Shift in representation and symbolisation of affective experience: A paradoxical outcome in therapy

Shift in representation and symbolisation of affective experience: A paradoxical outcome in therapyAbstract: Defining outcome represents a key challenge for psychotherapy theory, research and practice. The present paper uses a case study of a client with anorexia nervosa to contribute to the development of conceptual understanding of the nature of paradoxical outcome. In this case, different sources of outcome data offered different answers to the question of whether or not the therapy had been successful. Qualitative thematic analysis of therapy transcriptions was carried out, using Interpretative Phenomenological Analysis (IPA). Both the process of change that occurred in this case, and the conflicting outcome indicators, could be explained in terms of a model of affect elaboration. These findings are discussed in terms of the need for caution when interpreting results from outcome measures in psychotherapy research and practice.

Cambio en la representación y la simbolización de la experiencia afectiva: Un resultado paradójico en la terapia.

Introduction
Defining and measuring outcome represent two of the most challenging and complex tasks of modern psychotherapy research. In this study, we discuss a clinical case in which qualitative and quantitative data suggested different conclusions. While the client herself described her therapy as having been helpful, scores on key psychometric outcome measures indicated clinical deterioration. Through an analysis of therapy transcripts, we looked at the client's experience in therapy as a way of helping us understand how we can make sense of this type of paradoxical outcome. This case analysis is intended to shed light on a phenomenon that might be quite frequent in clinical practice, that is, a case where the therapy outcome measures indicate a paradoxical result. For a seasoned practitioner, this might not be an unfamiliar notion, as many patients might exhibit such patterns. However, in psychotherapy research, there seems to be a more general perspective on outcome that categorizes it as either good/positive or bad/negative. From this perspective, if symptom scales show deterioration, the therapy is understood as having been ineffective. An alternative position is to accept that outcome is a complex phenomenon, and that change might happen through different channels of expression and different levels of cognitive or emotional differentiation. For example, it has been found that as many as 20% of the clinical population might exhibit an 'illusion' of mental health (Shedler et al., 1993), which means that they might report good mental health, while actually being distressed. Our aim in the present study was to use an intensive analysis of a single case to explore this topic.
A mixed-methods theory-building case study was carried out, following the principles outlined by Stiles (2007). Both qualitative and quantitative data were analysed in order to develop a conceptual model of paradoxical outcome through intensive analysis of process and outcome in a single case. The purpose of theory-building research is to contribute to the development of frameworks for understanding that can be tested and further elaborated on other cases, as well as forming a basis for studies using other methodologies. To help us understand the observed changes in our material, we found it helpful to drawn on the Lecours and Bouchard (1997) theory of the process of mentalisation in psychotherapy because its explicit emphasis on how affects are transformed from the body to the mind seemed to be particularly relevant to patterns that the themes that were observed in this case.

The case: 'Lisa'
Lisa (not her real name), a young woman in her early twenties, had suffered from anorexia nervosa since her early teen years. At the onset of what she described as her 'illness', Lisa experienced two negative life events almost simultaneously. She regarded these episodes as antecedents to her illness and unhappiness. She went from being a spontaneous, untroubled and popular girl with a large social network to becoming a social outcast after one difficult event in elementary school. At the time, she felt that it was impossible for her to ask for support from her parents as they were in the middle of a divorce. Lisa had a hard time adjusting to life in this period. Prior to the treatment described here, Lisa had received several symptom-focused treatments, over a span of seven years. While these treatments had helped her to increase her body weight and to maintain the weight after treatment, at the beginning of the therapy described in the present study, Lisa remained preoccupied with food planning, which dominated her social life and waking thoughts. Lisa graduated from high school with excellent grades and had been accepted at a prestigious university. She came into the therapy feeling as if time had frozen for her, while other people's lives continued. She wanted to start feeling alive after seven years of illness, including overcoming a sense of being tired all the time.

Clinical setting
The client, Lisa, received 12 sessions of therapy at a psychodynamicallyoriented outpatient eating disorder clinic. The numbers of sessions were decided upon before treatment began, and were limited to twelve in order to fit the time frame of the professional training being undertaken by the therapist. The therapist (AA) was 30 years old at the time, a student at the professional program of psychology at a university in Norway, studying at the fifth year out of six years in total. The fifth year includes 5 months of fulltime clinical practice, in which this therapy was carried out, and included supervision by an experienced psychology specialist. Outcome measures were used routinely as a part of the diagnostic and outcome evaluation in the clinic, and recording of sessions was routinely used for supervision purposes. Following completion of therapy, retrospective ethical permission to use these data for research purposes was received from the Norwegian Regional Committee for Medical and Health Research Ethics (National Region South-East). The client gave written consent for the study, and commented on a draft of the present article.

