Firestone & Scholl (F&S) present a stimulating critique of putative empirical evidence for the cognitive penetrability of perception. In making their case, however, they focus exclusively on research on perception and cognition in neurologically normal populations. In doing so, they neglect potentially important sources of informative data afforded by research on abnormal perception and cognition. Cognitive neuropsychology and cognitive neuropsychiatry are scientific disciplines that draw inferences about aspects of normal cognition (such as reading, object recognition, belief formation, reasoning, decision-making, and theory of mind) by studying patients with cognitive deficits (Coltheart Reference Coltheart2007). We suggest that a comprehensive exploration of the relationship between perception and cognition should consider research from these disciplines. In particular, we demonstrate that the issue of cognitive penetrability looms large in contemporary debates about the formation and maintenance of delusions.

According to the two-factor theory of delusions, two distinct factors are causally responsible for the formation and maintenance of delusions (Coltheart et al. Reference Coltheart, Langdon and McKay2011). The first factor explains why the content of a delusional belief comes to mind, and the second factor explains why the belief is adopted rather than rejected. To date, the two-factor theory has focused on explaining specific monothematic delusions (delusions with one theme) associated with neurological damage, but some tentative suggestions have been made concerning how the two-factor theory might explain polythematic delusions (delusions with multiple themes) associated with psychiatric illnesses such as schizophrenia (Coltheart Reference Coltheart2013).

Consider the Capgras delusion, a monothematic delusion in which a patient believes that a spouse or close relative has been replaced by an impostor. This delusion is thought to stem from disruption to the autonomic component of face recognition, such that familiar faces are recognized as familiar but feel unfamiliar. Empirical support for this hypothesis comes from studies that have found that, unlike control participants, patients with Capgras delusion do not show a pattern of autonomic discrimination (indexed by skin conductance response) between familiar and unfamiliar faces (e.g., Brighetti et al. Reference Brighetti, Bonifacci, Borlimi and Ottaviani2007; Ellis et al. Reference Ellis, Young, Quayle and de Pauw1997; Hirstein & Ramachandran Reference Hirstein and Ramachandran1997). Other work, however, suggests that an anomalous autonomic response to familiar faces is not sufficient for the development of Capgras delusion. Tranel et al. (Reference Tranel, Damasio and Damasio1995) studied patients with damage to ventromedial frontal regions of the brain who also failed to show a pattern of autonomic discrimination between familiar and unfamiliar faces, yet were not deluded.

According to two-factor theorists, Capgras patients and Tranel et al.'s (1995) ventromedial frontal patients share a common first factor: anomalous autonomic responses to familiar faces. What distinguishes them from one another is that Capgras patients have a second anomaly: a cognitive deficit in the ability to evaluate candidate beliefs. Analogous two-factor accounts have been offered for several other monothematic delusions (Coltheart et al. Reference Coltheart, Langdon and McKay2011). Importantly, all two-factor accounts are predicated on a conceptual distinction (and empirical dissociation) between perception and cognition: abnormal perception as the first factor and a cognitive belief evaluation deficit as the second factor. Furthermore, two-factor accounts are not committed to perception being cognitively penetrable, meaning that the two-factor theory is consistent with the hypothesis F&S present.

In contrast to the two-factor theory, the prediction-error theory of delusions holds that delusion formation and maintenance are caused by a single factor: aberrant processing of prediction errors (mismatches between expectations and actual inputs). In particular, delusions are conceived as attempts to accommodate inappropriately generated prediction error signals (Corlett et al. Reference Corlett, Taylor, Wang, Fletcher and Krystal2010; Fletcher & Frith Reference Fletcher and Frith2009). Prediction-error theorists have tended to focus on delusions associated with schizophrenia, but they have also offered accounts of monothematic delusions associated with neurological damage (Corlett et al. Reference Corlett, Taylor, Wang, Fletcher and Krystal2010). Whereas the distinction between perception and cognition is critical for the two-factor theory, prediction-error theorists minimize or disavow this distinction:

Furthermore, according to prediction-error theorists delusions provide examples of cognition penetrating perception: There exist “interactions between perception and belief-based expectation” (Corlett et al. Reference Corlett, Taylor, Wang, Fletcher and Krystal2010, p. 357), and “delusional beliefs can alter percepts such that they conform to the delusion” (Corlett et al. Reference Corlett, Taylor, Wang, Fletcher and Krystal2010, p. 353). This position seems to be in tension with the hypothesis F&S present. Consequently, we suggest it would be useful for F&S to expand the scope of their review by critically examining whether there is empirical evidence from research on delusions that cognition penetrates perception. If empirical evidence is compelling, then there exists a counterexample to F&S's hypothesis.

In this commentary, we have shown that the relationship between cognition and perception is a major point of interest in contemporary research on delusions. This suggests that evidence from cognitive neuropsychology and cognitive neuropsychiatry may play an important role in testing the hypothesis F&S present.