Spatiotemporal Brain Signatures of Risk and Reward

Abstract

Rational economics and finance surmise that choices decree from a conscious arbitration between alternatives based on decision-theoretically computed values. Implicit in the computation of these option values are the perception and integration of different mathematical components (e.g., magnitude, valence, expected value and risk) of monetary rewards. Functional neuroimaging studies have consistently revealed that spatially distributed brain areas encode such mathematical moments that are the cornerstones of modern financial theory. However, the dynamics of the neurophysiological mechanisms whereby these computational parameters influence behavior remains poorly understood. To unravel the spatio-temporal electrophysiological (EEG) correlates of these computational statistics, we devised a novel experimental approach with sequentially presented information. Subjects decided whether or not to buy into a series of independent gambles while 128-channels EEG was acquired. Within a parallel factor (PARAFAC) analytical framework, constrained to take into account the physiological features of EEG signals and financial theoretic structure in the behavioral data, we identified for the first time and within a single experiment the EEG markers of mathematical moments and action values. Specifically, risk-related features - variance and skewness - are extracted within the first 350ms post-stimulus onset. EEG modulations predictive of choice and sensitive to expected value were observed 200ms prior to overt/registered choice. Distinct topographic maps indicate that the different processes are subserved by distinct networks of intracranial sources; the associated time courses occur at times in parallel albeit peaking at different latencies. Importantly, unlike option value computation and decision-making, information retrieval and choice implementation have time-consistent EEG signatures across subjects.