69
Introduction
Traditional neuroimaging analyses use information about the sensory stimulus or
behavioral state of the subject to calculate a measure of activation in a single brain
voxel at a time. Recently, techniques have been developed to measure distributed
patterns of activity across the brain, referred to as multi-voxel pattern analysis (MVPA)
(Norman, et al. 2006). With MVPA, the traditional analysis is reversed and
measurements of brain activity are used to decode the sensory stimulus presented to
the subject or the mental or behavioral state of the subject (Cox and Savoy 2003;
Haynes and Rees 2006; Kamitani and Tong 2005; Kriegeskorte, et al. 2006; LaConte, et
al. 2005).
Most distributed pattern analysis studies have focused on decoding visually
presented stimuli. Visual cortex is anatomically the largest of the early sensory cortices,
and even simple visual stimuli evoke activity in many visual areas (Grill-Spector and
Malach 2004). This distributed representation makes visual cortex an ideal laboratory
for MVPA, because it provides many active voxels across which to pool information.
However, it raises the question of whether Other sensory modalities whose cortical
representations are smaller or less distributed than visual cortex are amenable to
MVPA. We performed two experiments to investigate whether MVPA could be used to
decode individual stimuli presented in a different sensory modality, namely the
Somatosensorysystem.
In both experiments, a simple Vibrotactile somatosensory stimulus (touch) was
delivered to different locations on the body surface. In the first experiment, widely