent positive ornegative quality except as it might contain negation..It has meaning only in the presence of a verbal communitywhere this quality is largely conventional and highlysubjective. The point, then, is that all self-referencestatements constitute a natural response class while posi-tive or negative self-references do not. This seems tobe the most plausible reason for the ineffectiveness ofthe reinforcer. The decline in the response rate ofnegative self-references would be due to the extinctionof the response class, all self-references.

In order to 'approach the second question it winbe necessary to review the original research design. Thefollowing paragraph will provide a frame of referencefor further discussion. It is a statement of how theexperiment would have been conducted if conditioninghad occurred. ,

31

r

..,

"Subjects will be assigned at random to threeschedules of reinforcement: FR (fixed ratio),FI (fixed interval), and VI (variable interval).Two subjects will be assigned-to each schedulewith one subject each on fixed and variableinterval schedules being run in reverse topoint up possible sequence effects of theschedules. The number of values to be usedon each schedule will be determined by therelative time period remaining after responsestability is obtained at various phases ofthe experiment. Provisions will be made toprogram as many as three values on each schedule.The use of student volunteers restricts theduration of the research to a period coincidingwith the school term. This, together withthe requirement of response stability at suc-cessive levels, permits only a tentative timeschedule."

The subjects were assigned to the schedules at the startof the experiment. Although one subject (38) withdrewfrom school during the early stages of the experiment,this would have had no effect on the use of reinforcementschedules. The decision was made to eliminate the useof a VI schedule programmed in reverse. A table of valuesof progressions for generating VI schedules (Fleshier &Hoffman, 1962) was used in preparing the forward schedules.The Fleshier - Hoffman progressions are mathematicallyderived from a general equation in which the probabilityof reinforcement would remain constant as a function oftime since the last reinforcement. This negates anysequence effect. As pointed out earlier, 38 had thehighest rate of response and showed a conditioning effect.Had she continued, a VI 30" schedule would have been

introduced.

To what extent will the different reinforcementschedules produce different effects on the verbal behaviorof the subjects?

This question cannot be answered on the basis of

the results obtained. The only schedule employed wasa fixed ratio of two responses per reinforcement. Of

,

the tins subjects receiving this schedule, Si generateda pattern of mean responses that was similar to the subjectson continuous reinforcement. The results for allsubjects illustrated a steady decline in responding regard-less of whether or not reinforcement was continuous or

intermittent.

12e

To what extent will there be differences inresponding between subjects on the same schedule ofreinforcement?

The striking similarity between the subjectsreceiving FR reinforcement was the pattern of responseduring the FR period. Both subjects responded initiallywith a slight increase followed by a steady decrease.Subject 1-decreased rapidly to a point well below thebase rate of responding. Subject 2 decreased only slightlyto the previous level of responding where he remained.This subject's performance throughout the experiment wascharacterized by strict adherence to the instructions.He frequently verbalized his concern about this. The factthat his response rate remained relatively stable indicatedthat he might well have been behaving on the basis of aninformational set. His verbal behavior was probably underthe control of the original instructions. The resultsshow that it was not controlled by the reinforcing stimulus.

Operant Period

One of the initial problems was the extablishmentof a baseline. An attempt was made to observe the behaviorover an extended period of time in order to determine its

. characteristics. This served a two-fold purpose. First,it permitted the completion of transition states whichoften occur at the beginning of an experiment, i. e.,adjusting to the experimental-conditions. Second, itprovided a basis for developing a stability criterion.The fact that there is no general rule for identifyingstable behavior places the experimenter in a positionof deciding how much variability can be tolerated withoutmasking the experimental effects, If the manipulation ofthe independent variable produces a behavioral change,it can be described only by specifying the baseline fromwhich it occurred (Sidman, 1961). After an extendedobservation of the operant rate, a four session sequencewith limited variability appeared to be a realisticcriterion. Clusters of total responses appeared ingroups of 3 and 4 sessions with a range of 3-10 responses.A four session sequence with a range not exceeding 5responses became the criterion,

Figures 1-8 illustrate the variability for eachsubject during the operant period. It can be seen thatall subjects reached a stable rate of responding belowthe level of the first few experimental sessions. Subjects1, 4, 6, and 7 show considerable intrasubject variability.

