Position paper – CFT for psychosis
Abstract
Purpose
This paper outlines the theoretical and empirical basis for compassion focused therapy (CFT) for psychosis, the gaps in the current knowledge and research, as well as some of the challenges for addressing gaps. It will guide the direction of future work and the steps needed to develop and advance this approach.
Method
This paper reviews evidence of how evolutionary models such as social rank theory and attachment theory have greatly contributed to our understanding of psychosis and provide a clear rationale and evidence base for the mechanisms of change in CFT for psychosis. It reviews the evidence for outcomes of compassion training more generally, and early feasibility evaluations of CFT for psychosis.
Results
The process evidence shows that people with psychosis have highly active social rank and threat systems, and the benefits of switching into attachment and care systems, which can support emotion regulation and integrative mind states. The outcomes evidence shows that compassion training impacts not only psychological outcomes, but also physiological outcomes such as neural circuits, immune system, and the autonomic nervous system. Within the psychosis field, outcomes research is still in the early days, but there are good indications of feasibility and a clear path forward for the next steps.
Conclusions
CFT for psychosis is an approach that integrates biopsychosocial processes, an integration that's evidenced across each aspect of the model, from theoretical foundations (evolution-informed) to interventions (e.g., body/breath training and relational techniques), to evaluation. Future RCTs are required to understand the effects on biopsychosocial outcomes for people with psychosis.
Practitioner Points
- An overarching aim of CFT for psychosis is to help people notice when threat-focused patterns are active, and switch to compassionate patterns.
- The evolutionary approach offers a de-shaming and empowering framework that is important for this group who have often experienced trauma and threat from powerful others, coupled with high levels of social stigma and shame, which can amplify threat and dissociative processes.
- The current position of research in CFT for psychosis is that there are many strengths in the process evidence, but not yet in the outcomes evidence.
- Early research into the feasibility and acceptability of CFT for psychosis is established, as well as clinical manuals, training, and supervision, and RCTs are now required to advance the field.
INTRODUCTION TO COMPASSION FOCUSED THERAPY (CFT) FOR PSYCHOSIS
Compassion focused therapy uses an evolution-informed, biopsychosocial framework to understand the nature of minds, the causes of mental distress, and the therapeutic processes that might help. One of the main considerations in the history of CFT has been about developing a therapy process based on a scientific understanding of the mind and the functions of the mind, rather than on a science of treatment. A science of treatment starts with a pathology or difficulty and builds a science around the observed outcomes from its interventions and evaluations. CFT starts with an understanding of the nature of minds and why they function in the way they do.
A helpful point for readers of this special issue is to understand the difference between a model of therapy and therapeutic interventions and techniques. Although some of the techniques used in CFT may be similar to those used in other therapies described in this issue, this is not to say that CFT is the same model of therapy. The same technique can be used for very different aims and for targeting very different processes. The same technique can also be contextualised and textured in very different ways depending on the model and psychoeducation that precedes it. Consider exposure, for example, there are many therapies that use exposure to help build tolerance of a fear and to change a person's orientation to that which is feared. However, if a CFT therapist was using exposure, they would first ensure that the person has developed a compassionate mindset, which creates a psycho-physiological environment for exposure. They would have practiced compassionate inner dialogues with voice tones linked to validation and encouragement, along with facial expressions, body and breathwork that prepares them to approach exposure with caring intentions, as well as with wisdom and courage. This motivational stance and preparedness will have been further supported by the evolution-informed psychoeducation about our (‘tricky’) brains, offering a de-shaming framework for recognising that our fear patterns are understandable and not our fault. Some readers may have heard CFT sometimes being referred to as a ‘3rd wave CBT’, which suggests that it is the iteration of an existing therapy model. But that is not the case. While CFT does indeed use several cognitive and behavioural interventions, along with many others, it is based on a different theoretical model and science of process, which is what ultimately guides, patterns, and textures how its interventions are used.
This introductory section will outline the CFT model, particularly highlighting its theoretical roots in the evolution of social mentalities. The following three sections will then review the evidence that supports (i) the application of CFT in psychosis, (ii) the processes targeted in CFT, and (iii) the acceptability and usefulness of this approach.
