An Interacting Cognitive SubsystemsModel of Relapse and theCourse of Psychosis

Andrew Gumley*1, Craig A. White2 and Kevin Power1

1Anxiety and Stress Research Centre, Department of Psychology, University ofStirling, Stirling, Scotland, FK9 4LA2Department of Psychological Medicine, Academic Centre, University ofGlasgow, Gartnavel Royal Hospital, 1055 Great Western Road, Glasgow,Scotland, G12 0XH

This paper outlines a theoretical analysis of current psychologicalconceptualizations of psychosis that draws on Teasdale and Barnard's(1993) Interacting Cognitive Subsystems (ICS) model of depression.The ICS model differs from Beck et al.'s (1979) model of emotionaldisorders in that it specifies two qualitatively different levels ofmeaning. These different levels of meaning play a different role in theproduction, maintenance and modification of emotion, and indeed,also recognize the distinction between `intellectual' belief and`emotional' belief. Intellectual beliefs are at the level of specific pro-positional meaning: they have a truth value which can be assessed andthey are not directly influenced by sensory information such as voicetone, arousal and body state, or visual stimuli. Given this, propositionalmeaning does not have a direct link to the production of emotion.Rather the link to emotion is an indirect one, through the activation ofmore generic and holistic affect-related schematic models. These latterschematic models are hypothesized to have implicational meaning, andcorrespond to schematic models derived from emotive experiences. Inconsequence they do not have a specific truth-value and they reflectcontributions from a wide variety of sources, including multiplepatterns of specific meanings, and patterns of direct sensory input.This ICS perspective proposes that integration of these two levels ofmeaning can facilitate the psychological conceptualization of theinitiation, acceleration, and maintenance of positive psychotic sympto-matology during relapse. This theoretical analysis is illustrated by theuse of a case study, which describes the process and application of ICSto facilitating recovery from acute psychosis. This paper also provides adetailed discussion of the implications of this analysis for futureresearch and psychological therapy. In particular, ICS has relevance fortracing the evolution of schematic beliefs during the early episodes ofpsychosis, which are hypothesized to have a bearing on the futurecourse of psychosis. This has implications for the further developmentand expansion of psychological treatments for relapse and recoveryfrom psychosis. Copyright # 1999 John Wiley & Sons, Ltd.

CCC 1063±3995/99/040261±18$17.50Copyright # 1999 John Wiley & Sons, Ltd.

Clinical Psychology and PsychotherapyClin. Psychol. Psychother. 6, 261±278 (1999)

*Correspondence to: Andrew Gumley, Anxiety and Stress Research Centre, Dept. of Psychology, University of Stirling, Stirling,FK9 4LA.

Contract grant sponsor: The Chief Scientist's Office, Scottish Office.Contract grant number: K/RED/18/13.

INTRODUCTION

For many years it was believed that psychoticsymptoms were discontinuous from `normal'psychological functioning. This position is becom-ing increasingly untenable in the light of researchevidence that positive psychotic symptoms can beunderstood with reference to normal psychologicalprocesses. This paper will outline the evidence fromcurrent psychological conceptualizations thatpsychotic symptoms are closely related to aperson's psychological functioning and that theyare therefore amenable to psychological interven-tion (Yusupoff et al., 1996). Current psychologicalconceptualizations of hallucinations and delusionshold that the content of these symptoms containmaterial which is personally relevant to theindividual. Indeed, this personal relevance is criticalto the understanding of, and cognitive interventionswith psychotic symptoms.

In this paper we will propose that an individual'sattempts to assimilate and accommodate the changesassociated with psychosis are central to the devel-opment and maintenance of symptomatology,associated disability and distress. It will be arguedthat the process of relapse encapsulates this inter-action between an individual and their experience ofpsychosis. Indeed, given that relapse sees theemergence of positive symptomatology, there is aneed to develop an integrative model of psychoticrelapse that can accommodate existing psychologicalconceptualizations of hallucinations and delusions.These formulations have been predominantly linearconceptualizations of symptomatology. This paperwill outline a conceptualization of psychotic relapsethat draws upon Teasdale and Barnard's InteractingSubsystems Model (ICS; Teasdale and Barnard,1993). It will be argued that this is an improvementon the linear models, which are integral to most othercurrent cognitive theories of psychosis.

CURRENT PSYCHOLOGICALCONCEPTUALIZATIONS OF PSYCHOSIS

Hallucinations

Attribution theory (Weiner, 1985) has been helpfulin developing psychological models of hallucina-tions and delusions. Attributional models of hallu-cinations conceptualize an auditory hallucination asa mental event that is misattributed to an externalsource. This hypothesis has been supported byvarious studies which have shown that individuals

with hallucinations demonstrate a bias towardassuming a voice had been presented when it hadnot (Bentall, 1996). Indeed, Slade and Bentall (1988)demonstrated that individuals with hallucinationsare more inclined to judge a perceived event as realand make rapid and over-confident judgementsabout their perceptions. Bentall's (1990) cognitivemodel of hallucinations incorporates the concept of`metacognition': the process whereby individualsare able to reflect upon their own experience andmental processes. Bentall argues that in normalreality discrimination, individuals determine thesource (internal or external) of a perceived event bynon-consciously applying a set of criteria, such asthe properties of the event, contextual cues, andexpectations. Factors that may influence applicationof these criteria include environmental stimulationand style of information processing. Therefore,individuals will hallucinate when they wronglyinfer that internally generated cognitive events suchas thoughts are externally generated stimuli. Thismisattribution may be influenced by environmentalfactors (e.g. overstimulation) leading to high arousaland shallow information processing of stimuli.Furthermore, the misattribution of an aversivecognitive event to an external `non-self' sourcemay be negatively reinforcing through arousalreduction. This process may be influenced by theindividual's own schemata about self, world andfuture. Such schemata will influence individuals'expectations, interpretation and emotional responseto both internal and external events (Padesky andGreenberger, 1995).

Morrison et al. (1995) have elaborated a cognitivemodel of auditory hallucinations which proposesthat auditory hallucinations arise from an attribu-tional bias. Morrison et al.'s model draws on existingcognitive conceptualizations of intrusive thoughts(Rachman, 1978). Intrusions are defined as repetitivethoughts, images or impulses that are unacceptableor unwanted and are usually accompanied bysubjective discomfort. Salkovskis (1985) hypothes-izes that such thoughts are also ego-dystonic; i.e.they are incompatible with the individuals's ownbelief system. Salkovskis and Kirk (1997) hypothes-ize that the occurrence and content of intrusivethoughts in obsessive compulsive disorder, areinterpreted by individuals as meaning that theymight be responsible for harm to themselves orothers unless they take action to prevent it. It is theappraisal of responsibility, therefore, which is centralto the development of distress. This results inattempts to suppress or neutralize the intrusivethought. Morrison et al. (1995) note that there is a

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

262 A. Gumley, C. A. White and K. Power

similarity in the content of intrusions and auditoryhallucinations. Thus it is argued that the occurrenceof an auditory hallucination is accounted for by theresult of an attributional process whereby indi-viduals misattribute ego-dystonic, unwanted, anduncontrollable thoughts to an external source there-by leading to reduced distress. Cognitive dissonancetheory (Festinger, 1957; Beauvois and Joule, 1996)states that dissonance results when two cognitionscontradict each other. Therefore when positivethoughts are inconsistent with the individual'sown beliefs about self, dissonance will result. Anattribution to an external source (resulting in anauditory hallucination) will reduce dissonance. Thiswould account for the experience of pleasantauditory hallucinations in individuals with negativeself-schemata. Morrison et al. also hypothesize thattheir theory may also be useful in explaining thoughtinsertion, thought withdrawal, and thought broad-casting, which may be related to misattributions inresponse to uncontrollable and/or unwanted cogni-tive events.

Experimental support for this hypothesis isdescribed by Morrison and Haddock (1997) whoexamined the cognitive processes underlying audi-tory hallucinations in an experiment which investi-gated delayed and immediate source monitoring forpositive, negative and neutral verbal material. Theyfound that individuals experiencing auditory hallu-cinations had an external attributional bias for theirimmediate thoughts, but not for their memories ofthose thoughts. This bias was not demonstrated inindividuals with no auditory hallucinations butwith other psychotic symptoms. This implies a biasin moment by moment source monitoring specific toauditory hallucinations. Morrison and Haddockalso found that the emotional valence of verbalmaterial was a significant factor, which influencedthe bias in source monitoring. This is consistent withBentall's (1990) and Morrison et al.'s (1995) theorieswhich predict that emotional content of intrusionshas a direct effect on their misattribution to anexternal source.

