Theories of cognition, emotion and the social
world: missing links in psychosis
Paul Bebbington, David Fowler, Philippa Garety, Daniel Freeman
and Elizabeth Kuipers
Throughout the twentieth century, clinicians and researchers struggled to establish
a convincing account of psychosis and the processes and mechanisms underlying
its manifestations. This effort depended on refinements of classification and case
definition over the whole course of the century. However, despite the huge invest-
ment of intellectual energy and monetary resources, results have been slow in
coming and disappointingly piecemeal. It has become clear that psychosis is a
phenomenon of great complexity. Recent social and cognitive models of psychosis
are attempts to deal with some aspects of this complexity. We will argue for the
appositeness of such models, and place them within the broader research effort in
psychosis. Before doing this, we need to revisit some of the consequences of the
formulation of the concept of schizophrenia, the disorder that represents the core
of psychotic phenomena.
A failed category?
The idea that schizophrenia is a failed category emerges regularly in the psychiatric
and psychological literature, the most prominent current advocate being Bentall
(2003). This is essentially a criticism of approaching schizophrenia as a medical
entity. The medical strategy of investigation involves the identification of syn-
dromes, which, in turn, form the basis of theories. These include those relating to
aetiology, pathology, outcome and treatment (Wing, 1978). Syndromes are essen-
tially theoretical constructs: while they can be regarded as disease entities, they are
never really more than tentative. Nevertheless, there is a tendency in psychiatry to
Society and Psychosis, ed. Craig Morgan, Kwame McKenzie and Paul Fearon. Published by Cambridge
University Press. # Cambridge University Press 2008.
regard psychiatric syndromes as having more virtue than they actually possess.
They do, however, form a useful basis for research. If theories about syndromes are
refuted, consideration may be given to abandoning them. In medicine we generally
reject theories, of aetiology and so on, before rejecting the theoretical construct
represented by the syndrome. We may nevertheless eventually decide that the
syndrome has failed, in the sense that the knowledge built up in relation to it is
confused and imprecise, and there may be better ways of organising the clinical
information for the purposes of research. There is current debate about whether
we have reached this stage with schizophrenia.
One of the difficulties for a biomedical approach in psychiatry is the sheer
cussedness of the subject matter. It is, for example, extremely difficult to construct
a classification, as categories in psychiatry are notorious for overlapping one
another. This is not merely a problem of imprecise definition: the subject matter
is actually hierarchical rather than planar, and any attempt to construct a flat
classification is thus probably doomed to failure.
The nature of psychiatric phenomena creates three crucial problems for the
definition of categories like schizophrenia. The first is the threshold problem, that
is, the point at which the syndrome becomes recognisable and distinguishable
from normal experience. The second is the boundary problem, the difficulty of
drawing valid lines between, for example, schizophrenia and bipolar disorder.
Added to this is a third problem, which is both empirical and conceptual. This is
that the disorder seems to be inherently difficult to explain.
One of the early authorities to grasp the problem of the threshold was Jaspers
(1913). He saw schizophrenia as the battleground between opposing traditions of
explanation in science. The methods of the physical sciences were concerned with
causal explanation, whereas social sciences involved appreciation through a pro-
cess of understanding, through meaningful connections. His view was that where
understanding failed (as it seems to in people’s attempts to share the experiences
of those deemed to be mad), we are left only with resort to physical explanation.
The implication is that in such circumstances we are facing a physical illness
with physical causes. The conclusion is that there ceases to be room for expla-
nations of the schizophrenic experience in terms of social contexts, and that
cognitive accounts represent understanding at a different level from biological
explanations. The resulting biological hegemony held sway in psychiatry until the
last decades of the twentieth century, and is only now being replaced by a more
flexible and comprehensive approach to the problems besetting scientific research
Jaspers himself never regarded schizophrenia as a no-go area for social and
psychological research. He was quite happy with the idea that psychosis could be
meaningfully connected to circumstances, although reluctant to see the process
220 P. Bebbington, D. Fowler, P. Garety et al.
as wholly understandable. He did acknowledge that it was difficult to identify the
point at which paranoia became delusional, and there is now considerable evi-
dence confirming his view.
Psychosis as a continuum with normal experience
Modern cognitive models of psychosis actually start by postulating continuities
between psychosis and normal experience. The emergence of psychotic phenom-
ena thus reflects an abnormal concatenation of largely normal mechanisms. What
is weakness in the medical category, in this theoretical formulation becomes
strength. In the past, the urge to make a categorical distinction between psychosis
and normality almost certainly led to a Procrustean tendency to discount unusual
beliefs and experiences in people we would be reluctant to see as undergoing a
psychosis. However, it became apparent by the late 1980s that paranoid ideation
and anomalous experiences, such as hearing voices, were not confined to clinical
groups. It is clear that the frequency of auditory hallucinosis greatly exceeds the
accepted prevalence of psychosis (e.g., Wiles et al., 2006). The distinguishing
feature of those in contact with services is the level of distress occasioned by their
unusual experiences (Hanssen et al., 2003; van Os et al., 1999).
