Tuesday, 10 April 2012

CBT: A Philosophical Critique

A preliminary draft of a chapter to appear in the forthcoming Oxford Handbook of Philosophy and Psychiatry.           

                  CBT: A Philosophical Appraisal


Cognitive Behaviour Therapy (CBT) theorists propose that disturbances in the content of, or in our relationship with, our cognitions often underlie and maintain emotional disturbance. Accordingly the cognitive addition to behavioural therapy typically consists in the collaborative pinpointing and challenge of, or encouragement of disidentification from, these cognitions by the therapist and the patient. With the right group of problems, patients and therapists, the practice of CBT is well known to possess therapeutic efficacy. This chapter however primarily considers the theory rather than the therapy of CBT; in particular it looks at the central significance it gives to cognition in healthy and disturbed emotional function. It suggests that if ‘cognition’ is used to mean our belief and thought, then CBT theory provides an implausible model of much emotional distress. If, on the other hand, ‘cognition’ refers to the processing of meaning, then CBT risks losing its distinctiveness from all therapies other than the most blandly behavioural. The chapter also suggests that the appearance of distinct causal processes and discrete intervention sites in CBT’s causal models of psychopathology may owe more to the formalism of the theory than to the structure of the well or troubled mind. Finally it questions the CBT theorist’s conception of the value of self-knowledge, suggesting that the CBT theorist risks underplaying the value of some of the CBT therapist’s interventions. A consequence of these critiques is that there are reasons to doubt that CBT always works, when it does, in the manner it tends to describe for itself.

Introduction: CBT and Philosophy

CBT is a broad church embracing many theories, models and techniques, often now held in conjunction with other approaches to the mind in health and distress (e.g. compassion, mindfulness, dialectics, psychodynamics, etc.). Within this church there are denominations that complement, but also those that compete, with one another, both in their empirical claims and also in some of their underlying theoretical principles (Hofmann & Asmundson, 2008).

There are, for example, those who treat the ‘cognitive’ aspect of CBT theory as referring specifically to maladaptive thoughts and beliefs (e.g. Beck 1979). These psychological states are considered distinct from perception, emotion and behaviour. Precisely because they are seen as distinct, maladaptive cognitions can be appealed to in non-circular causal explanations of disturbances in these separate domains of perception, emotion and behaviour. However other theorists tie what is ‘cognitive’ not specifically to thought and belief, but rather to the processing of meaning (e.g. Teasdale 1997). Without assuming that meaning is an exclusive function of thought or belief, it is instead allowed to already belong properly to emotion, behaviour and perception. In such cases the explanatory force of a cognitive model may derive from its identification of the distinct structures of meaning immanent within emotion and behaviour themselves, rather than from showing how antecedent thoughts allegedly give rise to such disturbed emotion and behaviour.

For these and other reasons it therefore makes little sense to offer a philosophical critique ‘of CBT’ per se, and such a wide-ranging critique is not attempted in this chapter. Nevertheless, underlying several popular CBT models and treatments (henceforth ‘CBTs’) are, I shall suggest, several questionable assumptions about the significance of cognitive factors in psychopathology. In this chapter I add to previous critique (McEachrane, 2009; Whiting, 2007; Robertson, 2010) in specifying four related philosophical misunderstandings about the nature of the mind in health and distress that, I claim, can sometimes affect CBT theorist’s understanding both of psychopathology and of therapeutic action.

Before I begin I think it might be helpful if I set out, using some examples, what makes for a distinctly philosophical critique of a particular CBT theory. Consider first the following questions that cans be, and have been, asked of the CBTs:

For what range of difficulties, experienced by what kinds of populations, are CBTs effective? Are CBTs more or less effective than other therapies (e.g. psychodynamic, narrative, person-centred or drug therapies) and practical interventions (e.g. employment and dating advice, social sport and gardening)? How durable are the effects of CBTs compared with other treatments? When a CBT is effective, can this be explained using typical CBT models of psychopathology and therapeutic action, or must we account for it using ideas taken from other cognitive, systemic or psychodynamic theories of therapeutic action? When a CBT is found to be more or less effective than another therapy, is this to be accounted for ideologically (i.e. because of treatment effects due to the convictions of investigators, therapists or patients), or because of differential treatment efficacy, or because measures are being used which are tailored to what one but not another therapy would view as meaningful or valuable change (e.g. is it characterological or symptomatic changes that are being measured)?

These are all important questions for therapists, patients, and public health policy makers. Yet there is, I suggest, little that is distinctly philosophical about them. Part of what it means to say this is that they are not answerable through reflection alone, but must instead be answered by empirical investigation. The exception is the last: reflection on what counts as efficacy is a form of theoretical reflection on conceptual matters of meaning, rather than discernment of empirical matters of fact. Even so such theoretical reflection is not, in the sense in which I shall use the term here, itself particularly philosophical. For whilst all theoretical thought concerns itself with matters of meaning, what I have in mind by distinctly philosophical thought instead concerns itself with whether what is said can genuinely be understood in the way its author intends. Philosophical reflection in this sense concerns itself not simply with meaning but more particularly with meaningfulness; not with how, but rather with whether, something can really be understood in the way it invites us to understand it.

In what follows I articulate four ways in which CBTs can sometimes invite us to construe ourselves to ourselves, both in our healthy functioning and in our emotional distress, in ways that, it seems to me, go against what it means to be a human subject. More particularly, I shall consider certain conceptions, tacitly present within the psychopathological theories offered by various CBTs, of what it is to think, feel, and meaningfully respond. My claim will be that whilst these conceptions appear to motivate the practice of, conceptually dovetail with the theory of, and validate the distinctive scientific self-conception of, various CBTs, they nevertheless simply do not tally with what reflective understanding alone reveals as the character of human mental and emotional life. Given that such CBTs can nevertheless prove helpful to some patients, it follows that they must sometimes do so for reasons other than those they themselves suggest.