Standardized measures
There were two standardized self-report measures used: the SCL-90 and EDI-2. They were administered at two points in time, one month before the first session and between the 11th and 12th sessions. The Symptom Checklist 90 Revised (SCL-90-R; Derogatis, 1994) is a 90 item questionnaire with five point Likert-scale statements that seek to assess general psychological distress and symptoms over the past week. It is designed to measure nine symptom-specific factors (i.e. somatization, obsessive-compulsive symptoms, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, alienation/psychoticism) which can make up three composite scales (i.e. global severity index, positive symptom total and positive symptom distress index). Norms used in the analysis were based on information available in Siqveland et al. (2016). The Eating Disorder Inventory-2 (EDI-2; Garner, 1991) is a 91-item self-report questionnaire with six-point Likert-scale statements designed to measure three eating disorder scales (subscales of bulimia, body dissatisfaction and drive for thinness) and eight associated psychological scales (ineffectiveness, interoceptive awareness, social insecurity, impulse regulation, interpersonal distrust, asceticism, perfectionism and maturity fears). With no Norwegian norms for the EDI-2, Swedish norms were used due to cultural similarities between the countries (Nevonen et al., 2006;Nøkleby, 2014). Both of these questionnaires are widely used in research and clinical practice in the field of eating disorders and have been found in many studies to provide sensitive indicators of therapeutic change.

Therapy transcripts
Audio recordings were made of all sessions of therapy and were transcribed by the first author.

Qualitative analysis
As we were interested in the client's experiences in this therapy we used Interpretative Phenomenological Analysis (IPA; Smith et al., 2009) as a method of identifying themes within therapy transcripts. This method is specifically concerned with the systematic analysis of lived experience of the individual person, in its own terms. As well as giving a specific focus on the experience of the client, IPA also encourages the researcher's active interpretation of the material, making it possible to put the experience of therapy in a wider theoretical context (Smith et al., 2009). Although IPA is usually carried out on data collected in research interviews, it has also been used as a method for analysing therapy transcripts (Halvorsen et al., 2016). We used IPA because we were primarily interested in understanding Lisa's lived experience as expressed in her therapy discourse, and because IPA is a qualitative methodology that incorporates a phase of theoretical interpretation. In accordance with the principles of theory-building research, two of the researchers (LHKT and HHT) read the whole transcript and identified parts that were relevant to the theoretical purpose of the present study. Segments of the transcript were selected that addressed the following questions: how does the client describe the effect of therapy on her life?; how does the client describe her difficulties?; and, what changed during the therapy? The same researchers conducted a thorough thematic analysis of these transcript segments, in accordance with the procedures described in Smith et al. (2009). The two other researchers functioned as independent auditors. Divergent interpretations were discussed in a series of meetings, with the aim of reaching consensus. IPA methodology emphasizes the importance of using relevant theory to facilitate deeper understanding of themes, and enable transferability of findings to other contexts and cases (Smith et al., 2009). As a consequence, both the thematic analysis and the discussion section should be read as arising from the application of the IPA method.

Researchers
At the time of the study, LHKT and HHT were at the final stage of clinical training at the university. LHKT was the therapist for this case, collected all data, and negotiated client consent for the study. Data analysis was conducted by LHKT and HHT, and audited by ES and JM(research supervisors). LHKT and HHT espouse a psychodynamic approach to psychotherapy, ES is a psychoanalyst, and JM is an integrative therapist. ES and JM have extensive experience in the use of case study and qualitative methodologies in psychotherapy research. Throughout the project, different perspectives on the case were explored and resolved in a respectful manner through dialogue.

Results
Quantitative and qualitative findings are reported separately in the following sections, starting with the qualitative findings.

Thematic analysis of therapy transcripts
On entering therapy, Lisa described multiple on-going difficulties in her life, despite previous therapeutic interventions over a seven-year period. These difficulties centred on a pattern of restricted eating consistent with a diagnosis of anorexia nervosa. Our qualitative analysis of therapy transcript data enabled us to identify four core themes, along with nested subordinate themes, that characterized an overall process of recovery. Being understood was centrally important to the client. Taking up space was expressed in the client's account of difficulties associated with social situations, such as being afraid of taking too much space or influencing others in a negative way. The existence of a critical voice was a barrier to change. The process of recovery was represented in the theme To become alive.
Core theme 1. Being understood Understanding self. In session 1, Lisa told the therapist that she had ideas about underlying causes of her illness. She also expressed that previous therapies had not given her answers to questions of who she was and why she got sick: I have plenty of thoughts about what lies behind my illness, that is what is bothering me in a way, so I don't know. There was a lot of focus on food and meal-plans and those type of things. While not giving any answers to the questions about who I am, and why I got sick. Lisa explained that there was more to her than 'how she looks' and 'how thin she is', and that this had not been understood sufficiently in previous treatments: Being branded as a girl who is only preoccupied with how thin she is and how she looks and that does not take care of her body. That is also an area where I feel like I have been misunderstood. (. . .) That there isn't a room to talk about stuff like that in treatment (Session 1).

