_____________________________________________________________________________________________________________

Fam Proc 11:69-94, 1972

Empirically Unbinding the Double Bind: Review of Research and

Conceptual Reformulations
DAVID H. OLSON, PH.D.a
aAssociate Professor, Dept. of Family and Community Development, University of Maryland. Also, Family Development Section, Child
Research Branch, NIMH.

This article reviews the empirical research on the double-bind concept and describes the conceptual and
methodological limitations of the studies. Because of the difficulty in defining and operationalizing the concept, most
studies have not adequately tested substantive issues related to the double-bind hypothesis. The relationship of the
concept to other family dynamics concepts is discussed and a reformulation of some of the basic tenets are presented.
The article also attempts to clarify several of the theoretical and methodological issues related to the double-bind
hypothesis, and specific recommendations are made regarding future research on this topic.
One of the major problems with the double-bind concept generally is that it denotes and connotes so many things that
those attempting to unbind the concept find themselves unknowingly bound by it. The following is an illustrative example
of the way the concept can be misinterpreted and misused to describe anything and everything. The irony of the situation to
be described is that those investigating the double bind have been caught in a double bind (not, however, a true double
bind).
In the original double bind article by Bateson, et al, (2), the following three conditions were defined as creating a double
bind situation. First, "the individual is involved in an intense relationship; that is, a relationship in which he feels it is vitally
important that he discriminate accurately" (p. 254). A person doing research on the double bind is in a situation that
demands that he discriminate accurately what is being said by the theorists so that he can interpret and, thereby, investigate
the concept appropriately.
The second condition states that "the individual is caught in a situation in which the other person in the relationship is
expressing two orders of message and one of these denies the other" (p. 254). The theorists are first of all saying on one
level that the concept need not be tested because it is self-validating. In this regard, Bateson (4) stated that: "the theory itself
is highly abstract and, to this extent, is itself likely to be self-validating.... It thus becomes excessively difficult to test the
premises or the theory against empirical fact.... Personally I do not believe that the theory is at present subject to rigorous
empirical testing.... At best it can be vividly exemplified or illustrated by the phenomena of schizophrenia, humor, religion,
art, and the like" (p. 416-7). On a second level, however, there is the implicit message that more and better empirical
research should be done to test and validate the concept. In general, the second condition for a double bind has been created
for on one level there are explicit statements that the concept cannot be empirically investigated and at the same time an
implicit message on a second level that empirical research should attempt to validate the double-bind hypothesis.
The third and final condition for the situation to be binding is that "the individual is unable to comment on the messages
being expressed to correct his discrimination of what order of message to repond to, i.e., he cannot make a
meta-communicative statement" (p. 254). It is at this point that one can see the possibility of escaping from the double-bind
situation. First let me paraphrase what theorists of this concept have described as the defensive reactions of schizophrenics.
One could respond as a paranoid schizophrenic and react to this situation in a defensive manner to see how the situation is
constructed to destroy him. Or one could laugh off the double-bind situation as a hebephrenic does, or ignore the situation
and withdraw like a catatonic. However, as postulated by the double-bind hypothesis, these types of reactions to
double-bind situations are precisely the reason why schizophrenia develops and continues to persist. Other possible
reactions to this double-bind situation include reacting to only one of the messages and, thereby, either proceeding on the
assumption that valid research is desired or accepting the statement that it is impossible to test the concept. Another
alternative is to respond with incongruent messages that create further double binds. The most reasonable reaction,
however, is to unbind the situation by meta-communicating, i.e., communicating about the communication. More
specifically, this means that one makes an explicit comment about the mutually exclusive messages and requests
clarification. This later alternative will be the primary objective of this review of the empirical investigations of the double
bind.
According to the original hypothesis advanced by Bateson (2) and his associates in 1956, one would conceive of this
double-bind situation as one in which the research investigators are the "victims." But in accord with a revised
conceptualization of the double bind proposed by Watzlawick (26) there is no binder and bound but rather two victims.
This latter formulation is a more accurate description of what has happened in the relationship between the theorists and the
research investigators. The two victims in this case are those who have developed the concept and are hoping for empirical

_____________________________________________________________________________________________________________

support and those investigating the concept empirically who have had a difficult time operationalizing the concept and have
been unable to find such empirical support.
Ideally, the relationship between theory and research should be one that is reciprocal and mutually beneficial. Theoretical
formulations, such as the double bind, should direct one to yet unobserved phenomena and relationships and stimulate
research in these areas. Research ideally should insure that these formulations are operationalized and then systematically
test the postulated relationships. If the hypotheses are not supported, the findings should then be used in reformulating and
refocusing theory. Unfortunately, the two groups have been working with little apparent concern for the developments of
each other. Hence, rather than mutual gains, both have become victims of the very concept, the double bind, they have
attempted to investigate.

REVIEW OF LITERATURE
This paper will now review what has been found by those who have attempted to investigate the concept empirically.
The first and foremost task of research is to translate the phenomena into some type of operational form that can be studied,
without altering the nature of the concept so as to make the research irrelevant to the theory. This is a deceptively simple
task, as was found by Mishler and Waxler (17) in their intensive study of family interaction in schizophrenics. After
attempting to investigate the double-bind hypotheses, they reluctantly concluded that "We were not able to develop a direct
measure of double binds because there seemed no way of establishing clear criteria consistent with the original definition.
At best we were able to indicate that certain comparisons, such as between results for direct and indirect measure of affect,
seemed not to be inconsistent with the idea. However, this did not permit using our results as a direct test of the hypothesis
that the patient families would show a higher rate of double binds" (p. 274). They further stated that in regard to the double
bind "there is lack of precision and clarity ... that presents serious difficulties for an accurate understanding of the types of
interaction sequences that do and do not fall within the definition of the double bind. From the way the concept is used, it
sometimes appears that all communication sequences may be interpretable, at some levels of analysis, as double binds, and,
if this be so, the concept loses all usefulness" (p. 15). Bateson (4) also acknowledges that "They say with some justice that
the phrasings of the theory are sometimes ambiguous. They might have gone further and said that (like much of
psychoanalytic theory) the double-bind theory of schizophrenia is slipperyso slippery that perhaps no imaginable set of
empirical facts could contradict it" (p. 415).
In spite of, or perhaps because of, the serious conceptual difficulty in attempting to investigate this concept, several
research studies have been concerned with testing the validity of this phenomena. Although numerous publications have
been devoted to illustrative case descriptions and therapeutic uses of the concept, only quantitative studies will be reviewed
in this paper.
In general, these empirical studies have failed to develop a conceptual or operational scheme that accurately reflects the
double-bind hypothesis. In fact, several studies purporting to measure the double bind have used measures that seem to
have little relationship to the original concept. In addition, these studies often have included numerous methodological
problems that make the studies even more difficult to interpret.
Basically, there are two types of empirical studies that have attempted to investigate the double bind. One approach has
attempted to assess the extent to which the parents of schizophrenics send double-bind messages. They have predicted
these parents will create more double-bind situations than will parents whose children have not become schizophrenic. The
second approach has been to determine the schizophrenics' ability to discriminate double-bind messages and has predicted
they would do more poorly than other groups.

