Reply to Dr. Cutting and Dr. Fabrega

It is an honor to share this forum with Dr. John Cutting, as his work on psychopathology has been truly seminal to the field. We base our following comments not simply upon Cutting's remarks in this issue but on his book, The Right Cerebral Hemisphere and Psychiatric Disorders,¹ which offers a more detailed presentation of his views regarding delusions in schizophrenia. In all instances, we will try to make our abbreviated remarks self-contained so that reference to outside texts is not necessary; however, the reader is certainly encouraged to review Cutting's work for a more complete context.

Central to Cutting's view is the idea that a delusion's form provides critical clues as to the neurological impairments that give rise to it, and vice-versa. This is in marked contrast, as he points out, to the prevailing psychiatric tradition of using form simply for the purpose of diagnosing and categorizing a mental illness, or as a psychodynamic curiosity reflected by the individual patient. Cutting divides neurological impairments that potentially cause delusions in terms of their laterality, taking particular note of the respective brain regions' roles in distinguishing between classes and their members. Cutting argues that, consistent with the left hemisphere's responsibility for the abstraction of common properties, left-hemisphere aberrations are more likely to cause impairments in a patient's ability to group individual members by their common class. On the other hand, consistent with the right hemisphere's responsibility for recognizing individual variation, right-hemisphere aberrations are more likely to cause impairments in a patient's ability to recognize uniqueness within a common class.

These impairments, according to Cutting, dictate the form that a particular delusion may take. Left-hemisphere aberrations may cause delusions of persecution, reference, and influence (endowing inappropriate uniqueness or significance to objects and/or events that are, in fact, common), while right-hemisphere aberrations may cause delusions of misidentification or reduplication (endowing inappropriate generality to objects and/or events that are, in fact, unique). Cutting attributes a third type of delusion, characterized by “imminent misadventure to others, and bizarre happenings in the immediate vicinity,” to chronic generalized cortical dysfunction.

Schizophrenic delusions, Cutting argues, typically stem from a hemispheric imbalance in which impairment of the right hemisphere results in domination by the left. This is surprising, since—by Cutting's scheme—one would thus expect to see the majority of schizophrenic delusions of the misidentification/reduplication variety with a minority of schizophrenic delusions of the persecutory/reference/influence variety. Yet clinically, delusions of persecution, reference, and influence are by far the most common in schizophrenia. While Cutting states that studies suggest patients are more likely to create classes from individuals than are healthy control subjects (pp. 315–318), he also argues that schizophrenic delusions are characterized principally by a patient's tendency to imbue unimportant trivial events with importance—a tendency that he defines as “philosophical” or “religious” in his commentary to our article. Although intuitively appealing, attributions of this sort often suffer from vagueness and underdetermination. For example, the “philosophical” or “religious” nature of schizophrenic delusions may be understood as finding meaning (individual significance) in common (general) events, just as they may be equally understood as finding greater universal significance (generality) in unique events. Since Cutting identifies class-to-member and member-to-class reasoning with opposite neurological lateralities, and since he also views laterality as key to his conception of schizophrenic delusions, any ambiguity on this point would appear to weaken the efficacy of his method.

As Cutting points out in his book, the connection between schizophrenic cognitive symptoms and class-member processing has a venerable history, likely beginning with von Domarus's work in the 1940s on an alternative “schizophrenic logic” in which patients viewed two items as identical if they shared any properties in common. Subsequent work on this topic was inconclusive, which elsewhere we have attributed to a) the conflating of all schizophrenic patients in spite of varied cognitive symptoms (which makes it difficult to note how delusional patients differ from schizophrenic patients who are nondelusional); b) poor design of logical reasoning problems that often resulted in fatal floor effects for both patients and control subjects; c) lack of adequate control tasks (i.e., equivalent reasoning problems that were not of the class-member variety); and d) the differences in reasoning found between different patient groups when presented with affect-neutral versus affect-laden material—although certainly more progress has been made in this last area than in the other three.²