The distribution of unusual beliefs in the populace is also extensive. Many
people are convinced of the truth of ideas that are not supported by available
and accessible evidence. These include beliefs in astrology, alien beings, telepathy
or ghosts. Political beliefs are held with strong conviction even though they may be
untried, or indeed tried and found wanting. People who hold these cherished ideas
typically have a confirmatory bias, being unlikely to consider alternatives impar-
tially. These beliefs shade into what would be regarded as delusional, since the
thinking that underpins them is similar in style to that in people with acknowl-
Attempts have been made to define delusions in terms that would enable them
to be distinguished clearly and reliably from normal thinking. The form of words
varies, but generally implies the following: they are held with a basic and compel-
ling subjective conviction; they are not susceptible to contrary experience or to
counter-argument; they are impossible, incredible or false; they lie outside the
belief systems characteristic of the individual’s cultural group. However, because
of the overlap with ‘normal’ thinking, it has been impossible to construct criteria
that are, individually or jointly, both necessary and sufficient for an operational-
ised definition (Bebbington and Broome, 2004).
The continua of beliefs in the community have been demonstrated empirically.
One example concerns ‘paranoid’ beliefs, in other words, those relating to self-
reference and threat. Freeman et al. (2005b) found that 30% of an internet sample
221 Cognition, emotion and the social world
of students had such ideas. Moreover, the frequency distribution of individual
paranoid ideas followed an exponential curve (Figure 14.1), with the relationship
between them being non-hierarchical, such that more extreme ideas were predic-
tive of those that were less extreme but not vice versa. This finding has now
been corroborated using general population data (Bebbington, P. E. et al. The
paranoia dimension in the general population (submitted to the British Journal of
Psychiatry)). The pattern is much like that shown by affective symptoms (Melzer
et al., 2002; Sturt, 1981). At a single point in time the continuum is defined by
differences between individuals, who are thus located at individual positions on
the curve. However, people are themselves likely to vary in a way that would place
them at different positions on the curve at different times, dependent on changing
circumstances. In a sense, they would move along the curve.
These findings have considerable relevance to the aetiology of psychosis. They
imply that in some people movement along a continuum (indeed probably more
than one continuum) results in the emergence of psychosis. Thus the role of
aetiology is to explain exactly why particular people make this journey at particular
times in their lives. In the genetic arena, this suggests a focus on quantitative
analyses (Linney et al., 2003), along with the identification of quantitative trait loci
(Plomin et al., 1994). In the psychological domain, it implies the concatenation of
different psychological attributes, some cognitive, some emotional (Hanssen et al.,
Total number of
10 15 20
Proportion of population (%)
y = 24.474e
Figure 14.1 Distribution of paranoia scores in a student population (taken with permission from Freeman
et al., 2005b)
222 P. Bebbington, D. Fowler, P. Garety et al.
2005; Krabbendam and van Os, 2005; Krabbendam et al., 2005). There are also
implications for treatment, in particular psychological treatments, such as cogni-
tive behaviour therapy (Kuipers et al., 2006).
The boundary problem
The boundary problem, whereby schizophrenia is imprecisely separated from
other psychiatric conditions, has also been a source of discomfiture. The relation-
ship between bipolar disorder and schizophrenia, for instance, is certainly ambig-
uous. Clearly some individuals have mood dysregulation, and some have psychotic
experiences. However, many have both, and in varying proportions. While
attempts may be made to see affective psychosis and non-affective psychosis as
distinct, they really represent opposite ends of a spectrum, and this is reflected in
considerable overlap in the candidate genes associated with the two conditions
(Craddock and Owen, 2005; Craddock et al., 2005; 2006).
Mood disturbances of some kind are almost invariable in cases of schizophrenia.
Thus, up to 40% of people with schizophrenia also have clinical levels of depres-
sion (Birchwood, 2003; Sands and Harrow, 1999), 30% meet criteria for post-
traumatic stress disorder (Mueser et al., 1998), 20% have panic disorder (Turnbull
and Bebbington, 2001) and up to 25%, obsessive compulsive disorder (Berman
et al., 1995). For a classification seeking to distinguish affective and non-affective
forms of psychosis, this is an embarrassment. However, once we cease to worry
about the overlap and think of emotional changes as potentially an integral part of
the psychotic process, it becomes possible to see it as a doorway to explanation.
This is easy for cognitive models of psychosis, which all involve an acknowledged
role for affect (Birchwood, 2003; Garety et al., 2001).
What then of the problems of explaining schizophrenia in biomedical terms? We
define schizophrenia in relation to people’s reports of their unusual mental
experiences. It would make life much simpler if we had privileged access to some
underlying condition or entity. However, all we have to go on is the disorder as
defined, the phenomena, not the noumena. In other words, it is the strange beliefs,
the hallucinations, and their contents, that we have to explain.
There has long been a prima facie assumption about aetiological theories in
psychosis, a reductionist position placing genetic explanations at the beginning of
the aetiological process. Virtually all genetic research in schizophrenia has been
based on the assumption that genetic abnormalities lead to abnormalities in
protein function, with consequent distortions of enzymatic activity, and that
223 Cognition, emotion and the social world
these in turn lead to corresponding deficits in neuronal function. These deficits
then result in cognitive dysfunctions that form the substrate for the schizophrenic
experiences that permit the diagnosis of the disorder. This is obviously a very
useful working paradigm, but it must be treated with caution.
