The focus on, and study of, narrative in behavior analysis has been extremely limited (but see Hineline, 2018; Roche & Barnes-Holmes, 2003, for notable exceptions). Narrative is, of course, important in all of the ways that Hineline, for example, outlined. In the current article, we focus on one way in which narrative serves an intensely pragmatic concern, that of understanding and treating human psychological suffering in the wider context of modern behavior therapy, and acceptance and commitment therapy (ACT) in particular. It is important to emphasize that a focus on narrative or story-telling also appeared in the seminal volume on relational frame theory (RFT, Hayes, Barnes-Holmes, & Roche, 2001), which co-evolved as a basic science element of ACT.

In our view, narrative has become increasingly important in the application of RFT in clinical contexts and this will be the focus of the current article. Narrative and stories played a central role in narrative therapy (see Monk, 1997) and in the psychodynamic tradition more broadly (Book, 2004), in which the client’s story about themselves was taken very seriously (for a book-length review of narrative and psychotherapy, see Angus & McLeod, 2004). But, narrative-based therapies were rarely rooted in any form of experimental analysis of human language and showed little concern for how we get from the simplest units of language, such as naming, to the telling of full-length stories.

It is perhaps ironic, therefore, that the so-called third wave of behavior therapies places far greater emphasis on the client’s narrative in the context of the dialectic between therapist and client (Kahl, Winter, & Schweiger, 2012). Perhaps even more ironically, recent attempts to connect the laboratory-based experimental analyses of human language and cognition with psychotherapy have also focused increasingly on the dynamics of complex networks of relational responding, which may take the form of mini-narratives in the context of psychotherapy (see Barnes-Holmes, Barnes-Holmes, Luciano, & McEnteggart, 2017). In the current article, we attempt to show how this refocusing on narrative within RFT, and its connection to the assessment and treatment of psychological suffering, are unfolding. In particular, we explain how the analysis of language and cognition in RFT may be conceptualized in terms of an overarching framework that we refer to as the multi-dimensional multi-level framework (MDML; Barnes-Holmes, Barnes-Holmes et al., 2017). One of the purposes of the MDML is to provide a framework that allows one to appreciate how the simpler units of analysis specified in RFT, such as mutual entailment which is seen as directly relevant to naming, connect to the more complex units, such as the relating of relational networks, which is seen as critical to narrative and story-telling. For a detailed treatment of the MDML and its relationship to RFT, the reader is referred to Barnes-Holmes, Barnes-Holmes et al. (2017). For present purposes, we will provide a brief outline of the framework before illustrating how it can help to conceptualize the importance of narrative in the treatment of human psychological suffering.

Relational Frame Theory and a Multi-Dimensional Multi-Level Framework

The seminal (2001) text on RFT used the basic operant unit of the relational frame to provide a functional-analytic account of complex relational networks as a model for story-telling and narrative (Stewart, Barnes-Holmes, Hayes, & Lipkens, 2001). On balance, RFT-based experimental analyses tended to focus on individual frames, or relatively simple networks, at least in the initial stages of the research program. More recently, however, there has been a recognition, both conceptually and empirically, of the growing need to grapple with complex relational networks, especially if RFT is to connect with psychotherapy (Barnes-Holmes, Barnes-Holmes et al., 2018).

The MDML specifies five levels of relational responding: mutual entailing; relational framing (the simplest type of relational network); relational networking; relating relations; and relating relational networks. In addition, the framework conceptualizes four dimensions for each of these five levels: derivation, complexity, coherence, and flexibility (see Table 1). Where each level intersects with each dimension yields a total of 20 units of analysis which provide a conceptualization of the dynamics of relational responding. Let us briefly define each dimension. Derivation refers to the number of times a derived response has been emitted, where the first response is high in derivation because it is being derived entirely from a trained relation. Thereafter, derived responses gradually acquire their own history and are, therefore, less and less derived relative to the initial relation that was trained. Complexity refers to the detail or density of a pattern of relational responding, including the number or types of relations in a given relational network. Coherence refers to the extent to which relational responding is broadly predictable or is consistent with existing patterns of relational responding (whether they are directly trained or derived). Flexibility refers to the extent to which patterns of derived relational responding may be influenced or changed by contextual variables (e.g., when trained baseline relations are reversed).

