Portfolio Decisions and Brain Reactions

Abstract

Each decision is a multiplex mechanism based on integration and contrasting of elements of choice options. Understanding the human brain actions during decisions under risk and neural processes underlying (uncertain) investment decisions lies at the heart of neuroeconomics. In this paper, functional magnetic resonance imaging (fMRI) data from an investment decision study is analyzed. We propose a new method for identifying the activated brain regions. Our analysis is focused on the brain clusters rather than voxels units. Thus, we achieve a higher signal to noise ratio within the unit tested and a smaller number of hypotheses tests compared with the GLM method. The brain parcellation is done by spatially constrained Ncut spectral clustering. The information within each cluster is extracted by model-free DSFM dimension reduction technique and tested for activation. Our method allows for the model-free analysis of the local, BOLD signal. In particular, the risk attitudes of all subjects are predicted based on the estimated dynamics of anterior insula (aINS). The risk preferences obtained from subjects' investment decisions are successfully prognosed by the extracted, low-dimensional BOLD representation.