Parents of Schizophrenics and Double Bind Messages

There are several studies that have attempted to assess whether parents of schizophrencis create more double binds than
other parents. Beavers and his associates (7) interviewed nine mothers of hospitalized schizophrenic patients and nine
mothers of non-schizophrenic patients in order to test the hypothesis that the schizophrenic mothers would communicate
feelings in a more ambiguous manner during the interview. The taped interviews were transcribed and scored for the
number of definite responses, evasions, and shifts of meaning. The two groups were found to be significantly different on all
three variables with the control group having a higher number of definite responses and the mothers of the schizophrenics
having significantly more verbal shifts and evasion. Combining the number of shifts and evasions, it was possible to clearly
differentiate between the two groups. These findings indicate that mothers of schizophrenics do communicate in a more
elusive fashion as might be deduced from the double-bind hypothesis. However, the conceptual leap of counting the
number of shifts and evasions during an interview as adequate measures of the double bind is questionable.
One study that came closer conceptually to measuring double binds was done by Berger (8). He developed a
questionnaire composed of thirty double-bind statements. Subjects were asked to rate each statement on a four-point scale
in terms of how frequently they recalled their mother made these comments. The four groups of subjects consisted of
diagnosed schizophrenics and three control groups, i.e., patients not diagnosed as schizophrenic, hospital attendants and
2

_____________________________________________________________________________________________________________

kitchen personnel, and a group of college students. All groups were composed of white males between the ages of sixteen
and thirty-five. The schizophrenic group was found consistently to have recalled a higher number than the control groups,
but a significant difference was found only between the schizophrenics and the college students. This study, however, has
the serious shortcoming that it asked subjects to recall the extent to which these statements were made by their mothers.
Research by Yarrow, Campbell, and Burton (32, 33) have indicated the serious inadequacy of the retrospective method so
as to seriously affect the validity of this study.
Another more direct attempt to investigate this question was conducted by Sojit (24, 25). This research deals less
directly with the communication patterns between a parent and child but focuses on the husband and wife interaction. The
assumption is all parents create double-bind situations, but they develop various ways of coping with these situations and,
thereby, provide differing models for their children. It is further assumed that these parental styles relate to the type of
pathology presented by the child. The first study (24) had a total sample of 46 parents consisting of 20 parents who had
children with ulcerative colitis, 8 with delinquent children, and two sets of controls 9 with normal children and 9 with
children having cystic fibrosis. The second study (25) added 8 families in which one child was schizophrenic to the 46
parents used in the previous study (24).
The method used for measuring the double binds was to have the husband and wife discuss the "correct" meaning of a
proverb having two valid, but mutually exclusive, interpretations. Although this approach is characterized as fulfilling all
the essential conditions of a double-bind situation, there are serious conceptual and methodological problems with this
measure. The criteria of conflicting statements of different logical types is supposedly measured because the instructions
require that they determine the "correct" meanings of the proverb, of which there are two mutually exclusive possibilities.
Unfortunately, this measure has only superficial correspondence to a real double-bind situation because it does not deal
with issues that threaten the continuance of the relationship or have any emotional relevance. In addition, the data was
based only on the husband and wife discussion regarding these interpretations of the proverb and not their actual interaction
with the child. Lastly, there is some question regarding the statistical methods used with such small sub-samples and the
general data analysis.
In spite of the lack of an adequate measure of the double bind and other methodological limitations of the research, it was
found that parents of schizophrenics did give significantly more invalid interpretations of the proverbs and disaffirmation of
their own or their spouse's messages compared to all the other groups, except the delinquents (25). Also, there seemed to
be some logical correspondence between the parental style of relating and the type of problem exhibited by the child.
A more rigorous test of a similar hypothesis was done by Beakel and Mehrabian (6). They videotaped the parent-child
interaction in five families with a mildly disturbed adolescent and five families with a severely disturbed adolescent. They
then compared the parents' verbal communications with those that were non-verbal, i.e., postural. While it was predicted
according to the double-bind hypothesis that the parents of the severely disturbed child would make more statements in
which the messages were in conflict, this hypothesis was not supported. But the groups were significantly different in that
the parents of the severely disturbed children communicated significantly more negative feelings.
The study by Ringuette and Kennedy (20) attempted to provide a further test of the hypothesis. Previous work by
Weakland and Fry (29) indicated that letters from mothers of schizophrenics to their children contained good samples of
double-bind messages. Ringuette and Kennedy, therefore, attempted to determine whether persons with varying amounts of
experience with the double-bind concept were able to reliably rate double-bind messages contained in letters. Twenty
letters that hospitalized schizophrenic patients had received from their parents were obtained, twenty letters from parents of
non-schizophrenic patients, and twenty letters written by hospital volunteers as if they were writing to a hospitalized child.
Five types of judges were used and they included: (a) an expert group that was closely involved in the development of the
concept; (b) a group of first-year psychiatric residents trained in the double bind; (c) experienced clinicians untrained in the
double bind; (d) experienced clinicians trained in the double bind; and (e) a naive group untrained in the social sciences.
Each group was composed of three individuals that were asked to rate the sixty letters on a seven-point scale representing
the degree of double bind contained in each letter.
The findings clearly demonstrated that there was considerable difficulty in identifying double binds, for the average
interjudge reliabilities were consistently very low; i.e., expert group = .19; trained resident group = .26; untrained clinicians
= .13; trained clinicians = .44; and naive group = .39. Not only did these groups of judges fail to agree among themselves,
but none of the groups of judges was able to differentiate between the letters received by schizophrenics and
non-schizophrenic patients. In terms of the hypothesis, the experts and trained group also failed to differentiate between the
volunteer and patient letters, although this was done to a limited extent by the other three groups of judges. The failure of
the experts to agree poses a serious question regarding the conceptual definition of the double bind. The writters concluded
that the reasons for such negative findings indicate that: "(a) double-bind communication is not present in letters, in which
case a postulate of the theory is clearly invalid; (b) it is not presently a measurable phenomenon; (c) it actually does not
exist" (20, p. 141).
In summary, all of these studies are plagued with serious conceptual and methodological limitations that make their