Aspects versus Causes

The CBTs, along with other cognitive approaches in psychology, often present their understanding of psychopathology in a ‘box and arrow’ format. The arrows are intended to represent causal relations; the boxes represent what are intended to be understood as isolable cognitive, behavioural or emotional factors. Such schematic models are appealing because they are clear-cut and appear to offer a cogent rationale for investigation and intervention. Consider the following highly condensed summaries of typical cognitive models:

[***Note to OUP editor: I would be grateful if the three following paragraphs could be put into a box, or into three boxes (of the sort that the Handbook might use when providing examples embedded in the text), and if the diagrams which go with the first and third paragraph could be redrawn in the same style as each other. The second diagram does not need its heading. Thank you! ***]

One aspect of contemporary CBT models of panic can be elaborated as follows: A patient experiences a bodily sensation. They catastrophically misinterpret this as a sign of danger (e.g. that they are fainting or having a heart attack). This leads to a further perception of threat, giving rise to further fear, which in turn gives rise to further sensations that are in turn catastrophically misinterpreted. The patient is left in a rapidly cycling self-fuelling anxiety state (cf. Clark 1986).

The depressed patient as encountered by the CBT therapist is an individual with underlying pessimistic assumptions that organise his or her experience of the world. When activated by critical incidents these assumptions are thought to lead to the development of negative automatic thoughts (e.g. gloomy and non-deliberative thoughts about the current situation, self or future). These then give rise to negative emotions (e.g. guilt or fear), a lowering in mood, and consequently to a reduction in motivation and activity and engagement. A vicious circle of depressed thoughts and depressed mood results (cf. Beck, 1979; Fennell, 1989).

The patient with social anxiety often carries underlying assumptions (e.g. ‘unless someone shows they like me, they dislike me’). The CBT model describes how these can be activated in social situations, resulting in a belief that they are in ‘social danger’ (e.g. believing that others will treat me badly). Such beliefs may then cause anxiety, which in turn can cause various somatic and cognitive symptoms (sweating, mind going blank, shaking). These in turn can heighten a self-conscious mode of attention in which the patient considers that they are, for example, likely to make fools out of themselves (cf. Clark & Wells, 1995).

[***end of boxes***]

Models such as these are, I find, often very useful in clinical practice since they clearly depict what may be the main meanings and maintaining factors in the various difficulties with which patients present, and immediately suggest strategies for intervention. My objection at this stage, however, is not to their clinical or heuristic utility, but to an aspect of their implicit theorisation of the mind. The objection is that they tend to encourage us to mistake a fact about our representations of the mind for a fact about what is represented. The fact about our representations is that we can separately depict various aspects of (say) a depressive state, such as our behavioural dispositions, thoughts, motivation, mood, affects, etc. My claim here is that this cannot by itself be taken to indicate a fact about what is represented: that there are separately existing assumptions and perceptions and moods and motivational states – states ‘in us’, as we can hardly now but put it – linked by causal connections (the arrows in the diagrams).

Take the case of panic. The CBT model separates out a stimulus, a perception of it as threatening, a state of apprehension, and bodily sensations, and places these separated phenomena into a causal sequence. Yet whilst all these ingredients are clearly present in a panic attack, and whilst we can separate them for formal (descriptive) purposes, it is not obvious that they really are, in the normal run of things, best conceived of as separate states within a person which causally trigger one another. A better description of them may accordingly be as different aspects of the same (anxiety) state. We are ‘stirred up’ – in both our action-readiness and in a somatosensory way – by fearful stimuli, and our being stirred up in this way is itself of a piece with our registering of that stimulus as fearful. The components of our fear response are, I submit, not isolable causally related states in us, but rather are merely notionally separable aspects of the state which we are in.

Much the same can be said about the CBT conceptions of depression and social anxiety outlined above. Something of the passive and objectifying voice of depressive anxiety sometimes appears to have found its way in to the CBT conceptualisation of the nature of our mental life per se. Thus we now have assumptions that can be ‘activated’; meanings become reified into occurrent processes and states to be known as ‘cognitions’; the agent become a locus of ‘behaviours’. Everywhere we have to do with nouns, and nowhere with verbs: to do with nouns that bolster an appearance that we are referring to multiple isolable entities and processes, rather than with their root verbs that refer to the action and attitude of a single living entity. The living agent who makes assumptions, understands his experience a particular way, and undertakes actions becomes instead a locus of separate inner and outer states which causally trigger one another.

CBT models can be useful as heuristics for thinking about psychopathology, and when they do outline genuine causal interactions (two genuinely causal interactions in the above models include fear and bodily sensations in panic, and inactivity and depressed mood in depression) they significantly help to guide therapy. Furthermore, their use of mechanistic and objectivising rhetoric which reduces diverse (causal, dispositional, constitutive, etc.) relations to causal relations is not in itself a mortal sin; the CBT theorist is after all not pretending to provide us with a nuanced metaphysics or phenomenology of mind. The criticism at this juncture is more modest. First, that certain cognitive models have the somewhat misleading appearance of providing more by way of a scientific psychological explanation than is really available. Relatedly they can appear to possess a greater degree of scientific precision than found in other models of therapy – although this appearance may be due only to their having misleadingly dressed up description of different aspects of the patient’s disturbance as a causal-explanatory model of it. Third that, because of this, they can seem to overstate the number of apparent junctures for intervention. But to summarise: as often as not, what such models are really describing are different aspects of the same state we are in, rather than describing scientifically teased out, causally linked, interruptible, states and processes within us.