33

The graph points representing a peak number of responsesfor a particular session have no special significance.These sessions appeared to be just like any other exceptfor the unusually high response rate. Responses for 52varied over three sessions then stabilized. This initialvariability appeared to be a transition period as discussedearlier. Subjects 3 and 5 showed considerable similarityduring the operant period. The response rate of eachvaried only slightly from session to session with arather uniform decrease in response rate. The fact thatall subjects (except 32) decreased in total responsesmight have been due to the informational set brought aboutby the general instructions.

At first, subjects appeared to be making an effortto talk about themselves. This was characterized bysuperficiality and impersonal content of conversation.As the experiment progressed, the verbal behavior of subjectsbecame personal, introspective and centered on self.They began to discuss matters of concern and quite oftenit would be necessary to interrupt in order to terminatethe session. Many times subjects would ask advice eventhough they had learned that the experimenter would notrespond. This silence by the experimenter caused somesubjects to become irritated. This was associated withsuch complaints as "I can't think of anything to say"and "won't you help me." These eventually disappeared.Subjects would usually begin talking as soon as they camein and contine until time had expired. Negative state-ments such as the above quotations were included in theoperant rate, although they were generated by the experi-mental situation and might have been artifacts, On the otherhand, the subject of this research was the occurrence ofnegative self-reference statements under these conditions;that is, a quasi-counseling situation.

Reinforcement Period

In verbal behavior, it is extremely difficult tocreate a state of deprivation for a particular responseclass and, therefore, unlikely that a primary reinforcerwill be found. Since the strength of a verbal reinforceris acquired by association, it might be enhanced by pairingit with a strong social reinforcement. Simkins (1963)has suggested the use of pennies. Looking again at thegraphs of subjects' responses (Fig. 1-8), an initialincrease in responding can be seen for the CRF period.This was probably due to the experimenter's participation.That is, his silence during the operant period created

34

a certain amount of social deprivation. dhen he beganto respond to subjects' verbalizations the effect was toreinforce all verbal behavior, regardless cf what was said.

During the first three sessions reinforcement wasgiven for all self-reference statements. In the remainingsessions, only negative self-references were reinforced.This shaping appeared to have had little effect and mayhave been unnecessary except for the smooth transitionfrom numerous reinforcements of all self-references to

the less frequent reinforcement of negative self-references.The behavior to be manipulated was already occurring and

shaping was of dubious value.

The response rate of 58 was quite variable but

increased over a period of 8 sessions. Subjects 3 and 7gradually declined in response rate during the CRF period.Stability did not readily occur. The slope was similarto that of the operant period. For the other subjects,the response rate stabilized rather quickly. Subjects1 and 2 stabilized above the operant level indicating aslight conditioning effect. These subjects had previouslybeen assigned to an FR schedule of reinforcement whichwas then introduced.

The effects of fixed ratio reinforcement havealready been discussed although an additional comment is

in order. The very slight increase in responses duringthe CRF period created some doubt that the response rate

would rise during FR reinforcement. Subject 1 had only1 session in which the total responses were higher thanduring the CRF period, Subject 2 did not exceed the CRFperiod until the fourth FR session. Both began to declineafter these peak sessions.

The second CRF period for S1 was extremely variable.A high response rate was recovered during the firstsession but was immediately lost. A second high pointwas followed by the lowest stability level of all periods.

The continued ineffectiveness of the reinforcer wasobvious and it was therefore withdrawn. The pattern ofvariability and decrease in response rate continued asin previous periods.

