A DATA SONIFICATION APPROACH TO COGNITIVE STATE IDENTIFICATION

Abstract

The study of human brain functions has dramatically increased greatly due to the advent of Functional Magnetic Resonance Imaging (fMRI), arguably the best technique for observing human brain activity that is currently available. However, fMRI techniques produce extremely high dimensional, sparse and noisy data which is difficult to visualize, monitor and analyze. In this paper, we propose two different sonification approaches to monitor fMRI data. The goal of the resulting fMRI data sonification system is to allow the auditory identification of cognitive states produced by different stimuli. The system consists of a feature selection component and a sonification engine. We explore different feature selection methods and sonification strategies. As a case study, we apply our system to the identification of cognitive states produced by volume accented and duration accented rhythmic stimuli.