Statistical parametric mapping: a catalyst for cognitive neuroscience

Statistical Parametric Mapping (SPM) is a statistical framework and open source software package for neuroimaging data analysis. Originally created by Karl Friston in the early 1990s, it has been used by a vast number of scientific studies over the last three decades. SPM has not only revolutionized the analysis of neuroimaging data but also catalyzed the development of cognitive neuroscience. This short commentary reflects on key principles that have made SPM so enormously influential and successful: (i) the introduction of a principled general framework for statistical inference that applied to all neuroimaging modalities, (ii) the emphasis on open source code, transparency, and collaboration, and (iii) constant evolution over three decades, from a frequentist mass-univariate framework to generative models of neuroimaging, electrophysiological, magnetoencephalographic, and behavioral data.