Thinking versus Having Thoughts

Whilst all CBT theorists adhere to the centrality of cognition in emotional disturbance, not all theorists appear to agree on the meaning of the term, a term that in any case is most often left undefined. As described in the introduction, for some the term appears to refer to any aspect of psychological performance (e.g. in perception, action, emotion, judgement, etc.) that involves the conscious or preconscious processing or discrimination of meaning. Under consideration may be enduring attitudes – dispositions of thought, feeling and behaviour which may become sedimented in character – or passing emotions, perceptions, cogitations, imaginings, and behaviours. For others, including the founding fathers of CBT, the term refers more exclusively to our thinking, which in itself is understood as realised in the words and images that pass across our minds. Such thinking is conceived of as the second or third tier of a hierarchy of cognitive elements which move from the most subconscious and dispositional (the underlying schemata) level, through to more readily avowable maladaptive beliefs, up to occurrent automatic thoughts passing across the mind. What follows concerns only the latter sense of cognition (as occurrent thoughts passing across the mind), and draws in part on the thoughtful critique of McEachrane (2009).

Here are some examples of the idea that the content of (in traditional CBT), or the patient’s fused relationship with (in ‘third-wave’ CBTs such as Acceptance and Commitment Therapy or ‘ACT’), inner representational events maintains and mediates emotional disturbance:

CBT therapists encourage [patients] to become aware of their thoughts and thought processes. Cognitions are generally classified into negative automatic thoughts and dysfunctional or irrational beliefs. Negative automatic thoughts are thoughts or images that occur in specific situations when an individual feels threatened in some way. (Hoffman & Asmundson, 2008, p. 4)

When a person is able to fill in the gap between an activating event and the emotional consequences, the puzzling reaction becomes understandable. With training, people are able to catch the rapid thoughts or images that occur between an event and an emotional response. (Beck, 1979, p. 26)

ACT aims to alter the context in which thoughts occur so as to decrease the impact and importance of difficult private events. ... Clinically we want to teach clients to see thoughts as thoughts, feelings as feelings, memories as memories, and physical sensations as physical sensations. None of these private events are inherently toxic to human welfare when experienced for what they are. (Hayes et al, 2004, p. 8)

[In] cognitive fusion … the event and ones thinking about it become so fused as to be inseparable and that creates the impression that verbal construal is not present at all. ... A worry about the future is seemingly about the actual future, not merely an immediate process of construing the future. The thought "Life is not worth living" is seemingly a conclusion about life and its quality, not a verbal evaluative process going on now. (Hayes et al, 2004, p. 25)

In these examples we see clearly an idea that crops up not infrequently in the CBT literature: that to think something is to have a thought pass across one’s mind. We are, it is said, often oblivious to such inner events even occurring, perhaps (in the case of ACT) because we are over-identified or ‘fused’ with them. The therapist’s job is accordingly to help us notice and either challenge (CBT) or defuse from (ACT) these events that supposedly mediate between outer events and our feelings (on which more in the third section below).  But as McEachrane (2009, p. 86) suggests, ‘to say of someone that they ‘thought that p’ does not imply that they ‘thought of p’ or ‘thought about p’ or formulated p or that p occurred to them or was in their thoughts. … [If] a client says that in a particular situation they thought that, say, ‘I’ll never be like them’ or … ‘I’m not a likeable person’, then this does not necessarily mean that the client in that situation thought of these things, formulated these things to herself, that these things occurred to her or were in her thoughts.’ If you ask me what I think about the economic situation I need not wait on inner mental events to occur which I can then report. Instead I simply tell you how the situation seems to me, and in this telling I express what we could call my ‘attitudes’ rather than report on my occurrent ‘cogitations’.

It is important to be clear about the scope of this criticism. It is the aim of both CBT and other therapies to help the patient to take a more ‘ironic’ and flexible stance towards their own attitudes (Lear, 2003). Psychodynamic therapies, for example, aim to help a patient develop in their ‘mentalising’, to become aware that theirs is one among many possible ways of thinking about things, and to increase their mental playfulness or psychological flexibility (Holmes, 2009). Further, it is true that the depressed person can indeed become ruminatively lost in an inner world of depressed and depressing cogitations; in this case talk of ‘thoughts’ can indeed often be taken as talk of inner musings or imagery. Helping the patient become aware that this is going on, and encouraging them to instead harness their attentional resources to, make living meanings in, and increase their physical engagement with, the external world are well known to be therapeutically valuable. But treating everyday thoughts as if they are typically inner events not only falsifies the patient’s phenomenology, leading to possible ruptures in the therapeutic alliance, but also risks alienating the subject from their own attitudes.

This can in particular be seen with respect to the above-reported thoughts described by Hayes et al. (‘A worry about the future is seemingly about the actual future, not merely an immediate process of construing the future. The thought "Life is not worth living" is seemingly a conclusion about life and its quality, not a verbal evaluative process going on now.’) You ask me: ‘Why are you troubled?’ and I say ‘I’m worried about being made redundant, and the possible effect of this on my family life.’ Whilst it may indeed be possible for me to be overly worried and lose perspective, it is not at all clear that this is best described as a matter of fusion with ‘immediate processes of construing’ or with ‘verbal evaluative processes going on now’. By contrast with what Hayes suggests, we can and indeed should say that a thought that life is not worth living is indeed a judgment about life and its quality, regardless of whether the judgement is expressed in inner speech; and we also can and indeed should say that a worry about the future is indeed about the future, regardless of whether it manifests in an episode of worrying.

This difficulty does not adhere only to the third-wave CBTs such as Hayes’ ACT, but attends to any therapy which encourages the patient to change what he thinks (his attitudes) through changing his relation to or the content of his thinking (his cogitations). For what is surely natural is to suppose that our cogitations are often enough a function of our attitudes and not, on the whole, vice versa. If, that is, my attitude is that my life is not worth living then, when I am invited to attend only to my own cogitation I shall most likely be attending to it – say to my occurrent thought of ‘this is all hopeless’ – from just this attitude. In such a case my challenging of or distancing myself from cogitations that reflect my attitudes risks becoming a strategy promoting bad faith or self-alienation.