_____________________________________________________________________________________________________________

findings rather difficult to interpret and of questionable validity. In regard to the hypothesis that parents of schizophrenics
send more double-bind messages than other parents, three of the studies reviewed (7, 8, 25) did tend to find indirect
supportive evidence. Beavers (7) did find that mothers of schizophrenics communicate during an interview situation in a
more elusive fashion than other mothers. Berger (8) found that male schizophrenics do recall their mothers making more
double-bind messages than the control groups. Sojit (25) found that parents of schizophrenics do communicate with each
other in a more elusive manner than the control parents by disaffirming the content of their own and their spouse's
messages. However, the two studies (6, 20) that had fewer methodological limitations did not provide evidence supportive
of the hypothesized relationship. Although the study by Beakel and Mehrabian (6) did not include schizophrenic families,
they failed to find that parents of seriously disturbed children produced more communications in which the verbal and
non-verbal messages were in conflict than the control families. The Ringuette and Kennedy (20) study discovered that
expert judges were not only unable to agree about the frequency of double-bind messages contained in letters, but they were
unable to differentiate between letters from schizophrenic mothers and control mothers. In conclusion, there appears to be
very limited support for the hypothesis that parents of schizophrenics send more double-bind messages than other families.

Double-Bind Discrimination in Schizophrenics

The second approach empirical investigators have taken to test the double-bind hypothesis has been to determine if
schizophrenics themselves are less able than other individuals to discriminate double-bind messages. One of the first to
attempt to test this hypothesis was Ciotola (9) who conducted a study in which schizophrenics and a control group were
compared as to their reaction time to the discrimination of auditory tones when the task was made virtually impossible. The
hypothesis that schizophrenics would show longer reaction times was not supported. There is, however, some question
whether this is conceptually a valid test of the double-bind hypothesis anyway. In this regard, Watzlawick (26) has
criticized a parallel research study with rats in which an ellipse was gradually made more circular and a circle more
elliptical so that discrimination of the two became impossible. He felt that: "The double bind is not a failure in
discrimination" (p. 137). In addition, there is some question whether the hypothesis that schizophrenics would have a
longer reaction time can be legitimately predicted from the double-bind phenomena.
Another ostensible study of the double-bind hypothesis was done by Potash (19) in which he utilized the prisoner's
dilemma situation in a two-person, three-choice game. Subjects were male schizophrenics and a matched control group
composed of hospital employees. It was hypothesized that the schizophrenics would make more withdrawal responses to
the game such that they would not take the risk of maximizing their gain. Contrary to what they hypothesized, the
schizophrenic group did not show increased withdrawal responses to the double-bind task. Again, the conceptual
relationship of the prisoner's dilemma task for testing the double bind is not clear. This was more a study of trust and
cooperation in a non-zero sum game situation than a direct test of the double-bind hypothesis.
A more rigorous test of this double-bind hypothesis was done by Loeff (12). He tape-recorded 24 happy and 24 unhappy
statements in which the affect was varied in three ways to represent: (a) neutral affect; (b) appropriate affect, and (c)
conflicting affect. The subjects were 24 normal, 24 delinquent and 24 reactive schizophrenic adolescent girls. They rated
each statement on a scale indicating the degree of agreement between the voice and content, and completed a sincerity scale
and a semantic differential scale. It was found that all three groups were equally able to discriminate the conflicting
messages. However, the pathological groups were more affected by both the content and the affect aspects of the
communication than the control group who tended to miss (i.e. avoid, repress) the conflicting messages and placed greater
emphasis on the content component. These findings do not support the interpretation of the double-bind formulation which
predicts less discrimination by pathological groups.
However, results from a somewhat related study by Mehrabian and Wiener (13) did not replicate Loeff's findings that
normal subjects place greater emphasis on the content component of messages. Mehrabian and Weiner investigated the
relative influence of the content and tone on communication, using only a normal sample. One group of ten subjects heard
only the content component in which the main vocal frequencies were blocked out. Another group of ten heard the tonal
component but the content was unintelligible. The third group of ten heard both the content and tonal components. They
found that when the two components were both positive or both negative, the total message was judged more positive or
more negative than when either the content or tonal component was judged separately. However, contrary to what Loeff
found with his normal sample, when the two components were discrepant, the tonal component played by far the most
prominent role in the determination of the final message received. For example, when a negative tone was used to present a
positive word such as "dear," the total message was evaluated negatively.
Not only does this study by Mehrabian and Weiner lead one to question the results of Loeff's study, but it also challenges
two assumptions commonly made regarding the double-bind hypothesis. According to Mehrabian and Weiner's
interpretation of the double-bind hypothesis, it was first of all assumed that inconsistent components of a message are
decoded separately. Secondly, it was assumed that responding to the two conflicting messages is what creates the double
bind. The results of their study, however, do not support either of these assumptions. They found that inconsistent