In the CFT model, we are invited to consider our brains as ‘pattern generators’ (Gilbert, 2009), in that we have evolved a range of motives that organise our minds and bodies in ways that help us to pursue basic life tasks (such as avoiding harm, acquiring resources, and resting). It is these evolved motives that organise how we think, feel, pay attention, behave, etc., and in the case of social motives, how we organise our minds for roles and relationships (such as for competitive or for caring, cooperative relationships). A key insight in CFT is how these social motives not only organise our relationships with other people – for example, detecting if others will be threatening or friendly towards us – but the relationships we have with ourselves, and if we are voice-hearers, the relationships with our voices too.
In CFT for psychosis, a lot of the distress and suffering experienced by people is thought to be the result of being caught up in threat-focused patterns and motives that are evolved for organising social rank, i.e., whether someone is dominant (up-rank) or subordinate (down-rank). These competitive motives and mentalities are evolved to be very attuned to detecting and responding to threats from people and will be particularly active for those who have experienced trauma and harm inflicted by others. Within the social ranking system, there are either up-rank or down-rank protective strategies – either (a) assume the dominant role and protect ourselves by threatening subordinates and retaining power, or (b) assume the subordinate role and protect ourselves by becoming submissive. This pattern can also play out in our self-to-self relationship, for example, if we are being self-critical, we may be assuming both roles (the dominant and subordinate), essentially protecting ourselves by stimulating our own brain/body into submissiveness. The experience of distressing voice-hearing in psychosis has also been modelled in the same way, through earlier work by Birchwood and Gilbert (Birchwood et al., 2004; Gilbert et al., 2001) and in a theoretical review by Heriot-Maitland et al. (2019). The suggestion is that our brain and body can end up responding to its own internal (threat) creations, which is partly why people can get so caught and entrenched in these patterns.
If a person's social mentalities for threat are highly sensitised, as is often the case for people with psychosis (Gumley et al., 2010), their threat emotions will be easily stimulated. Social mentality theory (Gilbert, 2009) would suggest that when a threat emotion is switched on in the body, the brain will create a pattern of threat signals and threat responses that match or ‘fit’ this emotion. For example, if someone is walking through a dark forest at night and feeling very scared, their brain might generate anticipated threat signals – such as the sound of footsteps behind them – that ‘fit’ (or provide an orientating framework for) the emotion in their body. Not only does this threat signal provide a conceptual orientation for the emotion (to make sense of it), but it also sets up and orientates a protective response, e.g., preparing the body for escape and providing a direction in which to run. Self-generated threat signals can be attached to a person's internal or external world. In the internal world, sensory fragments of threat memories might become activated, such as sound or smell memories that fit with a threat emotion, or perhaps the memory of an attacker's felt presence. The example of self-criticism was already highlighted earlier as an internally-generated threat signal (Longe et al., 2010). Examples of threat signals in the external world might include, e.g.: superimposing threat or personally significant words into the muffled speech heard from another room; or reading hostile intentions into the brief glance from a stranger in the street. If there is a threat emotion switched on in our bodies, our brains will attempt to find a home for it – an orientating framework of threat signal detection, memory, and anticipation, even if these signals have nothing to do with what actually stimulated the emotion (Bargh, 2017; Haidt, 2001). The main thing is not how factually/objectively accurate these threat signals are, but how effective they are at orientating us and switching on our protective responses (Heriot-Maitland, 2023).
This paper will return to the research studies that support social mentality conceptions of psychosis in the evidence review sections that follow.
Guided by this evolutionary understanding of motivational systems and social mentalities (our brain's pattern generators), a key ‘position’ argument for CFT for Psychosis is that rather than trying to modify processes within threat-focused systems (as with many psychosis interventions), the therapeutic aims are instead to help people access and switch into care-focused motivation systems that organise bio-psycho-social processes (and relationships) differently. In CFT, we target building and activating the caring motivation – the idea being that if we can build up these inner motivational capacities, this will help us to care for, heal and resolve the harmful effects that trauma and threat have had on our psycho-physiological processes, as well as on our ability to feel safe and to socially trust.