To summarise: the CBTs invite the patient to notice their thoughts when they occur and then to question them or otherwise get some distance from them. However the thoughts often reported on thought records or in therapy sessions are frequently best understood as attitudes that obtain, rather than as cogitations that occur. They are of the substance of the person who thinks them, rather than being entrancing inner events taking place in their minds to which they stand in some kind of relation. So we need to bear in mind that to question such thoughts is to question the patient himself. That is certainly a desirable goal of therapy, but it will be obvious that it is a far more challenging goal than one of simply helping the depressed patient to notice and question or suspend his involvement in his mental processes. The latter task can be readily conceived as a partly didactic and partly collaborative process; the former more delicate task however will involve first forging a strong alliance with the healthier part of the patient, bolstering their sense of themselves as worthwhile and capable, so that the possibility can even arise of disidentifying with their critical or demoralising inner voice.

It is important not to overstate the case. What we will often find, especially in depression, are certain depressogenic cogitations (‘ruminations’, ‘negative automatic thoughts’) that have become habitual and fail to reflect the underlying attitudes of the subject’s better self. This in fact is something specific about depression rather than a function of having a mind per se. In such cases learning to take a step back from rumination will of course be important.[1] Furthermore, therapists of all stripes will naturally want to help the patient shift their underlying depressogenic attitudes, and not merely help shift the depressive cogitations that stem from them.[2] A point about the phenomenology of depression must be separated from one about the logical grammar of ‘thought’: whilst rumination may well promote depressive mood in depressed patients, this must be evidenced on empirical grounds, and not simply follow from an a priori misconstrual of all thought as cogitation. It is however just such a misconstrual that sometimes finds its way into the CBT literature. As quoted above, Hoffman & Asmundson (2008, p. 4) tell us that ‘CBT therapists encourage [patients] to become aware of their thoughts and thought processes. Cognitions are generally classified into negative automatic thoughts and dysfunctional or irrational beliefs. Negative automatic thoughts are thoughts or images that occur in specific situations when an individual feels threatened in some way.’ Yet often when we feel threatened in some way - for example when in a social situation I think ‘they will reject me’ – this involves little or nothing by way of anything like a negative automatic thought passing across my mind, and instead concerns how I see the social situation I am in. To invite the patient to record the thoughts which occur to them (say in a CBT ‘thought record’), when the kind of thoughts that are at issue are not occurrent cogitations but rather how they take matters to be, is to risk a misalliance through a counter-therapeutic falsification of their phenomenology.

Feeling and Believing

Perhaps the central claim of much CBT is that disturbances of emotion arise from disturbances of cognition. This is often presented as a modern day form of the stoic principle that ‘men are disturbed, not by things, but by the principles and notions which they form concerning things’ (Epictetus, 2004, section V; Robertson, 2010). As discussed in the introduction, much will turn on the question of whether we are to understand cognition as i) referencing any meaningful uptake or information processing of ‘things’ (in which case all (non-behaviourist) psychotherapies including psychoanalysis must, in this sense, be counted as cognitive therapies), or ii) whether we are to understand it as referring instead to cognitive items such as acts of interpretation, the application or formation of ‘principles and notions’, or to those of our attitudes which are most aptly described as ‘underlying beliefs’. In this section I shall consider ii) this more restricted latter sense of cognition, and ask whether emotional disturbance is aptly theorized as consequent upon it.

Within the CBT’s it is the ‘ABC’ model of emotional disorder, imported from Albert Ellis’s Rational Emotive Behaviour Therapy (Ellis, 1991; Wilson & Branch, 2005), that most explicitly embodies the notion of cognitions as thoughts or beliefs. ‘A’ here stands for the activating events or situations experienced; ‘Bs’ are the subject’s beliefs about or interpretations of these events; ‘Cs’ are the emotional or behavioural consequences. The ABC model accordingly construes the cognitive state of belief as a determining intermediary between experience and emotional response. The viability of conceiving of the ‘C’ in CBT along the lines of the ‘B’ in ABC has been questioned by CBT theorists on both theoretical and empirical grounds (Kohlenberg & Tsai, 1994; Teasdale, 1997; Rachman, 1997). One rather obvious empirical objection to the stoic claim that we are affected not by things but by our interpretation of them comes from medical and conditioning approaches to anxiety: I can be made mildly anxious directly by very strong coffee (even if I mistakenly believe it to be decaffeinated); I may be directly emotionally affected by hormonal changes or head injuries; I may feel compelled to perform compulsions to reduce obsessional anxiety even whilst knowing there is no rational basis for this; and I can be afraid of stimuli, such as spiders or birds, which I believe to be perfectly safe (Whiting, 2006, p. 239) – hence the relevance of behavioural therapies for certain phobic or compulsive behaviours. Another objection comes from the observation that, whilst certain thoughts and beliefs may be depressogenic or anxiogenic, this may only occur for people who are already caught up in a particular frame of mind, perhaps driven by certain unconscious desires and emotions (Whiting, 2006, p. 241). This, in fact, is an objection anticipated by a key founder of CBT, Aaron Beck, who claimed only to identify the maintaining factors, and not the underlying causes, of depression (Beck et al., 1979; Clark & Steer, 1996). The question to be answered in this chapter, however, concerns whether there are distinctly philosophical grounds for questioning the idea that disturbed emotions are typically driven by beliefs. My claim below is that the notion that this is so can sometimes be seen to arise from a misunderstanding, inscribed in certain CBT theories, of what it means to be an emotionally reactive human subject.

