Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

MINIMAL SOCIAL REINFORCEMENT AND INTERVIEW CONTENT*

FREDERICK H. KANFER AND JOHN F. MC BREARTY


Purdue University

PROBLEM
There has been increasing evidence in the recent literature that the verbal
productions of an S in the laboratory <^^ and in clinical interviews'*^ are affected by
subtle changes in the behavior of an interviewer. In demonstrating that increased
probability of occurrence of discrete verbal response units can be achieved by E's
intervention, the first step has been taken toward investigation of the specific effect
of minimal interviewer cues on verbal material obtained in clinical interviews.
The implications of these studies for psychotherapy have been discussed in
several papers ^^' *\ and it has been suggested that the same process of conditioning
may underlie more of the dynamic interaction between therapists and patients than
has yet been suspected. One important implication of these findings lies in their
contradiction to the long-standing clinical generalization that interview content is
determined mainly by the internal dynamics of the client, and therefore (a) the
content of a patient's "free" communication reveals those emotional areas which are
especially important in his life, and (b) either limited or excessive preoccupation
with a given topic during a free interview yields significant information about the
patient's life history and dynamics.
While conditioning of plural nouns and other isolated response classes in inter-
personal communications has previously been demonstrated, these responses may
have little relationship to S's past history of social reinforcement. This study raises
the question whether response classes which have been built up in connection with
S's personal life history, and which have been integrated into his total emotional and
verbal response pattern are equally modifiable. Only if this is the case can the
utilization of verbal conditioning principles in clinical interviews be justified.
This study used a semi-structured interview setting in which mihimal reinforce-
ment cues were studied in relation to the time which S devoted to a given topic, and
to the amount of talk on that topic. These variables were used as a measure of the
amount of information communicated; the importance of the content is obviously
not similarly indicated by duration. The "topics" were chosen to approximate those
verbal classes whose occurrence is noted in the interpretation oi clinical interviews
as significant and revealing of the interviewee's dynamics.
The specific purpose of this study was to investigate the effects of interviewer's
controlled attention and interest, manifested by physical attitude and minimal use
of such verbal responses as "I see", "mmh-hmm" on the duration of talk on the
specific content (topics), when S was free to select any one among a fixed number of
topics.
METHOD
Experimental design and procedure. The general experimental plan used a modi-
fication of a structured interview procedure reported elsewhere^''- ^K The first por-
tion of each interview consisted of a 12 minute pre-training phase (Phase I) during
which a three minute speech sample was obtained on each of four topics. Four topic
sequences (ABCD, BCDA, CDAB, DABC) were replicated by two jE's. Either sub-
set of two topics (AB or CD) was reinforced for alternate Ss, keeping constant the
time allotted to reinforced and non-reinforced topics. Topics A and B were combined
under the subset interpersonal, and topics C and D made up the subset self-exploring.
Following Phase I, S was directed to select freely any of the topics already discussed.
^This study was supported in part by Research Grant M-3700 from the National Institute of
Mental Health, U. S. Public Health Service. The authors wish to acknowledge the participation of
A. R. MarstoD in the statistical analysis of the data.
MINIMAL SOCIAL REINFORCEMENT AND INTERVIEW CONTENT 211