This is an important illustration of how, in CFT, the therapy model is key in guiding the therapy process, which in turn directs how the techniques and interventions are used. As mentioned, CFT would emphasise switching on the caring motivational system first, rather than doing the work within the threat system. This has already been demonstrated earlier in this paper with the example of exposure work, but it applies to all the many integrative bio-psycho-social techniques that are used in a CFT framework. If we were to stay in the social rank system when using our interventions, say a cognitive technique, then we may find that we are limited to only working within that one cognitive landscape (the landscape of threat). As our motives organise our thinking and attention, the issue here would be that we'd have a mind that's only able to think/pay attention in terms of up-rank or down-rank. This could impact all sorts of therapy processes. For example, our therapy goals may be generated from insecure striving (e.g., achieving goals to prove oneself or to avoid inferiority), as opposed to self-care. Or it could be that if something went wrong, such as we failed to reach a goal, or had a setback, we might be susceptible to becoming very critical of ourselves.
CFT would suggest against working solely within the threat system to try and somehow make it more caring. It's not evolved for caring. Its evolutionary ‘job description’ is something else, and this is not going to change however much evidence to the contrary we throw at it. So instead, from a CFT perspective, we need to try and move out of the threat system and switch on the caring system. In CFT, we do this by integrating Compassionate Mind Training (CMT) into therapy. So, for example, we might first slow down, bring our attention to the body, to a grounded posture and a soothing breathing rhythm – engaging the neurophysiological systems that evolved for caring, attachment and affiliative behaviour (e.g., oxytocin, vagus nerve, frontal cortex). A key phrase in CFT is ‘using the body to support a compassionate mind’ Gilbert and Simos (2022, p 283). This means establishing safeness in the body, which allows our mind to start activating processes and competencies that are conducive to compassionate relating, e.g., empathy, mentalising, and psychological flexibility. This opens up a new landscape of thinking and attention that supports compassion, whereby we can notice our distress and develop a caring intention and motivational stance towards ourselves.
This is why, CFT would argue, it's important to have a model that informs our therapy process, as opposed to just deploying various therapy techniques (even though these techniques are, themselves, evidence-based). It's not particularly helpful to just throw in a breathing exercise, for example, without an understanding of what we are trying to achieve with breathing in terms of brain / body functions, or without an understanding of how this might land with the person. In some cases, it may be unhelpful, or even harmful, to use a technique without a model of understanding why we are using them, especially if the person's reaction to this technique is unexpected. With breathing practices, for example, many people will have a ‘threat’ reaction the first time(s) they try it, and the concern is that the therapist abandons the technique, feeling that something has gone wrong. Another issue is that a therapist might try to use soothing breathing for anxiety when actually behind the anxiety is a fear of anger or assertiveness (Gilbert & Simos, 2022). Similarly, with compassionate imagery, these techniques might elicit all sorts of reactions, from fear to shame, to sadness. The therapy model is key in making sense of this and how to navigate this. In CFT, while the overarching aim is to activate the care and attachment motivation system (and the physiological processes that support this), a lot of the therapy is geared towards understanding and navigating the fears, blocks and resistances (FBRs) that arise (Gilbert & Simos, 2022; Matos et al., 2023). Each motivation system carries its own memories, and for many, the care and attachment system carries memories of trauma, either abuse (presence of threat) or neglect (absence of care). So, when the care system is activated, these trauma memories can start coming to the surface and elicit protective responses, such as fight, flight, freeze, and shut down. So, for these people, the care system needs to be detoxified and relearned as something safe, not threatening. This is a gradual process that takes time. In CFT, these are not seen as problems that interfere with therapy, nor signs that something has gone wrong – they are the therapy.
It is beyond the scope of this paper to describe how the CFT for psychosis model is operationalised in terms of formulations and interventions. A recent edited book by Gilbert and Simos (2022) covers this in-depth, including a chapter specifically on psychosis applications. Having introduced the model, the focus of this position paper is to now review the current state of the evidence that supports the model, its targeted processes, and the feasibility and usefulness of applying CFT in psychosis.