Chadwick and Birchwood (1994) have studiedindividual's beliefs about voices in relation toemotional and behavioural responses to voices.Using an ABC framework they formulated voicesas activating events (A) and found that there wascognitive content-specificity in that voices believed(B) to be malevolent were associated with negativeaffect and were always resisted (C), whereas voicesbelieved to be benevolent were associated withpositive affect and were engaged with. Conceptual-izations of auditory hallucinations by Bentall and

Morrison have centred on explaining psychoticsymptom occurrence in contrast to Chadwick andBirchwood's work which has focused upon theemotional and behavioural consequences of symp-tom occurrence.

The fact that these different models of auditoryhallucinations have focused upon different aspectsof the phenomenology of individuals' beliefssuggests that there may be a need for a multi-levelconceptualization of symptoms which encompassesa number of psychological factors responsible forthe initiation, acceleration and maintenance ofauditory hallucinations. A key initiating factor maybe the interaction between intrusive and unwantedcognitions, which conflict with core schemataproducing dissonant or discrepant internal experi-ences. The process of misattribution leading totemporary reductions in such discrepancies plays arole in the acceleration of symptomatology throughnegative reinforcement. As individuals' idiosyn-cratic beliefs about symptoms crystallize andemerge, these beliefs play a role in the maintenanceof symptomatology and the development of sec-ondary morbidity and distress. For example, in thecontext of voices, which are ascribed power andmalevolence, the experience of powerlessness andhelplessness will follow.

Delusions

Attributional models have also been utilized toexplain the development of delusions. Maher (1988)hypothesizes that delusions are formed by the samecognitive processes as normal theories and beliefs,and evolve as an explanation of puzzling anomalousexperiences. Like Morrison et al.'s (1995) formula-tion of hallucinations, Maher's theory incorporatesthe concept of cognitive dissonance. When adiscrepancy occurs between an expected sequenceof events and an observed sequence of events, thisleads to a sense of puzzlement and perplexity. Theadoption of a delusional explanation which ac-counts for personally significant and/or anomalousexperiences is accompanied by a reduction intension and uncertainty (Maher, 1988). Howeverthis model can be criticized on the grounds thatabnormal cognitive processes have been demon-strated in individuals with delusions (Garety, 1991).Garety found evidence of a judgmental bias in somedeluded individuals where these individuals areexcessively influenced by current information, andmake less use of past learned regularities in makinginferences. Garety also found an apparently contra-dictory bias in other individuals with paranoid

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

ICS and Psychosis 263

delusions, which was a tendency to rely excessivelyon prior expectations when processing new infor-mation. These contradictory biases are accommo-dated in her model of delusional formation (Garetyand Hemsley, 1994). She hypothesizes that thesetwo judgmental styles may reflect two stages ofresponse to an information-processing abnormality.In this model the initial process of delusionalformation arises from the excessive reliance oncurrent perception. The resulting delusional beliefwill then generate strong expectations, whichinfluence affective state and attention to `belief'-congruent information.

Attributional processes associated with delu-sional beliefs have also been the focus of recentexperimental investigation. Individuals with persec-utory delusions seem to be more likely to attributeblame for negative events to external causes andpositive events to internal causes (Kaney andBentall, 1989, 1992; Fear et al., 1996). They alsopreferentially attend to (Bentall and Kaney, 1989)and recall (Bentall et al., 1995) threatening informa-tion. Kinderman (1994) found that individuals withpersecutory delusions have a specific attentionalbias for information of relevance to self-concept. Ithas been suggested that these experimental findingsare consistent with the proposition that deludedindividuals have negative self-schemata, which areexperienced by the individual as a discrepancybetween their actual and ideal selves. The modelconceptualizes persecutory delusions as compensa-tory beliefs, which arise in response to this perceiveddiscrepancy. This concept of delusional beliefsserving such a protective function has also beensuggested for grandiose delusions.

Trower and Chadwick (1995) concur with thesuggestions of Bentall and other investigators thatpersecutory delusions have a defensive function inthat they are associated with low self-esteem which isoutside the individual's awareness. They haveexpanded this theory to encompass an interpersonalfocus. They suggest that sources of threat are a solelyinterpersonal negative evaluation and that thedefence, too, has this interpersonal focus in that it isan interpersonal evaluation of the other personwhich characterizes the paranoid `defence'. Theyalso suggest that paranoia can be divided into twotypesÐpersecution paranoia (the type researched byBentall and others) and punishment paranoia. Thesetwo types of paranoia are held to be characterized bydiffering attributional styles which relate to howinformation is processed with reference to self.Punishment paranoia is hypothesized to be charac-terized by an attributional style where the individual

attributes negative events to self and positive eventsto external causes. With punishment paranoia theindividual believes that they are bad and blame-worthy and that others are justifiably punishingthem. The defensive function of punishment para-noia is less clear than in persecutory paranoia.

Most of the work on cognitive theory and therapyhas emphasized the importance of conceptualizingdelusions with regard to core self-schemata. Anumber of psychological factors may play a role inthe initiation, acceleration and maintenance ofdelusional beliefs. The initiation of delusions arisesfrom the evolution of explanations for anomalousevents, which are experienced as puzzling, confus-ing and discrepant with expectations. Central to theprocess of acceleration of these explanations is theappraisal of threat and the reinforcement contin-gencies inherent in the adoption of delusionalexplanations. Individuals' core beliefs form a centralreference point for the appraisal of threat. Gilbert(1997) argues that attacks on, and/or losses of socialattractiveness are associated with shame andhumiliation. Gilbert's model proposes that bothshame and humiliation are associated with sensi-tivity to criticism, a desire to protect oneself,increased arousal, and complex affects. Howevershame and humiliation differ on a number ofdimensions including attributional style, view ofself, attentional focus, emotion, and behaviour. Weargue that the experience of shame or humiliationassociated with attacks on or loss of social attrac-tiveness is ontogenic in the development andacceleration of delusions. Humiliation, like persec-utory paranoia, is characterized by an externalattributional bias, where others are seen as badand are blamed for negative events. In bothhumiliation and persecutory paranoia there is astrong sense of injustice and the individual is morelikely to react to negative events with anger andhostility. On the other hand, both shame andpunishment paranoia are characterized by aninternal attributional style, where self is seen asbad and blameworthy of negative events. Shameand punishment paranoia are both associated withheightened self-consciousness, and no obvioussense of injustice. Individuals are more likely toreact with depression, submissiveness and with-drawal. Both shame and humiliation are associatedwith differing evolutionary strategies aimed at themaintenance of private social attractiveness (humi-liation) or public social attractiveness (shame). Themaintenance of private and public social attractive-ness may be implicit in the defensive functions ofpersecutory and punishment paranoia. This is not

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

264 A. Gumley, C. A. White and K. Power

currently addressed within current psychologicalconceptualizations of psychosis.

Relapse

Despite the advances in pharmacological manage-ment of schizophrenia, relapse remains a majorfactor in the development of illness chronicity andsocial disability. Indeed for the individual them-selves, relapse is critical in the development ofsecondary depression. Birchwood et al. (1993) foundthat perception of control over illness was the mostpowerful predictor of depression in schizophrenia.Recognition of the social, emotional and psycho-logical costs of relapse has led investigators toexamine approaches to the detection and preven-tion of relapse.

Retrospective studies of individuals and theirfamilies (Herz and Melville, 1980; McCandless-Glincher et al., 1986; Birchwood et al., 1989) showthat both groups are able to recognize reduced well-being. The most commonly reported early signs aresleeplessness, irritability, tension, depression, andsocial withdrawal. The consistency with whichthese early signs have been reported has led tothe development of prospective investigations ofearly signs. In essence, these investigations havesought to identify the sensitivity and specificity ofthese early signs as an indicator of emergingrelapse. Clearly if these early signs are sensitiveand specific to relapse, the monitoring of such signswould enable early intervention leading to theprevention and/or amelioration of relapse. Mallaand Norman (1995) in their systematic review of theearly signs literature found very few studies thatdirectly assessed the relationship between putativeprodromal symptoms or early signs, and the exacer-bation of relapse (Birchwood et al., 1989; Jolley et al.,1990; Marder et al., 1991; Tarrier et al., 1991;Gaebel et al., 1993; Marder et al., 1994; Jorgensen,1998). Table 1 below provides a summary of thestudies that have examined the sensitivity and/orspecificity of putative prodromes.

