Here we review the evidence for deficits in mindreading (or Theory of Mind [ToM]) tasks in schizophrenia and look at some work on true metacognitive tasks in schizophrenia, in which schizophrenia patients also display deficits. We argue that what psychiatrists refer to as a “lack of insight” is an example of a failure to make judgements of themselves that they do make of others.
Since the publication of Chris Frith's seminal Cognitive Neuropsychology of Schizophrenia (Frith Reference Frith1992), research investigating ToM in schizophrenia has been widespread. A deficit in mindreading is clearly demonstrable in schizophrenia, but whether this is specific and causally related to certain symptoms or merely further evidence of generalised cognitive impairment found in virtually all patients with the disorder remains unclear (Harrington et al. Reference Harrington, Siegert and McClure2005). Sprong et al.'s (Reference Sprong, Schothorst, Vos, Hox and Van Engeland2007) meta-analysis showed a large and statistically significant impairment on ToM tasks (Cohen's d=1.26) across all schizophrenia sub-types and tasks, larger than studies restricted to “paranoid” patients, and claimed that there was support for mindreading impairment as a trait marker of schizophrenia.
Studies apparently supporting Carruthers' claim that mindreading is intact in “passivity” schizophrenia have had small sample sizes in their group (e.g., 7 in Corcoran et al. Reference Corcoran, Mercer and Frith1995; 1 in Pickup & Frith Reference Pickup and Frith2001). Furthermore, these studies may be confounded by other common symptoms; that is, someone with passivity symptoms may well exhibit paranoid or negative symptoms. So claiming that passivity symptom patients “perform normally when tested on batteries of mindreading tasks” (target article, sect. 9, para. 1) is overly bold. Indeed, recent work comparing controls with “passivity” patients showed different neural activity in ToM tasks, even though responses were broadly similar (Brüne et al. Reference Brüne, Lissek, Fuchs, Witthaus, Peters, Nicolas, Juckel and Tegenthoff2008). This suggests that a different cognitive strategy is used by such patients in mindreading tasks.
We question whether passivity “experiences” are ever pure metacognitive failures. Although classical phenomenologists like Kurt Schneider attempted to separate what in today's parlance we would call the experience from the attribution, the tiny corpus of examples in the literature invariably link the experience with a “psychotic” explanation – that a person or organisation is doing the movement or speaking through me using some dastardly device or technology. When the experience alone is reported, we can't be sure that there is no abnormal attribution. Our clinical impression is that the two are inextricably linked – at least in people presenting to clinical services. We believe this is the essence of Frith's model: that delusions, such as those of alien control, may build on passivity experiences but are the result of ToM reasoning which has gone awry. One might even say that psychosis is the result of excessive attribution – usually of malign intent – to people or things. People with psychosis have a theory of mind – it is just the wrong theory.
We would also argue that there is evidence for impairment in metacognition in schizophrenia in a truer sense than the “weak” variety described. The Beck Cognitive Insight Scale (Beck et al. Reference Beck, Baruch, Balter, Steer and Warman2004) shows differences between schizophrenia patients and controls in their agreement with statements on subscales assessing self-reflection and self-certainty (Warman et al. Reference Warman, Lysaker and Martin2007). Furthermore, some studies show considerable deficits in psychotic patients' assessment of their own poor neurocognitive function (Medalia & Thysen Reference Medalia and Thysen2008), although accurate assessment of such deficits may coexist with lack of awareness of the implausibility of beliefs suggesting fractionation of metacognitive awareness (Gilleen et al., in press). However, metacognitive failure is not for a want of trying. Bedford and David (Reference Bedford and David2008) showed that patients actually place time and value in self-reflection, but just struggle to do it accurately. Additionally, patients' metacognitive performance on the Wisconsin Card Sort Task (i.e., the confidence they place in the decisions they make) predicts disrupted insight in schizophrenia, suggesting a link between a failure of metacognition in a narrow sense, to a broader one (Koren et al. Reference Koren, Seidman, Poyurovsky, Goldsmith, Viksman, Zichel and Klein2004).
What psychiatrists refer to as a lack of insight (the ability to judge oneself to have a mental disorder and to “relabel” abnormal experiences as related to such a disorder) is very common in schizophrenia and has even been conceptualised as a core aspect of the condition (David Reference David, Amador and David2004). It seems to be an example of a failure of metacognition. Patients fail to make accurate judgements about themselves – in relation to their thoughts and experiences – and instead choose often bizarre and impossible explanations. However, it does not seem to be the case that patients who lack insight into their own illness are completely unable to make accurate judgements about others. Their mindreading abilities remain intact to the extent that they are relatively unimpaired in their attributions of “madness” to others. Rockeach's Three Christs of Ypsilanti (Rockeach Reference Rockeach1964) reports the reactions of three patients presenting with broadly the same delusion. They all continued to believe in their identity whilst rejecting the others as mad: “Truth is truth, no matter if only one person speaks it” (p. 150), claims one, steadfastly defending his identity.
Indeed, empirical studies using vignettes have shown no differences between patients and controls in attribution of illness. Startup (Reference Startup1997) showed that patients could distinguish between normal and psychotic thoughts, feelings, and behaviours in the vignettes but this showed no correlation with the patients' insight into their own condition. McEvoy et al. (Reference McEvoy, Schooler, Friedman, Steingard and Allen1993) found that patients correctly labelled characters with psychotic symptoms in vignettes as suffering from mental illness, but failed to note their own similarity to the character – although it was obvious to the patients' doctor. Further studies assessing the relationship of mindreading and metacognition as related to their applications in clinical insight are required: a simple correlation between the two functions seems unlikely (Pousa et al. Reference Pousa, Duñó, Blas Navarro, Ruiz, Obiols and David2008).
We note that metacognition can be enhanced by presenting the self as if another person – for example, by replaying a video of himself when ill to the patient (Davidoff et al. Reference Davidoff, Forester, Ghaemi and Bodkin1998). A conceptually similar approach taken to neurological patients with anosognosia for hemiplegia seems to yield increased awareness (Marcel et al. Reference Marcel, Tegnér and Nimmo-Smith2004). These observations are not decisive in dissociating mind reading from metacognition but suggest the following hypotheses: That self-awareness may at least make use of cognitive mechanisms which afford awareness of others' intentions and beliefs; that this is not necessarily automatic and may be prevented, presumably by attentional mechanisms or strategic failures (not looking), but can be overcome by adopting a third-person perspective to the self. Thus, some practical benefit may accrue from considering metacognition and ToM as linked but separable.