EVIDENCE THAT SUPPORTS THE THERAPY MODEL IN CFT FOR PSYCHOSIS
Social rank theory and attachment theory
Gilbert et al. (2001) investigated social rank processes in 66 voice-hearers with a diagnosis of schizophrenia, finding that their voices were experienced as powerful and activated protective responses (such as fight/flight, as well as depression). This led the authors to consider how voice-hearing may be understood in the context of how our minds have evolved to organise social roles (particularly dominant-subordinate) – hence drawing attention to the voice-to-hearer relationship as a key focus of research interest. A series of subsequent studies by Birchwood and colleagues provided more support for this model, finding that voice-hearing relationships operate a lot like social relationships, that people's relationships with others mirrored their relationship with voices, and that the degree of powerlessness in their voice-to-self relationship was linked to levels of distress and depression (Birchwood et al., 2004). Using an evolutionary lens to interpret these findings prompts interesting questions about the function of voice hearing, which appears to switch on a particular social role orientation in people – one that elicits protective responses. So, the question is whether voice hearing might be functioning to protect individuals from threat and harm, particularly among those who have experienced harm from powerful others in the past.
The research on attachment theory in psychosis has typically focused on whether insecure attachment styles are associated with problems for people with psychosis, particularly due to difficulties in emotion regulation, and the impact this may have on psychotic symptoms and other outcomes. Evidence for the attachment model has been provided by several reviews (Berry et al., 2007; Gumley et al., 2014; Lavin et al., 2020). Liotti and Gumley (2008) highlight three mechanisms where attachment disorganisation may have an influence in psychosis: inhibiting emotion regulation, amplifying dissociative processes, and creating mentalising difficulties. In their Compassion Focused model of recovery after psychosis, Gumley et al. (2010) highlight how insecure attachment experiences can have consequences for the development of both care-focused and threat-focused social mentalities. They suggest that for these people who have not experienced secure attachment, not only will the affiliative care-focused systems be under-stimulated, but also the threat-focused systems (such as the social rank system) may be over-stimulated. After reviewing the evidence for these processes among people with psychosis, they summarise that “not only may the threat system be highly sensitized and vigilant to certain types of threat, the maturation of the soothing system, linked to, and developed via affectionate inputs from caregivers and others seems compromised, greatly interfering with threat regulation” Gumley et al. (2010, p 196). Liotti and Gilbert (2011) highlight the role of attachment in building social safeness experiences (linked to oxytocin systems and the vagus nerve), and the importance of social safeness capacities in regulating threat responses and developing successful mentalisation.
De-pathologising, de-shaming and empowering
An evolutionary approach to psychosis, and mental health conditions more broadly, offers a viewpoint that considers people's difficulties as being the processes of ‘normal’ brains operating in, and adapting to, environments, rather than the result of ‘abnormal’ brains. This helps us to distinguish between processes that have adaptive functions, and those that are pathological – a distinction that has often been overlooked in medicine more generally. For example, viruses are clearly abnormal, but the way our immune system tries to flight them and causes us symptoms is not. Breaking a leg is abnormal, but the intense pain we experience is not. Other symptoms like coughing, sweating, vomiting, and diarrhoea are all defence reactions – strategies that our bodies use to protect us in some way, e.g., from toxic substances that we may have taken in. Their occurrence is not a pathology, but rather a sign that our body is functioning perfectly well and adapting. So, it is important in physical health to distinguish between what is a genuine pathology and what are our adaptive reactions. And the same is true for mental health. With an evolutionary approach, we are invited to consider whether mental health symptoms may be understandable reactions to toxic situations or environments that have limited our ability to grow in certain ways. From a CFT perspective, in particular, the emphasis would be on our evolved needs for attachment and caring relationships. Therefore, the abnormality for people struggling with mental health would not be located not so much in their brain or mind, but more in the social contexts and environments where these relationships were not available.
The unpleasantness of defences is just part of their utility (Nesse, 2019 p 41)
This kind of evolution-informed psychoeducation is common for people who access services for help with anxiety, but far less common for those with psychosis. For people with psychosis, the traditional (and still dominant, mainstream) view, is that experiences like voice-hearing and delusions are seen as the manifestations of a brain abnormality or dysfunction. This can often lead a person into attempts to fight, suppress or control their experiences, which are seen as bad or wrong. It can also lead to alienation/disconnection (‘I am different to other people’), shame and self-stigma (‘there's something wrong with me’). In short, there is often a considerable amount of threat about threat (or put another way, struggle with a struggle).