Understood by others.
Lisa's acute sensitivity around being understood encompassed a fear of negative responses from others. For example, in session 1 the therapist chose to self-disclose his reactions when Lisa had talked about her low self-worth: T: It sounds like. It sounds sad, but you're smiling. C: Yes, because it is a bit funny (T: Okay), or I think that, it's not funny. Or it sounds like. I just always smile (T: Yes). Because it's the easiest thing you can do to create a good atmosphere. No, I do not know why I smile but. No.
T: No. You smile, and I notice that I am also smiling. (C: Mhm) Ehm, but when you told me, I notice it makes me sad. C: Maybe I do it to ease your sadness (T: Okay). Because I do not want you to hear that . . . I do not want you to get upset.
She went on to explain that she imagined that the therapist's perception of her was: 'Oh Lisa is so sad, I do not want to be with her'. His statement had also made her aware of something being sad in herself: 'Yes, but it was nice that you said it, because if, or if people did not tell me (. . .) then I wouldn't, that maybe one would not have found out that it is sad. (. . .)' (Session 2).
Self-understanding, particularly around emotional states, was problematic for Lisa: at several points in the therapy she described having a hard time figuring out what she felt: 'very often I don't know how I am doing' (Session 4), and sensing bodily impulses: 'there's been such a long time since I felt hunger or the feeling of being full' (Session 7).
Uncertainty around self-understanding and being understood by others represented an area of change within therapy: toward the end of therapy Lisa informed the therapist that she had become more able to express 'this is actually how I feel' and had become 'more conscious' of parts of her experience that had gone unnoticed before.

Core theme 2. Taking up space
Lisa reported that she was accustomed to taking space in elementary school by being a 'girl who spoke her mind'. However, her experience was that being visible and taking up space was hazardous, because it could lead to ostracism: 'Nobody wanted to be friends with me anymore' This sense of being an 'outsider' was also an element in her eating disorder: T: What do you think I would think, then? C: That Lisa does not keep her appointments. Or that she thinks that this is not important, or, yes. It is not wrong to be late for an appointment (T: No), but I think that it becomes very overdimensioned.
T: So, even though you understand that something is okay and that it is not really a big deal, you still get that feeling that you do something that is wrong, and that it defines you. That feeling you get, the feeling of being wrong. Do you sometimes relate that to food and stuff? (. . .) Do you have anything related to body with that emotion? C: Mhm, yes I have.
T: In what way? C: That I am not within a boundary, perhaps.
T: And what would you have to do to become within that boundary, then? C: Look like everybody else (Session 7).
The theme 'taking up space' represented Lisa's sense of being the focus of attention of other individuals in a way that was experienced as 'negative' or not fulfilling the needs of another person, for example, by not helping, being useful, making other people happy or listening to others. A recurring focus of therapy involved Lisa exploring her fantasies about being understood in unfavourable ways outside of the therapy room.

Core theme 3. The critical voice
Client-therapist discourse involved considerable use of metaphor. One metaphor that the client repeatedly used was the experience of a 'critical voice' that described parts of her that were both negatively self-critical and controlling while at the same time having a protective function. How she described this voice gradually shifted over the course of therapy. She began to talk more about a benign voice, or side of her. In the process of therapy, the client and the therapist made use of toy figurines to elaborate and explore the meaning of these metaphors.
Making mistakes. The critical voice tended to be activated in situations where Lisa saw herself as having 'made mistakes': (. . .) The critical voice can also appear in retrospect of things and then it's not . . . then it's more like, strict, but not just strict, more like strict in a punishing way, rather than "you must do this", it's like commanding, "I told you so", and then that bad voice gains power. It draws energy from situations like that. Yes. Because then it is even more apparent that "you actually need a voice like that" (Session 8). This excerpt indicated that her critical side not only punished her, but also maintained itself by convincing her that she needed it.
The purpose of the critical voice. The critical voice was not all bad, but also served a protective function for Lisa by keeping her from experiencing something 'bad'. She further elaborated on the critical voice and its functional aspects: C: That bad voice makes sure that I am on time, that I do what I'm told. It is a kind of self-discipline that one is dependent on, too.
T: So it can't just be gone, it can't be thrown out.
C: I think it is good for people having a balance between the two. If one has a good balance between the two then maybe they can melt together and become one. But the stronger they become, the more polarized they become, in a sense (. . .) (Session 9). The existential consequences of this level of control over herself were substantial, in terms of limiting her freedom to live her life and making her feel lonely: That I am lonely, that I don't have anyone to be with. (. . .) But one of the reasons why I think it is best to be home is that I become tired of being social, because I have to watch out all the time (Session 5).
She also described a more benign voice within herself, which we understood as representing a positive life force in her, a striking contrast to the critical voice: The stronger the good voice is, the harder it is to stay with that emotion that one is not taking part . . . that time just passes by . . . one feels that one does not have the right to take part in life . . . one thinks that it's best if one just lets other people live their lives so that one is put on the back burner. But when the healthy, strong voice says that: "you have the right to take part in life, you can eat and have fun and stuff like that", then that feeling disappears a little. Yes. The healthier I become, the more I feel that (Session 9).
The critical voice kept her away from life activities to gain control, while the benign voice was related to her approaching life activities.