The performance of 52 was atypical in that it wasrelatively stable when compared with other subjects.The second CRF period was much the same as the first.Response stability occurred at about the same level withonly slight variability. Subject 2 was the only individualwho did not show a general decrease in response rate.

35

.,C._

o.

In the experimenter's judgment, 52 did not enter into acounseling relationship as did the other subjects. Thesecharacteristics might have bean due to the controllingeffect of the instructions and the cognitive nature ofhis verbal behavior.

Response stability occurred for Ss 3, 4, and 5at a level no higher than that of the operant period and,with no increase in response rate, schedules of reinforce-ment could not be employed. In order to verify theapparent ineffectiveness of the reinforcer, an extinctionperiqd was begun. Continuous reinforcement was againintroduced after stability occurred during extinction.There was essentially no change in response rate forall three subjects. This suggested the possibility thatthe operant level was false and that the first extinctionperiod more closely approximated a true operant rate.

Extinction

The extinction period for S1 was marked by anincrease in response rate. Subject 2 remained relativelystable showing little sesi.tivity to the change in experi-mental conditions. This was also true of Ss 3 and 4.Subject 5 had a three session period below the rate of thepreceeding CRF period. Variability and a continued declinein Tesponse rate occurred.for Ss 6 and 7. The extremehigh session for Se was brought about by depression. It

was a one session effect (Fig. 6) characterized by consi-derable negative verbal behavior and crying. There wereno indications, however, that this behavior was any morethan a "normal" reaction to the stresses of academicpressure. In the strictest sense of the term, the absenceof a conditioning effect precluded any extended discussionof behavior during extinction. This was merely a periodduring which no attempt was made to reinforce the responseclass.

General Ubservations

During the experiment several observations weremade reoarding subjects' verbal behavior, the counselingrelationship, and the verbal conditioning effect.

Subjects often spoke as if they were merelyexpressing a stream of consciousness or musing over pastexperiences. Such verbal behavior gave rise to manyincomplete statements; that is, sentence fragments which

36

N

were negative self-references but not negative self-reference statements. This was particularly a problem inidentifying members of the critical response class.

The maintainance of negative verbal behaviorappeared to be due, in part, to self-reinforcement. Ithas been suggested that one's own behavior is a signifi-cant aspect of his environment (Kanfer, 1965) and thatawareness of such may effect the probability of subsequentbehaviors. Observation of subjects during the sessionsindicated an awareness of the content of conversationand the effect of this awareness on further speech.This appeared to be evaluative in nature since subjectswould sometimes anticipate the experimenter's reactionsto their statements.

With regard to the nature of the counselingrelationship, the experimenter felt that all subjectsexcept S2 behaved in much the same way as true counselees.These people were volunteers but they talked about problemsmuch of the time. At the conclusion of the experimentmost subjects felt that they had made positive gains inself-understanding.

All subjects were seen over a period of approxi-mately 60 sessions but absences and tardiness reducedthis number to some degree. The data show the net numberof sessions per subject. The division of the summerschool term into two parts proved to be a hardship conditioninsofar as it decreased the number of useable days andemployed unusual class scheduling which was a factor intardiness.

The verbal conditioning effect is somewhat of anelusive phenomenon, although there seems to be no doubtthat the counseling process can be investigated throughthe use of verbal conditioning techniques. The furtherinvestigation of counselee verbal behavior may providesignificant information on many aspects of the counselingprocess. The failures to achieve conditioning point upweaknesses in design and suggest a re-evaluation ofcurrent methodology. Further research is imperative.

CONCLUSIONS AND IMPLICATIONS

The data indicated that reflection did not functionas a positive reinforcer when it was made contingentupon a negative self-reference statement. The net effect

37

..... -411.

was that no self-reference statements were reinforced,creating an extinction condition. This, in part, accountedfor the rather uniform decrease in response rate. Themajor dterminant was the punishing effect of reflectionwhich, when contingent upon a negative self-reference,became a negative stimulus.

