This dataset consists of fMRI data from 24 participants that were exposed to an experiment consisting of auditory cues and abstract visual shapes. The experiment consisted of 6 functional runs (~13 minutes per run), described below.

In two out of six runs (run 1 and run 6), one of five shapes was presented on each trial, in a pseudorandom manner. In the other four runs, auditory cues (falling or rising tones, respectively) predicted which shape (either shape 2 or shape 4) would be presented on that trial, with 75% validity. In two of the four prediction runs, an unexpected shape was presented on the 25% invalid trials (e.g. shape 2 was predicted, but shape 4 presented). In the other two runs, no shape at all was presented on the 25% invalid trials (omission trials).

All imaging data were acquired using a 3T Siemens Prisma MR scanner, with a 64-channel head coil. Functional images were obtained with a multiband EPI sequence: TR = 1000 ms; TE = 32.6 ms; 60 transversal slices; voxel size = 1.5 x 1.5 x 1.5mm; 55 degree flip angle; multiband factor = 6. Additionally, anatomical T1 (TR = 2300 ms;
TE = 2.27 ms; voxel size= 1 x 1 x 1 mm; 192 transversal slices; 8 degree flip angle; GRAPPA acceleration factor 3) and T2 (TR = 11,390 ms; TE = 90 ms; voxel size = 0.44 x 0.44 x 1.5mm; 54 coronal slices, perpendicular to the long axis of the hippocampus; distance factor = 20%; 150 degree flip angle) weighted images were acquired. To correct for susceptibility-induced distortions in the echoplanar images, a pair of spinecho volumes was acquired in opposing phase-encode directions (anterior/posterior and posterior/anterior) with matching slice prescription, voxel size, field of view, bandwidth, and echo spacing (TR = 8000 ms; TE = 66 ms).