Playing with the critical and benign inner voices to develop insight.
In session 9, the therapist brought in some toy figurines and invited Lisa to explore the meaning of these voices by assigning one for the critical voice and one for the more benign voice in herself. An inner drama between forces then unfolded, with a lot of 'war language', perhaps highlighting the pain associated with these experiences: 'a voice in my head that is very against me, (. . .), I must think hard to fight, to defeat that voice' (Session 8). The benign voice was represented by a lion, which she associated with the lion from the C.S. Lewis Narnia books: Aslan was crucified, or he wasn't crucified, but nailed to a stone block. And killed by the evil witch, and then he resurrected. So I kind of connect the lion to everything that is good, and to Jesus in a way. So he is just good (Session 10).
By contrast, her negative sides, or the critical voice, were represented by a (male) spider that had '(a) web around itself, so that no one can come close' (Session 10).
The therapist interpreted this as depicting the way that her critical side created a distance between her and others: C: . . . maybe he builds that web so that no one can come close to the real him, because alone he stands a little by himself, wary. Yes, it is not enough with just the spider, he needs to have something around himself.
T: Because without that web and threads of protection around him he becomes weak (Session 10).
Therapist and client continued using these metaphors over subsequent sessions. For example, the following segment of active metaphor use, in session 11, highlighted a co-constructed process of meaning-making around points of contact between the lion and the spider: T: Yes, if the lion was to say something like 'When you say that, you make me feel . . . ' C: . . . insignificant and without value. And that I don't, that it makes me feel invisible maybe. And that, but then also, the lion says: 'it's really me who's working for Lisa to do well, not the spider'. T: What do you think the spider needs to hear or know to understand the lion's perspective better.
C: That he is destructive. Because he doesn't know it himself.
T: If the spider is to hear that he is destructive. How would he feel about that? C: He would probably be even more angry and just make more spider web. So I think maybe the most important thing for the lion is to cooperate with the spider.
An outcome of this type of exploration was that it became apparent that the lion was not necessarily all good either: the lion might have been less visible and powerful when the spider was in control, which disappointed Lisa: T: In what situations is he not so pronounced? C: When the spider is very pronounced. Yes. It's the fact that they must cooperate. Or they must be able to think that the both of them can exist together. And that, I wonder why, when he sees that the spider becomes so strong, why would he pull back? Why doesn't he do more for . . . if it is like that for him, if he sees it as lost? If he even bothers to try. That he rather sees . . . if I have like the upper hand I'll do what I can, but that there is no use in wasting energy when I see that it won't go my way.
T: So he's kind of like 'choose your battles'. C: So then sometimes he disappoints me, leaves me a little alone. So why is he there, I wonder? (Session 11.) In this exchange, the client seemed to be telling the therapist that she felt let down by the lion, or the benign parts of herself, because it was not being persistent to fight the critical parts of her. With this and the previous excerpts, we clearly see how the therapist and the client actively used metaphors in the sessions, elaborating on them together.  The use of figurines to explore the meaning of her critical voice was experienced as illuminating for Lisa, as highlighted by the excerpts above. It seemed especially important for her to get to know the critical part of her; its fears, intentions and even resources. The process of talking about it made her uncomfortable, 'the spider does not want to be talked about' (Session 11). But by the end of therapy she acknowledged the importance of understanding the critical voice: '(. . .) I am kinder towards myself and I have learned to get to know my critical voice(. . .)' (Session 12). If we were to interpret this, but in the metaphoric language of the client, we might say that the lion had gained more power within Lisa as it was better at handling the critical voice.