The failure to achieve conditioning in a quasi-counseling situation indicated that adventitious reinforce-ment of negative client statements does not occur. Thereis some doubt that negative statements can be increasedby verbal reinforcement in a counseling interview, andit is relatively certain that counselor reflection will notstrengthen such behavior. On the contrary, reflectionbrought about a decrease in negative self-references.This not only supports current counseling practice butprovides empirical evidence of the relationship betweenclient statements and counselor responses. This exemplifiesthe fact that the behavioral model, on which the presentstudy was based,was a conceptual framework within whichcounseling procedures can be experimentally tested.

It seemed that the research method employedhad certain limitations. One of the most serious was theoperant period during which a baseline was determined.The variability for each subject required an extendedperiod of observation in order to select a stabilitycriterion against which changes could be compared. On theotheehand, an extended period of observation createdan artificial condition in which the experimenter's silencebegan to affect the subjects' behavior. The characteristicsof human behavior during an operant period are not fullyspecified as yet. Additional research is needed concerningthe effects of prolonged silence by the experimenter,the feasibility of accepting a normal conversationalperiod, and the pattern of variation associated withdifferent response classes. Additional information aboutthese problems might also be helpful in identifying trueresponse classes.

One major difficulty in attempting a verbal condi-tioning study is the specification of the response classto be reinforced. In the present case of negative self-reference statements, there is good reason to suspectthat the only common element is self-reference. Thesubjective nature of positivism or negativism is associatedwith emotional feeling and does not appear to be acharacteristic of verbal behavior per se. The difficultyin specifying affect in behavioral terms points up theneed for an investigation of the physiological (autonomic)

38

activation that is associated with patterns of verbalbehavior. A measure of autonomic activity such as thegalvanic skin reflex; might be employed. If some relation-ship could be found between negative self-reference state-ments and level of activation, the verbal conditioningparadigm would take on even greater significance as acounseling tool.

In addition to response class, informational setis another variable that warrants further investigation.

.

The pre-experimental instructions have an unspecifiedeffect which may vary from subject to subject. Thedifferent outcomes associated with a particular counselingtechnique might well be a function of the information theclient has about counseling. This suggests the ppssibilityof a standardized initial interview.

The pattern of partial success in attempts tocondition verbal response classes was attributed to theeffects of unspecified variables, the arbitrary natureof the response class, and methodological problems associatedwith the use of human subjects. It has been shown thatone counselor technique, reflection, has no undesirableeffect on negative self-reference statements. This facttends to indicate that the verbal conditioning paradigminvolves far more than the simple contingency of student-counselor responses, and makes explicit the need forfurther research. This research is another demonstrationof the practical application of a behavioral model tocounseling.

SUMMARY

This study was an attempt to simulate a desirablecounseling process with the exception of the systematiccounselor reinforcement of student negative self-referencestatements. By investigating the tendency of some counselorsto respond to negative client statements, it was determinedto what extent such methods might produce undesirableclient behavior.

The objective was to demonstrate that definitemeasurable counseling procedures could be experimentallytested and had practical consequences. The verbal operantconditioning paradigm was employed in an attempt to answerseveral basic questions. Can the frequency of negativeself-reference statements be modified by verbal reinforce-ment in a quasi-counseling situation? M1 schedules of

39

e

..,

reinforcement produce different effects? Would subjectsrespond differently to the same schedule of reinforcement?A functional research design permitted a focus on individualbehavior.

Eight undergraduates were seen daily for approxi-mately three months. All subjects were given an operantperiod during which a baseline was obtained. This wasfollowed by a reinforcement period. Reinforcement wasreflection (paraphrasing) contingent upon a negativeself-reference statement. Only three subjects showedan increase in rate of response. One withdrew fromschool. The remaining two were placed on a fixed ratio (FR)schedule of reinforcement. The other subjects were placedon extinction during which time no reinforcement was given.With three of these, a second attempt was made to reinforcenegative self-references followed by a second extinctionperiod.