Table 1 The Multi-Dimensional Multi-Level Framework for analyzing the dynamics of relational responding
Full size table

We will now focus on how the MDML helps us to better conceptualize complex relational responding (or narrative) in psychotherapy. Imagine a client who comes into therapy and says, “I am a terrible person.” The therapist asks, “Is this really true?” and the client answers “Definitely, I am terrible right through.” The therapist then asks, “How long have you felt like this?” and the client replies, “I’ve known this since I was very young.” The therapist then asks, “What makes you feel you are so terrible?”, and the client replies, “I don’t know really, I just know that I am.” Finally, the therapist says, “You don’t seem like a terrible person to me,” and the client reacts harshly with, “You don’t know me, if you did, you’d know that I am terrible.”

Within the framework of the MDML, we could conceptualize this therapeutic interaction as follows. The client’s first statement, “I am a terrible person” involves mutually entailing the verbal self (i.e., words and terms, such as “I” coordinated with self) with “terrible”. The next statement “Definitely, I am terrible right through” suggests that the mutual entailing is high in coherence in the sense that it coheres strongly with other self-statements. The answer to the question about how long this has been the case (“I’ve known this since I was very young”) suggests that the mutual entailing is also low in derivation, because the client thinks this a great deal. When asked why, the statement “I don’t know really, I just know that I am” suggests that the mutual entailing is low in complexity. The harsh reaction to the therapist’s final statement suggests that the mutual entailing is highly inflexible.

The precision available with the MDML is highlighted with subtle differences in a client’s narrative. Imagine again the client described above, who now offers a list of reasons why he is terrible (rather than simply saying, “I just know that I am”). For example, imagine he said “My wife has left me, my kids are ashamed of me, I eat too much, and I’m lazy”). Rather than defining this as mutual entailing, the response might be better considered as relational networking or relating relational networks. Now imagine when the therapist asks “Is this really true?” and instead of answering “Definitely, I am terrible right through,” the client says “I don’t think I’m completely terrible,” this suggests responding that is low in coherence (rather than high), because it is inconsistent with other parts of the relational network. Imagine too after being asked how long this has been the case, the client says instead “I’ve been thinking about this a lot recently” (rather than “I’ve known this since I was very young”), this suggests responding that is high in derivation, because it emerged only recently in the verbal repertoire. The list of reasons used to confirm that the client is terrible also suggests that the relational networking is high in complexity. Finally, imagine if in response to the therapist’s last statement, the client now says “I know that’s true at least sometimes” (instead of “You don’t know me, if you did, you’d know that I am terrible”), this suggests higher flexibility.

In the foregoing, we have offered an interpretation of a client’s narrative in the rather abstract relational terms of the MDML. Doing so may help to connect the concepts and language of the basic experimental science with therapeutic practice, but we recognize that more is needed to fully capture the functional properties of the relational responding highlighted within the MDML. More informally, the client’s narrative is more than just a complex story, but may have important functional properties for how they live their lives. To capture this, we have begun to conceptualize the functional properties of any pattern of relational responding within the MDML as possessing relatively appetitive versus aversive functions. As will become clear, this strategy may be important in helping us to understand the role of narrative in the psychotherapeutic context.

Specifically, the MDML framework highlights an important distinction between the relational and functional properties of particular parts of a relational network or narrative when conducting functional assessments and analyses in the context of therapy. That is, the modern behavior therapist may focus on the life narrative that the client brings to therapy and in so doing engages in a type of verbal functional assessment of the narrative itself. Parts of this narrative may have relatively strong appetitive functions for the client, whilst other parts may have relatively aversive functions. An important part of therapy is to identify the aversive functions, because these are initially hidden or avoided by the client. Once these aversive parts of the narrative or network are revealed, the therapist may then begin to encourage approach behaviors toward the aversive parts of the narrative, thus creating a broad and flexible behavioral repertoire where a previously narrow and inflexible repertoire existed. One of the key roles of modern behavior therapy, therefore, is not to treat the client as somehow broken, sick, defective or faulty, but instead to help the client to understand as a human being how their narrative pertaining to their life history either facilitates or hinders living a full and rich life.