The first author's results from research in this area, although preliminary at this point, do not readily support the notion that class-member reasoning per se constitutes the area of delusional patients' reasoning that is disproportionately affected, but rather that class-member reasoning is disproportionately affected by emotional arousal, and that emotional arousal most affects delusional patients.³ On a Logical Reasoning test that required drawing propositional inferences, as well as choosing relevant versus irrelevant information using both class-member reasoning and propositional reasoning, there were significant differences between groups of healthy control subjects (n=16), delusional patients (n=10), thought-disordered patients (n=5) and schizophrenic patients who were neither delusional nor thought-disordered (n=13) on all three types of reasoning (P=0.005, 0.024, and 0.012, respectively). On all three types of reasoning, healthy control subjects performed best (approximately 60% correct), followed by delusional patients, non-delusional/non–thought-disordered patients, and finally thought-disordered patients. Post hoc analysis revealed that for each test, the significant differences were mostly due to differences between healthy control subjects and patients who were thought-disordered rather than between delusional patients and any of the other three groups.

In the same Logical Reasoning task, the items described above were additionally presented in identical form but using violent language, which was designed to induce mild emotional arousal in subjects while they performed the task. Interestingly, in the emotionally arousing context, delusional patients' performance showed marked differences from both that of healthy control subjects and those of other patients, and in this context, class-member reasoning did in fact stand out. The decline in performance from nonaroused to aroused conditions (counterbalanced for order) were the most striking for delusional patients (P=0.004; df=9) when compared both to patients with other symptoms and to healthy control subjects. In choosing relevant from irrelevant information, the decline in performance under arousal for delusional patients was significantly greater in the class-member reasoning section than in the propositional reasoning section (P=0.033; df=9 for decline in class-reasoning). For healthy control subjects, performance under arousal was minimally improved for propositional inferences. In choosing relevant from irrelevant information under arousal, healthy control subjects' performance declined slightly for the propositional reasoning, but reached the trend level (P=0.083; df=15) for the class-member reasoning. Nondelusional/non–thought-disordered patients displayed a pattern similar that of to delusional patients, although less exaggerated (decline under arousal was significant only for one section, again class-member reasoning, P=0.017; df=11). Patients with formal thought disorder showed significant improvement under arousal, but only for the class-member reasoning (P=0.035; df=4). Our data suggest not only that class-member reasoning, as opposed to other forms of reasoning, may be unusually sensitive to emotional arousal, but also that delusional patients stand out from other patients in that they are most likely to feel the deleterious effects of that arousal. Therefore, even under minimal emotional stimulation, delusional patients appear to be most likely to suffer deficits in class-member reasoning, consistent with Cutting's neurological predictions.

Where Cutting views the distinction between class-to-member versus member-to-class to be important to understanding the neural foundations of delusional content, we suggest that either type of class-member reasoning may be reducible to a more fundamental problem in sorting data and recognizing counterexamples—in this case, appropriately identifying the relevance or irrelevance of information in forming and amending beliefs. As we have tried to point out in our article, the ability to identify an experience as “general,” for instance, critically depends upon the ability to provide counterexamples to its uniqueness, just as the ability to identify an experience as “unique” critically depends upon the ability to provide counterexamples to its generality. Further analysis of the results described above showed that thought-disordered patients committed their errors almost entirely by taking in too much information (assigning inappropriate relevance to irrelevant information), whereas all other patient groups as well as healthy control subjects committed their errors generally by blocking out too much information (treating relevant information as if it were irrelevant). These patterns were particularly pronounced during arousal.

Our article focused on providing diagnostic criteria for delusions that would be at least consistent with a neurobiological explanation, rather than focusing on the neurobiological explanation itself. This does not mean that such an explanation is lacking, or that the theory that we advance is, as Cutting alleges, “untestable”: adequately recognizing counterexamples is certainly as biologically brain-based an ability as Cutting's identification of classes, and neurobiology surely “flattens out” the exact content of delusions just as surely and appropriately as do considerations of information processing. Current neuropsychological and neuroimaging (functional MRI) studies by the first author have centered on testing whether the recognition of counter-examples may be grounded in a more fundamental deficit in “filtering,” related to other, preattentive or minimally attentive deficits in sensory gating significantly associated with schizophrenia, combined with a strong and inappropriately generated affective component. If so, then the problem is not simply the failure to recognize counterexamples, but the combination of a strong, inappropriately activated emotional state (such as fear, anxiety, or euphoria) in need of explanation, coupled with a generalized failure to adequately discriminate between relevant and irrelevant information, interacting with a relatively intact ability to reason. We say “relatively intact” because, as with all individuals both healthy and not, reasoning ability—and particular that involving class-member reasoning—seems to decrease proportionally to the amount of emotional arousal; however, delusional patients seem, by our results, to require significantly less emotional stimulation in order to become aroused. Since we postulate that the foundational problem is the delusional patient's inappropriate choice of the information from which he will infer, rather than an impaired inference process itself, we are in agreement with Cutting's suspicion that at least one of the neurological problems to be identified concerns “input,” rather than “processing.” Our hypothesis that delusional patients possess an emotional vulnerability that further compromises their ability to correctly identify the relevance of information provides needed consistency both with the emotional themes that are characteristic of delusions and with a wide range of evidence that the formation and maintenance of delusions are exacerbated by emotional stress.