_____________________________________________________________________________________________________________

components in a message are not decoded separately but in a holistic manner such that if the tonal and content aspects are
in conflict, the tonal component determines the final message. For example, praising a person in a negative tone, such as in
saying "Thanks a lot," communicates a negative message, i.e., sarcasm. It should be remembered, however, that this
research was not done with schizophrenics, and so this study should be replicated with a patient population. A crucial
factor, which they failed to take into account, is the relationship component of the double bind. If two messages from two
different channels, i.e., verbal and non-verbal, contradict each other, this is not necessarily an example of the double bind
unless there is an indication of some relationship component present in either of these messages. As Wynne (31) has stated:
"Contradictions at a single level of commmunication, both of one logical type, for example in parallel but different channels
of communication may sometimes reverberate at the relationship level, but more often are simply examples, not
invalidations, of the complex character of the relationship" (p. 4).
Three of the four studies just described did not provide an adequate test of the hypothesis that schizophrenics have
greater difficulty identifying double binds. The studies by Ciotola (9) and Potash (19) used research methods that do not
adequately relate conceptually to the double-bind phenomena, and the study by Mehrabian and Weiner (13) used only
normal subjects. The only study that most adequately tested the hypothesis that schizophrenics have greater difficulty
discriminating double-bind messages was done by Loeff (12), and his study did not support the hypothesis.

Summary of Empirical Studies

This review of the research literature has revealed that the present studies have not provided a very good test of the
double bind. In addition to the many methodological problems in these studies, most of them failed to develop a conceptual
or operational scheme that accurately reflected the double-bind phenomenon. By the time the concept was operationalized
and an instrument was devised to measure it, the final measure had little resemblance to the actual double-bind
phenomenon. So while the results are generally negative, the double bind still has not been rigorously tested.
There are two studies, however, that are particularly significant because they do provide the most rigorous tests of the
double bind to date. First there is the study by Mehrabian and Weiner (13) challenging an often made assumption about the
double bind that conflicting components of a message are interpreted separately, thereby creating a double bind. They
found, however, that when they presented conflicting message components to normal subjects, the final message
communicated was not conflicting or confusing.
The second study that bears further attention is by Ringuette and Kennedy (20) in which they found that there was very
little agreement within various groups of individuals. Because of this finding, it it not at all surprising that in most of the
empirical studies cited, a fundamental problem was the difficulty of operationalizing the double bind. In addition, others,
such as Mishler and Waxler, who have tried to utilize the concept in research have had such difficulty operationalizing it
that they did not incorporate it into their study. The predominant difficulty with the present formulation of the double bind is
that it remains so abstract that it is elusive. Abstractness is a necessary component of a theoretical formulation, but
abstractness must also have operational relationships to natural situations to be useful. As Muzafer Sherif (22) has
emphasized, "If divorced from actualities, abstraction becomes a game; it...becomes inner gymnastics for a select group of
people who are 'in' on the secret and exclusive lingo" (1966, p. 4). It appears that, at this time, the double bind continues to
be a concept that even the 'in' group has difficulty agreeing upon, as indicated in the research by Ringuette and Kennedy in
which the reliability among the experts was, indeed, very low. To retreat in a humorous note to the psychologizing of
Charlie Brown, one might not too inappropriately respond to the double-bind concept the same way that Lucy does to
Charlie Brown: "You're a good concept, Double Bind, if only you weren't so wishy-washy."

THEORETICAL AND METHODOLOGICAL CLARIFICATION OF THE DOUBLE-BIND

CONCEPT
The present review has demonstrated that there needs to be increased attention given to clarify some of the
"wishy-washy" qualities of the double-bind concept that have become most apparent when researchers have attempted to
investigate the concept.
Before investigations can proceed further regarding the substantive questions regarding the double bind, it is essential
that the conceptual and operational definitions of a double bind be clarified. In order to more adequately accomplish this
first step, this paper will now specify the relationship of this concept to other family process concepts, describe some of the
central principles of the concept, and then propose extensions and reformulations of the double-bind concept.

Relationships of Double Bind to Other Clinical Concepts

Although the double bind has generally been considered a unique concept, it is clear that it has considerable similarity to
the report-command components of communication proposed in earlier works by Bateson (1) and others (27) and to the
concept of pseudomutuality proposed by Wynne and associates (30). Since all three of these concepts were developed from

_____________________________________________________________________________________________________________

clinical work with schizophrenic families by individuals concerned with communication patterns in these families, it should
come as no surprise that the concepts appear interrelated.
The reason for the relationship between the double bind and the report-command description of communication is
primarily explained by the fact that Bateson was the major contributor in the development of both concepts. Historically
speaking, Bateson first described the report-command idea in 1951 and published the original paper on the double bind in
1954. In describing communication, Bateson (1) felt that in every message there is both a report and command component.
The report component relates the content of the message, whereas the command component defines the nature of the
relationship between the two parties. In instances where the relationship between the two parties is established and clearly
defined, the relationship component plays a less significant part in the interpretation of the message and, consequently, the
content can be more accurately understood. Conversely, in cases where the relationship is not clearly delineated, the content
of the messages often can be interpreted in a variety of ways, thereby increasing the possibility of misinterpretation.
Therefore, within this model it is important to have the type of relationship sufficiently defined so as to minimize the
possibility of misinterpretation.
The similarities between the report-command model and the double bind-concept now become more apparent. If we
consider the report-command as two separate components that represent two different logical types, we have the
essential ingredients of the double bind. What the double-bind concept describes is the situation in which the report and
command components relay conflicting information within the same message. Or, as Wynne (31) has stated: "A
relationship constitutes the context, which is of a different 'logical type' from the component tasks or specific messages that
may be found within the relationship. Double binding involves an impasse in which a relationship is partially invalidated or
contradicted by new events within the relationship" (p. 4); i.e., people are caught in a system of relating in which the
definition of the relationship is brought into question. Logically, this would occur most often when the nature of the
relationship is not clearly defined, which is often the case in the relationships of schizophrenics.
As was previously mentioned, when the relationship component is relatively clear, there is less chance for messages to
be sent or received that will be misinterpreted, and even those that are incongruent with the definition of the relationship
will be interpreted in a way to minimize the chance that they would create a double bind.
The similarity between the double bind and pseudo-mutuality concepts has recently been described by Wynne (31). He
states that the concept of pseudo-mutuality emphasized "the nature of the relationship context and mentioned more in
passing the contradictions between specific messages. Conversely, the double-bind theorists emphasized the contradictions
of messages within the relationship but did not really spell out how these messages fit into the building up and invalidating
the relationship" (p. 5). In other words, while pseudo-mutuality describes the guiding principle that creates and maintains
the schizophrenic and his family in a pseudo-mutual relationship, the double bind describes the communication processes
that help to explain how pseudo-mutuality is maintained in these families.