Core theme 4. To become alive
Lisa said that she thought 'it was sad' and 'scary' that the therapy was coming to an end, and both the therapist and client recognized the importance of the sessions. The main outcomes that she described, in the closing sessions of therapy, referred to gains in self-awareness, kindness towards self, an improved relationship with food, and reduced fatigue.
More aware. In session 11, Lisa reflected upon the fact that: 'I am on the right track, and that I've figured out a lot of things I didn't know before the summer'. In the final session, she stated that: 'I think we have talked about a lot of good things here, that has made me stronger and more aware (. Kinder towards herself. Several times at the end of treatment, she described how she had become kinder towards herself, and not as self-critical: Yes, I think the good voice is stronger. I am better at talking to the bad voice. (. . .) And it is maybe that I feel like the strong voice -no I mean the good voice becomes stronger because I am not that self-critical. Yes, maybe I look at myself like I am my best friend or a little sister or something like that (Session 10).
In the last sessions, she still seemed to be favoured the 'good voice' or the lion over the 'bad voice' or the spider, by calling them good and bad voices. One way of understanding the conflict between her voices was that talking about the critical voice or spider and better understanding it made the conflict clearer, as described under 'critical voice'. She said 'the good voice becomes stronger because I am not as self-critical'. Improved relationship with food. Although during the course of the therapy her weight was stable and without gains, another important aspect of the relationship to food in an eating disorder is lack of flexibility in food intake. Lisa reported that her mother had noticed changes in her eating patterns, which might relate to flexibility: I don't know how [the treatment] has affected . . . I think that it has affected how I eat. A week ago, my mom sent me a message saying 'I am so proud of how things are with the eating, I can tell that it's going in the right direction'. It is so nice to get a comment like that (. . .) that comments on the symptoms of the eating disorder, so it might have had an effect on me without me noticing (Session 12).
She also explained that she herself felt more flexible than before: I am no longer restricted to just eating at home. I am very much on the run, and I have to figure out what to eat. So I like fix a lot of food myself, I can go out and eat with friends and eat lunch and stuff like that (Session 12). One way of looking at this is that she was that therapy gave her energy (made her less tired) because it helped her to solve problems that had previously seemed insurmountable.

Analysis of data from symptom measures
The client's scores on the EDI-2 and SCL-90-R symptom measures before and after treatment were compared to published norms for clinical and nonclinical populations (Tables 1 and 2).
Previous studies have reported significant differences in both EDI-2 total and sub-scale scores between non-clinical and clinical samples of eating disorder patients (Nevonen et al., 2006;Nøkleby, 2014), and in patients who have received therapy for an eating disorder (Schlegl et al., 2016). Lower scores on the EDI-2 are indicative of healthy eating, while high scores are evidence of disordered eating. At the start of therapy, Lisa's EDI-2 total score fell within the non-clinical range, while at the end of therapy it was clearly within the eating disordered population range. On the basis of these results, if this had been a case within a psychotherapy outcome study, Lisa's eating disorder would be categorized as having deteriorated over the course of therapy. Analysis of EDI-2 sub-scales indicated a mix of areas of improvement and deterioration of the course of therapy (Table 1). The most significant negative shift was in the area social insecurity (the extent to which the individual perceives that social relationships in general are disappointing, insecure and that people give little back). The next most significant area of deterioration was recorded in interpersonal distrust (difficulties in the establishment and the maintenance of close relationships, and the capacity to tell other people about thoughts and feelings). Other subscale deteriorations were recorded in relation to perfectionism and body dissatisfaction. By contrast, while there was a substantial improvement on the sub-scales assessing bulimia (items include: 'I have gone on eating binges where I felt that I could not stop'; 'I stuff myself with food'), ineffectiveness, drive for thinness and impulse regulation.
Over the course of therapy, Lisa recorded improvements in relation to a number of subscales of the SCL-90 R: phobic anxiety, indicating fewer difficulties relating to avoidance of different situations or places or experiencing difficulties in crowded places; somatization, indicating less distress arising from bodily symptoms such as tiredness, dizziness, nausea, troubled stomach, hot flashes and chills, numbness, tingling and bodily weakness; anxiety, indicating reduction in difficulties with nervousness, feelings of fear, tension and uneasiness (Table 2).