The evolutionary approach involves starting from a premise of adaptive function, not pathology. A premise of pathology considers that if someone is experiencing distressing symptoms, it means something is wrong. A premise of adaptive function considers that there might be reason for having these experiences, that we all have brains capable of these experiences given the right environments, and that these may be strategies of some kind. This immediately invites us to be more curious about what has happened in this person's life: What situations have you been in? Which brain systems have been activated/exercised from being in these situations? (And which have not?). Importantly, this approach can also lead the person to develop an understanding of their lived experience that is both de-shaming (it's not my fault that we have evolved brains like this, nor that my brain has become shaped by social contexts I did not choose) and empowering (I have choices about which contexts to start building to allow my brain to shape/grow in different ways). Striking this balance between de-shaming and empowering is a crucial part of CFT and is greatly served by the evolutionary, bio-psycho-social framework.
EVIDENCE THAT SUPPORTS THE PROCESSES TARGETED IN CFT FOR PSYCHOSIS
Evidence for psychosis difficulties being located in threat patterns
There is considerable evidence that people with psychosis have heightened levels of threat-information processing. In an early review of evidence from cognitive, psychophysiological, neuropsychological neuroimaging studies, Green and Phillips (2004) highlighted that in clinical populations with paranoia, there is an initial (automatic) attention bias towards threat stimuli and a later (responsive) attention bias away from these stimuli. They suggest that this may be due to both vigilance towards, and active avoidance of, the threat. This fits with the social rank theory of psychosis described earlier – that a psychotic experience may involve a simultaneous activation of both the stimulus and response aspects of the threat system (Heriot-Maitland et al., 2019). In CFT, Gilbert refers to social mentalities as evolved algorithms, which he describes as simple stimulus–response profiles of “if A then do B” Gilbert (2020, p 3). So, in the same way that voice-hearing was described earlier as switching on both the threat stimulus (e.g., a critical voice) and the protective response (e.g., a submissiveness elicited in the hearer), the Green and Phillips (2004) review suggests a similar process may be occurring in paranoia – i.e., an increased stimulation of both threat-detection (if A) and -response (do B).
A series of virtual reality (VR) studies by Freeman and colleagues have further supported models of paranoia that focus on threat-biased information processing. One study specifically tested social rank by experimentally manipulating perceptions of social status by lowering the height (tallness) of participants using VR. They found that lower height was related to lower perception of social rank and higher levels of paranoia, and changes in paranoia were fully mediated by changes in social rank (Freeman et al., 2014). Again, linking these findings back to CFT terminology, the experience of paranoia is seen to be directly related to the activation of the evolved mentalities for organising social rank relationships.
Another line of research has investigated shame, self-criticism, and self-stigma among people with psychosis. As described in the introduction, CFT regards these patterns of self-to-self relating as aspects of social rank mentality as well. The same algorithms that organise our relationships with others can also become internalised and adopted by our own self-monitoring. Essentially, the dominant-subordinate roles ‘play out’ in the relationship with ourselves (one part of us in the dominant/threatening ‘up-rank’ role, while another is in subordinate/threatened ‘down-rank’ role). Therefore, further support of the CFT model of social rank mechanisms in psychosis is provided by evidence that people with psychosis often struggle with shame (Keen et al., 2017; McCarthy-Jones, 2017; Michail & Birchwood, 2013; Turner et al., 2013; Upthegrove et al., 2014; Wood & Irons, 2016), internalised stigma (Pyle et al., 2015; Wood et al., 2016) and self-stigma (Corrigan et al., 2009; Watson et al., 2007).
An exciting new development in the literature – with real potential in guiding our future understanding of psychosis – is the research into the dissociative properties of shame (Dorahy et al., 2017; Kouri et al., 2023). If shame is indeed dissociative, as these studies suggest, then shame could be a prime candidate for future research into the pathways of causation/maintenance of psychosis (Heriot-Maitland et al., 2021).
Evidence for benefits of switching to compassion patterns
In CFT, the aim is to pattern-switch from our ‘threat mind’ to our ‘compassionate mind’, guided by the understanding that these are distinct patterns of neural circuitry that have evolved for different functions. At the level of self-to-self relating, these patterns are also expected to manifest differently, for example, switching from self-criticism, self-stigma, and shame (threat) into self-compassion (compassionate). Neuroimaging (fMRI) research has identified differential patterns of neural activity between self-criticism and self-reassurance (Longe et al., 2010) and neural networks associated with threat are found to be heightened when participants are being self-critical, but reduced when practicing compassion (Kim et al., 2020), which is further validation of the ‘pattern-switching’ rationale. Using heart-rate variability (HRV) measurements, the same researchers also found that compassion training was associated with increased parasympathetic responses (Kim et al., 2020).