Meta-Communication and Unbinding Double Binds

In the original paper on the double bind by Bateson (2) it was hypothesized that double-bind situations were a necessary
but not sufficient condition for explaining the etiology of schizophrenia. Double binds were also discussed as a by-product
of schizophrenia. It is also apparent that double binds can occur in any relationship, but it is assumed that the frequency of
this type of communication is so great in certain families that persons begin reacting to all communication as if it were
double-binding. Once this style of relating is learned, discrimination of other types of interaction becomes impaired, and the
person may act in ways that appear schizophrenic. While this process may occur, it should not be assumed that all
double-bind situations actually prove to be binding or necessarily pathogenic. Double binds can be unbound.
The primary way of preventing or unbinding a double-bind situation has been described in the literature as
meta-communication. A description of meta-communication is illustrated in the following case. Imagine two individuals
playing a game according to past rules that were explicit but now have become implicit over time. During the game, one
person violates these rules and they cease playing and begin discussing the issue. This discussion about the rules of the
game is at a different level of abstraction than the actual playing. When the play violated the rules that had become implicit,
it was necessary to point out the discrepancy and make the rules explicit again. This reaffirmation of the rules is a different
type of discussion than the communication that took place while playing the game. Now the discussion involves the process
and rules of the game. Considering the discussion during the game as one type of communication, the discussion of this
communication can be called meta-communication. Let us assume that by meta-communicating, communicating about their
communication, these two individuals were able to resolve their disagreement. This might have been done by learning that
they had different definitions of the rules of the game or by a variety of other ways. What is important is that they were able
to meta-communicate about the conflict that arose during the game and resolve their difficulty.
Using this same model, one can understand how a potential double-bind situation can be prevented from occurring. In
the original double-bind article (2), a situation was described in which a hospitalized male schizophrenic was visited by his
mother. While trying to embrace his mother, she stiffened, pulled away from him, and asked, "Don't you love me any
6

_____________________________________________________________________________________________________________

more?" On the verbal level she was implying love, but non-verbally she was rejecting him. Because of his dependent
relationship to her, he was unable to verbally respond to the double-binding situation she had created. Since he could not
make explicit what he implicitly felt, i.e., meta-communicate, he remained trapped by the double-bind situation.
Sometimes it is easier for a child to pick up these conflicted messages and respond to them than an adult. For example, I
visited the home of a friend who has a five-year-old son with whom I have a mutually close relationship. One day we talked
about his dislike for girls, and I commented that I really liked girls but disliked boys, and, therefore, since he was a boy, I
really disliked him also. After being rather perplexed by my statement for a moment, he then enthusiastically responded,
"You're teasing," whereupon we both laughed. What had happened was that in spite of the conflicting message I had given
him about our relationship, this child was able to respond and point out the incongruence. Because I acknowledged this
discrepancy, we were both able to see the humor in the situation and it, therefore, never proved to be a double-bind
situation.
In general, however, it is more difficult to respond to a double-bind message when it occurs with someone with whom
one has a close and dependent relationship. Because of the significance of the relationship, the possibility exists that
meta-communicating might destroy that relationship. Meissner (14) describes the dilemma faced by someone in a
dependency situation who is confronted by a double-bind message. Emotionally he is caught between the need to
communicate in order to clarify the message and the need to avoid clarifying communication in order to preserve the
relationship. He resolves the dilemma by masking and qualifying his communication so that the possibility of a rejecting
response is diminished. This maneuver diminishes the probability of a clear positive response from the other. It appears
that those who care most about a particular relationship are the least interested in seeing it change and will, therefore,
exhibit all sorts of behaviors, even change their own perceptions (cognitive dissonance) in order to maintain the
relationship as it is. Paradoxically, the more dependent the relationship in which double-bind messages occur, the
greater the resistance to clarifying these messages because of the potential risk of changing or losing the relationship.
There are other ways of responding to a potentially double-binding situation than meta-communicating about the
conflicting messages. Sluzki (23) described four approaches that a respondent could use to avoid a double bind. In addition
to commenting on the incongruent messages, an individual can withdraw, either physically or into silence; he can accept
only part of the message; or he can disqualify the message by a statement or by symptomatic behavior. Weakland (29)
refers to similar ways of escaping the double bind; these are commenting on the incongruity, giving a dual message in reply,
or making a humorous remark about the message. He also points out that although such responses might prove useful, the
sender will often counter by concealing, denying, or inhibiting any responses. This, in essence, minimizes the opportunity
for clarifying the communication.The receiver, therefore, begins to feel less able to trust his own perceptions and might
have increasing difficulty discriminating the messages he receives. His world, therefore, would become increasingly
complex and confusing, and he might begin repressing various experiences and withdrawing into his own world. This is, in
part, the explanation given for the connection between double bind and schizophrenia.

Are Double Binds Necessarily Pathogenic?