The results of some of these studies have beendisappointing. However, Birchwood (1995) pointsout that individual variations in the nature andtiming of early signs will act to reduce their apparentamplitude in group studies. Group studies fail tocapture the qualitative and quantitative differencesbetween individuals in their early signs. Therefore itmay be more appropriate to think of early signs asan individualized configuration of symptoms whichBirchwood refers to as a `relapse signature'.

Birchwood (1995) accounts for this variation inthe nature and timing of early signs by integrating

individual's own idiosyncratic response to emer-ging relapse. This cognitive explanation for thevariation in early signs suggests that dysphoricsymptoms such as anxiety, tension, withdrawal,depressed mood, suspiciousness, and sleeplessnessarise from the way in which individuals explain andinterpret internal and external events. Birchwood(1995) offers a compelling cognitive analysis ofearly relapse which draws upon Maher's (1988)model of delusional formation and Weiner's (1985,1986) attribution theory. Birchwood proposes thatthe attributions made by individuals to account forand explain the emergence of disturbing symptomscan serve to either accelerate or retard the process ofrelapse. In this model, dysphoria is seen as aresponse to the fear of impending relapse (perhapsfor those with previous experience of relapse) or afailure to explain symptoms and experiences(perhaps for those with less experience of relapse).

This model might therefore predict that thoseindividuals with extensive prior experience ofrelapse and its associated negative repercussionswould respond with high levels of fear and perhapshelplessness leading to depression and withdrawal.On the other hand, those with less experience mayrespond with puzzlement, confusion and perplexity.The model may also help explain the speed at whichrelapse proceeds by specifying the cognitive mech-anisms and associated emotional consequencesresponsible for acceleration.

This model highlights the role of individual'sattributions for and meaning ascribed to symptomsand internal experiences. One might hypothesizethat where such attributions have an impact uponthe acceleration of emotion, they will relate topersonally relevant beliefs and assumptions held by

Table 1.

Author(s) Relapses Sensitivity%

Specificity%

Subotnik andNeuchterlein (1988)

17 59 NR

Birchwood et al. (1989) 8 63 82Hirsch and Jolley

(1989)10 73 NR

Tarrier et al. (1991) 16 50 81Gaebel et al. (1993) 162 8 90

14 7010 93

Marder et al. (1994) 42 37 NR48

Malla and Norman(1994)

24 50 90

Jorgensen (1998) 27 81 79

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

ICS and Psychosis 265

an individual. These beliefs might concern theindividual themselves (I am worthless, I am afailure), the world (other people will criticize me,other people will harm me) or their illness (myillness is god's retribution on me, I will nevercontrol this). A model of relapse, which integratessuch beliefs, may be more powerful in explainingthe initiation and acceleration of relapse.

For example, the experience of visuo-perceptualchanges or increased arousal associated withemerging psychosis may be interpreted by anindividual as a sign of impending relapse. If thisindividual believes that they have no control overtheir illness, their response to these changes is likelyto be increased vigilance, anxiety, worry, sleepless-ness and perhaps withdrawal. Therefore thisindividual's beliefs about self and others and theirmeta-beliefs about illness and symptoms may serveto accelerate the speed at which relapse proceeds.The activation of individuals' beliefs about them-selves, others, and/or their illness serves togenerate a range of cognitive, emotional, behaviour-al and physiological changes which represent anindividualized signature or configuration of symp-toms that characterize the early relapse phase.Cognitive formulation and intervention may there-fore provide an additional means of identifying,preventing and/or ameliorating relapse.

Within this analysis we propose that a number ofpsychological factors play a role in the initiation andacceleration of relapse. Initiation of relapse may bethrough the activation of core schemata about self,world, and future. Idiosyncratic beliefs about illness,which have evolved during, and following the firstepisode will have close relationships with theseschemata. Activation of schemata and beliefs maybe through external stressors (e.g. increasing tensionand hostility at home) or internal cues (idiosyncraticsymptoms which have psychological significance inthat they resemble or foretell the emergence ofillness). From the point of initiation, the activation ofschematic models of self (including illness), worldand future drive the emergence of a range ofthoughts, attributions and inferences which accele-rate the development of an acute episode.

INTERACTING COGNITIVESUBSYSTEMS MODEL

Implicit in our analysis of the preceding literatureof current psychological conceptualizations ofpsychosis is the construct of initiation, accelerationand maintenance. Within this construct we have

proposed a number of psychological factors whichcould play a critical role in the evolution ofpsychosis. Teasdale and Barnard's (1993) Interact-ing Cognitive Subsystems model provides anexisting theoretical framework, which may holdour emerging conceptualization.

The concept of persecutory and punishmentparanoia poses similar issues to those associatedwith the understanding of auditory hallucinationswhich have a positive content as opposed to thosewhich have a negative content. Critical to theattributional process are individuals' underlyingbeliefs about themselves, and the attribution ofresponsibility for negative and positive (internal orexternal) events. Single level approaches, whichattempt to explain the role of personal meaningin the development of symptoms, are insufficient.Indeed, single level approaches have come underincreased criticism within the literature (Power andChampion, 1986; Brewin, 1989; Greenberg andPascual-Leone, 1997; Power and Dalgleish, 1997).Also, from a clinical viewpoint, Teasdale (1997a)argues that within cognitive therapy, the useof `rational' argument is frequently ineffective inchanging the emotional response, even when theindividual is able to acknowledge the intellec-tual power of an argument. An analysis of attribu-tional style in the development of psychoticsymptoms provides an illustration of the limitationsof single level approaches. Purely addressingattributions implicit in the development and main-tenance of psychotic symptoms made by indi-viduals with psychosis will have limitedtherapeutic value without recourse to underlyingbeliefs about self and others. As such, attributionscan be thought of as providing a marker forpersonal meanings.

This is central to Teasdale and Barnard's (1993)Interacting Cognitive Subsystems (ICS) model ofdepression. They propose that negative thoughtsand images may be useful markers for the stateof `parent' schematic models. In consequence,cognitive-behavioural or other therapeutic proce-dures, which address negative thoughts and images,will not necessarily lead to emotional change unlesssuch thoughts and images relate to the parentschemata. Teasdale and Barnard (1993) and morerecently Teasdale (1997a) provide an account of therelationship between meaning and emotion. The ICSmodel differs from Beck et al.'s (1979) model ofemotional disorders in that it specifies two qualitat-ively different levels of meaning. These differentlevels of meaning play a different role in the produc-tion, maintenance and modification of emotion, and

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

266 A. Gumley, C. A. White and K. Power

indeed, also recognize the distinction between `intel-lectual' belief and `emotional' belief. Intellectualbeliefs are at the level of specific propositional mean-ing: they have a truth value which can be assessedand they are not directly influenced by sensoryinformation such as voice tone, arousal and bodystate, or visual stimuli. Given this Teasdale arguesthat propositional meaning does not have a directlink to the production of emotion. Rather the link toemotion is an indirect one, through the activation ofmore generic and holistic affect-related schematicmodels. These latter schematic models are hypoth-esized to have implicational meaning, and corre-spond to schematic models derived from emotiveexperiences. In consequence they do not have aspecific truth-value and they reflect contributionsfrom a wide variety of sources, including multiplepatterns of specific meanings, and patterns of directsensory input. In this analysis the sensory inputsfrom body state have an important role in the pro-duction of affect-related schematic models. The heartof the ICS model provides an eloquent explanation ofthe development and maintenance of depression.

Affect-related schematic models are derived frommultiple sources including patterns of specificmeanings arising from current environmentalevents and information processing, and patterns ofinternal and external sensory input. In depression,the sensory source will be internal, for exampledepressed posture, facial expression, retardation,anhedonia etc. The configuration of these multipleelements leads to the generation of idiosyncraticmodels of depression, where self is viewed as failureand the depressed state as aversive, uncontrollableand persistent. In this respect, depression might betriggered by any number of the elements in theconfiguration and is maintained by what Teasdaleand Barnard (1993) refer to as the `depressive inter-lock'. This interlock represents the interactionbetween depressive configurations and theindividual's higher order view of depression. Forexample relapsing depression would be accountedfor within an ICS model by initial (minor) depress-ive reactions associated environmental events whichevoke schematic models of depression where self isviewed as powerless and depression as uncontrol-lable resulting in a self-perpetuating and accelerat-ing depressive interlock.