In Phase II (Free period) reinforcement for the same topic subsets was continued,
and S was allowed to go on until she completed the topic started after the ninth
minute of the free period.
The major interest of the study was in the effectiveness of minimal social re-
inforcement in conditioning the duration of communication on preselected topics.
Furthermore, it was of interest to test the generality of any demonstrated effects
over different topics, different Es, and topic sequences. Therefore, the main statisti-
cal analyses used a difference score of the total amount of time devoted to reinforced
topics minus the amount of time devoted to non-reinforced topics during Phase II
as each S's measure of conditionability. Further analyses of variance were carried
out to examine the effects of topics and Es on these conditionability scores. For
theoretical reasons it is useful to differentiate between the amount of time that S
spends on a topic even though he may not be talking continuously, and the absolute
amount of time spent in talking on a topic. Parallel analyses were carried out with
each of these measures: (a) total time on a topic, and (b) talk time, i.e., total time
minus silence.
Several secondary hypotheses were evaluated in an analysis of the relationship
between conditionability and awareness, and the correlation between condition-
ability and total talk time. In order to assure that Es followed instructions during
the pre-training Phase (Phase I), analyses were also carried out to demonstrate that
variations in pre-training could not be held to account for subsequent reinforcement
effects.
Subjects. Thirty-two female undergraduates participated in this study. All
were unmarried and under 22 years of age. In a preceding pilot study each E gave
ten practice interviews, using the same procedure. Data from these Ss were not used.
The interview procedure. Each S reported to E's office and was introduced by the
secretary. S was told that the purpose of the research was to study different tech-
niques of interviewing in order "to find out how we can conduct interviews best, so
that we can improve our methods". S was asked to talk freely about anything which
came to mind about a topic posed to her by E. She was asked to talk about the
same topic until E suggested a new one. S was assured of the confidentiality of the
interview. S was told that the sessions would be tape-recorded.
Prior to Phase I each S was gi\ en a five minute warm-up topic. On a basis of
the pilot research the warm-up topic selected consisted of questions about her cur-
rent school work, her attitudes about university life, etc. None of these questions
showed any overlap in content with the experimental topics given in Phase I. During
the warm-up S learned that E did not participate freely in the conversation and that
she had to continue talking about the suggested topic.
In Phase I, E asked each of the following questions in predetermined order,
allowing at least three minutes of uninterrupted speech by S. At the end of each
topic E made a general statement about the topic, identical for all Ss. The topics
were presented verbatim as follows:
A. Getting along with your parents. What do you feel and think about them?
B. Your dating experiences. What do you feel and think about them, how do
you get along with men?
C. Your hobbies and sports. What do you enjoy doing most in your spare
time?
D. Your ambitions in life. What do you want to accomplish in the next few
years and how do you feel about it?
Minimal reinforcement consisted of a practiced, intentional posture of interest,
manifested by smiles, by attending to S, and by the phrases "I see", "mm-hmm",
and "yes". E refrained from any other verbal comments and gave verbal reinforce-
ment whenever appropriate in the natural flow of the interview. On non-reinforced
topics, E attempted to minimize cues of body movement, changes in facial ex-
212 FREDERICK H. KANFER AND JOHN F. MC BREARTY