A standard cognitive model of depression has it that aversive early experience can lead to the formation of dysfunctional assumptions which, when activated by a critical incident, lead to cycles of negative automatic thoughts and other behavioural, motivational, affective, cognitive and somatic symptoms (Fennell, 1989, p. 171). These assumptions or beliefs are said to take such forms as ‘If someone thinks badly of me, I cannot be happy’, or ‘I must do well at everything I undertake’, ‘I am inferior as a person’, ‘My worth depends on what other people think of me’ (Fennell, 1989, pp. 171-8). Therapy then consists in strategies such as teaching the patient to become aware of depressing thoughts as they occur, or undertaking tasks to test the truth of fixed negative beliefs (Williams, 1997, p. 265). The first thing to be said about such propositions is that, whilst the therapist probing for core beliefs may well elicit them, they are often far more naturally taken to express a patient’s feelings. McEachrane (2009, p. 92) quotes Ellis (1994, pp. 32-3) attempting to convince a patient of the opposite of this:

‘... I know I’m doing better of course, and I’m sure it’s because of what’s gone on here in these sessions. And I’m pleased and grateful to you. But I still feel basically the same way – that there’s something really rotten about me, something I can’t do anything about, and that the others are able to see. And I don’t know what to do about this feeling.’

‘But this “feeling”, as you call it, is largely your belief – do you see that?’

‘How can my feeling be a belief? I really – uh – feel it. That’s all I can describe it as, a feeling?’

‘Yes, but you feel it because you believe it. If you believed, for example, really believed you were a fine person, in spite of all the mistakes you have made and may still make in life, and in spite of anyone else, such as your parents, thinking that you were not so fine; if you really believed this, would you then feel fundamentally rotten?’

‘Oh, Hmm. No, I guess you’re right; I guess I then wouldn’t feel that way.’

However the logical error here seems to be Ellis’s rather than his patient’s. Ellis appears to be assuming that, since I am unlikely to feel rotten about myself if I truly believe I am a good and fine person, then it must be the case that my feeling rotten about myself is a product of my believing myself to be rotten. But what, we might ask, if we start with the idea that beliefs are often products of feelings rather than vice versa: might we not expect to arrive at the same state of affairs? That is to say: if I felt – really felt – good about myself then may I not, as a consequence, be disposed to form only positive beliefs about myself? And then might it not also be true that sometimes people feel that there is something rotten about themselves without having formed any associated belief – just as, say, a phobic person may find herself feeling scared of spiders or birds without believing that they are in any way dangerous? It may be hard to imagine truly believing that one is a fine person whilst also feeling one is rotten, but this may reflect nothing more than the powerfully constraining impact of feelings on belief formation. Furthermore there may be many cases in which feeling and belief are indistinguishable. Certain patients may defend against their painful emotions by couching their self-talk in non-affective idioms (sticking to ‘believe’, ‘think’ etc.), but this shouldn’t lead us to overlook the sometimes genuinely interchangeable roles that ‘feel’ and ‘believe’ may occupy. We should also not forget those uses of ‘feel’ which function precisely to draw a contrast with ‘believe’ or ‘know’: ‘I feel lucky today – although I don’t believe in luck’; ‘I know that you were in the right but I still feel hurt’.

One way in which a CBT theorist might try to salvage the idea that it is maladaptive beliefs that drive emotional disturbance could be as follows. Beliefs, they may say, possess intentionality and carry mental content (they are about this or that). Feelings, on the other hand, are merely ‘positive’ or ‘negative’ (‘hot’ or ‘cold’) emotions, sensations, or states of being. Feelings therefore need to be driven by beliefs before they can properly be said to be about anything or to express any kind of understanding. The reply to this is that it is hard to understand why we should accept such an impoverished conception of our feelings. For feelings often precisely are about something: I feel furious that he has wasted my time; she is sad because he has left without saying goodbye; he is delighted by her having accepted his proposal. And since our feelings are very naturally taken to be already about states of affairs, it is hard to see what help they need from intermediary beliefs.

Imagine that I always dreamed of being a psychologist but then fail my exams and become morose. This emotional state of mine worsens and I start to believe that I cannot be happy unless I am a psychologist. A CBT therapist who becomes wedded to the idea that intermediary beliefs or adherence to rules underpins emotional distress may propose that it is precisely this belief of mine that is driving my despondency. It is indeed possible that such additional beliefs may serve to maintain or worsen depression. What isn’t clear though is why much of my sadness may not be thought of as arising simply because I have not realized my life’s ambition. What is significant here is not an intervening belief about what I need to be happy, but more generally how I see my life, my outlook on or evaluative perception of my self, my situation and my future (cf. McEachrane, 2009, pp. 94-5). In a depressive frame of mind I may arrive at beliefs such as ‘I cannot be happy unless I am a psychologist’, but such beliefs are, I suggest, in the run of things more likely to result from my sense of my situation rather than vice versa.

What then are we to make of the not infrequent CBT insistence that beliefs have a key pathogenic role to play in depression and other emotional disorders? My own belief is that, whilst certain CBT therapists may sometimes be driven by an overly intellectual (belief- or rule-mediated) conception of our emotional perception of our worlds, what has more often happened is that they have stretched the meaning of ‘belief’ to fit the phenomenology and save the theory. For example, I once asked my then CBT supervisor what she meant by her talk of her patients’ ‘beliefs’, and she replied that you knew that what you had to do with was a belief when a patient cries in articulating it. This clearly has little to do with the normal notion of belief, given that the infinity of our everyday beliefs (that it is not raining, that most people have legs, that the next symbol is a closing parenthesis…) hardly provokes an infinity of tears. What instead we have to do with here may rather be the result of a common enough situation in psychiatry and psychotherapy generally: that the practical meanings of clinical terms become constituted more by their actual clinical uses than by associated theory, where such clinical uses are themselves a product of a distinctive lexicon being forced to carry whatever are the communicative and therapeutic burdens of the clinical encounter.

This explanation should not, of course, be mistaken for an exculpation. CBT theorists have as much a professional duty to communicate clearly and accurately as all psychologists, and misusing ordinary terms (‘belief’, ‘thought’) may be no less unhelpful than deploying the language of science when it adds nothing to what is better stated in everyday language.[3] And the fact remains that we do ordinarily distinguish between beliefs and feelings, and use the distinction to mark situations in which, say, we feel scared of a cat even whilst believing it to be safe. Furthermore cognitive neuroscience has now provided several models outlining different streams of ‘information processing’ for propositional and embodied understandings (cf Power & Dalgleish, 1999; Teasdale, 1997). Rather than think of the emotionally significant meanings which arise as a patient experiences their world as encoded in propositionally structured interpretations or beliefs, such theorists instead invite us to consider the prime significance of non-linguistic, somatosensory, forms of meaning. The clinical upshot is that therapy must aim at helping a patient change their way of being in the world through changing such non-belief-based meanings.