AN ICS ANALYSIS OF RELAPSE ANDCOURSE OF PSYCHOSIS

We propose that ICS theory confers two distinctadvantages in understanding psychosis. First, the

use of its multilevel theory enables the integrationand elaboration of existing psychological concep-tualizations of psychotic symptoms. This will bedemonstrated through introducing four classes ofoverlap between these models and ICS. Second, theICS model contains a number of consequences forthe development of further empirical work and therefinement of current therapeutic practice. Again,this will be demonstrated through examining theparticular predictions implied by ICS, as opposedto other multilevel models, and through a caseillustration.

To summarize, the ICS model, outlined byBarnard (1985) proposes that no single systemcontrols the processing of information. Rather, it isthe interaction of a number of subsystems, whichprocess and encode information from multiplesensory sources (visual, acoustic and body-state).Information processing depends on informationflowing from one subsystem to another, and thatregular recurring patterns of information enablesbasic recognition, comprehension, processing ofstimuli (see Barnard, 1985; Teasdale and Barnard,1993, chapter 5). This is critical within the ICStheory. In ICS the processing configuration ismotivated to reduce depression, by reducing dis-crepancies (dissonance) between current and highlydesired states of affairs ( for example, `alwaysseeking the approval of other people is the roadto happiness'; Teasdale et al., 1995). We propose thatthe initiation, acceleration, and maintenance ofpsychosis can be explained by individuals' attemptsto make sense of, and reduce discrepancies arisingfrom multiple and interacting sources of infor-mation. Indeed, this represents an area of overlap orconvergence with current psychological conceptu-alizations of psychosis which have emphasized therole of cognitive dissonance and its reduction in theevolution of psychotic experiences such as auditoryhallucinations and delusional beliefs.

Initiation of relapse

In a clinical setting the prediction of relapse is basedupon the identification of a range of signs andsymptoms which have been demonstrated in retro-spective and prospective studies. These signs andsymptoms are subject to considerable variance intheir character and timing. This reduces theirapparent amplitude in group studies, and increasesthe likelihood of false positive or false negativepredictions of relapse in clinical practice. In ICSinformation processing relies on the recognition ofrecurring patterns of information which are derived

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

ICS and Psychosis 267

from multiple sources. These recurring patterns areembodied in the creation of propositional repres-entations reflecting, on the one hand, intended and,on the other hand, current states of affairs. Thecreation of generic and holistic implicational modelswhich reflect the discrepancy between intended andcurrent states can then effectively monitor progresstowards goals attainment. In this sense implicationalmeaning abstracts over sensory, body state, andsemantic sources of information.

Within ICS the occurrence of a pattern orconfiguration of internal and/or external eventswhich have a strong similarity with previousrelapses will access implicational meaning morerapidly than if the configuration has a lowersimilarity. The activation of implicational meaningstructures derived from previous experiences ofpsychosis will initiate the process of relapse. Aprimary source of information for a relapse con-figuration is internal information on body state andarousal (Tarrier and Turpin, 1992), and cognitive-perceptual change (Neuchterlein and Dawson, 1984;Frith, 1992). Indeed, feelings of lack of control overcognitive and perceptual processes during relapsehave been reported by a number of authors (Bowers,1968; Freedman and Chapman, 1973; Donlan andBlacker, 1975). There are two implications of thisanalysis.

First, definitions that are currently used tocapture sensitivity of early signs could well benefitfrom definitions more closely allied to alternativemultilevel views of how implicational meaning isconstitutedÐrather than relying on a more closelydelineated set of individual signs and symptomsalone. For example, for an individual a combinationof signs and symptoms such as increased sleepdisturbance, reduced attention and concentrationand increased agitation will be experienced as adiscrepancy from intended internal states andrepresented at an implicational level as a `sense oflosing control' or a `sense of not feeling right'. Thismight then be accompanied by a sense of impend-ing disaster or threat to oneself. As this sense isrepresented at an implicational level this will nothave a specific truth-value, rather it emerges from acombination of sensory, body-state and semanticinformation. If the clinician pays sole attention tothe occurrence of specific signs and symptoms,they risk failing to capture the more holistic andgeneric meaning experienced by the individual.Therefore, configurations of early signs which aremore closely allied to the schematic meaningsachieved during early relapse may increase sensi-tivity, and reduce the apparent variance in the

nature and timing of experiences signalling futurerelapse.

Second, ICS proposes that information derivedfrom different sources is managed by the evaluationof discrepancies between systems. In ICS, informa-tion processing is motivated to reduce thesediscrepancies. It is this process which we proposeis responsible for the acceleration of relapse and thedevelopment of psychotic symptoms.

Acceleration of Relapse and the Development ofPsychotic Symptoms

In order to examine how relapse is acceleratedwithin an ICS perspective it is worth consideringfour classes of overlap between ICS and existingpsychological conceptualizations of psychosis. First,psychological models of delusions and hallucina-tions imply that separate but related information-processing systems are involved in the developmentand maintenance of symptomatology. One system,such as attributional bias, manages specific momentby moment meanings. The second information-processing system relates to meta-cognition (Bentall,1990), underlying schematic beliefs (Morrison et al.,1995), or self-concept (Kinderman, 1994; Trowerand Chadwick, 1995). ICS specifies to subsystemsthat are responsible for the management of mean-ing; one manages specific moment by momentpropositional meaning, the other higher orderschematic or implicational meaning. It is theinteraction between propositional meaning andimplicational meaning subsystems that constitutesthe `central engine' of cognition.

Second, underlying schematic beliefs that areproposed within existing psychological conceptual-izations are utilized to explain the emotional valenceof an event to an individual. For example, in the caseof auditory hallucinations. Bentall's (1990) andMorrison et al.'s (1995) theories both predict thatthe emotional content of intrusions has a directeffect on their misattribution to an external source.In ICS, depression is produced by the processing ofdepressogenic schematic models. Specific meanings,negative automatic thoughts, and images only con-tribute indirectly through their influence on higherorder meanings. We propose that during relapse,acceleration of the central engine is through theprocessing of schematic models of self (e.g. `I haveno control', `I'm a bad person'), world/others (e.g.`people want to harm me', `I'll be punished') andillness (e.g. `my illness is a punishment', `my illnessis a weakness').

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

268 A. Gumley, C. A. White and K. Power

Third, beliefs about the power, purpose, malevo-lence/benevolence of symptoms are directly linkedto the emotional and behavioural consequences ofauditory hallucinations. In ICS, depression is main-tained by the continuing regeneration of depresso-genic schematic models where depression is viewedas aversive and uncontrollable. Therefore, beliefsabout symptoms such as auditory hallucinationsare critical to the maintenance of symptomatologythrough regeneration of implicational meaningstructures. In Chadwick and Birchwood's (1994)study using cognitive therapy to modify beliefsabout voices it was predicted that this approachwould lead to a reduction in the levels of distressassociated with voice-hearing but not a reductionin frequency. The reduction in frequency of audi-tory hallucinations following cognitive therapy,which was observed in this study, would beaccounted for within an ICS perspective. If beliefsabout voices are directly linked to the implicationalsubsystems responsible for the initiation and accel-eration of symptoms, then modification of thesebeliefs will lead to a modification of underlyingimplicational meaning. ICS predicts that this wouldtherefore lead to a reduction in the frequency ofauditory hallucinations, as well as associateddistress.

Fourth, current psychological models of psycho-sis, predict that the information processing implicitin the development of psychotic symptoms is moti-vated to reduce the dysphoria arising form dis-sonant or discrepant experiences, and to protect selfin relation to self (e.g. persecutory paranoia) or selfin relation to others (e.g. punishment paranoia). InICS, the processing configuration is motivated toreduce depression by reducing discrepancies(dissonance) between current experience and in-tended experience. Critical to our analysis of theinitiation, acceleration and maintenance of relapse isthe role of the interlock between propositional andimplicational meaning as the central engine ofrelapse. This engine is motivated to reduce dis-crepancies between intended and actual experiencethrough the construction of implicational meaningstructures. The activation of relevant implicationalmeaning serves to initiate the process of relapse, theinterlock between propositional meaning and im-plicational meaning serves to accelerate relapse, andthe regeneration of implicational models serves tomaintain relapse. This is illustrated in Figure 1.