pression, or other attention-responses. E made no verbal response on the non-


reinforced topics.
After the pre-training each S was given the following instructions; "Now we will
go over these topics again, but this time I will make it easier for you. You can select
any of these topics and talk about them in any order that you like, and in any way
you wish. You may shift from one topic to another; but when you do, please call
out the letter preceding the topic so that I can follow you. I have written these
questions down on separate cards and you will notice that each question has a letter
before it. 1 want you to go ahead and talk as before, except that now you can shift
these topics around whenever you wish. I'll let you know when our time is up. Now
here are the topics to choose from."
S was handed four randomly ordered cards with the same questions as in Phase
I, preceded by a letter which had been randomly associated with each topic. S was
allowed to continue to talk until she completed that topic which had been started
after the ninth minute of Phase II.
Post-experimental interview. E indicated that the interview was completed and
attempted to get S's inipression about the interview and her awareness of its purpose.
S was asked which topic she had enjoyed most and which topic she found hardest to
talk about. She was also asked whether she felt that E was equally interested in
everything, or whether he tended to be more interested in one topic than another.
If S indicated uneven interest by E, she was asked in what way this difference in
interest had become apparent to her. Finally, S was asked whether she thought that
the interview helped her to organize some things differently than she had before. If
S said yes, she was asked on which topics these effects were noted.
Data-recording. All interviews were tape-recorded and data were obtained by
playing back the tapes. Talk time was measured from onset of a continuous utter-
ance by S to its termination. Silence constituted any failure by S to emit any verbal
response for an interval of two seconds or more. Duration of silence describes the
time of non-verbalization including the initial two second pause. Total time on a
topic comprised silence plus talk.
Awareness ratings were scored from S's answers to the post-experimental
questions on a five point scale, as follows: 1. Unaware of E's interest. 2. Aware of
E's general interest only. 3. Aware of unequal interest by E but cannot state topics
reinforced or presents false hypothesis. 4. Aware of unequal interest by E and men-
tions one topic which was reinforced and/or one which was not. 5. Aware of unequal
interest by E and names reinforced and non-reinforced topics correctly.
RESULTS
Reliability of coding."^ The dura,tions of talk, silence, and total time spent on
each topic, were tabulated from indivdual tapes. In order to test the accuracy of
these coding operations, two recorders independently tabulated all measures on a
sample of 38 verbal units, derived from 3 randomly selected Ss. For duration of talk
a product moment r of .99 was obtained. For silence and total time spent the r's
were .91 and .98, respectively. These results indicate adequately high reliability
for the coding. Awareness ratings were scored from S's answers to post-experimental
questions. Two independent raters scored these responses along a five-point scale
for 16 randomly selected Ss. Perfect agreement was obtained in 14 cases.
Phase L The training phase allotted at least 3 minutes for each topic. In order to
ascertain whether Es had shown bias in terminating or prolonging exploration on any
topic during Phase I, <-tests were run to compare total time on reinforced vs. non-
reinforced topics, subsets Interpersonal vs. Self-exploring, and Ei vs. Ej. None of
the ^-values were significant. Thus, Ss were given equal time in Phase I by both Es,
and on all topics, regardless of content or reinforcement condition.
'The authors wish to acknowledge the assistance of Beth M. WoUand and Marcia M. Bradley
who coded the tapes and rated awareness.
MINIMAL SOCIAL REINFORCEMENT AND INTERVIEW CONTENT 213

Although the mean duration of total time per topic did not differ in Phase I,
the reinforcing operations might have affected the proportion of time spent in silence
by Ss on different topics. In order to determine whether Es and topics differentially
affected the distribution of silences, an analysis of variance was run on arc-sine
transformations of the silence quotient differences
silence on non-reinforced topics silence on reinforced topics
total time on non-reinforced topics total time on reinforced topics
On these quotient differences an F ratio of 9.16 with 1, 28 df (p < .05) was found for
topic-subsets (Interpersonal vs. Self-exploring). The F ratios for Es and for the
interaction of topic-subsets X Es were not significant (F < 1). This analysis in-
dicated that on interpersonal topics minimal social reinforcement by either E re-
sulted in relatively less silence, while on Self-exploring topics it produced more
silence.
Phase IJ. The major hypothesis predicted an increase of verbal communication on
all reinforced topics, regardless of content or interviewer. The mean duration of talk
on non-reinforced topics was 329.1 sec, SD 101.46. On reinforced topics the mean
duration of verbal output was 429.8 sec, SD 169.23. A t-iest for correlated measures
yielded a t of 2.52, p < .02. The conditionability scores (time on reinforced topics
minus time on non-reinforced topics) were subjected to an analysis of variance. No
differences among Es (F < 1) and over topic-subsets (F = 2.60 with 1, 28 df,
p > .10), nor in the interaction (Es X Topic-subsets) were found. These analyses
indicate that minimal reinforcement resulted in a significantly higher verbal output
and that this effect was not significantly affected by the individual characteristics
of E and his demeanor, nor by the specific topics selected for reinforcement.
In contrast to Phase I it was possible for the effects of reinforcement to manifest
themselves in Phase II either in an increase of verbal output, or simply in an increase
in the total time spent on the topic with increased silence but no increase in verbal
output. A t-tesi for correlated measures indicated that silences did not significantly
differ between reinforced and non-reinforced topics {t = .18). As in Phase I an
analysis of variance of the arc-sine transformations on silence quotients was carried
out. The results showed no significant differences among Es and topic-subsets. A
t-ie&t also compared the total time (talk plus silence) on reinforced and non-rein-
forced topics. The significant value {t = 2.13, p < .05) indicated that more time
was spent on reinforced topics. Since talk-time and total time are correlated, this
finding taken together with the analysis of silence, essentially confirms that the
greater time spent on reinforced topic was mainly spent in talking rather than in
silence. An analysis of variance on the conditionability scores for total time yielded
no significant F values for Es, topic-subsets or their inter-action, demonstrating that
total time was also unaffected by E's individual contributions and by topics.
In Phase I four different topic sequences were employed. Since onset of re-
inforcement may have differential effects as a function of the position of the topic
for which it was first introduced, a simple analysis of variance was run on condition-
ability scores, separated by topic sequences. An F of 0.49 indicated that sequence of
reinforced topics in Phase I had no effect on differential duration of verbal output
in Phase II.
An incidental question of interest is the correlation between S's ease to condi-
tion and the total time talked, A product moment r of 0.498 (F = 9.9, p < .005)
was obtained between conditionability scores and total verbal output in Phase II.
This correlation indicates a significant relationship which cannot be interpreted
unequivocally in the present study. It suggests either that Ss who condition more
easily also talk more; or that effective reinforcement on some topics may also general-
ize to increase verbal output on all topics.
E's verbal responses. Since a separate record of the verbal components of the
reinforcing operations was available, additional information about the role of verbal
214 FREDERICK H. KANFER AND JOHN F. MC BREARTY

