Code associated with the paper Social-affective features drive human representations of observed actions. The repository contains:

This repository contains a set of functions (that have expanded over a few years) to run SVM decoding of sensor-space and source-space MEG data using the LibLinear/LibSVM libraries and the Fieldtrip toolbox. 

If you mainly use Fieldtrip, they give you a quick way to add some multivariate analyses, including time-resolved & searchlight decoding with iterated cross-validation and various preprocessing options, randomization-based statistics with correction for multiple comparisons, temporal generalization, accuracy/classifier weight plots, and representational similarity analysis.

A demo with some example data is available here