Although the relationship between double binds and schizophrenia was the emphasis of the original article and most of
the writings by others to date, Bateson (3) does say that play and humor also contain two levels of messages, but they are
more clearly differentiated than occurs in a double bind. In a recent paper, Bateson (5) emphasizes that, although the
double bind is still pathogenic in some cases, the essential characteristic of the double bind is dealing with the
transcontextual nature of the communication, and this is also a central component of play, humor, and creativity. Since he
feels they share this transcontextual nature, "there is nothing to determine whether a person shall become a clown, a poet,
schizophrenic, or some combination of these" (p. 3). He further points out that in situations where there are messages of
conflicting logical types, "if the pathology can be warded off or resisted, the total experience may promote creativity" (p.
10). Little empirical attention has been given to this orientation regarding double-bind phenomena.
In a recent reformulation of paradoxical experiences by Kafka (10, 11), he suggests that there is an essential relationship
between paradox and individuation. This is so because becoming distinct and socialized involves paradoxical experiences.
Communication with another depends, however, on the degree of correspondence between the range and levels of
abstraction in the two individuals. Kafka proposes that rather than emphasize the ways of preventing and escaping from
double-bind situations, it would be more appropriate to indicate that growth and individuation can take place by learning to
tolerate the paradoxes created. He further proposes that individuals in whom dissociative "splitting" phenomena are
prominent, schizophrenic patients among others, have been less well prepared to integrate paradoxical experiences, and
thus individuation has been limited. He conceptualizes the therapeutic double bind as a useful tool precisely because it
exposes patients at appropriate moments and in appropriate doses to paradoxical experiences, thus belatedly increasing
their ability to cope with and tolerate these experiences.
Along these same lines, the writer also feels that double-bind messages can prove to be beneficial to a relationship. They
can, first of all, create a situation in which a person feels the need for a better understanding regarding the definition of the

_____________________________________________________________________________________________________________

relationship; this can be accomplished by discussing the nature of that relationship directly. Such meta-comunication does
not necessarily mean escaping from the relationship but could entail a clarification and re-negotiation of it. Often such a
possibility is perceived as a threat that might entail the risk of losing the relationship. A second paradox related to the one
previously described is the following: In order to feel free enough to improve a relationship, one must be willing to risk
losing the relationship in the process. Consequently, the more one has invested in a relationship, the less one is willing to
directly discuss or change that relationship, even though this is often necessary if the relationship is to remain vital or grow.
If this does not occur, the relationship stands the risk of becoming increasingly static, routine, and boring. Because
individuals are not willing to take this "existential risk" regarding the relationship, their relationship with others becomes
increasingly predictable and routine. Their interaction style, therefore, becomes very rigid, and the destructive aspects of
their relationships are maintained and even increased. It is at this point in time that the relationship issues need to be
discussed and clarified. Unfortunately, this is something that is very difficult for individuals to do, and it is rarely seen as an
opportunity for enhancing a relationship. A third paradox might, therefore, be that: Relationships that are most in need of
change are the ones most resistant to change, even if the change might mean improving the relationship.

Reformulation of Double-Bind Postulates

In addition to the methodological difficulties encountered by those investigating the double bind, some have falsely
assumed that schizophrenics would have greater difficulty discriminating a double-bind situation (9). However, the
double-bind formulations do not describe the schizophrenic dilemma as one in which they have difficulty discriminating
double-bind situations, but the formulations do emphasize that schizophrenics have difficulty meta-communicating about
these conflicted messages.
Rather than assuming that schizophrenia is associated with a lack of discrimination, the schizophrenic's difficulty could
be that he discriminates the conflicting messages too well. Research to date has not demonstrated that schizophrenics or
other pathological groups are less able to discriminate conflicting messages. Evidence is available, however, indicating that
pathological groups are more affected by both the content and the affective components, whereas the normals tend to miss
the conflicting message because they pay primary attention to the content alone (12). If schizophrenics and other
pathological groups have greater powers to discriminate these types of messages, they would be more susceptible to
double-bind messages when they occur.
The next step is to attempt to understand why normals would tend to rely more exclusively on purely the content of the
messages whereas pathological groups would be attuned to both the affective and content components of messages. This
can be partially understood by using the model previously described, which considers the command and report components
of communication (1). First of all, it could reasonably be assumed that normals would have their relationship with
significant others (command component) more clearly defined. When this occurs, the command aspect becomes less
prominent, and the content (report component), therefore, plays a more significant role in the interpretation of the message.
In pathological groups, where the relationship is not as clearly defined, the command component plays a more important
role in the interpretation of the messages. Since these relationship issues are not generally openly discussed, they are
communicated more frequently in an affective and non-verbal manner. As a result, there would be a greater tendency for
pathological groups to develop a style of interaction with others that is attuned to both the content and non-verbal aspects of
the messages.
Assuming that relationships among family members are less clearly defined in pathological than other families, we might
predict that there will also be a greater number of double-bind messages communicated in these families.
If part of the dilemma is that pathological individuals are more able to perceive double-bind messages and that there are
more such messages, another part of their problem might be that they have great difficulty in communicating their
awareness of the conflicting messages to the sender; i.e., they do not or cannot meta-communicate. This could be because
they are so dependent on the relationship that they are afraid to discuss this issue because it might entail losing the
relationship in the process. It could also be that, if an attempt is made to clarify their feelings, the other person, such as the
mother, might negate this discrimination. As Weakland (29) has pointed out, the sender can refute attempts at
meta-communication from the reciever by concealment, inhibition, or denial. The receiver, therefore, is punished for his
correct discrimination and learns to deny his awareness and sensitivity because of the response of the other. His problems
are made progressively worse because he "tunes in" on these messages more frequently than a normal would in the same
situation and, therefore, is increasingly caught in the double-bind situations. When this style of relating becomes habitual,
increasingly all situations might be indiscriminately seen as double-binding. It is only in these extreme cases that
pathological groups might be less able to discriminate double-bind situations than normals.
Unfortunately, investigations of the double-bind concept to date have not determined whether it is an individual's
inability to discriminate or whether the difficulty is with the meta-communicative process between the sender and receiver.
It would be important first of all to determine the relative ability of various groups to discriminate double-bind messages
and to also assess their relative ability to meta-communicate about these messages.