cues in this study could be obtained. A product moment r between number of verbal
responses by E in Phase I and conditionabihty scores was obtained. An r of 0.157
(p > .10) indicated that the number of E's verbal responses during Phase I did not
relate significantly to S's differential output in Phase II. For Phase II an r of 0.769
(p < .001) showed a significant relationship between E's verbal responses and in-
creased interview productivity on reinforced topics. The latter correlation reflects
the fact that increased talk on reinforced topics in Phase II, i.e., conditionability,
also provided greater opportunity for reinforcement by E. On the other hand, the r
of 0.157 for Phase I is free of this confounding factor.
Awareness and conditionability. The average awareness rating for all Ss was
2.3. Only two Ss obtained a score of 5, while ten Ss were given the lowest rating.
This suggests that Ss were generally unable to indicate the response class for which
reinforcement was given. 19 Ss were not even aware of E's "unequal interest" (score
3). An analysis of variance on awareness scores indicated that neither Es nor re-
inforcement on either topic-subset significantly affected awareness. A product-
moment r between conditionability scores and awareness of —0.09 also indicated
that ease of conditioning and high awareness scores did not show any consistent
relationship.

DISCUSSION
Our findings clearly indicate that minimal social reinforcement results in in-
creased communication on those topics for which it is given. Since the response class
used here consisted of a wide variety of heterogeneous verbal responses which shared
only the property of reference to a specific topic, the results demonstrate the feas-
ibility of expanding a reinforcement-anchored framework to the manipulation of
complex behavioral classes encountered in clinical interviews.
The exact nature of our reinforcing stimuli is difficult to describe. In contrast
to laboratory studies ^^^ with restricted response classes, more subtle and possibly
equally powerful cues have been used here. The attitudes and cues which we have
described as minimal social reinforcers have been treated by Skinner ^'^ as important
generalized reinforcers. The correlations between number of E's verbal responses
and conditionabihty do not support the suggestion that it was solely the verbal
component of the treatment which affected Ss' increased productivity. At least in
the present study the reinforcing operations, similar to those described for a "per-
missive" atmosphere, appeared to have been somewhat more powerful than use of
verbal responses alone. Replies on the post-experimental questionnaire further
attested to the importance of the attitudinal component in E's behavior. Of those
Ss who were aware of E's "unequal interest" several said that this was conveyed by
some class of non-verbal responses, e.g., E's interested look, smile, or E's indication
that "he seemed to understand me". As researchers focus their attention on inter-
viewer actions, objective recording of all components of E's behavior become in-
creasingly important.
Observation of Ss' behavior suggested the possible influence of personality
variables on sensitivity to reinforcement. Both Es noted that some Ss interpreted
the instructions to indicate that they were to talk "as if E were not present at all."
These Ss tended not to look at E, and therefore were less susceptible to the reinforc-
ing operations. If such differences in the interpretation of the instructions or in S's
set can be demonstrated, it would suggest that a therapist's structure of his expecta-
tions for the therapeutic sessions may have strong effects on the patient's output.
For example, simply because it minimizes the patient's exposure to the therapist, the
psychoanalytic free association method would predictably differ from a face to face
counseling procedure in the extent to which the therapist's behavior would affect
the specific content of the client's communication from moment to moment. This
method, in fact, represents an early awareness of the importance of minimal social
reinforcement in determining interview content.
MINIMAL SOCIAL REINFORCEMENT AND INTERVIEW CONTENT 215