It was found that no subjects acquired a sustainedincrease in response rate as a consequence of reinforce-ment. Of the seven subjects who completed the experimentonly one failed to show a relatively uniform decreasein total responses per session.

Results indicated that reflection was riot areinforcing stimulus when made contingent upon a negativel

self-reference. This was attributed to the response class,informational set, and the problem of baseline stability.It was shown, however, that counselor reflection of

negative statements had no undesirable effect. In fact,it seemed to contribute to a decline in such statements.An examination of the methodology employed and certaincritical variables provided reasons for the partial successof verbal conditioning studies. Results identify problemareas in which further research is needed.

It was demonstrated that the effects of counselorreinforcement on a student's negative responses are not

necessarily undesirable. In addition, evidence wasprovided in support of the behavioral model as a basis

for counseling research.

40

REFERENCES

Adams, H. E. C: Frye, R. L. Psychotherapeutic techniquesas conditioned reinforcers in a structured interview.Psychol. Beg., 1964, 14, 163-166.

Adams, J. S. :c Hoffman, E. The frequency of self-referencestatements as a function of generalized reinforcement.J. abnorm. soc. Psychol., 1960, 60, 384-389.

Babbitt, H. G. An attempt to produce changes in attitudestoward the self by means of verbal conditioning,J. verbal Lrnq. verbal Behay., 1962, 1, 168-172.

Bremer, L. w. Shostrom, H. L. Therapeutic s choloo .

Englewood Cliffs, W. J.: Prentice-Hal , 1 6.3.

butler. J. m. Weasuring the effectiveness of counselingand psychotherapy. Personnel Quid; J., 1953, 31,88-92.

Cahoon, D. D. 4enrich, 'W. W. Verbal reinforcement.Where do we go from here? Psychol. Rec., 1965,15, 141-143.

Centers, R. A laboratory adaptation of the conversational. procedure for the conditioning of verbal operants.

J. abnorm. sac. Psychol., 1963, 67, 334-339.

Clance, Pauline R. & Dixon, T. R. The effect of previouslylearned verbal habits upon two response classesin verbal conditioning. J. Psychol,., 1965, 60,271-276.

Cohen, .d. D., Kalish, H. I., Thurston, J. R., & Cohen, E.Experimental manipulation of verbal behavior.J. ea. Psychol., 1954, 47, 1.'1-110.

Cole, C. W. Effects of verbal stimuli in a counselinganalogue. J. counsel. Psychol., 1965, 12,408-413.

Daily, J. k. Verbal conditioning without awareness.Dissertation Abstr., 1953, 13, 1247-1248.

41

Eysenck, H. J. Handbook of abnormal psychology. NewYork: Basic 3ooks, 1961.

Farley, J. A. ::: Hokanson, J. E. The effect of informationalset on acquisition in verbal conditioning.J. verbal Lrnn. verbal Behay., 1966, 5, 14-17.

Ferster, C. a. a Skinner, 3, F. Schedules of reinforcement.New York: Appleton-Century :UFani, 1957.

Fleshier, Hoffman, H. Si A progression for generatingvariable-interval schedules. J. 222. Anal. Behay.,1962, 5, 529-530.

Friedman, L. F. Verbal conditioning and attitudes towardconditioned autonomic activity. DissertationAbstr., 1964, 25, 1357-1358.

Gonzer, V. J. w Sarason, I. G. Interrelationships amonghostility, experimental conditions and verbalbehavior. J. abnorm. soc. Esychol., 1964, 68,79-84.

Greenspoon, J. The effect of verbal and non-verbal stimulion the frequency of members of two verbal responseclasses. Unpublished doctoral dissertation,Indiana Univer., 1951. Cited by Greenspoon, J.Verbal conditioning and clinical psychology.In A. J. 3achrach (Ed.) Experimental foundationsof clinical psychology. New York: Basic Books,

Greenspoon, J. The effects of two non-verbal stimulion the frequency of members of two responseclasses. Amer. Psychologist, 1954, 9, 284.