In this section I have added to such empirical critique by providing philosophical reasons for thinking that an ‘ABC’-style model falsifies what it actually means to be an emotionally reactive, meaning sensitive, human subject. If, when a CBT theorist tells us that emotional reactions are shaped by our ‘cognitions’ they mean by ‘cognitions’ something like our beliefs, and if by our ‘beliefs’ they mean something other than how we feel about or see our situations, then the CBT theorist appears to run the risk of radically misconstruing the foundations of our affectivity by putting the cart of our belief before the horse of our emotion.

On Articulating Our Assumptions

In this section I inquire into the relation between i) a patient’s verbal articulations in therapy of their underlying self-understandings, and ii) the nature of such underlying self-understandings. Several typical CBT models expound something like the following: i) that the first, rather preliminary, step of cognitive therapy is to help the patient clearly identify their emotionally problematic core beliefs, rules and assumptions. And ii) that the second task is to encourage them to quasi-scientifically test out these assumptions, either through rational engagement (e.g. Kuehlwein, 2002) leading to what is sometimes called ‘cognitive restructuring’, or more practically through ‘behavioural experiments’ (e.g. Bennett-Levy, 2004).

In what follows I will challenge such a conception of what is happening in therapy when assumptions are articulated, questioned or put to the test. My aim is not to question the CBT practice, but rather to question the conception of self-understanding that it encourages: that articulation of one’s own self-understanding in therapy is a rather preliminary stage occurring prior to the genuine transformations that can occur only when such self-understandings are challenged and healthier alternatives considered and embraced. The criticism offered here will be that those CBTs operating according to the above description risk underplaying both i) the transformative nature of the acknowledgement of one’s deepest fears, and ii) the way in which verbal, behavioural and experiential therapeutic techniques function not merely to test hypotheses, but rather to effect a more fundamental change in the form of the patient’s thought. A more fundamental change, that is, that first of all enables it to become penetrable by the light that experience and rational enquiry can shine into the fearful recesses of the mind.

Consider first what happens when an underlying dysfunctional assumption gets put into words, perhaps through the application of the ‘downward arrow’ technique. (This technique has the therapist repeatedly ask of, say, a patient voicing self-critical thoughts, questions such as ‘and if that were true, what would that say about you?’ until a definitive underlying negative fear is articulated.) Here is how Fennell (1989, pp. 204-5) articulates the therapeutic strategy:

Rather than challenging the thoughts themselves, the therapist asks: ‘Supposing that was true, what would that mean about you?’ This and similar questions… are repeated until it is possible to formulate a statement general enough to encompass not only the original problem-situation, but also other situations where the same rule is operating. … Once a dysfunctional assumption has been identified, questioning and behavioural experimentation are used to find a new, more moderate and realistic rule.

Fennell’s own assumption is that dysfunctional assumptions can be hard to unearth because, ‘rather than [being] discrete events occurring in consciousness, they are generalized rules which may never have been formulated in so many words’ (1989, p. 204). Why such assumptions should not have been put into words is not considered by her, nor is the emotional experience of the patient who articulates them - which, in my clinical experience at least, typically involves a quite particular mixture of distress and relief.[4] This empirical issue is not however our key concern here – which is instead philosophical and concerns what happens to such tacit assumptions when they are voiced. Fennell’s presentation, which is quite standard in the CBT literature, encourages the view that their voicing involves merely their being put into words. But reflection on what is meant by the voicing of our deepest troubling thoughts – reflection that is philosophical in so far as it concerns itself with how the objects of psychological investigation ought to be characterised, rather than itself depending on the results of empirical enquiry – reveals something richer. The articulation of one’s previously unarticulated fears involves, that is, not merely voicing but also acknowledgement, a correlative increase in self-understanding, an emancipation which comes from making visible or thinkable those assumptions which otherwise continue to invisibly constrain our meaningful experience and, so long as the resultant fears can be contained until they dissipate, an increased capacity to tolerate reality.

To take up the last of these: the patient who presents under the influence of a dysfunctional assumption which they have not yet articulated to themselves can often be seen to be failing to fully distinguish between fear (or wish) and belief. Whether they always really believe, for example, that others laugh at them as they walk down the street, or whether this is simply what they fearfully imagine or think, is not always entirely clear. An important reason for this is that they can be presenting in a state of mind which somewhat ablates the very distinction between fearing and believing.[5] Their ‘safety behaviours’ (which they perform to prevent (what they imagine will be) the experienced realisation of their fears – e.g. never meeting the gaze of others for fear of meeting a hostile look, or never failing to grip onto the shopping trolley for fear that they would collapse) similarly keep in place the constitutive lack of clarity of the anxious state. However when the CBT therapist encourages them to test out their fears by surveying the beliefs of others, or by undertaking a behavioural experiment (which may involve seeing if matters really do deteriorate if they don’t perform their safety behaviours), what happens is not merely that the truth of an assumption is evaluated, but that a clear distinction between fear and genuine belief begins to be instated in the mind. A fearful state of mind that demotes reality contact becomes substituted for by an empirical hypothesis.