For experimental purposes the comparison between reinforcing operations and


their absence was carried out. It is obvious that non-reinforcement, expressed by E's
silence and his failure to respond in "accepting" ways might better be considered as
potentially aversive. Silence has also been treated by others ^^' as punishment rather
than simple absence of reinforcement.
SUMMARY
This study investigated the reinforcing effects of minimal social cues by E on
interviewees' duration of talk on selected topics. Thirty-two Ss were given a semi-
structured interview in Phase I and talked freely for three minutes on each of four
topics. On two topics E showed interest and attention and made brief verbal agree-
ing responses. In Phase II, S was free to talk about the same topics. E continued to
reinforce the same topics. The procedure was replicated by two Es and with re-
inforcement for either subset of two topics.
The results indicated that only silence quotients in Phase I differed significantly
for topic sets. Reinforcement increased duration of silence on Self-exploring topics
while it decreased silence for Interpersonal topics.
In Phase II, duration of talk on reinforced topics significantly exceeded verbal
output on non-reinforced topics. These results were not affected by topics or Es.
They were similar for duration of speech and total time spent on a topic (talk plus
silence). Awareness ratings yielded evidence that most Ss could not verbalize the
response-reinforcement contingencies. The ratings were not affected by the in-
dividuahty of Es, nor by the topics which were selected for reinforcement. The data
support the contention that reinforcement-oriented behavior theory can be extended
to cover clinical interviews, if such extrapolations continue to be substantiated at
the same level of complexity as the phenomena to which the theory is to be applied.

REFERENCES
1. FRANK, J. D. Some effects of expectancy and influence in psychotherapy. In Masserman, J. H.
and Moreno, J. L. (eds.). Progress in Psychotherapy. New York: Grune and Stratton, 1958, Vol.
3, 27-43.
2. KANFBR, F . H . Verbal rate, content, and adjustment ratings in experimentally structured inter-
views. J. abnorm. soc. Psychol., 1959, 58, 305-311.
3. KANFEK, F . H . Verbal rate, eyeblink, and content in structured psychiatric interviews. J.
abnorm. soc. Psychol., l%0, 61, 341-347.
4. KiiASNEU, L. Studies of the operant conditioning of verbal behavior. Psychol. Bull. 1958, 56,
148-170.
5. MURRAY, E . J. A content analysis method for studying psychotherapy. Psychol. Monogr., 1956,
70, No. 420.
6. SALZINGER, K . ; and PISONI, S. Reinforcement of verbal affect responses of normal Bubjects dur-
ing the interview. / . abnorm. soc. Psychol, I960, 60, 127-130.
7. SKINNER, B . F . Science and Human Behavior. New York: The Macmillan Company, 1953,
pp. 402.
8. STEVENSON, I. The psychiatric interview. In Arieti, S. (ed.). American Handbook of Psychiatry.
New York: Basic Books, 1959, Vol. 1, 197-214.

You might also like