This model attempts to illustrate the process ofinitiation, acceleration and maintenance of relapse.Box 1 to 2 illustrates the relationship between earlysymptomatic changes in body-state and information

processing and how these changes access implica-tional meaning structures. Within box 2, theinterlock between implicational and propositionalmeaning which forms the central engine of relapseis illustrated, where the development of psychoticsymptoms (box 3) serves to reduce some of theemotional and affective experiences produced bythis interlock. The process of maintenance ishypothesized to be accounted for through thecontinuing regeneration of implicational meaningillustrated in the direct relationship between psy-chotic symptoms (box 3) and implicational meaning(box 2).

Persistent reactivation of implicational meaningwill present as increased chronicity and resistanceto treatment, whereas intermittent reactivation willpresent as a relapsing course with periods ofremission. A range of factors may be responsiblefor this process of reactivation including a range ofinternal or external experiences, which have specialand specific psychological significance for individ-uals. We believe that such a model may provide auseful means of identifying and preventing earlytreatment resistance by guiding clinicians in theirconceptualization of the interaction between theindividual and their beliefs, their experience ofpsychosis, and the subsequent development ofsecondary beliefs about self, illness, environmentand future. We hypothesize that these elementshave their basis in the early episodes of psychosis.Indeed, Jackson and Birchwood (1996) refer to thefirst few years of psychosis as the critical period. Itis this period, which sees the emergence ofsecondary co-morbidity and treatment resistance.This period can be characterized by high levels oftraumatization (McGorry et al., 1991) and negativeconsequences associated with reduced social attrac-tiveness, status and rank.

PREDICTIONS AND APPLICATIONS

The key feature that distinguishes ICS fromother multilevel models such as the SchematicPropositional Associative Analogue RepresentationSystems (SPAARS; Power and Dalgleish, 1997)approach, is the contribution of the interactionbetween propositional and implicational meaningin the process of change (be it relapse or recovery)and how self-related schematic models play a rolein those processes. ICS offers a detailed view ofwhat these representations are, how they relate tospecific patterns of information flow, and how those

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

ICS and Psychosis 269

exchanges between subsystems are managed bydiscrepancies among representations.

Specifically, SPAARS seeks to integrate the keycomponents of ICS including implicational, prop-ositional and other subsystems (visual, olfactory,gustatory, proprioceptive, tactile and auditory).These latter subsystems are considered withinSPAARS as analogical representations. ICS providesa detailed analysis of how discrepancies betweenthese systems are managed by the interaction ofpropositional and implicational meaning. Thisdetailed view of how subsystems interact hasimportant consequences for the formulation andprediction of the impact and involvement ofpsychological factors in the initiation, acceleration,maintenance and recovery from psychosis.

Early Intervention

Birchwood (1995) offers a cognitive conceptualiza-tion of relapse that specifies that attributions reflect

the individual's idiosyncratic response to earlyrelapse and that these attributions play a role inthe acceleration of an acute psychotic episode. ICSprovides a more detailed analysis by specifying howrelapse is initiated and accelerated. We likenedrelapse to an engine that is initiated by discrepanciesbetween subsystems, which activate implicationalmeaning. Acceleration of relapse then occursthrough the interlock between propositional andimplicational meaning. This interlock is gearedtowards the reduction of discrepancies betweensubsystems. For example, if the attribution `I feel asif something bad will happen', is coupled withexternally-derived sensory input (e.g. harsh tones orfrowns from friends and family), externally-derivedsemantic contents (e.g. critical remarks) and bodilystates (e.g. physical agitation and unease), theimplicational meaning derived from these multiplesources of information might be `Other people aretrying to harm me'. Therefore during relapse, whena vulnerable individual is repeatedly exposed to

Figure 1.

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

270 A. Gumley, C. A. White and K. Power

particular combinations of sensory inputs this servesto regenerate implicational and schematic modelswhich will progressively come to dominate andcontrol mental life.

This analysis predicts that the speed of relapse isgoverned by the extent to which implicationalmodels are activated and regenerated. This processwill be an idiosyncratic one which depends uponthe particular schemata, assumptions and beliefs ofindividuals. In this sense the ICS analysis isconsistent with Birchwood's (1995) view that earlysigns represent a continuous process as opposed toprodromal conceptualizations which see early signsas a discrete stage before acute psychosis.

In terms of its clinical application, ICS has anumber of implications. First, early signs configur-ations which are more closely allied to concepts ofimplicational meaning may provide a means ofreducing the apparent amplitude and variation inearly signs, and increase the sensitivity andspecificity of relapse prediction. Second, falsepositive increases in early signs imply incompleteactivation of implicational meaning and thereforeprovide an opportunity for individuals and clin-icians to rehearse coping scenarios for early relapse.Third, enabling individuals to increase their mind-fulness of implicational meaning may facilitatedelayed acceleration of relapse thus widening thewindow of early intervention. Fourth, ICS providesa detailed view of the initiation and acceleration ofrelapse, which may facilitate the development ofcognitive therapy as an effective early interventionstrategy.

Recovery from Psychosis

Within ICS the central engine of cognition is theinteraction between propositional and implicationalsubsystems. This interaction is a motivated systemwhich can be utilized to explain the maintenance ofpsychotic symptoms. In our analysis the develop-ment of psychotic symptoms such as delusions andhallucinations is a consequence of the centralengine's attempts to reduce discrepant experiencesbetween subsystems. In this sense delusions resultfrom the interaction between externally-derivedsensory inputs (e.g. voice tone, criticism, and visualimages), propositional meanings (e.g. `I believe hXithinks this of me'), and internally-derived sensoryinputs (e.g. body-state and information processing).Hallucinatory phenomena such as voices, thoughtinsertion, withdrawal or broadcasting, result fromthe interaction between internally-generated events(e.g. thoughts and images), internally-derived

sensory inputs and patterns of propositionalmeaning. Implicational meaning provides aschematically-derived and holistic level of meaningwhich abstracts over these subsystems. In depres-sion, implicational meaning is motivated to reducedepression. We propose that in psychosis, implica-tional meaning offers a vehicle which serves toprotect important evolutionary strategies associatedwith private and public social attractiveness. Privatesocial attractiveness is maintained through theexternal attribution of discrepancies as in persecu-tory delusions or negative auditory hallucinations.Public social attractiveness is maintained throughinternal attribution of negative events (as in punish-ment delusions), or external attribution of positiveinternal events (as in pleasant auditory hallucina-tions). According to this analysis, psychotic symp-toms could be thought of as providing a marker forthe status of `parent' schematic models that containinformation about self- and other-related models(including illness). This has important implicationsfor the role of psychological therapy in facilitatingrecovery from acute or chronic psychosis. Thefollowing case illustration aims to demonstrate therelevance of ICS in the course of treatment andrecovery.

CASE ILLUSTRATION

Referral

Miss X is a 23-year-old female with a 212-year history

of Schizoaffective Disorder (DSM-IV; AmericanPsychiatric Association, 1994). During this periodshe has had two acute episodes of psychosis. Herfirst episode occurred during her final year studyingchemistry at university. She experienced a probable6-month duration of untreated illness before she wasseen by her general practitioner and referred to aconsultant psychiatrist. She was initially diagnosedas having manic depression. Her second episodeoccurred almost 6 months later and required a6-week period of hospitalization. At discharge shewas referred for assessment for cognitive therapybecause of persisting depressed mood, delusionsand ideas of reference. The decision to refer forcognitive therapy was based on the case manage-ment goals identified prior to discharge to facilitaterecovery from acute psychosis.