Greenspoon, J. The reinforcing effects of two spokensounds on the frequency of two responses. Amer.J. Eachol., 1955, 68, 409-416.

Greenspoon, J. Verbal conditioning and clinical psychology.In A. J. Bachrach (Ed.) Experimental foundationsof clinical psychology. New York: Basic Looks,M2,

Grossberg, J. Behavior therapy: a review. Psychol.Bull., 1964, 62, 73-88.

42

4

Hartman, C. H. Verbal behavior of schizophrenic andnormal subjects as a function of types of socialreinforcement. Dissertation Abstr., 1955 15,

1652-1653.

Hildum, D. C. a Brown, R. W. Verbal reinforcement andinterviewer bias. J. ebnorm. soc. aychol.,1956, 53, 108-111.

nce, L. P. The effect of three schedules of intermittentreinforcement on the frequency of a verbal responseclass in a quasi-counseling situation. Unpublisheddoctoral dissertation, Florida State Univer., 1965.

Insko, C. A. Verbal reinforcement of attitude. J. Pers.soc. Psychol., 1965, 2, 621-623.

Kanfer, F. The effect of partial reinforcement onacquisition and extinction of a class of verbalresponses. J. 22.42. Psychol., 1954, 48, 424-432.

Kanfer, F. Verbal conditioning: reinforcement schedulesand experimente/, influence. Bachol. B22.,1958, 4, 443-452.

Kanfer, F. a Marston, A. R. Control of verbal behaviorby multiple schedules. Psychol. B22., 1962, 10,703-710.

Krasner, L. Studies of the conditioning of verbal behavior.Psychol. Bull., 1958, 15, 148-171.

Krasner, L. Behavior modification research and the roleof the therapist. In L. A. Gottschalk L A. H.Auerback (Eds.) Methods of research in psycho.-therapy. New York: Hoeber Harper, 1964.

Krasner, L. Verbal conditioning and psychotherapy. In

L. Krasner and L. P. Ullmann (Eds.) Research inbehavior modification. New York: Holt, RineFirtc dinston, 1965.

Krumboltz, J. U. The parable of the good counselor.Personnel Quid. J., 1964, 43, 118-124.

Krumboltz, J. D. & Thoreson, C. E. The effect of behavioralcounseling in group and individual settings oninformation seeking behavior. J. counsel. Psychol.,1964, 11, 324-335.

43

Lubin, I. Sources of differences in effectiveness amongcontrollers of veral reinforcement. DissertationAbstr., 1965, 25, 4831-4832.

McNair, D. N. Reinforcement of verbal behavior. J. exp.Psychol., 1957, 53, 43-46.

Matarazzo, J., Saslow, G., & Pareis, E. Verbal conditioningof two response classes: some methodologicalconsiderations. J. abnorm. soc. Psychol., 1960,61, 190-206.

Merbaum, M. The conditioning of affective self-referencesby three classes of generalized reinforcers.J. Pers., 1963, 31, 179-191."

Merbaum, M. & Southwell, E. A. Conditioning of affectiveself-references as a function of the discriminativecharacteristics of experimenter intervention.J. abnorm. Psychol., 1965, 70, 180-187.

Meyer, t. J., Swanson, B., a Kauchack, N. Studies ofverbal conditioning: I. Effects of age, sex,intelligence and "reinforcing stimuli." Child.Devel., 1964, 18, 499-510.

Michael, J. & Meyerson, L. A behavioral approach toguidance. Harvard educ. Rev., 1962, 32, 382-402.

Patterson, C. H. Control, conditioning, and counseling.Personnel quid. J., 1963, 41, 680-686.

Powell, Jr., 11. J. A comparison of the reinforcing effectsof three types of experimenter response on twoclasses of verbal behavior in an experimentalinterview. Dissertation Abstr., 1965, 25, 4260-4261,

Portnoy, S. Cc Salzinger, K. The conditionability ofdifferent verbal response classes: positive,negative, and nonaffect statements. J. gen.Psychol., 1964, 70, 311-323.

