Schematic of predictions tested in the present adaptation study - from neuronal to voxel-level responses. The top row shows examples of the experimental conditions (red: visual stimuli, blue: auditory stimuli). Rows 2-4 show schematic illustrations of three possible voxel types in multisensory STS: containing only unisensory subpopulations (type 1), a mixture of uni- and multisensory populations (type 2) or only multisensory neurons (type 3). All circles represent one neuron (visual, auditory or audiovisual), the thickness of the lines indicate the neuron's activity (see index bottom right). Activation/adaptation strength is predicted for all neurons in the three voxel types for the different experimental conditions, and the resulting predicted fMRI response of the voxel in it's entity is shown in the column on the right (X-axis: 0 = No-Ad; 1 = Ad-C 1; 2 = Ad-C 2; 3 = Ad-I). Importantly, the relative activity for the Ad-C vs. Ad-I (marked by the black arrow) will distinguish between voxels containing multisensory neurons (type 2 or 3 → differential adaptation) or voxels consisting only of unisensory neurons (type 1 → identical adaptation). Hypothetical representations of three different letters ("a", "o" and "d") are indicated by encircled regions. To ensure equal averaged adaptation strength for different letter exemplars (Ad-C 1 vs Ad-C 2), each letter and each speech sound exemplar was equally often presented in congruent and incongruent audiovisual combinations (see also figure 2).