Initial Presentation

Miss X's initial presentation was characterized bydepressed mood. She reported symptoms of low

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

ICS and Psychosis 271

energy, poor motivation, excessive sleep, poorappetite, and anhedonia. She described spendingmuch of her day in bed, and was frequently pre-occupied with the supernatural. The main emotionsreported were guilt, fear, shame and embarrass-ment. She reported feeling fearful in a range of socialsituations and avoided going outside, meetingfriends, and playing golf and swimming. Miss Xsaid that she had got `involved' with the super-natural at university after she used a Ouija board.She described feeling ashamed and embarrassed ofher behaviour and feared that some retributionwould be carried out against her. During hospital-ization, she had been admitted with punishmentdelusions. She believed that she was being talkedabout in the media and that the Queen, governmentand the church were operating a conspiracy againsther. At initial presentation she continued to reportfears that this retribution was continuing against herand indeed cited evidence from newspapers andtelevision supporting her beliefs.

The First Episode

Miss X's first episode occurred during the final yearat university. In the November, her mother andfather noticed that she was becoming increasinglywithdrawn. They attributed this to the long dailytravel to university and so arranged for her to share aflat closer to the university. Miss X became increas-ingly isolated, and her work at university deterio-rated. She recalls becoming increasingly worriedabout her exams as the year progressed andconsulted a Ouija board in order to gain reassuranceabout her exams. Approximately 2 weeks later, shedescribes experiencing passivity phenomena, whichshe attributed to the act of an evil spirit, which shemust have accessed by the Ouija board. At the timeof experiencing passivity she was ripping up heruniversity textbooks. In an effort to seek forgivenessfrom God, she visited her local church at night andcarved crosses into the grass. Following a consul-tation with her general practitioner a referral wasmade to a consultant psychiatrist. Following suc-cessful pharmacological treatment, Miss X con-tinued to have intrusive recollections of theseevents, particularly the memory of her body havingbeen taken over. These memories were associatedwith increased fear and somatic manifestations ofanxiety. She became increasingly worried aboutothers finding out what had happened and becamehypervigilant for any references to religion and thesupernatural.

The Second Episode

Miss X became increasingly socially avoidant. Shewas encouraged to take part-time work in a localbar. During this period other employees bullied her.Miss X attributed this to her use of the Ouija boardand making crosses in the ground. She becameincreasingly vigilant, and began selectively attend-ing to any evidence of others knowing about herbehaviour. She started noticing programmes ontelevision and articles in newspapers, whichreferred to religion and the supernatural. Shedeveloped the belief that information about herwas being passed to the media, and that the sourceof this information was the church and government.She believed that the purpose of publishing thisinformation was to assist some retribution againsther. She experienced a range of intrusive imageryrepresenting this retribution. For example, sherecalled an image of being boiled in a bath, afterwhich her body was broken up and fed to crows.

She was seen by her consultant psychiatrist, whonoted marked depressed mood characterized bydelusional preoccupation, ideas of reference, andthought disorder. She was admitted informally to alocal inpatient psychiatric unit for 6 weeks.

Background history

Miss X lives at home with her parents and youngersister. She described her family background as closeand strongly Christian. She was encouraged toperform well at school and was given considerableapproval on her academic performance. She recallsperforming badly at school in English. This was asource of considerable tension within the family.Miss X recalled being told that she was doing wellbut not good enough. Her parent's response todifficulties at school was disappointment, whichMiss X interpreted as failure to meet her parent'sexpectations. This gave rise to increased studyand reduced contact with friends and peers. Sherecalls high levels of competitiveness with hersister, perceiving her sister as more talented thanher. Some of these issues were observed duringassessment of the family. There were strongpressures on Miss X to recover her social function-ing. At times she perceived that she was beingcriticized by her parents for her lack of motivationand progress in recovery.

Formulation and Treatment

Figure 2 below provides a diagrammatical formu-lation of the initiation, acceleration and maintenance

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

272 A. Gumley, C. A. White and K. Power

of Miss X's psychosis. The patterns of implicationalmeaning which were hypothesized to abstract overthese processes are detailed alongside the associatedpropositional and sensory-derived inputs (emotion,

behaviour and physiology). For the purposes offacilitating recovery, intervention initially targetedmaintenance processes before proceeding toaccelerating and initiating factors.

Figure 2.

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

ICS and Psychosis 273

MaintenanceICS proposes that psychotic symptoms are main-

tained by the continuing regeneration of implica-tional meaning which abstracts over multiplepatterns of internally- and externally-derived infor-mation. In addition, according to our model psy-chotic symptoms represent thoughts and beliefswhich are markers for the state of these `parent'schematic models. In the case of Miss X a number ofinformation sources appeared to be relevant to themaintenance of her delusions and ideas of reference.Miss X reported a range of thoughts and imagesincluding intrusive memories of being bullied, usinga Ouija board, and being possessed. These imageswere provoked by externally-derived sensorysources of information including religious or super-natural references contained in television pro-grammes, newspapers, books and radio. Miss X'sdelusional beliefs reflected a pattern of propositionalmeanings represented by, `I believe people areplanning to harm me', `I believe the church, media,and government are involved in these plans', and `Ibelieve that this is because I have offended society bycommitting crimes'. The implicational meaning ofthese beliefs (self as bad, others as punishing) wasassociated with experiencing fear, shame, guilt anddepression. The combination of these multiplesources of internal and external information servedto continually regenerate implicational meaning,and thus represents the interlock maintaining hersymptomatology.

Initial treatment sought to reduce the conviction,preoccupation, and distress associated with Miss X'sdelusional beliefs. Strategies were selected whichwould directly or indirectly address implicationalmeaning structures. These strategies were the form-ulation of an alternative rationale, which accountedfor the evidence, maintaining her beliefs, cognitiverestructuring of intrusive memories and images,and modification of beliefs concerning televisionand other media.

Careful analysis of evidence cited by Miss X toaccount for her beliefs led to the identification of aseries of critical incidents. These are illustrated inFigure 2 as consulting a Ouija board, passivityphenomena (ripping up textbooks), making crossesin the church grounds, and bullying at work. Herinitial appraisal of these incidents was that she hadsinned against God and the Church by using aOuija board. In consequence she described experi-encing shame and guilt. These emotions weretransformed to fear after she ripped up her text-books. She externally attributed this act to beingpossessed by an evil spirit, which she had accessed

during her use of the Ouija board. Her subsequentcarving of crosses in the church grounds com-pounded her increasing shame and fear associatedwith her actions. She later perceived the bullying atwork as direct evidence of the intention of others tokill her as punishment for her sins. The therapist(A.G.) formulated these incidents as `traumaticevents', which challenged her pre-existing assump-tions. These assumptions are illustrated in Figure 2,for example, `Good things happen to good people,bad things happen to bad people'. The formulationproposed that Miss X's attempts to assimilate theseevents to her pre-existing beliefs lead her to assumethat because these negative events occurred, sheattributed blame to herself. Thus her pre-existingassumptions were preserved.

The formulation was also utilized to explain theoccurrence of cognitive events including intrusivememories, personalization of media references, andthe co-existing agitation, depressed mood, andavoidance/withdrawal. Subsequent cognitiverestructuring of Miss X's intrusive memoriesenabled her to reattribute her explanations concern-ing her role in these events from evidence of herbadness to evidence of her efforts to cope with theincreasing fear and uncertainty generated by herexpectations of the forthcoming university examina-tions. This enabled Miss X to consider alternativeexplanations for the occurrence of religiousand supernatural references and articles in themedia.

AccelerationCriticism from family members was hypothes-

ized to play an accelerating role in the developmentof psychosis. Miss X was exposed to a pattern ofcommunication within the family, which impliedthat she was a failure. This implicational meaningwas generic across her prior academic performanceand indeed her rate of recovery and social func-tioning achievements. Perceived criticism fromparents was addressed within family sessions. Thedevelopment of expressed emotion was traced toher parents' attributions of her behaviour. Theyperceived her withdrawal as a sign that Miss X hadnot been motivated to `get better'. Their evidencefor this was her persistent rumination over materialconcerning the supernatural. Initially they weresupportive of her recovery, but lack of progress intheir eyes led to feelings of disappointment andanger. Miss X interpreted this as disapproval, whichled to increased depression. Miss X believed thatshe was a failure to her family and that there waslittle that she could do. Family sessions focused

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

274 A. Gumley, C. A. White and K. Power

upon developing a shared information of symp-toms and the family's response to Miss X's illness.The family and Miss X were taught strategies tomanage these problems and difficulties.