Raimy, V. C. Self-references in counseling interviews.J. consult. pachol., 1948, 12, 153-163.

Rogers, C. R. The place of the person in the new worldof the behavioral sciences. Personnel quid. J.,1961, 39, 442-451.

44

Rogers, C. R. a Skinner, E. F. Some issues concerningthe control of human behavior: a symposium.Sci., 1956, 1249 1057-1066.

Rogers, j. h. Operant conditioning in a quasi-therapy

setting. J. abnorm. soc. Psychol., 1960, 6i,

247-252.

Ryan, T. A. Krumboltz, J. D. Effects of planned rein-forcement counseling on client decision making

behavior. J. counsel. Psychol., 1964, 11, 315-323.

Salzinger, K. & Pisoni, Stephanie. Reinforcement of

affect responses of schizophrenics during the

clinical interview. J. abnorm. soc. Psychol.,

1958, 57, 84-90.

Salzinger, K. Experimental manipulation of verbal behavior.

J. nen. Psychol., 1959, 11, 65-95.

Salzinger, K., Portnoy, Stephanie, & Feldman, R. S.Experimental manipulation of continuous speech

in schizophrenic patients. J. abnorm. soc. Psychol.

1964, 68, 508-516.

Shoben, E. J. Comment on Ueitz, H. Guidance as behavior

change. Personnel quid. J., 1961, 39, 550-560.

Shoben, E. J. The therapeutic object: men or machines.

J. counsel. Psychol., 1963, 10, 264-268.

Sidman, M. Tactics of scientific research. few York:

basic Looks, 1961.

Simkins, 1. Instructions as discriminative stimuli in

verbal conditioning and awareness. J. abnorm.

soc. Psychol., 1963, 66, 213-219.

Skinner, 3. F. Science and human behavior. New York:

Skinner, Ls. F. Verbal behavior. New York: Appleton-

century-CrOME,75g77-

Solley, C. 0. fc Santos, J. F. Perceptual learning with

partial reinforcements. Percept. mot. Skills,

1958, 8, 183-193.

Southwell, E. A. Conditioning of hostile and neutral

verbs in neurotics and normals. J. consult.

Psychol., 1962, 26, 257 -.262,

45

Speilberger, C. D. Deaike, L. D. Descriptive behaviorismversus cognitive theory in verbal operantconditioning. Psychol. Rev., 1966, 73, 306-326.

Spivak, Para john, J. The effect of the scheduleof reinforcement on operant conditioning of averbal response in the auto-kinetic situation.J. abnorm. sot:. Psychol., 1957, 54, 213-217.

Strong, S. R. Verbal conditioning and counseling research.Personnel quid. J., 1964, 42, 660-669.

Ullmann, L. P. & Krasner, L. (Eds) Case studies inbehavior modification. ;:ew York: Holt, Rinehart aWinston, BM'

Ullmann, L. P., Krasner, L., & Knowles, J. B. Effect ofverbal conditioning of attitudes on subsequentmotor performance. J. Pers. soc. Psychol.,1965, 1, 407-412.

Verplanck, 3.. The control of the contents of conversation:reinforcement of statements of opinion. J. abnorm.sac. Psychol., 1055, 51, 668-676.

Verplanck, U. S. A glossary of some terms used in theobjective science of behavior. Psychol. Rev.,Suppl., 1957, 64.

L'Jaskow, Irene E. Reinforcement in a therapy like situationthrough selective responding to feelings orcontent. J. consult. Psychol., 1962, 26, 11-19.

Jeitz, H. Guidance as behavior change. Personnel add.J., 1961, 39, 550 -560.

46

GLOSSARY

ACQUISITION Progressive increments in response.strength observed over a series of occasions onwhich the response is measured (Verplanck, 1957).