Paradoxical outcomes
On the basis of comparison between qualitative clinical information from transcripts of therapy sessions and data from pre-and post-therapy symptom measures, it is possible to identify several areas in which contradictions between different outcome data sources occurred. The main paradox was associated with overall estimate of outcome. Pre-post change on Lisa's total score on the EDI-2 suggest that her overall pattern of eating disorder symptoms had become more problematic by the end of therapy. This evidence is of considerable clinical relevance, given that the primary goal of therapy was to address Lisa's anorexia. By contrast, on the basis of both qualitative data and SCL-90-R scores, it appeared that therapy had been helpful for Lisa. A further area in which a contradictory finding was apparent was around high pre-therapy bulimia scores on the EDI-2, which stood in stark contrast with a presentation in therapy of a pattern of eating disorder characteristic of anorexia nervosa rather than bulimia. An additional contradictory or paradoxical outcome could be observed in major increases in EDI-2 measures of interpersonal difficulties (interpersonal distrust and social insecurity; on both sub-scales Lisa shifted from the normal range at start of therapy to the clinical range by the end), contrasting with SCL-90-R scores for interpersonal sensitivity that remained in the high clinical range over the course of therapy, and qualitative data that indicated some gains in relationship functioning.

Discussion
In terms of clinical and qualitative data generated by the client, this case can be defined as having a successful outcome. Lisa herself expressed that the therapy had been helpful, she reported that a close observer of her life (her mother) had noticed significant gains, there were clear links between key elements of therapy process (e.g. therapist interpretation of transference and use of metaphor) and eventual outcomes, and there were no extra-therapy factors that could account for these results. These qualitative indicators satisfy the criteria developed by Elliott (2002) to determine efficacious outcomes within single case qualitative data. By contrast, the psychometric results were mixed. While Lisa's scores on standard outcome measures indicated reduction of somatization symptoms, they also recorded that she became a lot less trusting and a lot more socially insecure. These negative shifts were clinically significant -upon entering therapy her scores on these factors had been very low, and increased to very high levels. The overall change profile derived from quantitative measures suggests that Lisa's eating disorder had worsened over the course of therapy, while her well-being had improved. Taking this set of qualitative and quantitative as a whole, it is therefore possible to identify the case as an example of paradoxical outcome.
The paradoxical outcomes observed in the Lisa case can be understood in different ways. In their programme of research into illusory mental health, Shedler et al. (1993) identified a subgroup of patients that rated themselves as mentally healthy, whilst physiological measures and clinical judgment indicated that they were distressed. Shedler et al. (1993) hypothesised that psychological defenses caused these patients to deny the severity of their emotional distress when filling out the mental health scale. However, since Lisa reported high levels of at least some symptoms before the beginning of treatment, she does not easily fit the profile of the group that Shedler and colleagues termed 'defensive deniers' (Shedler et al., 1993).
From the first session of therapy, Lisa stated that in her everyday life she generally endeavoured to appear more positively than actually was the case. The low scores on the scales of social insecurity and social distrust at the beginning of treatment might have been influenced by her need to present a positive image to others. By contrast, it might have been easier for her to communicate distress related to experiences of the body, compared to the more sensitive interpersonal area. These processes might be understood both in the context of the operation of mechanisms of psychological defense (Shedler et al., 1993), and the social context of the questionnaire administration (McLeod, 2001). These two perspectives do not seem to be theoretically separate, as defenses are considered somewhat dependent on the social context, and social contexts influence people's defenses. A preliminary interpretation of the pattern of results in this case therefore suggests that changes in psychological defenses might have influenced changes in the psychometric data. Helpful therapy might have made it possible for Lisa to provide a more authentic estimate of the severity of her eating-related relationship difficulties in her end-of treatment measures. By contrast, her intake scores around somatic concerns were high, so it is possible that the observed reduction in these scores may also be attributed to helpful therapy.
In terms of what might be regarded as a widely-understood typical process of change observed by clinicans offering therapy for anorexia, it is likely that the stabilisation of eating patterns, adoption of healthy eating, and amelioration of associated processes such as secrecy and perfectionsm, would have had a positive impact on Lisa's well being and general mental health as measured by the SCL-90-R. However, this process would not explain why the EDI-2 total score shifted in the direction of apparent exacerbation of Lisa's eating disorder. Studies using the EDI-2 as an outcome measure have found improvements in both eating disorder symptoms and well-being occur in tandem -the oppositive pattern to that observed in the present study (Schlegl et al., 2016).
We believe that the closer examination of the process of affect elaboration has the potential to provide a useful theoretical framework to understand this specific case, in which the change was characterized by reduction of somatization symptoms and emergence of interpersonal symptoms (e.g. heightened interpersonal insecurity and distrust). Of particular relevance is the model developed by Lecours and Bouchard (1997), and based on the concept of mentalisation. The key proposition within this model is the suggestion that 'affective experiences must first be represented and symbolized before they can become part of psychic life' (Lecours et al., 2000, p. 47). In summary: with respect to the Lisa case, we propose that for this client, using therapy to engage with emotions associated with painful life experiences resulted in a new ways of representing and symbolizing these emotion, and that different information sources (outcome measures and analysis of what she said in therapy) were sensitive to different aspects of this process.