InitiationAs Miss X progressed in therapy the formulation

was used to develop a model of relapse which con-tained a configuration of symptoms, thoughts andbehaviours. This configuration was given meaningby the formulation itself by specifying the beliefsresponsible for the development of emotions suchas shame, anxiety, depression and the paranoiaitself. Early intervention strategies focused upon anumber of factors. First the identification of earlysigns of relapse. Second the use of cognitivetechniques to weigh evidence for and against thepossibility of imminent relapse. This is particularlyimportant in reducing the likelihood of falsepositive predictions of becoming unwell, and there-by reducing oversensitization to relapse. Third, theuse of formulation to provide meaning to theappearance of early signs. Fourth, the delivery ofcognitive and behavioural interventions aimed atdecelerating relapse, reducing symptomatology andrestoring functioning.

In terms of outcome at 24 months Miss X reportsno paranoid or depressive symptomatology. Miss Xhas commenced mainstream full-time employment,and there have been no further relapses.

Discussion

The central element which distinguishes thisapproach from approaches using similar cognitivebehavioural treatment strategies for psychoticsymptoms is a more dynamic case conceptualiz-ation which serves to explain process variablesoutlined in the ICS model, as well as the contentvariables from other approaches (e.g. attributions)to formulating psychotic symptoms. The ICSapproach to formulation enables the clinician totrace the development and evolution of implica-tional meaning structures through the encapsula-tion of salient pre-morbid cognitive structures andtheir action on the acceleration and maintenance ofpsychosis. In the case of Miss X, pre-morbidassumptions regarding acceptance by others andachievement, played a key role in the externalmisattribution of discrepant behaviours during theacute episode, e.g. ripping up textbooks wasattributed to external spiritual forces. In addition,ICS shows how multiple patterns of sensory inputhave an impact on the maintenance and acceleration

of implicational meaning. For example, duringrecovery the pattern of criticism expressed withinthe family regenerated the implicational meaningthat Miss X was a failure. Therefore, therapeuticstrategies, which targeted this pattern of expressedemotion within the family were hypothesized to bea more effective means of modifying implicationalmeaning, as opposed to use of cognitive techniquesaimed at restructuring this belief within individualsessions.

CONCLUSIONS

The conceptualization presented in this paper aimsto provide an integrated model of relapse and thecourse of psychosis which is both theoretically andclinically relevant. In order to achieve this aim, theconceptualization draws on Teasdale and Barnard's(1993) Interacting Cognitive Subsystems model ofdepression. The ICS approach enables a detailedview of how multiple sources of informationinteract to establish self-organizing, self-perpetuat-ing, processing configurations that act to maintainpersistent cognitive-affective states. The modelpredicts that implicational meaning is criticallyinvolved in the processes of initiation, accelerationand maintenance of relapse in psychosis.

The main predictions of the ICS model would bethat it should be possible, for example, using singlecase methodology, to trace the evolution of im-plicational meaning regarding the illness process,and the gradual integration of self-schemata withinthe overall configuration. These testable predictionswould seem to be the main ways in which theorycould drive therapeutic endeavours. In clinicalpractice, patients could be enabled to understandthe gradual evolutionary process that is developing.Indeed, this process would in itself influenceimplicational meaning, providing as it does a newmeaning framework of the experience of prodromalsigns and emerging psychosis.

Further research could classify illness trajectoriesinto stages characterized by differences in therelative predominance of implicational meaning ininitiation, acceleration or maintenance stages. In-deed, it should also be possible to test this modelusing cognitive psychology paradigms whichwould provide evidence on which meaning systemis primary in the processing of information, andindeed, the relative transparency of implicationalmeaning. One might hypothesize, that this wouldbe different according to illness history, and degree

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

ICS and Psychosis 275

of critical overlap in schematic models of self,others, illness and world.

In addition, this model provides a basis for thedevelopment of psychological early interventions forrelapse. This is an area, which has remained under-developed. In this respect we are currently under-taking a randomized controlled trial of a cognitivelyorientated intervention for relapse. In this trial we arecurrently recruiting individuals who are consideredto be relapse prone. Relapse proneness is defined byone or more of the following factors; those with ahistory of recent relapse, non-compliance withmedication, socially isolated, and living in a highstress environment. This is a client group thathas traditionally challenged services and we hopethat such an approach as described in this paper willoffer hope in the management and amelioration ofpotentially negative illness trajectories.

ACKNOWLEDGEMENTS

The Chief Scientist's Office, Scottish Office GrantNumber K/RED/18/13, has supported this work.The views expressed are those of the authors' andnot necessarily those of the Chief Scientist's Office.We are also very grateful to Dr Philip Barnard forhis very helpful comments on earlier drafts of thispaper.

REFERENCES

American Psychiatric Association (1994). The Diagnosticand Statistical Manual of Mental Disorders (4th edn).Washington, DC: American Psychiatric Association.

Barnard, P. (1985). Interacting Cognitive Subsystems: Apsycholinguistic approach to short term memory. In: A.Ellis (Ed.), Progress in the Psychology of Language (Vol. 2).Lawrence Erlbaum Associates Ltd, pp. 197±258.

Barrowclough, C., Johnston, M. and Tarrier, N. (1994).Attributions, expressed emotion and patient relapse: anattributional model of relatives response to schizo-phrenic illness. Behaviour Therapy, 25, 67±88.

Beauvois, J. and Joule, R. (1996). A Radical DissonanceTheory. London: Taylor and Francis.

Beck, A. T., Rush, A. J., Shaw, B. F. and Emery, G. (1979).Cognitive Therapy of Depression. Chichester: Wiley.

Bentall, R. P. (1990). The illusion of reality: a review ofintegration of psychological research on hallucinations.Psychological Bulletin, 107, 82±95.

Bentall, R. P. (1996). From cognitive studies of psychosisto cognitive-behaviour therapy for psychotic symp-toms. In: G. Haddock and P. D. Slade (Eds), Cognitive-

Behavioural Interventions with Psychotic Disorders.London: Routledge.

Bentall, R. P. and Kaney, S. (1989). Content specificprocessing and persecutory delusions: an investigationusing the emotional Stroop test. British Journal ofMedical Psychology, 62, 355±364.

Bentall, R. P., Kaney, S. and Bowen-Jones, K. (1995).Persecutory delusions and recall of threat-related,depression-related and neutral words. Cognitive Therapyand Research, 19, 331±343.

Birchwood, M. (1995). Early intervention in psychoticrelapse: cognitive approaches to detection and manage-ment. Behaviour Change, 12, 2±19.

Birchwood, M., Mason, R. and Macmillan, F. (1993).Depression, demoralization and control over psychoticillness. Psychological Medicine, 23, 387±395.

Birchwood, M., Smith, J., Macmillan, F., Hogg, B., Prasad,R., Harvey, C. and Bering, S. (1989). Predicting relapsein Schizophrenia: The development and implementa-tion of an early signs monitoring system using patientsand families as observers. Psychological Medicine, 19,649±656.

Bowers, M. B. (1968). Pathogenesis of acute schizophrenicpsychosis: an experimental approach. Archives of Gen-eral Psychiatry, 19, 348±355.

Brewin, C. R. (1989). Cognitive change processes inpsychotherapy. Psychological Review, 96, 379±394.

Chadwick, P. and Birchwood, M. (1994). The omnipotenceof voices: A cognitive approach to auditory hallucina-tions. British Journal of Psychiatry, 164, 190±201.

Donlan, P. T. and Blacker, K. H. (1975). Clinicalrecognition of early schizophrenic decompensation.Disorders of the Nervous System, 36, 323±330.

Fear, C., Sharp, H. and Healy, D. (1996). Cognitiveprocesses in delusional disorders. British Journal ofPsychiatry, 168, 61±67.

Festinger, L. (1957). A Theory of Cognitive Dissonance.Stanford: Stanford University Press.

Freedman, B. and Chapman, L. J. (1973). Early subjectiveexperience in schizophrenic episodes. Journal of Abnor-mal Psychology, 82, 45±54.

Frith, C. D. (1992). The Cognitive Neuropsychology ofSchizophrenia. Lawrence Erlbaum Associates.

Gaebel, W., Frick, U., Kopcke, W., Linden, M., Muller, P.,Muller-Spahn, F., Pietzcker, A. and Tegeler, J. (1993).Early neuroleptic intervention in schizophrenia: areprodromal symptoms valid predictors of relapse?British Journal of Psychiatry, 163(Suppl. 21), 8±12.

