This data set was analysed for the publication "Acoustic and higher-level representations of naturalistic auditory scenes in human auditory and frontal cortex" by Lars Hausfeld, Lars Riecke and Elia Formisano (10.1016/j.neuroimage.2018.02.065). Anatomical and functional MRI was acquired at 7 Tesla and over the course of 3 sessions for each participant. Participants listened to natural auditory scenes consisting of three sound sources: voice, instrument and pure tone. To reveal effects of selective attention, participants were asked to listen to one of the three sources by performing a task. The dataset is arranged as follows: - MRI data for each participant are contained in the participant-specific folders S[participant id]_MRIdata.zip - Information and design, stimulation protocols and sounds are contained in DesignProtocolsSounds.zip - README files contain necessary information on the MRI acquisition, experimental design and the coding of conditions Please note additional information in the original publication Abstract of corresponding manuscript In everyday life, we are often confronted with auditory scenes comprising multiple simultaneous sounds. When listening to two simultaneous talkers, the neural representation of the attended talker is selectively enhanced in auditory cortex. However, it remains unknown whether and how this selective attention mechanism operates on representations of different natural sound categories. In this high-field fMRI study we presented participants with simultaneous voices and musical instruments while manipulating their focus of attention. We found an attentional enhancement of neural sound representations in temporal cortex at locations that depended on the attended category (i.e., voices or instruments). In contrast, we found that in frontal cortex the site of enhancement was independent of the attended category and the same regions could flexibly represent any attended sound regardless of its category. These results are relevant to elucidate the interacting mechanisms of bottom-up and top-down processing during real-life audition. Author contact: lars.hausfeld@maastrichtuniversity.nl

We thank Peter de Weerd, Niels Disbergen and Martha Shiell for discussions on the experimental setup, Federico De Martino for support with data acquisition, Milene Bonte, Inge Timmers and Tahnée Engelen for help with sound creation, Giancarlo Valente and João Correia for support on statistical testing. This work was supported by Maastricht University and the Netherlands Organization for Scientific Research (NWO; VICI grant 435-12-002 to E.F.).