Other areas of evidence that support the CFT rationale of switching to a compassionate mind comes from a range of studies into the benefits of self-compassion and Compassionate Mind Training (CMT). In the non-clinical population, the associations of self-compassion have been researched extensively, identifying a range of psychological benefits, such as reduced rumination and worry (Raes, 2010), increased self-improvement motivation (Breines & Chen, 2012) and well-being (Neff & McGehee, 2010). Self-compassion (compared to self-esteem) is associated with a more stable sense of self-worth (Neff & Vonk, 2009), and of particular relevance for the current focus on psychosis mechanisms, self-compassion is also negatively related to social comparison, public self-consciousness, and a need for cognitive closure (Neff & Vonk, 2009). A meta-analysis of the relationship between self-compassion and psychopathology found a large (negative) effect size of r = −.54 from studies that used measures of depressive, anxiety, and stress symptoms, (MacBeth & Gumley, 2012).
Self-compassion has also been investigated in clinical populations. In a sample of those with problematic substance use, self-compassion was found to mediate the relationship between childhood trauma and emotion regulation difficulties (Vettese et al., 2011), and in a sample of people with trauma, those experiencing more types of interpersonal trauma were characterised by avoidant attachment and lower self-compassion, and were more likely to have severe posttraumatic symptoms (Bistricky et al., 2017). For people with depression, self-compassion was found to be an effective regulator of depressed mood (Diedrich et al., 2014). In psychosis, higher self-compassion is found to be associated with reduced severity of, and less distress from, voices (Dudley et al., 2018), and to lower scores on the PANSS measure for positive symptoms, excitement, and emotional discomfort, but not negative symptoms (Eicher et al., 2013).
Evidence for Compassionate Mind Training (CMT), and addressing fears of compassion
A number of studies have explored the impact of CMT on shame and self-criticism; for example, in a group of people with high shame levels, CMT was found reduce self-criticism, shame, inferiority and submissive behaviour (Gilbert & Procter, 2006). CMT groups for the general public are also found to improve scores on social rank and well-being, as well as self-criticism, attachment anxiety, and distress (Irons & Heriot-Maitland, 2021). In a study that used biomarkers to assess the impact of self-compassion training on social evaluative threat, the group engaged in self-compassion training, compared to control groups, showed reduced sympathetic nervous system reactivity following a social stress task (salivary biomarkers), more adaptive parasympathetic responses (heart rate variability), and milder subjective anxiety (self-report; Arch et al., 2014). The authors suggest that self-compassion “may activate a biological caregiving system implicated in both bonding and stress regulation” (p 55), which aligns with the theoretical standpoint of CFT and attachment theory described in this article. Steffen et al. (2021) found that for participants attending a 12-week group CFT intervention, increases in self-compassion were associated with increased HRV. Other biomarker studies have demonstrated that compassion training also impacts a range of neural circuits, immune system, and the autonomic nervous system (Singer & Engert, 2019; Weng et al., 2013, 2018). An important study by Trautwein et al. (2020) showed that specific types of training (e.g., whether it's practice in mindfulness, compassion, or a perspective-taking) will have specific effects on plasticity in different cognitive and social functions (e.g., attention, compassion, and theory of mind). This reinforces the point made in the introduction that it may not be helpful for therapists to add in interventions and techniques without a guiding model of process of what patterns are being targeted and why.
The evidence for fears of compassion being associated with poor mental health outcomes is summarised in a recent meta-analysis by Kirby et al. (2019), using data from 4723 participants in 19 studies who have competed the Fears of Compassion Scales (Gilbert et al., 2011). They found strong associations between fears of compassion with mental health outcomes such as shame, self-criticism, and depression. These associations were significantly stronger in clinical populations, compared to non-clinical. This provides a clear rationale for why targeting fears of compassion is an important and integral part of CFT. The psychometric testing of these scales has already confirmed good reliability and validity in a sample (n = 196) of people with psychosis (Carvalho et al., 2021), so this may be a useful tool for future research into understanding the common blocks and barriers when training compassion in this group.