Where Dr. Cutting argues that our formulation lacks neurobiological rigor, Horacio Fabrega, Jr., suggests just the opposite—that we rely unproductively on neurobiology in clarifying a problem that is inextricably cultural. Dr. Fabrega additionally finds that our method may suffer from a form of hypocrisy, if putting the ostensibly culturally neutral criteria into practice requires that, for all intents and purposes, one still must implicitly use cultural considerations during evaluation. Obviously, this scenario would be worse than the current DSM standard, since cultural considerations would be used but not openly recognized and identified.

Dr. Fabrega's first claim, that delusions are inherently cultural constructs, “the amalgam of meaning-creating systems referred to earlier that includes syntax, grammar, semantics, metaphor, metonymy, imaginative models of reasoning, and conventionally named objects,” functions in his critique more as a self-evident assumption than a derived conclusion. Our article, of course, was written around the very question of whether delusions could be explained as constructs of inefficient information-processing without reference to syntax, grammar, semantics, metaphor, metonymy, etc.—the motivation for which was precisely an attempt to move toward neurobiology in diagnosing (and eventually, in treating) delusions. The thrust of our methodological differences with Dr. Fabrega seem to stem from our conviction that, in order to find the biological cause of delusions, or any other symptom for that matter, it is first necessary to abstract foundational deficits that could ground the random complexity that these deficits may entail. Such foundation deficits would thus occur independently of the complicated culture-context in which delusions actually are observed. This would mean, essentially, moving from a complicated set of culturally bound symptoms to a more simple set of biologically understood causes, a process that we view as absolutely necessary if psychiatry is to be absorbed into the rest of the natural sciences.

Dr. Fabrega's second concern, regarding the potential for hypocrisy, is certainly apt and a potential problem to which we have given some consideration. We have said that in testing model-restriction, the clinician needs to focus on how the patient formed and maintained his beliefs, rather than on the beliefs' specific content. In order to do so, the clinician must estimate to the best of her abilities the following: a) the number of potential counterexamples to the patient's belief, and b) the degree to which the patient appropriately responds to counterexamples that could disprove or cast doubt on the belief. Answering the latter can be determined solely with respect to the patient. Answering the former—i.e., determining the number of counterexamples to which he could reasonably be expected to have been exposed—arguably must be dependent on knowledge of the patient's culture.

Although we are very sympathetic to this criticism, we have also tried to make a compelling case in our article that the opposite is often true: that taking cultural considerations too seriously actually can obscure an accurate understanding of the real exposure that a patient has had to counterexamples. As one of our case examples, we presented a scenario in which a patient lives in a culture in which voodoo is widely practiced and believed. We pointed out that exposure to possible counterexamples to voodoo, even within this single well-described culture, can vary enormously depending upon different participants' roles in that culture. Information about actual counterexamples, in this case, can and should be determined only by specifically questioning the patient and people who are close to him about these, because general knowledge of the culture may presume more or less exposure to actual counterexamples than is actually present for that individual. As we explained, a patient from this culture may have had personal experience with voodoo, or he may have simply heard about it from friends, family, or neighbors. Even if he has had personal experience with voodoo, he may or may not have been in a position to be exposed to counterexamples to his practice of voodoo, either through inability to assess the results or because the hoped-for results may actually have been confirmed. Contextual (in contrast to actual, or “concrete”) counterexamples may also be evaluated outside the cultural context, even more easily, by questioning that determines the degree to which everyday laws of probability are maintained in various types of scenarios, both more and less abstract.