The Distinction Between the Relational and Functional Properties of Narrative: Implications for Conducting Functional Analyses in Therapy

In this section, we present two general approaches to psychotherapy, which we refer to as verbal functional analysis and the drill-down (see Barnes-Holmes, Boorman et al., 2018, for a full treatment of these approaches in the context of two case studies). We present these two approaches here because they have been very much motivated and directed by our knowledge of, and ongoing research activity in, RFT and most recently by our work with the MDML.Footnote 1

Verbal Functional Analysis

Verbal functional analysis focuses on the functions of stimuli and responses that are verbal, as defined by RFT. In conducting a verbal functional analysis, we typically operate at the level of complex relational networks or narratives, rather than specific relational frames. We have found that conducting verbal functional analyses at this level provides a coherent understanding of client behavior and useful directions on how these relational networks can be altered.

Specific verbal stimuli that are observed in psychotherapy may be conceptualized as participating in complex relational networks that generate narrow and inflexible responding. For instance, the word “mistake” (i.e., the relational networks in which the word participates) may elicit defensive or avoidant reactions by the client, such as turning their face downwards. As a result, the therapist may reason that the verbal stimulus “mistake” has significant functional properties for the client’s behavior, and thus these functional properties and the relational networks in which they participate, will become the focus of the verbal functional analysis.

In our verbal functional assessments, we often distinguish between appetitive and aversive relational networks in which the deictic-IFootnote 2 participates. Appetitive relational networks have dominant approach or S+ functions, while more aversive networks have dominant avoidance or S- functions. Consider a client who comes to therapy to talk about his despair. Although despair is indeed problematic in his life, and very distressing, the self-label “depressed” may facilitate avoidance of the more complex issue of fear of abandonment. Verbal functional assessment allows the therapist to distinguish the S+ from S- functions of the client’s self-labeling. Specifically, “depressed” may have more positive functions than “abandoned”. Indeed, self-labeling as “depressed” perhaps permits the client to avoid the more accurate description of his behavior which is to avoid being abandoned. Thus, we refer to ‘depressed’ as the S+ network (appetitive functions) and to ‘abandoned’ as the S- network (predominantly aversive functions). We use verbal functional assessments to first deal with the S+ (e.g., depressed) network, with which clients engage more readily. Thereafter, we begin to talk about the S- network, where client discomfort and defense will be greater.

Once behavioral momentum in discussing both S+ and S- relational networks has been established, the therapist uses verbal functional analyses to create causal (if-then) relations between these networks. Using the example above, imagine the therapist saying, “Being depressed probably keeps you apart from the people you care about.” This relates the S+ and S- networks, thus putatively transferring the appetitive functions of the ‘depressed’ network to the more aversive ‘abandonment’ network, so that the client becomes more willing to talk about abandonment. The therapist might then ask rhetorically “What if being depressed forces you to abandon others? What if being depressed causes you to be abandoned? If you could choose between being depressed and being abandoned, would you choose being depressed?”

To conclude, the primary aim of verbal functional assessment is to identify the key verbal stimuli that participate in the relational networks that possess approach and avoidance functions. This move enhances the therapeutic relationship, such that even the most sensitive issues can be shared, including those that have been avoided, and which are necessary to approach in order to build psychological flexibility.

The Drill-Down

It is no surprise that verbal functional assessment and analyses involve a strong therapeutic relationship, and indeed this relationship is part of the verbal functional analysis itself. We use the drill-down as a metaphor to describe the key behavior that is involved in constructing and developing this relationship. Specifically, the drill-down involves relational responding between the deictic-I and deictic-Others. This is achieved by creating “coordinated narratives” between the client and therapist. In other words, this involves the therapist validating the client’s narrative, and we often do so with phrases such as “I can completely see where you’re coming from.”

