'Decoding' of brain data (e.g. fMRI) has become very popular, because it offers much higher sensitivity in detecting discriminatory brain signals that encode particular stimuli, categories, motor-related signals and cognitive states. There are only few standard decoding software packages, and most existing studies are based on different and unpublished analysis pipelines. This makes it difficult to reproduce results, opens the door to programming errors, and leads to problems for new users to quickly participate in the use of decoding analyses for their data.

We developed a Matlab toolbox that we want to make available to a broad audience, ranging from users with minimal programming experience to users that want to quickly expand existing methods. The toolbox is optimized for use in SPM and AFNI but can easily be extended to other image analysis software packages. It allows searchlight decoding, region-of-interest decoding and whole brain analyses. Algorithms that are integral part of the toolbox include support vector classifiers, Haxby-style correlation classifiers, logistic regression, and linear discriminant analysis for classification, as well as support vector regression for continuous variables. You can easily add new algorithms. The toolbox includes data scaling, feature selection, parameter selection and feature transformation (e.g. PCA).

For more information, please visit the toolbox homepage.