Multivariate discrimination in natural and urban scene viewing

INTRODUCTION The discrimination of visual scenes is a complex task that the human mind can achieve rapidly. Using biofeedback techniques we may be able to decode neural responses to visual scenes in a real-time manner. Multivariate pattern classification and prediction offers an alternative approach to standard univariate analysis techniques, and has recently been applied in MR imaging using support vector machines (SVM) [1], and to improve real-time imaging [2]. Previous work [3] has investigated scene identification using MVPA, but required large numbers of training examples. This study applied SVM techniques to predict the viewing of natural and urban scenes, and to calculate prediction accuracy based on the number of training examples as a first step towards real-time biofeedback.