In the course of childhood, we propose that the deictic-I develops in a highly shared and cooperative context through which significant others help to construct the verbal sense of self. Indeed, early on, young children cannot distinguish themselves verbally from others, but learn to do so, physically and psychologically, across time. Now imagine if this shared and cooperative context with significant others is relationally incoherent, such that the relationship between deictic-I and deictic-Others is unstable, unpredictable, and discontinuous. Imagine, for example, a child who is the center of attention when the relatives visit, but who is ignored when there are no visitors. This situation would likely weaken the relational coherence pertaining to the deictic-I by undermining the extent to which it can be used as a locus to relate hierarchically with the child’s psychological events. That is, if ‘I’ is to serve as a constant locus for constructing a psychologically healthy narrative, ‘I’ must develop in a relationally stable and consistent environment. Where this is not the case, a deictic-I that is clearly verbally distinct from others will likely not emerge (McEnteggart, Barnes-Holmes, Dillon, Egger, & Oliver, 2017). Individuals with this type of history often report in therapy that they do not really know who they are – a mini-narrative that is in the broad functional class of the verbal relations in which they were raised as children.

Our core proposition is that the therapeutic relationship must provide the predictability and consistency that were historically absent with significant others. The therapist must provide a highly shared and cooperative verbal context or narrative in which a clearly distinct, stable deictic-I can be established across time. This may seem paradoxical if we conceptualize it as coordinating the deictic-I (the client) with the deictic-Other (the therapist), with the use of phrases such as “I completely get that,” “If I were you, I would have done/felt just the same,” and “I can see how lonely you really are.” Indeed, experienced therapists can often ‘absorb their clients’ self-narratives’ in a rich and full way without pulling back, or being reactive or defensive. In more technical terms, the therapist establishes specific contextually controlled coordinate relations that are relationally coherent between the client’s deictic-I and the therapist’s deictic-I, in order to build trust and safety into the therapeutic relationship. It is important to emphasize that we are not suggesting full coordination between I and Others (therapist and client), which is neither feasible nor desirable. Instead, the therapist to some extent sees what the client sees, feels, etc., but does so from a context of hierarchical relational responding from the therapist’s own deictic-I. In other words, the therapist embraces the client’s narrative fully and without defense, but within the wider context of the shared narrative that is the therapeutic relationship.

It is pivotal in the therapeutic relationship to establish a repertoire in which the client relates the deictic-I as located in the here and now to the deictic-I as located there and then. In simple terms, the therapist shares with the client different ways to talk about the deictic-I. This repertoire is established by coordinating the therapist’s deictic-I and the client’s deictic-I (both located in the here and now), so that they share their narratives in a cooperative way. As a result, all psychological events and experiences in the client’s life become (even momentarily) an ‘it’ (there and then) that is separate from both the client and the therapist as coordinated deictic-Is (here and now). Put another way, the client and therapist develop a narrative on the client’s sense of self as an event or object that can be observed and talked about in numerous ways. This might also be seen as the therapist and client building a new narrative about the client’s older narrative that previously hindered living a rich and full life.

It is important to emphasize that the drill-down is intertwined with verbal functional analyses, but focuses specifically on the relational processes that seem central to the therapeutic relationship. The drill-down metaphor helps to describe how we use the therapeutic relationship to ‘dig deeper’ in a verbal functional sense into the self-narrative. For example, a verbal functional assessment might identify “mistake” as a critically important verbal stimulus for a client. Verbal functional analyses then enable the therapist to assess the therapeutic relationship itself on an ongoing basis. For instance, the therapist might then ask the client if they are willing to explore the word “mistake” as uttered aloud and to notice its impact. Being willing to do so on behalf of the client permits further verbal functional analyses. The therapist might say in the context of the drill-down “I would have made the same mistakes if I had been in that place at that time.” The key point is that verbal functional analyses and the drill-down are dynamical, and should ebb and flow with each other in the course of therapy in the service of exploring existing narratives and building new ones.

Concluding Comments

When behaviorism and behavior therapy emerged as a dominant force in psychology, the importance of narrative could be seen as a baby that got thrown out with the bathwater. Focusing on narrative, as Hineline has done in the current volume, brings narrative back into view, but through a behavior-analytic lens. We certainly share his enthusiasm in this regard. Moreover, we have attempted to show in the current article how a behavior-analytic focus on narrative appears to be important in developing a modern behavioral approach to human language and cognition, and in particular the assessment and treatment of human psychological suffering through psychotherapy. We have offered a framework that was developed directly from laboratory research on RFT and its application in the clinical domain. As such, it should be possible to explore the value or utility of what we offer, in basic and applied research, and in clinical practice itself.