Or perhaps the patient is invited to put a probability figure to their fear: what is the actual likelihood that you will be scoffed at if you show your face in the town centre? 90%? 20%? Once again the CBT strategy invites the transformation of what would once have been called a ‘neurotic’ state (in which an avoidance of reality has conspired with and inspired a fusion of fear, belief and experience – a state which may sometimes be compensated for by structurally similar fusions of wish, belief and experience) into an empirical hypothesis. Various CBT strategies have the clarifying effect of helping the patient shape up their fearful state so that it can be brought into contact with reality, first expressing it in the necessary bivalent (true/false) or probabilistic form of a genuinely reality-oriented proposition. They have this effect, that is, despite perhaps understanding themselves as merely being in the business of promoting the voicing then the testing of beliefs. What is significant here is not so much the disproving, but the shift that occurs within the structure of such fears when they are rendered provable or disprovable.

In this section I have suggested that whether we are dealing with the verbal articulation, or with the testing, of a patient’s deepest dysfunctional assumptions, what is principally significant is a change in the form of the patient’s fears. For what renders such assumptions dysfunctional is not simply their content but also their neurotic form – their insulation from empirical testing and rational thought, their fusion with wish or fear – and it is a change in such form which ultimately makes for the possibility of a change in unhelpful content. The CBT therapist’s armoury of tools for unearthing and engaging with such assumptions has the effect of shining the light of reason into the darker recesses of the patient’s mind, encouraging their thought to be governed not by fearful fancy but rather by what psychoanalysts would call ‘the reality principle’. In the process their thought becomes articulated – which is to say, not merely voiced but structured so that it now more clearly embeds a distinction between appearance and reality, is less insulated from reality testing, and is less a hostage of their deepest fears and wishes.

A much debated concern in the scientific CBT literature is whether the benefits of CBT are mediated by cognitive changes in the patient, whether such benefits may occur before cognitive techniques have even been applied, and whether cognitive techniques really add much to behavioural therapy (Longmore & Worrell, 2007). One way of putting the philosophical point I am making here is that expressive and behavioural techniques can already be seen to be (in the broadest sense) cognitive – in so far as they involve a change in the form of the patient’s fearful or depressive cognitions (cf. Carey & Mansell, 2009). The various techniques of CBT serve to free the patient from those inchoate fears that hitherto have not found adequate, clearly delineated, expression. Making the fears less inchoate and more thinkable risks making them feel more real, and the avoidance of this scary possibility often seems to have been a significant part of the patient’s difficulties. Therapy, however, provides an emotionally ‘containing’ environment in which the therapist’s clarity regarding the relatively benign and possibly even nourishing nature of social reality – by contrast with what the patient fearfully imagines to be the case – can become internalised by the patient. It is for this reason that a good therapeutic relationship must not merely be collaborative (Leahy, 2008) but also be emotionally containing, and an effective CBT therapist’s containing manner shows itself in her confidence in the empirical methods of CBT, her pragmatism, clarity, reality orientation, compassion and warmth.


To summarise the above four objections: the first proposed that certain CBT formulations conflate internal relations of meaning with causal relations between discrete inner states, resulting in models of psychopathology that appear more scientific than they really are. The second had it that some CBT models construed emotionally laden perspectives too much as an occurrent inner process, and too little as a subject’s attitudes. The third considered that such attitudes can sometimes be misdescribed in CBT models as beliefs – when what we really have to do with here are feelings with intrinsic meanings. The fourth argued that CBT models can underplay the significance of changes in the form of (a subject’s ownership of) such attitudes when they focus instead on changing their content. In these concluding paragraphs I will consider whether these various distortions to what it means to be a living human subject can be understood as aspects of a single common tendency of thought to which the CBTs may at times be prey.

My suggestion here is that what unifies the above disturbances of vision is their being expressive of what could be called an ‘alienated’ conception of human subjectivity. I will now spell out what I mean by this. Imagine that the self were no longer an expressive bodily being located immanently by its feelings within an already meaningful intersubjective world – but had retreated inwards, away from the world, the body, and even the mind, becoming instead a disengaged inner spectator trying to make sense from the outside of a world onto which it looks. Perhaps such a self is a prototype of a scientist-observer who is in the business of trying to control and predict the world by constructing inner representations or interpretations of it.[6] The effects of such a retreated conception of the self would be several. The mind and body, having been denuded of subjectivity, will now more naturally appear as domains of merely causally (rather than meaningfully) inter-related, objectified states and processes. Models of psychopathology constructed within such a vision start to look as if they possess some kind of scientific edge, since the description of the inner life of their subjects seems to provide an account of a scientifically describable network of causally inter-related isolable states and processes – rather than a phenomenological description of how we manifest and are embedded in our intentional worlds. The self will no longer speak from its attitudes, but will rather be reduced to speaking about them. Our minds and bodies become not so much what we could call the flesh or substance of the self, but instead show up as domains of inner and outer processes that are in interaction with the self. Therapy, accordingly, would get seen as no longer in the business of self-transformation, but instead becomes a technology – grounded in what will appear to be, amongst the therapies, uniquely scientific causal models of inter-related inner processes – for helping us manage our minds. Our relationships too will be reduced to a merely external form – which is to say, that they will become not constitutive of who we are, but rather seen merely to causally connect us to that which is essentially other. So too the being of the patient will no longer be thought of as partly immanent in the emotional flux of the therapeutic relationship; that relationship will instead risk being reduced to something which is merely a collaboration between distinct relata.

Precisely such a mechanistic and self-alienated conception of the mind can, it seems to me, be what we often find in the patient who presents wanting to know how to better ‘control their anxiety’, ‘change their thoughts’ or ‘manage their feelings’. Such a patient has, we could say, become alienated from their own inner life which, accordingly, is seen as an independent domain painfully afflicting them and requiring management or excision. Now it bears recollection that in this chapter I have not been concerned to critique CBT per se, but instead to scrutinise some of the ways in which some models of some CBTs may sometimes be inflected in ways that go against what it means to be an emotionally alive human subject. The risk for the CBTs that I have identified could be described as one of joining the above-described patient in such a de-subjectivised vision of the psyche. By way of corrective what we can remember is that by helping the patient to give articulate structure to his fears, to think, to be nourished by reality contact, and to distinguish fearful fantasy from genuinely representational belief, the CBT practitioner can be understood as doing far more than, say, helping her patient to test out his hypotheses; she is helping to restore her patient’s subjectivity.