EVIDENCE THAT SUPPORTS THE FEASIBILITY AND USEFULNESS OF CFT FOR PSYCHOSIS
Evidence for CFT
There is increasing evidence for the benefits of CFT in targeting these processes across various clinical populations, including complex difficulties (Gilbert & Procter, 2006), personality disorders (Lucre & Corten, 2013), eating disorders (Gale et al., 2014), and bipolar affective disorder (Gilbert et al., 2022). The bipolar study used a mix of self-report and HRV measures, and focus groups, finding that HRV significantly improved during the group, and concluding from focus groups that “participants were able to switch from competitive focused to compassion focused processing with consequent improvements in mental states and social behaviour” (Gilbert et al., 2022). For systematic reviews of the therapeutic benefits of CFT, see Leaviss and Uttley (2015) (14 studies, 3 RCTs) and (29 studies, 9 RCTs), and for Compassion-Based Interventions (more generally, including CFT), see Kirby et al. (2017). A recently published meta-analysis included 15 studies of CFT in clinical populations, reporting that CFT improved scores of self-compassion (effect sizes 0.19–0.90), self-criticism (effect sizes 0.15–0.72), and fears of self-compassion (effect size 0.18; Millard et al., 2023).
Evidence for CFT for psychosis
For psychosis populations specifically, the CFT evaluation research is still in the early days, and mainly focused on feasibility and acceptability, with small samples. In a case series (n = 3) of CMT for people who hear malevolent voices, Mayhew and Gilbert (2008) found that working with self-critical thoughts (as opposed to with critical voices directly) led to a reduction in the malevolence of voices. In a group of people with psychosis, CMT was found to reduce levels of social rank comparison and shame (Laithwaite et al., 2009). Braehler et al. (2013) also studied CFT as a group intervention for people with psychosis, finding that the CFT group was feasible, with participants reporting pre- to post-group reductions in depression associated with psychosis compared to a control group. Brown et al. (2020) used a VR social environment to test the impact of compassionate imagery exercises on scores of paranoia in a non-clinical sample. In one study, the imagery intervention was designed to target self-compassion, and in another, it targeted compassion to others. Both interventions successfully improved scores on compassion to self and others respectively, and both interventions also led to a reduction in paranoia. In study 1, the change in self-compassion explained 57% of the change in paranoia, and in study 2, the change in compassion for others explained 67% of the change in paranoia.
Another brief compassion-focused imagery intervention was found to improve paranoid thoughts in a group of non-clinical participants, an effect that was mediated by reduced negative emotions (Lincoln et al., 2012). However, in a sample of clinical participants with paranoia, the same brief compassion-focused imagery (vs. a control imagery) intervention only showed significant effects of positive self-relating (self-reassurance), but not on negative self-relating (self-criticism) or paranoia. The authors suggested that the intervention was likely too brief (10 min) to impact self-criticism and paranoia scores in a clinical population and that longer and more intensive interventions would be needed (Ascone et al., 2017). In a later study of clinical participants with persecutory delusions, a longer (four-session) compassionate imagery intervention was found to improve scores on self-compassion, negative beliefs about the self, and paranoia (Forkert et al., 2022).
Several single case studies provide evidence for the feasibility and acceptability of longer term (formulation-based) CFT for psychosis (Ellerby, 2013a, 2013b, 2014a, 2014b; Heriot-Maitland & Levey, 2021; Heriot-Maitland & Russell, 2018). In a case series study of people with distressing positive symptoms (Heriot-Maitland, 2020; Heriot-Maitland et al., 2023), we delivered a 26-session (manualised) individual CFT for psychosis. At the group level of analysis (n = 7), there were significant improvements in process and outcome measures, with interesting trends in when the improvements occurred: voices, stress, social rank, shame, and self-criticism all improved quickly (first half of therapy); whereas delusions, depression, and self-compassion all improved slowly (second half of therapy). The finding that self-compassion was slow to change might reflect the process of addressing fears of compassion, which, as mentioned earlier, takes time. Measures of anxiety and dissociation did not improve pre- to post- at the group level; however, at the individual level, a session-by-session measure of dissociation was found to improve significantly in half of the cases that provided scores. Distress levels associated with voices and delusions came down for every single case in the study; however, there was an intriguing trend for when distress came down. For over half of the cases with voices, voice-related distress went up first (midpoint) before it came down (endpoint). Similarly, with delusion-related distress, improvements were far more pronounced in the second half of therapy (sessions 13–26). An important implication is that CFT should not be shortened or rushed, particularly for complex populations where fears of compassion are likely to be a primary feature of the work. Further research is urgently needed in larger-scale effectiveness (RCT) studies to evaluate and advance the field of CFT for psychosis.
