Prolonged complex naturalistic stimulation is arguably more likely to elicit brain responses that are representative of naturally occurring brain states and dynamics than artificial, highly controlled experiments with a limited number of simplified conditions. One particularly interesting aspect of natural stimulation using movies is the temporal synchronicity of changes in the response pattern across individual brains, presumably caused by synchronous temporal dynamics of underlying neuronal processes.
Algorithms have been developed that utilize inter-individual synchronicity of changes of the brain state over time to align fMRI data from individual brains into a group space based on functional connectivity patterns11 and BOLD time-series correlation12. Haxby and colleagues were able to demonstrate that temporally synchronous patterns can be used to transform brain response patterns of individual brains into a high-dimensional representational space with common dimensions for all brains13. This technique enables group analyses of distributed activation patterns at the same level of detail and accuracy as the analysis of idiosyncratic patterns of an individual brain. Uniformly, these studies find that deriving inter-individual alignment from fMRI data recorded while participants watch movies yields transformations that are of greater general validity when tested on data from controlled experiments. This is further evidence that movies elicit brain response patterns and dynamics that are representative for naturally occurring neuronal processes.
We hope that our reproducible stimulus and acquisition procedure will be used by independent researchers to extend the scope of this dataset by adding data from participants with different cultural backgrounds, a different spoken language or an audio-visual stimulus. We have already begun to acquire more data and aim to add additional modalities in the future. We invite interested researchers to coordinate with us on creating a high-dimensional brain response based annotation of this movie stimulus.
However, as it becomes increasingly difficult to identify functionally corresponding voxels across brains at high spatial resolution, we did not only assess pattern similarity using voxel-wise time-series correlations of anatomically aligned data. Instead, we employed representational similarity analysis (RSA)33 to identify 2nd-order isomorphisms in the response patterns across brains. The premise of RSA is that any two segments of the audio movie which elicit similar neural signal patterns in one brain also elicit similar patterns in another brain; whereas two segments, which lead to distinct patterns in one brain, also evoke dissimilar patterns in another brain.
How to cite this article: Hanke, M. et al. A high-resolution 7-Tesla fMRI dataset from complex natural stimulation with an audio movie. Sci. Data 1:140003 doi: 10.1038/sdata.2014.3 (2014). 2b1af7f3a8