Despite significant research efforts, the progress in understanding and prevention of suicide has been slow. This is due at least in part to the vagueness of current suicide theories that rely on verbal specifications of all constructs and consequent predictions and thus provide correlative rather than mechanistic accounts of suicidal behavior (SB). In contrast, formal theoretical frameworks grounded in learning theory and decision neuroscience, such as reinforcement learning (RL), afford a productive approach for uncovering cognitive and decision processes involved in suicide through the use of explicit mathematical ideas to define constructs, hypotheses, predictions, and assumptions of interest. Further, when faced with a crisis, a transition from thinking about to attempting suicide involves the need to make a decision under increased cognitive demands imposed by high information load (multiple decision options), uncertainty, and time pressure (current crisis). However, although real-life decision-making is dynamic and is influenced by situational demands, existing neural models of suicidal diathesis are relatively static and not yet integrated with the advances in decision neuroscience. In contrast, electrophysiology augmented with computational modeling can reveal the dynamic unfolding of within-trial neural events and is well suited for answering mechanistic questions about cortical oscillations that underlie real-time decision-making. This project will experimentally manipulate cognitive demands on decision-making using an RL task with a validated computational model to examine decision processes in SB (Aim 1). It will also build on the basic electrophysiology literature that implicates alpha, beta, and theta oscillations in working memory and cognitive control by examining neural encoding of reinforcement history in attempted suicide (Aim 2). Exploratory Aim 3 will allow for a preliminary examination of the effects of self-reported mood, stress, and task-related neural activity on behavior. Participants will be 120 adults (40 suicide attempters, 40 suicide ideators, and 40 healthy controls) recruited from inpatient, outpatient, and community settings. Expert mentorship team will provide training in behavioral and computational modeling, basic neuroscience, and electrophysiology. The work will take place at the Department of Psychiatry of the University of Pittsburgh, which has a long track record of supporting career development of its junior faculty. This K23 will promote the Candidate’s long-term goals of contributing to suicide research and prevention by identifying the neurocomputational mechanisms of deficient decision-making in SB. The project is consistent with the NIMH’s goals and funding priorities and builds on the Candidate’s prior NIMH-funded RDoC-informed work. It also falls into several areas of emphasis for NIMH’s Computational Psychiatry Program (“translating basic computational models into clinical research by informing the models with experimental data from clinical population,” “understanding how individual differences in neuronal activity contribute to the transmission and processing of pathophysiological information in the central nervous system”).

When faced with a crisis, some people but not others transition from thinking about suicide to acting on these thoughts. This project will use a formal theoretical framework combined with electroencephalography to test the hypothesis that individuals at risk for suicide become cognitively overwhelmed under challenging life circumstances and make poor choices as a result. The anticipated impact of this proposal includes the identification of more precise targets for cognitive clinical interventions.