Imagine going to see your M.D., describing how you feel unwell, and being told that you have ‘deficits in your immune system’. You would presumably inquire about the nature of the deficits, and whether they involve, say, autoimmune disorder or reduced levels of leukocytes. Now imagine a trip to the psychiatrist, whereupon, after a battery of tests, you are told that you have ‘deficits in social cognition’. What have you learned?

Pinkham et al. (Reference Pinkham, Morrison, Penn, Harvey, Kelsven, Ludwig and Sasson2019) compared individuals with autism to those with schizophrenia, for a large set of psychological tests designed to assess social-cognitive performance. Their comprehensive study was the largest to date in terms of sample sizes and numbers of tests, for such comparisons. They found that performance levels did not differ significantly between these two clinical groups for most of the tests. Their primary inference, couched in well-justified caveats, was an indication of ‘the potential benefit of applying treatments transdiagnostically’.

An alternative interpretation of this and similar studies is that poor performance on standard psychological tests of social cognition provides little information about the diagnosis, causes, or treatments of psychiatric conditions. This interpretation is based on three main points.

First, similar levels of deficits in social cognition, as indexed by self-report and task-based psychological tests, can result from similar, partially overlapping, independent, or opposite biological causes. Table 1 lists studies in which patients with autism and schizophrenia were compared for both deficit-associated psychological traits (measured, or based on previous work) and their neurological mechanisms assessed by EEG and/or MRI. In each case, similar psychological deficits showed evidence of opposite underlying causes. Such results suggest that information on levels of psychological deficit is of limited usefulness, for understanding or treatment, without information on neurological causes or correlates. Apparently, there are too many ways for complex systems, like human social-cognitive development, to undergo alterations in function, for processes to be inferred using levels of deviation from optimality. Pinkham et al. (Reference Pinkham, Morrison, Penn, Harvey, Kelsven, Ludwig and Sasson2019) acknowledge that shared deficits need not indicate similarity, but this point is not applied to their inferences, nor to the more general question of what information quantitative levels of deficits actually provides. A broader issue, as regards interpretation of social deficits found in both autism and schizophrenia, is that they are typically considered as evidence of etiologic overlap (e.g. Couture et al., Reference Couture, Penn, Losh, Adolphs, Hurley and Piven2010). This belief is belied by the apparent complete lack of evidence in the literature for shared social deficits in autism and schizophrenia (or psychosis) being demonstrated to be underlain by shared neurological mechanisms.

Table 1. Studies that compare subjects with autism and schizophrenia (or schizophrenia spectrum disorders) using the same protocol, and including neurological information from EEG and/or MRI

More-extensive comparisons of autism with schizophrenia, for diverse phenotypes, are provided in Crespi and Go (Reference Crespi and Go2015, Table 2).

Can disorder-specific tests salient to social-cognitive abilities help to alleviate such limitations? The Autism Quotient (AQ) test, one of the most commonly used metrics of autistic traits, shows elevated scores among individuals with anorexia (Westwood et al., Reference Westwood, Eisler, Mandy, Leppanen, Treasure and Tchanturia2016), borderline personality (Dudas et al., Reference Dudas, Lovejoy, Cassidy, Allison, Smith and Baron-Cohen2017), suicide attempts (Richards et al., Reference Richards, Kenny, Griffiths, Allison, Mosse, Holt, O'Connor, Cassidy and Baron-Cohen2019), and schizophrenia spectrum disorders (De Crescenzo et al., Reference De Crescenzo, Postorino, Siracusano, Riccioni, Armando, Curatolo and Mazzone2019). These results can be construed either as evidence of psychological overlap between autism and each of these conditions, or as indicating that the AQ broadly quantifies social problems (among other traits) with limited regard to their diverse underlying causes. Is there neurological overlap that grounds such psychometric data in biology?

Psychological tests that are based not on deficits per se, but on patterns of qualitative psychological differences, should, by contrast, provide information that can link more readily to potential causes. For example, the Movie for Analysis of Social Cognition permits differentiation of social deficits as due to under- v. over-mentalizing (Dziobek et al., Reference Dziobek, Fleck, Kalbe, Rogers, Hassenstab, Brand, Kessler, Woik, Wolf and Convit2006; Fretland et al., Reference Fretland, Andersson, Sundet, Andreassen, Melle and Vaskinn2015), which can in turn be connected with different patterns of neural activation (e.g. Backasch et al., Reference Backasch, Straube, Pyka, Klöhn-Saghatolislam, Müller, Kircher and Leube2013).

Second, quantification of psychological deficits, in the context of psychiatric categories, departs from Research Domain Criteria (RDoC) core recommendations to focus on adaptive biological phenotypes and how they malfunction, rather than on diagnoses. Approaches based on RdoC indeed fit with the standard medical model for the disease, which involves determining what biological system has become maladaptive, and how (Nesse and Stein, Reference Nesse and Stein2012). Application of the standard medical model of disease to psychiatry requires recognition that diagnoses are typically reified (falsely considered as ‘real’), as opposed to conceptualized as what they are: societal constructs that are useful for scientific communication. Pinkham et al. (Reference Pinkham, Morrison, Penn, Harvey, Kelsven, Ludwig and Sasson2019) contend that ‘shared dimensions of observable behaviors may help to pinpoint common mechanisms' in the RdoC framework, but such pinpointing may be more likely through analysis of how biological adaptations connect to psychological maladaptations, rather than through computation of psychological impairments.

Third, the quantification of deficits in psychiatry, in conjunction with the application of reified DSM diagnoses, encourages treatments based on categories rather than individuals. Current diagnostic categories can instead be considered as starting points for differential diagnosis of the biological causes of each individual's psychological difficulties. Such individual-based, differential-diagnostic procedures are, of course, the norm in non-psychiatric medicine, such as immunology, but they are extremely challenging in psychiatry given the complexities of how the brain develops and works. However, remarkably, even given sets of well-validated causes and correlates of, for example, autism, procedures for differential diagnoses of its biological etiology have yet to be developed, aside from genetic testing. Self-report and task-based psychological tests for social cognitive domains will certainly have their place in individualized diagnostic protocols, but only to the degree that they indicate differences rather than just quantifying deficits, and thereby connect relatively directly with biological causation.