The social defeat hypothesis of schizophrenia: issues of measurement and reverse causality

Eleven years ago, two of us1 published the social defeat hypothesis of schizophrenia, in an attempt to find a common denominator for several schizophrenia risk factors. The hypothesis posits that the long‐term experience of being excluded from the majority group leads to an increased baseline activity and/or sensitization of the mesolimbic dopamine system, putting the individual at increased risk for the disorder1, 2.

The hypothesis may explain to a certain degree why a history of migration, membership of a disadvantaged ethnic minority group (e.g., African‐American ethnicity), urban upbringing, low IQ, childhood trauma, drug abuse, hearing impairment, homosexuality3, 4, and perhaps also autism are schizophrenia risk factors.

We noted that the experience of defeat is neither a specific nor a sufficient or necessary risk factor for schizophrenia, and that other factors, including genetic vulnerability, co‐participate in determining the nature of the outcome. Interestingly, neuroreceptor imaging studies reported evidence of dopamine sensitization in non‐psychotic subjects with hearing impairment or with a history of childhood trauma, thus supporting the hypothesis5, 6.

However, there are at least two good reasons to criticize the hypothesis. First, it is difficult to measure social defeat in humans, because assessments based on interviews or questionnaires are biased by a tendency to give socially desirable replies. Second, one could argue that many children who go on to develop schizophrenia exhibit motor, cognitive and social impairments and that social defeat, therefore, is not a causal factor, but a consequence of a disorder in neurodevelopment, already present before the onset of psychosis and mainly driven by genetic factors.

As for the first issue, we recognize that the social defeat hypothesis is based on an interpretation of group comparisons (e.g., migrants versus natives, deaf subjects versus normal hearing individuals) and that we do not know with certainty whether individuals who develop schizophrenia are more “defeated” than others. This situation entails the risk of an ecological fallacy, which would be the case if, for example, successful migrants were found to be at equal risk of schizophrenia as non‐successful migrants. However, we contend that the social defeat hypothesis is the most viable interpretation of the available data. The pattern of findings for ethnic minorities in Europe, for example, shows the highest risks for the least successful and most discriminated groups: African‐Caribbeans and Black Africans in the UK, Inuit in Denmark and Moroccan‐Dutch in the Netherlands.

As to the second point of criticism, we agree that schizophrenia likely “begins” long before the onset of psychosis. Studies of the Philadelphia Neurodevelopmental Cohort, for example, have shown that individuals aged 11 to 21 years who endorse psychotic symptoms (but do not meet the criteria for schizophrenia) are cognitively delayed, have a diminished whole brain grey matter volume, and grey matter volume deficits in frontal, temporal and parietal cortex7. It is true that these individuals are more likely to develop schizophrenia than others. However, given the fact that about 16% of all cohort members endorse psychotic symptoms, it is also evident that the majority will not develop the disorder and that motor, cognitive, social or anatomic impairments are merely risk factors or risk indicators of disorder, not hallmarks.

We propose that the epidemiology of schizophrenia supports a role for social exclusion, because it is unlikely that the genes that contribute to a defective neurodevelopment also code for migration, disadvantaged ethnic minority status, urban upbringing, low IQ, childhood trauma, drug abuse, homosexuality, hearing loss and autism. The social defeat hypothesis offers a more parsimonious explanation for this pattern of findings and deserves further development and testing.

First, since only two studies examined the risk of schizophrenia among individuals with a non‐heterosexual orientation, further investigations of this topic are required. The hypothesis can also be tested in various other discriminated groups, such as those who are physically less attractive, who harbor a congenital or acquired handicap, a gender identity disorder, etc..

Second, it is important to examine whether “defeated” individuals who develop schizophrenia differ from other defeated subjects in the way they cope with defeat. Are they more likely to deny the very occurrence of defeat or do they attribute their problems to external causes? If they deny any problem, can implicit association tests reveal that they are implicitly aware of an inferior position?

Third, it is possible to conduct experiments in the laboratory. One can expose individuals to a negative evaluation or rejection and examine which subjects react by developing an increase in subclinical psychotic symptoms.

Fourth, any of these approaches can be examined in models of gene x social defeat interaction which, if apparent, would add to the validity of the notion of “defeat” underlying environmental effects.

Finally, using neuroreceptor imaging, dopamine function can be compared between non‐psychotic members of excluded and non‐excluded groups. For example, a prospective study of dopamine function among migrants, shortly after arrival and after an interval of several years, and a comparison of the results to those obtained from a native control group, would be most informative.

In sum, the hypothesis seems to provide many promising avenues for investigating epidemiological patterns that are still lacking a satisfactory explanation.

Jean‐Paul Selten1,2, Jim van Os1,3, Elizabeth Cantor‐Graae4


1Department of Psychiatry and Psychology, University of Maastricht, Maastricht, The Netherlands; 2Rivierduinen, Leiden, The Netherlands; 3King's College London, King's Health Partners, Department of Psychosis Studies, Institute of Psychiatry, London, UK; 4Division of Social Medicine and Global Health, Department of Clinical Sciences in Malmø, Lund University, Malmø, Sweden