STRENGTHS, CONTROVERSIES, AND PRIORITIES FOR FUTURE RESEARCH
In this section, two tables are presented to summarise the (a) strengths, (b) challenges, gaps and controversies, and (c) questions for future research. Table 1 focuses on the model of therapy, psychoeducation, and formulation, and Table 2 on the processes targeted, interventions, and evaluation. Many of the strengths of CFT for psychosis lie in its consilient (bio-psycho-social) model and how the model is underpinned by a science of psychological processes. The priorities for future research are to run larger-scale RCTs of this approach, with measures that are anchored to the specific biological, psychological, and social (preferably all three) processes outlined in this paper.
|
(a) Strengths
|
(b) Challenges, gaps, and controversies
|
|
(c) Questions for future research
|
|
(a) Strengths
|
|
(b) Challenges, gaps, and controversies
|
(c) Questions for future research
|
CONCLUSION
CFT is a transdiagnostic approach to understanding and helping people with mental health difficulties. It is transdiagnostic because its processes and interventions are not guided by any specific mental health condition, pathology, or difficulty per se, but by an evolutionary understanding of the nature of minds and why they function in the way they do. A full review of the evidence for these processes, and descriptions of how the theory–research–practice links are applied in CFT, is beyond the scope of this paper, but are reported in a recent publication for clinicians (Gilbert & Simos, 2022). The focus here has been on the evidence most relevant to understanding key mechanisms in psychosis, namely social rank/shame, attachment, and dissociation, and how CFT is suited to targeting these mechanisms.
This paper has highlighted that the strengths of CFT for psychosis lie in the process evidence, and the strong theoretical grounding in the science of biopsychosocial processes. For outcomes evidence, CFT is behind some of the other therapies in this issue in terms of number of RCTs conducted; however, CFT evaluation research is growing all the time, and in CFT for psychosis, all the feasibility/acceptability phases are now in place to run a trial. One of the challenges going forward is for psychosis researchers to distinguish between the models of therapy so that new approaches do not simply get incorporated under the existing umbrellas of ‘CBT for Psychosis’ or ‘third wave CBT’. CFT for psychosis is a popular approach among clinicians and people with lived experience, and professional training/supervision as well as community events have been running consistently over the last 9–10 years, in the UK and internationally. There may be a danger that if the RCT evidence does not catch up soon with the interest (even though the process evidence is very strong), then therapists will find themselves in the tricky situation of having to call what they are doing ‘CBT for Psychosis’ anyway, which compounds the problem of not having a clear orientating model of how attachment and compassion operate in brain patterns and in therapy, and how to navigate the barriers that arise from doing this work. Several research directions are suggested in Tables 1 and 2, but perhaps the most urgent need is for a randomised controlled trial (RCT) of CFT for psychosis, building on the feasibility research, manual, and clinician interest and training/supervision structures that are already in place.
Finally, this paper has highlighted the value of building therapy for people with psychosis upon the science of biopsychosocial processes and ensuring that these processes are guiding each stage of the therapy, from sharing psychoeducation, through texturing techniques and interventions, to measuring outcomes.
AUTHOR CONTRIBUTIONS
Charles Heriot-Maitland: Conceptualization; writing – original draft.
ACKNOWLEDGEMENTS
Most of the ideas and content for this paper directly resulted from the author's work studying CFT for psychosis at King's College London and University of Glasgow and put together as a position paper while working at Balanced Minds. For the purposes of open access, the author has applied a Creative Commons (CC BY) licence to any Accepted Author Manuscript version arising from this submission.
CONFLICT OF INTEREST STATEMENT
The author is a practitioner, supervisor and trainer of CFT and has received fees for providing these services. The author receives royalties from a book they have published on CFT.
Open Research
DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.