BASE RATE OF RESPONDIWG The ongoing behavior, orcontinuous and continuing performance of a singleindividual which serves as a baseline from whichto measure the effects of experimental operations(Sidman, 1961).

CONDITIONING The strengthening of behavior whichresults from reinforcement (Skinner, 1933).

CONTINUOUS REFORCEPENT (CRF) The experimentalprocedure of reinforcing a response each time itoccurs (Verplanck, 1957).

EXTINCTION The operation of nonreinforcement. Aresponse whose frequency has been decreased bynonreinforcement is said to have been extinguished(Sidman, 1961).

FIXED-INTERVAL SCHEDULE (FI) A schedule of intermittentreinforcement where the first response occurringafter a given interval of time, measured fromthe preceding reinforcement, is reinforced(Ferster (: Skinner, 1957).

FIXED-RATIO SCHEDULE (FR) A schedule of intermittentreinforcement where responses are reinforced uponcompletion of a fixed number of responses counted-from the preceding reinforcement (Ferster aSkinner, 1957).

FREE OPERANT SITUATION A free responding situationin which the only restrictions placed upon thesubject's behavior are those inherent in thelaws of behavior (Sidman, 1961).

FUNCTIOWAL DESICP: An experimental design in which theparameters of control are varied to generate aquantitative description of the interactionspectrum (Sidman, 1961).

47

4

(

wETHOD OF SUCCESSIVE APPROXIATION The process whichincludes the differentiation of new forms ofresponse, and the development of appropriatestimulus control, so that a given response isgenerally emitted only upon an appropriate occasion(Ferster & Skinner, 1957).

OPERANT A term used to describe behavior which operatesupon the environment to generate consequences.It is used both as an adjective (operant behavior)and as a noun to designate the behavior definedby a given consequence (Skinner, 3.953).

OPERANT LEVEL The rate of occurrence of an operantresponse before the response has been experimentallyreinforced (Verplanck, 1957).

OPERANT PERIOD The period of time during which noexperimental restrictions are placed upon thesubject's behavior (Sidman, 1961).

PUNI3WEWT Presentation of an aversive stimuluscontingent upon a response or the withdrawal ofa positive reinforcer contingent upon a response.In bath cases the behavioral effect is a shortterm response suppression.

REINFORCE.IENT The operation of presenting to theanimal in operant conditioning, after it hasmade a response (and therefore contingent on'itsoccurrence), a reinforcing stimulus or of with-drawing a negative reinforcing stimulus(Verplanck, 1057).

REI6FORCER Any stimulus which, when made contingentupon a response or class of responses, is fol-lowed by a modification in the frequency ofoccurrence in that response or class of responses.

RE3P00:SE (1) An instance of an identifiable partof behavior. (2) A class of such instances.In this sense, response is equivalent to operant(Ferster L?: Skinner, 1957).

RESPOASE CLASS A response class consists of a numberof different responses which are similar to oneanother, in that the occurrence of one may besubstituted for the occurrence of another(Salzinger, 1959).

48

SCHEDULE OF REINFORCEWENT A plan or procedure wherebythe experimenter determines which one of a seriesof responses will be reinforced (verplanck, 1957).

SHAPIG bringing a response under experimental controlby the process of successive approximation.

STABLE RATE OF RESPONDUG A state of respondingin which the behavior in question does not changeits characteristics over a period of time(Sidman, 1961).

STEADY STATE BehaVior whose characteristics do notchange for long periods of time (Sidman, 1961).

VARIABLE INTERVAL SCHEDULE (VI) A schedule of inter-mittent reinforcement where reinforcements areprogrammed according to a random series ofintervals having a given mean and lying betweenarbitrary extreme values (Ferster e: Skinner, 1957).

VERBAL BEHAVIOR (1) Behavior involving the vocalizationor writing of words, or response to spoken orwritten words. (2) Behavior whose reinforcementis contingent upon stimulation of and responseby another individual (verplanck, 1957).

---

49