_____________________________________________________________________________________________________________

GUIDELINES FOR FUTURE DOUBLE-BIND RESEARCH

One of the most significant aspects of the double bind that has rarely even been considered in double-bind research is the
importance of the relationship among the family members. As was previously mentioned, one of the central tenets of the
double-bind situation is that the individual is involved in an intense relationship and is "caught up in an ongoing system
which produced conflicting definitions of the relationship" (3, p. 157). The intensity of the relationship has generally failed
to be considered in these projects, and as a result the idea that the double-bind dilemma is caused by the conflicting and
changing definitions of that relationship has not been seriously considered either.
Wynne (31) has also emphasized that while "true double binds occur only within intense relationships, the authors have
usually not made it clear that one pole, as it were, of the double bind is the relationship itself .... Double binding involves an
impasse in which the relationship is partially invalidated or contradicted by new events within the relationship.... At any
rate, one must make assumptions about the relationship context, or better, have data about it" (p. 4).
This means that some assessment must be made of the status of the relationship and the way the relationship changes
over time. Although Bateson (3) rejects the idea of measuring subjective reactions, it is difficult to obtain definitions of a
relationship without relying on subjective judgments of the individuals involved. Mishler and Waxler (15) point out that
there has been some work done on describing the relationship, but it has been done from the point of view of observers
rather than the participants themselves. "This is a difficult problem for research and analysis, since messages an observer
judges to be incongruent ... may carry with them implicit meanings ... that make the messages congruent to the family
members" (p. 383-4). The opposite can also be true. As a starting point, perhaps the measures of the relationship should be
obtained from both the participant and the observer. But it seems most important that research on the double bind should
incorporate measures on the nature of the relationship between the two or more parties involved in the actual double bind.
To date, this aspect of the double bind has not been considered, and most of the studies have also failed to include
significant others in the studies.
One of the potentially most useful methods for measuring the double bind is by focusing on observable and repetitive
communicative patterns in families. In an early paper on this topic, Bateson (3) stated that emphasis should be placed on
"observable communication, behavior, and relationship contexts...(and)...upon circular systems of interpersonal relations...
(in contrast)...to a more conventional emphasis upon the behavior of individuals alone or single sequences in the
interaction" (p. 157).
Unfortunately, researchers have seldom attempted to locate double binds in verbal communication between family
members. In those cases where they have attempted to look at such communication, they have done it through parental
letters by Ringuette and Kennedy (20), parental interaction on proverbs by Sojit (24, 25), retrospective reporting of
parental interaction by Berger (8) and interview styles of mothers by Beaver (7). Only the study by Beakel and Mehrabian
(6) attempted to obtain measures of actual family interaction, but this study also failed to obtain any measure of the nature
of the relationships between the parent and child.
It is important to emphasize that while double-bind studies should focus on actual family interaction between the various
numbers, it should not be limited to single interaction sequences but should consider repetitive patterns over time.
Although the examples often given to illustrate a double bind are useful, one should not assume that most double-bind
situations are as easy to locate; they will probably not often occur in these short interaction sequences. A further
clarification of this point of view has been described by Wynne (31). He feels it would be best "not to refer to two
contradictory messages...but to specifiy that a message is so structured that it raises questions about the validity of an
established, intense relationship.... Only in extreme examples would one expect to find restatements about a complex
established relationship context which would within a brief excerpt make it obvious that the relationship is being
invalidated" (p. 7).
In summary, future research on the double bind should include measures of the intensity of the relationship among family
members, and it should be sensitive enough to assess changes in the relationship over time. Secondly, it should be focused
on the way in which communication patterns in families invalidate the definition of the relationship. The measures must,
therefore, be obtained of actual family interaction. The design of these projects should be such that family members will be
observed as they interact in a variety of settings using a range of instruments to assess these variables, rather than relying
exclusively on self-report data obtained from one party. It does necessitate obtaining both behavioral and self-report data
from each family member. The behavioral data would focus more on the actual interaction of the family, whereas the
self-report data would tap the relationship component among the family members. Although such projects will be more
comprehensive and difficult than past attempts, they will permit a more adequate assessment and investigation of the
previously elusive double-bind situation.

Topics for Future Double-Bind Research

In evaluating the research on the double-bind concept to date, it is apparent that researchers have had considerable

_____________________________________________________________________________________________________________

difficulty operationalizing the concept and, as a result, have not adequately tested most of the major propositions relating to
the concept. Good research can only begin if adequate research methods can be developed.
A brief review of some basic questions that need to be investigated would include the following methodological and
substantive issues:
1. How can a double bind be operationally measured?
2. How can it be reliably and validly assessed?
3. What are the various types of double bind situations?
4. How frequently do double-bind messages occur in families with schizophrenic, neurotic, and normal offspring?
5. Is the double bind necessarily pathogenic?
6. Is the double bind a cause or symptom of schizophrenia?
7. Why does one child in a family become schizophrenic and others appear not affected?
8. Is the schizophrenic more or less able than other types of subjects to discriminate a double-bind situation?
9. Is the schizophrenic more or less able to meta-communicate about double-bind situations than other types of
subjects?

SUMMARY
In conclusion, let us return to the introductory analogy regarding the double-bind situation created for those who have
attempted to empirically unbind the concept. The investigators of the double bind have discharged part of their
responsibility and at the same time escaped from a potential double bind by meta-communicating. That is, they have
communicated about confusing communication regarding the double bind, both by word and deed. Their responsibility does
not end there, however, since the research has primarily only indicated that the concept is difficult to investigate. As the
double bind becomes more clarified, investigators need to be more careful and creative in their attempts to develop
adequate measures of it. Future investigators should also pay greater attention to the other conceptual problems that have
not been adequately dealt with in the previous studies. In other words, investigators have proceeded with some vigor to
study the double bind, and they should now preceed with greater vigor.
This review has demonstrated that the double-bind concept has generated considerable interest. It is, however, still a
concept in the process of being more fully defined and refined. It is precisely at this point in time that constructive criticism
should be directed at the research and the present state of the concept. In fact, when the concept is still being developed, it
is even more important that the theorists have reliable empirical findings to further clarify and refine the theory. Hopefully,
this paper has helped to clarify some of the ambiguities of the concept and has pointed out some guidelines for future
empirical research. Until more creative and valid measures are obtained, it will be impossible to empirically unbind the
double bind.

REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.

10

Bateson, G., "Information & Codification," in Ruesch, J. & Bateson, G. (Eds.), Communication: The Social Matrix
of Psychiatry, New York, W. W. Norton Company, 1951.
Bateson, G., Jackson, D. D., Haley, J. and Weakland, J., "Toward a Theory of Schizophrenia," Behavioral
Science, 1, 251-264, 1956.
Bateson, G., Jackson, D. D., Haley, J. and Weakland, J. H., "A Note on the Double Bind1962", Family Process,
2, 154-161, 1963.
Bateson, G., "Critical Evaluations," International Journal of Psychiatry, 2, 415-417, 1966.
Bateson, G., "Double Bind1969," Paper presented at annual meeting of the American Psychological
Association, Washington, D.C., 1969.
Beakel, N. G. and Mehrabian, A., "Inconsistent Communications and Psychopathology," Journal of Abnormal
Psychology, 74, 126-130, 1969.
Beavers, W. R., Blumberg, S., Timken, K. R. and Weiner, M. F., "Communication Patterns of Mothers of
Schizophrenics," Family Process, 4, 95-104, 1965.
Berger, A., "A Test of the Double Bind Hypothesis of Schizophrenia," Family Process, 4, 198-205, 1965.
Ciotola, P. V., "The Effect of Two Contradictory Levels of Reward and Censure on Schizophrenics," Doctoral
Dissertation, University of Missouri, Ann Arbor, University Microfilms, No. 61-2278, 1961.
Kafka, J. S., "A Critique of the Double Bind Theory & Its Logical Foundations," Paper presented at the annual
meeting of the American Psychological Association, Washington, D.C., 1969.
Kafka, J. S., "Ambiguity for Individuation: A critique and Reformulation of the Double Bind Theory," Archives of
General Psychiatry, (In press, 1971).

_____________________________________________________________________________________________________________

12.

13.
14.
15.
16.

17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.

Loeff, R. G., "Differential Discrimination of Conflicting Emotional Messages by Normal, Delinquent, and
Schizophrenic Adolescents," Doctoral Dissertation, Indiana University, Ann Arbor, University Microfilms, No.
66-1470, 1966.
Mehrabian, A. and Wiener, M., (1967) "Decoding of Inconsistent Communications," Journal of Personality and
Social Psychology, 6, 109-114.
Meissner, S. J., (1964) "Thinking about the Family Psychiatric Aspects," Family Process, 3, 1-40.
Mishler, E. G. and Waxler, N. E., "Family Interaction Processes and Schizophrenia: A Review of Current
Theories," Merrill-Palmer Quarterly, 11, 269-315, 1965.
Mishler, E. G., "Problems of Interpretation in Research on Family Processes and Schizophrenia, or the Life,
Death, and the Life, Death and Rebirth of Clinical Concepts." Paper presented at the Association for Research in
Nervous and Mental Disease, New York City, 1967.
Mishler, E. G. and Waxler, N. E., Interaction in Families: An Experimental Study of Family Processes and
Schizophrenia, New York, John Wiley and Sons, Inc., 1968.
Mishler, E. G. and Waxler, N. E., Family Processes and Schizophrenia, New York, Science House, 1968.
Potash, H. M., "Schizophrenic Interaction and the Concept of the Double Bind," Doctoral Dissertation, Michigan
State University, Ann Arbor, University Microfilms, No. 65-2052, 1965.
Ringuette, E. L. and Kennedy, T., "An Experimental Study of the Double Bind Hypothesis," Journal of Abnormal
Psychology, 71, 136-141, 1966.
Schuham, A. I., "The Double-Bind Hypothesis a Decade Later," Psychological Bulletin, 68, 409-416, 1967.
Sherif, M., "If Basin Research is to have Bearing on Actualities," Invited address, 4th Annual Psi Chi Day,
Pennsylvania State University, 1966.
Sluzki, C. E., Beavin, J., Tarnopolsky, A. and Veron, E., "Transactional Disqualification," Archives of General
Psychiatry, 16, 494-504, 1967.
Sojit, C. M., "Dyadic Interaction in a Double-Bind Situation," Family Process, 8, 235-259, 1969.
Sojit, C. M., "The Double-Bind Hypothesis and the Parents of Schizophrenics," Family Process, 10, 53-74, 1971.
Watzlawick, P., "A Review of the Double Bind Theory," Family Process, 2, 132-153, 1963.
Watzlawick, P., Beavin, J. H. and Jackson, D. D., Pragmatics of Human Communication, New York, W. W.
Norton and Co., 1967.
Wearland, J. H., "The 'Double-Bind' Hypothesis of Schizophrenia & Three-Party Interaction," in D. D. Jackson
(Ed.), The Etiology of Schizophrenia, New York, Basic Books, 1960.
Weakland, J. and Fry, W., "Letters of Mothers of Schizophrenics," American Journal of Orthopsychiatry, 32,
604-623, 1962.
Wynne, L. C., Ryckoff, I., Day, J. and Hirsch, S., "Pseudo-Mutuality in the Family Relations of Schizophrenia,"
Psychiatry, 21, 205-220, 1958.
Wynne, L. C., "On the Anguish, and Creative Passions, of Not Escaping Double Binds: A Reformulation," Paper
presented at the annual meeting of the American Psychological Association, Washington, D.C., 1969.
Yarrow, M. R., Campbell, J. D. and Burton, R. V., "Reliability of Maternal Retrospection, A Preliminary Report,"
Family Process, 3, 207-218, 1964.
Yarrow, M. R., Campbell, J. D. and Burton, R. V., Child Rearing: An Inquiry into Research and Methods, San
Francisco, Jossey-Bass, Inc., 1968.

Reprint requests to David H. Olson, Ph.D., Family Development Section, Child Research Branch, NIMH, Building 15
K, Bethesda, Maryland 20014.

11