5.1. Inferring non-social intentions

In terms of performance, overall, the only between-group difference was observed in the superordinate condition, in which participants had to appreciate distal goals of an agent performing a sequence of basic motor acts. In that condition, patients with schizophrenia were found to perform significantly less well than healthy participants. Importantly, these difficulties were associated with an abnormally high effect of the bias, with patients providing significantly less responses than controls toward the non-preferred intentions than toward the preferred ones. Last, in the schizophrenia group, performances improved to a lesser extent than in healthy controls as the visual information increased, further suggesting that patients’ tendency to over-rely on their internal, prior expectations arise together with a disposition to neglect the external, visual information conveyed by the action scene.

This pattern of performance might be the consequence of an impaired ability to revise expectations which one progressively formed during the task. Increasing the probability of some intentions indeed led participants to favour specific expectations about which type of intention was the most likely to be accomplished, given past observation. Because preferred and non-preferred intentions were randomly alternated in the sequences, participants had to continuously revise their expectations regarding the type of intention to come. The patients’ inclination to perseverate in selecting preferred intentions may suggest an impaired ability to revise these expectations, i.e. to update their beliefs about the agent’s current intention. Concomitantly, schizophrenic patients showed an information-gathering bias that led them to mistrust the visual information which could potentially disconfirm these initial beliefs.

Such an inability to disengage from prior, self-generated expectations, together with a tendency to ignore outer information, echo specific biases observed in schizophrenia in a wide range of studies, such as a ‘confirmatory bias’ (Jones et al., 1999), or a tendency to make hasty decisions (Colbert et al., 2002; Huq et al., 1988; Brankovic & Paunovic, 1999). In tasks of probabilistic reasoning, patients with delusions indeed tend to make a decision on the basis of less evidence than healthy controls. They are also overconfident in their own judgments relative to healthy participants (Huq et al., 1988) and preferentially seek evidence that confirms their initial beliefs (Brankovic & Paunovic, 1999).More generally, deluded patients are prone to give excessive credit to their own internal productions, and, as a consequence, to hold them despite little evidential support (Jones et al., 1999).

These observations fit particularly well with the present results: in the superordinate condition, schizophrenic patients are prone to give massive priority to biased intentions at the expense of non-biased ones and to underweight visual evidence arising from the action scene. Consistently, we found this profile of performance to significantly predict the severity of both the productive and disorganisation symptoms of the disease. Previous evidence showed that productive patients may be characterised by an impaired ability in decoupling their own intentions from those of others (Frith & Corcoran, 1996), while disorganised individuals exhibit specific difficulties in organizing their actions in the context of their own, but also other people’s goals (Zalla et al. 2006 ; Chambon et al., 2008). The present results further suggest that productive, as well as disorganised patients, exhibit a mentalizing profile that may be characterized by a disposition to hold prior beliefs about others’ intention with an abnormal degree of certainty, making them incorrigible by external evidence; thereby outdated beliefs about others’ mental states are not replaced by contextually appropriate ones. Finally, we show that such a profile (over-reliance on priors, disconfirming visual evidence) is characterised by specific difficulties in inferring distal, rather than proximal intentions. The reason for this specificity may be that inferring superordinate intentions requires participants to refer to an internal representation of the goalwhich is not directly available from observation, and, as a result, tends to be less challenged by visual evidence.

In total, depending on outdated expectations could characterise an abnormal style of reasoning that would contribute in both reasoning biases observed in deluded patients and specific mentalizing disorders shown in productive patients in general. As such, failure to revise prior expectations about other people’s goals and intentions could further account for productive patients misattributing actions, especially in situations that require continuously monitoring visual signals arising from the action scene (Franck et al., 2001). Finally, patients’ difficulties in disengaging from their prior expectations might echo problems that schizophrenics with passivity symptoms have in decoupling outer from inner world experience (Blakemore, 2003; Frith, Farrer et Franck, etc.), resulting in an abnormal inclination to attribute intentionality where there is none (‘over-TOM’: Abu-Akel & Bailey, 2000; Blakemore et al., 2003; Franck et al., 2001), or to perceive moving geometrical objects as endowed with agency (Frith, 2005).