Garety, P. and Hemsley, D. R. (1994). Delusions: Investi-gations into the psychology of delusional reasoning. Oxford:Oxford University Press.

Garety, P., Hemsley, D. and Wessely, S. (1991). Reasoningin deluded schizophrenic and paranoid patients. Journalof Nervous and Mental Disease, 179, 194±201.

Gilbert, P. (1997). The evolution of social attractivenessand its role in shame, humiliation, guilt and therapy.British Journal of Medical Psychology, 70, 113±147.

Greenberg, L. S. and Pascual-Leone, J. (1997). Emotionand the creation of personal meaning. In: M. Power andC. R. Brewin (Eds), The Transformation of Meaning inPsychological Therapies: Integrating Theory and Practice.Chichester: Wiley.

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

276 A. Gumley, C. A. White and K. Power

Herz, M. and Melville, C. (1980). Relapse in Schizo-phrenia. American Journal of Psychiatry, 137, 801±812.

Hirsch, S. R. and Jolley, A. G. (1989). The dysphoricsyndrome in schizophrenia and its implications forrelapse. British Journal of Psychiatry, 155(Suppl. 5),46±50.

Jackson, C. and Birchwood, M. (1996). Early interventionin psychosis: Opportunities for secondary prevention.British Journal of Clinical Psychology, 35, 487±502.

Jolley, A. G., Hirsch, S. R., McRink, A., et al (1989). A trialof brief intermittent neuroleptic prophylaxis forselected schizophrenic outpatients: clinical outcome atone year. British Medical Journal, 298, 985±990.

Jolley, A. G., Hirsch, S. R., Morrison, E., et al. (1990). Atrial of brief intermittent neuroleptic prophylaxis forselected schizophrenic outpatients: clinical and socialoutcome at two years. British Medical Journal, 301,837±842.

Jorgensen, P. (1998). Early signs of psychotic relapsein schizophrenia. British Journal of Psychiatry, 172,327±330.

Kaney, S. and Bentall, S. R. (1989). Persecutory delusionsand attributional style. British Journal of MedicalPsychology, 62, 191±198.

Kaney, S. and Bentall, S. R. (1992). Persecutory delusionsand the self serving bias: evidence from a contingencyjudgement task. Journal of Nervous and Mental Disease,180, 773±780.

Kinderman, P. (1994). Attentional bias, persecutorydelusions and the self-concept. British Journal of MedicalPsychology, 67, 56±66.

McCandless-Glincher, L., McKnight, S., Hamera, E.,Smith, B. L., Paterson, K. and Plumlee, A. A. (1986).Use of symptoms by schizophrenics to monitor andregulate their illness. Hospital and Community Psychiatry,37, 929±933.

McGorry, P. D., Chanes, A., McCarthy, E., Van Reil, R.,McKenzie, D. and Singh, B. (1991). Post traumatic stressdisorder following recent onset psychosis. An unrecog-nized post psychotic syndrome. Journal of Nervous andMental Disease, 179, 253±258.

Maher, B. A. (1988). Anomalous experiences and delu-sional thinking. In: T. F. Oltmanns and B. A. Maher(Eds), Delusional Beliefs. New York: Wiley.

Malla, A. K. and Norman, R. M. G. (1994). Prodromalsymptoms of schizophrenia. British Journal of Psychiatry,164, 487±493.

Malla, A. K. and Norman, R. M. G. (1995). Prodromalsymptoms of relapse in schizophrenia: A review.Schizophrenia Bulletin, 21, 527±536.

Marder, S. R., Mintz, J., Van Putten, T., Lebell, M.,Wirshing, W. C. and Johnson-Cronk, K. (1991). Earlyprediction of relapse in schizophrenia: An applicationof receiver operating characteristic (ROC) methods.Psychopharmacology Bulletin, 27, 79±82.

Marder, S. R., Wirshing, W. C., Van Putten, T., Mintz, J.,McKenzie, J., Johnson-Crink, K., Lebell, M. andLiberman, R. P. (1994). Fluphenazine vs placebosupplementation for prodromal signs of relapse inschizophrenia. Archives of General Psychiatry, 51,280±287.

Morrison, A. P. and Haddock, G. (1997). Cognitive factorsin source monitoring and auditory hallucinations.Psychological Medicine, 27, 669±679.

Morrison, A. P., Haddock, G. and Tarrier, N. (1995).Intrusive thoughts and auditory hallucinations: acognitive approach. Behavioural and Cognitive Psycho-therapy, 23, 259±264.

Neuchterlein, K. H. and Dawson, M. E. (1984). A heuristicvulnerability/stress model of schizophrenic episodes.Schizophrenia Bulletin, 10, 300±309.

Padesky, C. A. and Greenberger, D. (1995). CliniciansGuide to Mind over Mood. New York: Guilford Press.

Power, M. J. and Champion, L. A. (1986). Cognitiveapproaches to depression: A theoretical critique. BritishJournal of Clinical Psychology, 25, 201±212.

Power, M. J. and Dalgleish, T. (1997). Cognition and Emo-tion: From Order to Disorder. Hove: Psychology Press.

Rachman, S. J. (1978). An anatomy of obsessions.Behaviour Analysis and Modification, 2, 235±278.

Salkovskis, P. M. (1985). Obsessive-compulsive problems:a cognitive-behavioural analysis. Behaviour Research andTherapy, 25, 751±783.

Salkovskis, P. M. and Kirk, J. (1997). Obsessive compul-sive disorder. In: D. M. Clark and C. G. Fairburn (Eds),Science and Practice of Cognitive Behaviour Therapy.Oxford University Press.

Slade, P. D. and Bentall, R. P. (1988). Sensory Deception: AScientific Analysis of Hallucinations. London: Croom-Helm.

Subotnik, K. L. and Neuchterlein, K. H. (1988). Prodromalsigns and symptoms of schizophrenic relapse. Journal ofAbnormal Psychology, 97(4), 405±412.

Tarrier, N. and Turpin, G. (1992). Psychosocial factors,arousal, and schizophrenic relapse. British Journal ofPsychiatry, 161, 3±11.

Tarrier, N., Barrowclough, C. and Bamrah, J. S. (1991).Prodromal symptoms of relapse in schizophrenia. SocialPsychiatry and Psychiatric Epidemiology, 26, 157±161.

Tarrier, N. (1992). Management and modification ofresidual positive psychotic symptoms. In: M. Birch-wood and N. Tarrier (Eds), Innovations in the Psycho-logical Management of Schizophrenia: Assessment,Treatment and Services. Wiley.

Teasdale, J. D. (1997a). The transformation of meaning:the interacting cognitive subsystems approach. In: M.Power and C. R. Brewin (Eds), The Transformation ofMeaning in Psychological Therapies: Integrating Theory andPractice. Chichester: Wiley.

Teasdale, J. D. (1997b). The relationship betweencognition and emotion: the mind in place in mooddisorders. In: D. M. Clark and C. G. Fairburn (Eds),Science and Practice of Cognitive Behaviour Therapy.Oxford University Press.

Teasdale, J. D. and Barnard, P. J. (1993). Affect, Cognitionand Change: Re-modelling Depressive Thought. Hove:Lawrence Erlbaum Associates.

Teasdale, J. D., Taylor, M. J., Cooper, Z., Hayhurst, H. andPaykel, E. S. (1995). Depressive thinking: shifts inconstruct accessibility or in schematic mental models?Journal of Abnormal Psychology, 104, 500±507.

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

ICS and Psychosis 277

Trower, P. and Chadwick, P. D. J. (1995). Pathways to thedefence of self: A theory of two types of paranoia.Clinical Psychology: Science and Practice, 2, 263±278.

Weiner, B. (1985). Spontaneous causal thinking. Psycho-logical Bulletin, 97, 74±84.

Weiner, B. (1986). An Attributional Theory of Motivation andEmotion. New York: Springer-Verlag.

Yusupoff, L., Haddock, G., Sellwood, W. and Tarrier, N.(1996). Cognitive behavioural therapy for hallucina-tions and delusions: current practices and future trends.In: P. M. Salkovskis (Ed.), Trends in Cognitive andBehavioural Therapies. Chichester: Wiley.

Copyright # 1999 John Wiley & Sons, Ltd. Clin. Psychol. Psychother. 6, 261±278 (1999)

278 A. Gumley, C. A. White and K. Power