Beck, AT 1976, Cognitive therapy and the emotional disorders, Meridian Press, New York.

Beck, AT, Rush AJ, Shaw, BG & Emery, G 1979, Cognitive therapy of depression, Guilford, New York.

Bennett-Levy, J, Butler, G, Fennell, M, Hackmann, A, Mueller, M & Westbrook, D 2004, Oxford guide to behavioural experiments in cognitive therapy, Oxford University Press.

Carey, TA & Mansell, W 2009, ‘Show us a behaviour without a cognition and we'll show you a rock rolling down a hill’, The Cognitive Behaviour Therapist, vol. 2, pp. 123-133.

Clark, DM & Steer, RA 1996, ‘Empirical status of the cognitive model of anxiety and depression’, in Salkovskis, PM (ed.), Frontiers of cognitive therapy, The Guilford Press, New York, pp. 75-96.

Robertson, D 2010, The philosophy of cognitive behavioural therapy (CBT): Stoic philosophy as rational and cognitive psychotherapy, Karnac Books, London.

Clark, DM 1989, ‘Anxiety states: panic and generalized anxiety’, in Hawton, K, Salkovskis, P, Kirk, J & Clark, D (eds.) Cognitive behaviour therapy for psychiatric problems: a practical guide, Oxford University Press, Oxford.

Ellis, A 1991, ‘The revised ABC’s of rational-emotive therapy (RET)’, Journal of Rational
Emotive and Cognitive Behavior Therapy, vol. 9, no. 3, pp. 139-172.

Epictetus 2004, Enchiridion. Dover.

Hayes, S, Strosahl, K, Bunting, K, Twohig, M & Wilson, K 2004, ‘What is Acceptance and Commitment Therapy?’, in Hayes, S & Strosahl, K (eds.) A practical guide to Acceptance and Commitment Therapy, Springer, New York.

Holmes, J 2009, Exploring in security: towards an attachment-informed psychoanalytic psychotherapy, Routledge, Hove.

Kelly, G 1955, The psychology of personal constructs, W. W. Norton & Co., New York.

Kohlenberg, RJ & Tsai, M 1994, ‘Improving cognitive therapy for depression with functional analytic psychotherapy: theory and case study’, The Behavior Analyst, vol. 17, pp. 305-319.

Kuehlwein, KT 2002, ‘The cognitive treatment of depression’, in Simos, G (ed.) Cognitive behaviour therapy: a guide for the practising clinician, Brunner-Routledge, Hove.

Leahy, R 2008, ‘The therapeutic relationship in cognitive-behavioural therapy’, Behavioural and Cognitive Psychotherapy, vol. 36, pp. 769-777.

Lear, J 2003, Therapeutic action: an earnest plea for irony, Other Press, New York.

McEachrane, M 2009, ‘Capturing emotional thoughts: the philosophy of cognitive-behavioral therapy’, in Gustafsson, Y, Kronqvist, C & McEachrane, M (eds.) Emotions and understanding: Wittgensteinian perspectives. Palgrave Macmillan, Basingstoke.  

Newman, MG & Borkovec, TD 2002, ‘Cognitive behavioural therapy for worry and generalised anxiety disorder’, in Simos, G (ed.) Cognitive behaviour therapy: a guide for the practising clinician, Brunner-Routledge, Hove.

Mollon, P 2007, ‘Debunking the ‘pseudoscience’ debunkers’, Clinical Psychology Forum, vol. 174, pp. 13-16.

Power, MJ & Dalgleish, T 1999, ‘Two routes to emotion: some implications of multi-level theories of emotion for therapeutic practice’, Behavioural and Cognitive Psychotherapy, vol. 27, pp. 129-141.

Robertson, D 2010, The philosophy of cognitive-behavioural therapy (CBT): Stoic philosophy as rational and cognitive psychotherapy, Karnac Books, London.

Teasdale, J 1997, ‘The relationship between cognition and emotion: the mind-in-place in mood disorders’, in Clark, D & Fairburn, C (eds.), Science and practice of cognitive behaviour therapy, Oxford University Press, Oxford.

Wilson, R & Branch, R 2005, Cognitive behavioural therapy for dummies, John Wiley & Sons, Chichester.

Whittal, M, Rachman, S & McLean, P 2002, ‘Psychosocial treatment for OCD: combining cognitive and behavioural treatments’, in Simos, G (ed.) Cognitive behaviour therapy: a guide for the practising clinician, Brunner-Routledge, Hove.

[1] What it takes to effect such learning remains a matter of contention. Psychodynamic therapists suggest that the patient develops such mental space through internalising the therapeutic relationship; they suggest that the high relapse rates after short-term therapy for depression reflect a failure of such internalisation. CBT therapists, by contrast, typically conceive of a collaborative therapeutic relationship as merely facilitative of intrinsically non-relational learning of good inductive habits, and suggest that relapses are due to a failure of such learning.
[2] How these attitudes are to be conceived – as beliefs (as the CBT theorist often conceives them) or as feelings (as the psychodynamic therapist often suggests) – will be considered in the next section.
[3] For example Mollon (2007, p. 13) complains of the scientistic rhetoric of the CBT literature, where ‘facing your fears is called ‘exposure’, refraining from an activity is called ‘response prevention’, learning to relax is called ‘stress inoculation’, and revising your thoughts is called ‘cognitive restructuring’’.
[4] The psychodynamic theorist has a ready explanation of such phenomenology: that the articulation of such assumptions involves the lifting of the repression of emotionally painful material.
[5] In psychodynamic terms, their thinking is more a matter of ‘primary’ than ‘secondary process’.
[6] George Kelly, psychologist of ‘personal constructs’, proposed just such a scientist-observer conception of the human subject (Kelly, 1955). His work significantly influenced the early CBT theorist Albert Ellis.