Within the Lecours and Bouchard (1997) model, drive-affects are experiences that are felt, and that represents biological innate needs, e.g. impulses, wishes, feelings. The most important developmental challenges for the human infant is to learn how to meet its needs in the world. A healthy mind is one that is able to make accurate predictions for how needs most likely will be met in the real world (Solms, 2018). Lecours and Bouchard (1997) argued that degrees of drive-affect elaboration can be conceptualized along two orthogonal, theoretically separate, dimensions: (1) modalities of expression/representation and (2) levels of affect tolerance. The first dimension proposes four levels of expression of drive-affect from a low to high level of mental elaboration: (1) somatic (e.g. expression of drive-affect through the body; physiological sensations, bodily disturbances, etc.); (2) motor expression (e.g. expressions of drive-affect using voluntary bodily movements); (3) imagery (e.g. expression of drive-affect through, images, dreams, metaphors) and (4) verbal expression (e.g. labelling of drive-affect using language). The second dimension proposes five levels of affective tolerance/containment, here presented in 'chronological' order from least to most tolerated: (1) disruptive impulsion -uncontrolled or untolerated (often unconscious) expression of affect; projective identifications, conversion symptoms, somatization (e.g. sudden stomach ache during a session, walking out of the room and physical fights); (2) modulated impulsion -more adaptive than the first level, but drive-affect not yet tolerated and is refused -content is not reflected upon (e.g. 'Oh no, I won't make the appointment!'); (3) externalization -more tolerated drive-affect, less intense, but not fully owned as one's own affect (e.g. 'she made me feel so down'); (4) appropriation -drive-affect is fully tolerated and owned as one's own (e.g. 'I was afraid'), and (5) meaning association -drive-affect is tolerated and put in a wider context making it more complex and meaningful to the individual (e.g. 'I always feel anxious before a doctor's appointment'). The degree of affective tolerance can be expressed through any mode in the first dimension.
The Lecours and Bouchard (1997) model draws upon a wide set of notions from the psychoanalytic literature, most notably from two French psychoanalysts; Pierre Marty and Pierre Luquet, and the British psychoanalyst Wilfred Bion (Lecours & Bouchard, 1997). They also draw upon Freud's bindung-concept, that bodily affective activation that has not yet been processed can become more manageable through a binding-process between words and experiences (Bouchard et al., 2008, p. 50).
The Lecours and Bouchard (1997) mentalisation model provides a framework that accounts for the apparently paradoxical outcomes observed in the Lisa case. As shown in the IPA analysis, the client stated that she had become kinder towards herself, more self-aware, less tired and more flexible with food. By contrast, psychometric data indicated a decrease in symptoms related to phobic anxiety, somatization and anxiety, and an increase in symptoms of social insecurity and interpersonal distrust related to her eating disorder. This pattern suggests that Lisa's symptoms and distress may have been more related to bodily experience in the beginning of therapy, over the course of therapy she became more able to make sense of them in interpersonal terms. The Lecours and Bouchard mentalisation model implies that helpful therapy involves a transformative process through which somatic affective experiences are represented and symbolized to become accessible to conscious reflection. Even though the SCL-90-R scales are not made to measure levels of mentalized affect, one could argue that the SCL-90-R scales indirectly capture this type of transformation.
From the perspective of the Lecours and Bouchard mentalisation model, the ability to mentalize is considered to be a tool to make a person's life more meaningful (Lecours et al., 2007). The intensity of a bodily experience (such as anxiety) is an indication of level of mentalization: anxiety-laden bodily experiences will be experienced as less intense, the more mentalized (i.e. more meaningful) they are. Lisa's scores suggest that these types of experiences had become less intense over the course of therapy, as she gives lower rating on the somatic perceptions, anxiety and phobic anxiety scales after treatment compared to before. Assuming that the scales completed by Lisa actually measure what they are supposed to measure, then how could we make sense of her being more socially insecure? How could she both feel like the therapy had helped by making her less tired, more conscious and kinder towards herself, and also feel much more socially insecure and less trusting? We would argue that her somatization symptoms and anxious experiences were transformed by gradually being more able to be expressed through verbal communication. From this perspective, her high pre-therapy score on bulimia could be interpreted as reflecting a level of preoccupation with the concrete physical activity of eating that was replaced by a capacity to reflect on the meaning of eating.
We believe that the Lecours and Bouchard's mentalisation model (Lecours & Bouchard, 1997) also makes it possible to understand the process through which therapeutic collaboration through shared metaphor was helpful in this case. At the beginning of the treatment, Lisa described herself as being concerned with appearing more positively than was reflected in reality. During the course of therapy, the critical voice became more evident. It seemed as though it became easier to talk about this voice, and to describe it in a more elaborate manner by using the lion and spider figurines. Collaboration use of (and play with) metaphors can be considered a method for working within the imagery channel as defined in the mentalisation model. This channel is considered to be an intermediate level between somatization and verbalization (Lecours & Bouchard, 1997). Somatic experiences that are not available to the verbal channel of communication, become more easily accessible through the imagery channel before being verbalized. This process was apparent in how Lisa used metaphors when talking about events from her everyday life. By using images and metaphors Lisa was able to talk about her critical voice, and her eating disorder, in an indirect way that protected these sensitive areas of experience from direct scrutiny. It also seems relevant to point out that it seemed like it was easier to talk about the resistance to this treatment through the use of metaphors: the spider did not like to be talked about.
Many clients with anorexia nervosa appear to be ambivalent about changing in therapy, because their condition also has some benefits to them, such having a sense of mastery (Nordbø et al., 2012). Other clinical features are alexithymia and deficient interoceptive awareness, which impede a general ability to sense and label one's emotions, and to sense satiety and hunger (Pollatos et al., 2008;Sifneos, 1973). Rizzuto (2001) suggests that imagery and metaphor can give access to experiences otherwise too painful to talk about directly, supporting the notion that metaphor is a general tool or mechanism of the mind (Enckell, 2002). The sources of our mental representations are fundamentally bodily sensory experiences, which are subsequently integrated in higher-order representations that are detached from the primary experience (Rizzuto, 2001). Bodily self-experiences are hard to convey in a straightforward language, but doing this through metaphors can help to verbalize and share something 'unshareable' in the self-experience. Metaphors, Rizzuto (2001) holds, are better able to grasp and 'tie up' basic bodily representations, compared to the other channels of expression, which might make them applicable for elaborating on somatized experiences. For clients with anorexia, Skårderud (2007) has argued that their bodies might act as concretized metaphors for inner experiences, especially when the representational function of the mind is diminished and the client struggles with verbalizing their emotional experiences. The body might represent their inner conflicts, hence becoming a concrete metaphor for the mind.