The high heritability of schizophrenia is easy to demonstrate from twin studies
(Craddock et al., 2005). However, recent research has served to emphasise that not
only is schizophrenia a complex disorder to the clinical observer, it is also complex
in genetic terms. Such disorders have relatively high population prevalence, are
non-Mendelian and are imprecisely distinguished from the normal range. The
search for genes linked to schizophrenia has been long, arduous and expensive
(Norton et al., 2006; Owen et al., 2005). Candidates have been unearthed, but their
association with schizophrenia remains tentative. The genes encoding dysbindin
(DTNBP1) and neuroregulin (NRG1) are the strongest contenders, with evidence
for other genes (disrupted in schizophrenia (DISC1), D-aminoacid oxidase activator
(DAOA), regulator of G-protein signalling 4 (RGS4) and V-AKT murine thymoma
viral oncogene homolog 1 (AKT1)) being at least suggestive (Norton et al., 2006).
Some have been associated with abnormalities of brain structure and enzyme
function, which, albeit plausible, would seem to have a tenuous connection to
the schizophrenic phenotype (Callicott et al., 2005; Harrison and Weinberger,
2005; Meyer-Lindenberg et al., 2005). Given the modern synthesis of evolutionary
theory, genetics and developmental biology, focusing solely on encoding for
enzymes is likely to prove simplistic, even though links with schizophrenia may
be found. After all, only two percent or so of the DNA in human cells produce
proteins that act as enzymes, while much more is involved in complex regulatory
systems that switch the functions of protein-encoding genes on and off, something
clearly of potential relevance to an acknowledged developmental disorder like
schizophrenia. These issues have at last started to inform genetic research in
schizophrenia (Glaser et al., 2006; Law et al., 2006; Lipska et al., 2006).
Adjacent elements in the aetiological process seem likely to fall far short of a
one-to-one relationship, with the implication that at every step on the way the
specificity of relationships is progressively lost. The consequence is that the
association of individual genes with schizophrenia must inevitably be attenuated,
with small effect sizes (Norton et al., 2006).
Moreover, the genes identified so far seem to be associated with neurocognitive
deficits (Hariri et al., 2003; Meyer-Lindenberg et al., 2005). These are plausible
antecedents of negative symptoms of schizophrenia, and there is appreciable
empirical support for the association between them (Pinkham et al., 2003). Thus
genetic explanations so far lean towards an explanation of negative symptoms.
However, the heritability of schizophrenia is the heritability of positive symptoms,
since it is from these that the condition is identified. The potential association of
224 P. Bebbington, D. Fowler, P. Garety et al.
the Val/Val allelic form of COMT with vulnerability to psychotic responses to
cannabis might be one way round this difficulty (Caspi et al., 2005). However, we
fall far short of a comprehensive explanation in genetic terms for the positive
symptoms responsible for the heritability of the condition in the first place.
There are further problems with the aetiological process. The implicit unidir-
ectionality is an act of faith that almost certainly cannot be sustained. Thus the way
in which heritability is calculated means that it actually includes gene–environment
interaction. Moreover, gene–environment interaction covers environmental gene
induction, and the inducing environment may be social as well as physical. Thus,
natural variation in the competency of mouse mothers affects their offspring’s
responses to stress in later life (Meaney, 2001), apparently mediated by epigenetic
changes involving histone regulation of DNA expression. This is then maintained,
and affects the offspring’s own parenting behaviour (Fish et al., 2004; Weaver
et al., 2004). Thus the causal direction may sometimes run from the social to the
physical, rather than the other way around. Furthermore, our knowledge about
putative aetiological factors is based on induction. Thus, because some people with
schizophrenia have, for example, a family history of the same condition, an
inductive leap is made to the conclusion that all schizophrenia has a genetic
basis. The frailty of the logic underpinning this inference is apparent. As the same
caveat applies to social theories, aetiological modesty is imperative on all sides.
The final problem relates to the nature of schizophrenic symptoms themselves.
Most of the experiences that form the basis of the identified symptoms of schizo-
phrenia are about something: they carry within them representations of the social
world (paranoid ideation is an obvious example of this). This is the characteristic
of intentionality identified by Brentano (1874). Defining a medical condition in
terms of experiences that have intentionality means that explanations in purely
physical terms will always be incomplete.
This is the context in which models of psychosis involving social, emotional and
cognitive elements have been developed. They do not deny an important role for the
physical matrix of disorder, but they do add to the complexity and probably to the
potential of explanation. They form one part of the overall model. There are several
variants, and all are better described as social-cognitive-emotional models. They
focus on the sorts of explanation that can be developed for considering the person
with psychosis as an agent in a social and societal context. They also seek to explain
psychosis by trying to explain single symptoms or coherent groups of symptoms.
Analysing single symptoms
The study of single psychotic symptoms, or single types of symptom, is assisted by
a clear account of the symptoms themselves. Thus cognitive models encourage a
225 Cognition, emotion and the social world