Cognitive impairment has long been recognized as a symptom of schizophrenia, affective, and psychotic disorders. Until relatively recently, most research has focused on the severity, type, and course of impairment. As is reviewed in this series by Dr Alice M Saperstein and Dr Matthew M Kurtz,1 we have learned that cognitive deficits are evident in most people with schizophrenia and schizoaffective disorders, evident even in the prodromal phase but seem to worsen around the time of first episode, and stable for many years, even as positive symptoms, such as hallucinations and delusions, wax and wane. Dr Christopher R Bowie, Ms Maya Gupta, and Ms Katherine Holshausen2 report in this series that cognitive symptoms in the affective disorders appear to be more state related, though more than 60% of people with bipolar disorder (BD) continue to manifest cognitive dysfunction, even when clinically stable. Most studies comparing people with schizophrenia to nonpsychiatric samples find that cognitive test scores are on average 1.3 to 2.0 standard deviations below the mean, which means that 91% to 98% of the general population score higher. These deficits put patients at a considerable disadvantage when seeking employment and more generally facing the challenges of independent, productive living. Repeatedly, the research indicates that cognitive deficits are associated with functional outcome. It is for this reason that there is tremendous interest in finding ways to treat cognitive dysfunction.In this In Review are 2 papers that provide an overview of the behavioural approaches that have been developed to address cognitive dysfunction in the psychotic and affective disorders. Dr Saperstein and Dr Kurtz1 focus on cognitive remediation (CR) for schizophrenia, while Dr Bowie and colleagues2 address how CR is being used to treat people with psychotic and nonpsychotic affective disorders. We learn from these 2 papers that the field of psychology is at a crossroads. No longer is there debate about whether people with schizophrenia can learn, or if cognitive skills are malleable. The research overwhelmingly supports the efficacy of CR for schizophrenia and schizoaffective disorder-when it is done in the context of a psychiatric rehabilitation model. The evidence base for using CR for people with nonpsychotic affective disorders is relatively smaller; nevertheless, CR for mood disorders is emerging as an effective nonpharmacological treatment option for improving cognitive performance in people with major depressive disorder and BD. Further, CR impacts not only cognition but also, as intended, it helps to improve psychosocial functioning and reduce disability.Having established that cognitive deficits do respond to behavioural interventions, researchers are now tackling how to enhance the ability of CR to improve cognition, how to promote the transfer of cognitive gains to real-world functioning, and develop personalized treatments that are easily disseminated with fidelity. A tension has emerged as these issues are addressed. While there is tremendous enthusiasm for harnessing emerging knowledge about neuroplasticity to engineer more effective and efficient computer-based treatments, there is, simultaneously, a groundswell of support for integrating humanistic psychiatric rehabilitation principles into CR approaches to enhance functional outcomes. At their most extreme, these tensions emerge as camps.1 One side, fuelled by the technology industry, advocates a medical model approach that uses US Food and Drug Administration-approved computer devices that dose cognitive exercises like medications. The other side, fuelled by person-centred advocacy, emphasizes psychotherapeutic approaches to address the many factors (for example, beliefs, motivation, and emotions) that promote transfer of learning to real-world behaviours.The first large multisite study of CR in schizophrenia4 chose the model of dosed, device-based computer exercises to test the effectiveness of CR, and the results failed to replicate the promising findings from a single-site trial of device-based CR. …