A Multi-Subject Deconvolution Algorithm for the Analysis of Naturalistic fMRI data

Collecting fMRI data during naturalistic paradigms has drawn considerable attention in human neuroscience as a way to investigate brain function in ecologically valid conditions. We introduce a novel method (multi-subject paradigm free mapping) to decipher BOLD events in a temporally agnostic manner, and explore concordant group activations and individual idiosyncrasies. Besides, it can operate at the fastest temporal and spatial resolutions of the data. We validate it on simulated and real naturalistic fMRI data, revealing events that track expected features of the stimulus. Overall, this technique substantially increases sensitivity in linking moment-to-moment brain activity to its underlying cause(s).