This is the author manuscript accepted for publication and has undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as

A growing body of literature demonstrating the malleability of critical higher-order cognitive functions by means of targeted interventions has incited widespread scientific interest, most notably in the form of cognitive training programs. The results are mixed and a point of contention: It has been argued that gains observed in cognitive training are mainly due to placebo effects. To address this, we examined the effect of participant expectations on one type of cognitive training that has been central to the controversy, namely n-back training, by inducing beliefs about expected outcomes. Participants receiving n-back training showed improvements in non-trained n-back performance regardless of expectations, and furthermore, expectations for positive outcomes did not result in any significant gains in an active control group. Thus, there was no detectable expectancy effect in either direction as a function of the cognitive intervention used, suggesting that training-related improvements are unlikely due solely to a placebo effect.

Smartphone-based ecological mobile cognitive tests (EMCTs) can measure cognitive abilities in the real world, complementing traditional neuropsychological assessments. We evaluated the validity of an EMCT of recognition memory designed for use with people with serious mental illness, as well as relevant contextual influences on performance. Participants with schizophrenia (SZ), schizoaffective disorder, and bipolar disorder (BD) completed in-lab assessments of memory (Hopkins Verbal Learning Test, HVLT), other cognitive abilities, functional capacity, and symptoms, followed by 30 days of EMCTs during which they completed our Mobile Variable Difficulty List Memory Test (VLMT) once every other day (3 trials per session). List length on the VLMT altered between 6, 12, and 18 items. On average, participants completed 75.3% of EMCTs. Overall performance on VLMT 12 and 18 items was positively correlated with HVLT (ρ = 0.52, P < .001). People with BD performed better on the VLMT than people with SZ. Intraindividual variability on the VLMT was more specifically associated with HVLT than nonmemory tests and not associated with symptoms. Performance during experienced distraction, low effort, and out of the home location was reduced yet still correlated with the in-lab HVLT. The VLMT converged with in-lab memory assessment, demonstrating variability within person and by different contexts. Ambulatory cognitive testing on participants’ personal mobile devices offers more a cost-effective and “ecologically valid” measurement of real-world cognitive performance.

Background Cognitive tasks delivered during ecological momentary assessment (EMA) may elucidate the short-term dynamics and contextual influences on cognition and judgements of performance. This paper provides initial validation of a smartphone task of facial emotion recognition in serious mental illness. Methods A total of 86 participants with psychotic disorders (non-affective and affective psychosis), aged 19–65, were administered in-lab ‘gold standard’ affect recognition, neurocognition, and symptom assessments. They subsequently completed 10 days of the mobile facial emotion recognition task, assessing both accuracy and self-assessed performance, along with concurrent EMA of psychotic symptoms and mood. Validation focused on task adherence and predictors of adherence, gold standard convergent validity, and symptom and diagnostic group variation. Results The mean rate of adherence to the task was 79%; no demographic or clinical variables predicted adherence. Convergent validity was observed with in-lab measures of facial emotion recognition, and no practice effects were observed on the mobile facial emotion recognition task. EMA reports of more severe voices, sadness, and paranoia were associated with worse performance, whereas mood more strongly associated with self-assessed performance. Conclusion The mobile facial emotion recognition task was tolerated and demonstrated convergent validity with in-lab measures of the same construct. Social cognitive performance, and biased judgements previously shown to predict function, can be evaluated in real-time in naturalistic environments.