Conclusion
The aim of this case study was to understand the paradoxes that emerged between the psychometric scales and the client's experience of change. In some respects, psychological defenses seemed to be relevant in explaining the observed pattern of outcome in this case. However, because the client's symptoms shifted from bodily experiences to social issues, the case could not be categorized as a straightforward example of illusory mental health (Shedler et al., 1993). Instead, the Lecours and Bouchard (1997) model of affect elaboration seemed to offer a more comprehensive account of how changes in psychometric scores might have been related to the therapy process. Although we hope to have demonstrated the potential relevance of the Lecours and Bouchard (1997) model, we acknowledge that other theoretical perspectives could similarly be applied to the task of making sense of the apparently paradoxical pattern of evidence regarding the outcome of therapy for Lisa.
We believe that it is likely that paradoxical and contradictory evidence is frequently encountered in therapy practice (see, for example, Truijens et al., 2019). The key learning point that we would wish to emphasise on the basis of this case analysis, is that theoretical reflection on how it might be possible to make sense of different sources of evidence can lead to new insights about a case. Little is known about strategies used by practicing clinicians handle apparently paradoxical client outcome and process data. Further research on this topic would be valuable.
Our results question the validity of any single psychometric measure as a sole source of information about treatment outcome, and support the use of a diversity of information sources. We have not been able to find any other research that has implicated mentalisation of affect as a contributor to apparent increases in symptoms in mental health scales. Further research needs to be conducted to investigate other cases of paradoxical outcome to explore how this process unfolds in different contexts.

Limitations of the study
The therapist was also one of the researchers, which might have given him some preconceptions and tacit knowledge that might have guided analysis and interpretation of data. However, the three other researchers had never met the client and could function as critical auditors of the therapist's views. The most important limitation of the study is that we could not ask Lisa herself to offer her own interpretation of outcome scores, for example, in respect of how she understood specific questionnaire items. We were unable to collect follow-up data, which means that we cannot know if the changes that she reported at the end of therapy were sustained over time. A follow-up interview and further psychometric assessment could have helped us in illuminating important aspects of our interpretation of case data. As the idea for this study was decided upon after the end of the treatment, we did not have data that could be used to analyse countertransference processes within the case. Finally, while IPA is an effective methodology for analyzing themes within lived experience, it is not a suitable tool for understanding interaction patterns between the client and the therapist.

Disclosure Statement
No potential conflict of interest was reported by the author(s).

Notes on contributors
Lars Henrik Kaasen Thoresen is a clinical psychologist at Nydalen DPS, Oslo University Hospital.