Full Program »
148. Reliability And Construct Relationships of Working Memory and Reinforcement Learning Paradigms in CNTRACS
The Cognitive Neuroscience Test Reliability and Clinical Applications for Serious Mental Illness (CNTRACS) consortium adapted cognitive paradigms from the cognitive neuroscience literature for use in clinical trials. Using data from 65 controls, 52 people with schizophrenia, 39 people with schizoaffective disorder, and 52 people with bipolar disorder with psychotic features, we assessed test-retest reliability and intertask relationships of these paradigms. In the domains of cognitive control, visual integration, and relational encoding, intraclass correlation coefficients (ICCs) replicated good test-retest reliability (ICCs: 0.52-0.77; controls only). Working memory paradigms also had good reliability (ICCs: 0.56-0.85). However, reinforcement learning tasks fared worse overall (ICCs: 0.00-0.68). Follow-up analyses with Cronbach’s α suggest implicit learning tasks have good to excellent internal consistency at initial testing (α’s: 0.71-0.92), but explicit learning tasks had largely unacceptable internal consistency (α’s: 0.05-0.62). Generally, task measures correlated with other measures from the same cognitive domain but not measures from other domains. This was not true for the reinforcement learning tasks, where measures within the tasks did not correlate. These findings reveal that most CNTRACS tasks appear to be good candidates for use in clinical trials, but highlight the challenges of understanding change over time for learning tasks.