Delusions may be sub-served by the striatal habit system because the computationally intensive (but flexible) goal-directed system is impaired. A recent account of reinforcement learning from the point of view of predictive processing actually suggests that the goal-directed system is associated with processing higher up the Bayesian hierarchy; bootstrapped from simpler habitual reflexes lower down (Pezzulo et al., 2015). In this account, behavioural control is a function of the degree of precision of prediction error and control is ceded to the level of the hierarchy with the highest precision (of priors and prediction error)—there is no need for separate competing systems (Pezzulo et al. 2015).
We propose a single impairment in prediction-error–driven learning (across the hierarchy) in three stages: beginning with a delusional mood; next, the delusion forms in an a ha moment, in which explanatory insight occurs. Finally, the delusion is applied to do further explanatory work. Delusions may be an adaptive product of the shear pin breaking because they enable patients to remain in connection with their environment and to exploit its regularities to some degree.
Prior to delusions, there is a prodromal delusional mood in which attention is drawn toward irrelevant stimuli, thoughts, and associative connections which are distressing and unpredictable (Kapur 2003). This reflects an impairment in the brain’s predictive learning mechanisms, such that inappropriate prediction errors, are registered (Corlett et al. 2010).
People in this state feel uncomfortable and confused (Kapur 2003; McGhie and Chapman 1961). Biochemical markers of stress increase: cortisol, the endocrine signifier of stress, increases by as much as 500% (Sachar et al. 1963). Both emotional stress without narrative (Rosenberg et al. 2002), and neurohumoral markers of stress worsen health outcomes. Women with Borderline Personality Disorder have increased markers of cardiovascular plaques, likely due to increased cortisol (Greggersen et al. 2011). By some accounts, emotional stress can drive the emergence of auto-immune disease, perhaps via cortisol (Greggersen et al. 2011). It would seem to be important for an organism to have a mechanism in place to escape this state. Stress is also known to impair goal-directed behaviour and promote habitual responding (Schwabe & Wolf 2009)
If we conceive of delusions as habits that maintain some instrumental engagement in the world, they do have some adaptive value.
However, not all delusions motivate action. Some people with delusions double-bookkeep, both claiming a delusional belief and yet acting otherwise (Bleuler 1908; Sass 1994). For example, someone might claim their food is being poisoned yet continue to eat (Sass 1994). This is of particular interest in light of the above-discussed marked fixity of many delusions: though the person does not recant the delusional claim, he is able to eat the supposedly poisoned food. Bortolotti and Broome solve this paradox by appealing to biological motivation (Bortolotti and Broome 2012). Indeed, it is not difficult to see how acting in spite of one’s delusions could be biologically adaptive (maintaining nutrition, housing, etc.). In situations in which their negative symptoms and apathy are low, patients with delusions may be more likely to act.
This is consistent with the impact of incentive motivation on instrumental responding in the associative cybernetic model put forward by Dickinson and Balliene (1993). Here, instrumental behaviour is invigorated by a motivational system that codes the current value of an outcome (e.g. a reward)—thus allowing for the effects of changes in outcome value by satiation. This system might well be instantiated in the orbitofrontal cortex and ventral striatum (Daw et al. 2005), two structures that have been associated with apathy, anhedonia, and amotivation (Lebreton et al. 2009). The interaction of this motivational system with learning systems may dictate the degree to which delusions are acted upon (Bortolotti and Broome 2012).
Epidemiological studies highlight stress as an important predictor of delusion-consistent action (Coid et al. 2013). People more distressed by their odd beliefs are more likely to commit delusion related violence. Stress is likewise a facilitator of habitual responding (Schwabe et al. 2011; Schwabe and Wolf 2009). Here, we posit that stress further impairs the computationally expensive model-based prefrontal processing, increasing reliance on simpler, model-free control in the striatum. These effects correlate with cortisol levels.
Thus, stress may be both a key motivator (on a biologic level) and mechanism for a doxastic shear pin break that leads to delusions—a salve that permits some instrumental engagement with an unpredictable world to ensure continued survival.