The meaning of group differences in cognitive test performance

What does it mean that groups of individuals given a diagnosis of schizophrenia get fewer items correct and respond more slowly than comparison groups on tests of memory, attention, executive functioning, and social cognition? Green et al's answer1 is unequivocal: the weaker performance is caused by a “cognitive impairment” , which is a “primary deficit” , arising from a specific neuropathological cause. This “im‐pairment” is viewed as both a vulnerability factor and a “core feature” of schizophrenia. This is a position currently held by a majority of the field and supported by a large evidence base.

However, the question “why do individuals diagnosed with schizophrenia tend to demonstrate weaker neurocognitive performance?” actually remains without a clear answer. The basic data represent group differences on test performance, but this performance is multidetermined. Factors that have been found to influence the cognitive test performance of individuals given a diagnosis of schizophrenia include motivation, effort, dysfunctional attitudes, asociality, stress, negative emotion, and conceptual disorganization2. While it is an empirical question the extent to which these factors contribute to poor test performance, it seems fair to assume that accounting for their aggregate impact would significantly reduce the group differences.

The authors – along with the field – claim external validity for the tests: that weaker performance represents a “pervasive and stable” deficit of cognitive processes across situations in everyday life. However, variability in cognitive performance is a common observation3. When individuals who perform in the “severely impaired” range on neurocognitive tests are engaged in satisfying and personally meaningful activities – such as playing chess, driving a car, or preparing a complex recipe – they are observed displaying high levels of cognitive function. During these meaningful activities, motivational and cognitive resources become energized and are activated for use. This contrasts with the neurocognitive testing environment, which is rife with factors – such as beliefs about being incapable, inferior, or judged; and completing tasks that are not engaging or meaningful – that de‐energize motivation and resources.

Seeing everyday cognition in this more dynamic way across tasks and situations echoes recent work by Cohen et al4 demonstrating situational variability in alogia and flat affect, which are features that have also traditionally been viewed as stable. It also implies that treatment efforts may be better suited to helping people access their cognitive resources rather than remediate neurocognition. Ultimately, terms like “deficit” and “impairment” are inaccurate as applied to cognition, especially as the literature is dependent on testing conditions that are denuded of the very qualities that allow variability in performance to emerge. Given the multitude of factors responsible for test performance, group differences in performance on these tests cannot simply “reflect” isomorphic representations of “core deficits” in brain pathophysiology.

Turning to the neurological evidence, a similar pattern emerges. Post‐mortem studies showed group differences in brain volume between individuals given a diagnosis of schizophrenia and those from the general population. Rather than being evidence of a core cognitive impairment, as had been claimed, the reductions in brain volume turned out to be the result of long‐term exposure to antipsychotic medications5. Notably, non‐human primates exposed to antipsychotic medications exhibited the same degree of brain volume loss observed in the post‐mortem human studies6. This is yet another example of group differences not being obvious indices of underlying neuropathology.

Similarly, group differences in functional magnetic resonance imaging and EEG measures have been interpreted as representing a core neurocognitive impairment. Though widespread practice, the assumption that differential neurophysiological activation during neurocognitive tasks indicates neurocognitive impairment is a fallacy that has plagued the neuroscience literature broadly7. For example, in Green et al's description of dorsolateral prefrontal cortex (DLPFC) activation, they indicate that some studies find hypoactivation and others hyperactivation of the DLPFC. It is unlikely that both hypoactivation and hyperactivation suggest impairment, yet this is the interpretation that is made. Additionally, group‐level neurophysiological differences do not indicate impairment. Musicians show increased neural tissue volume in some regions and decreased volume in others8. This does not suggest that musicians have a neurocognitive impairment, but that they have specialized knowledge from repeated practice. Presumably individuals given a diagnosis of schizophrenia have had experiences – such as trauma, exclusion, and positive symptoms – to a greater degree than the general population, which would be expected to manifest in differential neurophysiology.

The developmental course of test performance is often cited as evidence that impaired neurocognition is a core and stable feature of schizophrenia. As Green et al describe, poorer performance is observed prior to onset of the first episode of psychosis and remains relatively stable after a diagnosis is given. However, negative symptoms are also present prior to the onset of the first episode, and both poor neurocognitive performance and negative symptomatology appear to emerge at similar periods of development (approximately age 9) in individu‐als who later develop schizophrenia9. Thus, it is possible that poor neurocognitive performance represents a consequence of negative symptoms such as amotivation, or that other variables (e.g., negative bias and beliefs) may lead to the development of both negative symptoms and poor neurocognitive performance. For example, childhood trauma has been associated with later performance on neurocognitive tests, and laboratory induced social exclusion impairs subsequent neurocognitive performance in healthy individuals10.

It is also worth emphasizing that neurocognitive performance has failed to predict who will develop a psychotic disorder among individuals at clinical high risk. If neurocognition were a core feature, then it should be associated with the development of the disorder. It would seem more likely that group differences in test performance do not reveal a distinct feature of schizophrenia, but instead an epiphenomenon that arises from amotivation, negative attitudes, trauma history and other aspects of the disorder.

Finally, language matters. It is unfortunate that the authors refer to “schizophrenia patients” throughout the manuscript. Similar to the other absolute terms – “deficit” and “impairment” – this term has inaccurate connotations. Nobody is just a patient, just a diagnosis, or just a collection of deficits. The science of schizophrenia would benefit from focusing on the whole person, with the mental health challenges that make up the diagnosis being just part of the full picture. Knowing the person at his/her best; and being able to accurately and dynamically assess his/her strengths, positive attributes, and beliefs will all be invaluable in this effort. Ultimately, the distance between the person given a diagnosis of schizophrenia and the typical citizen might grow very small indeed.