Figure 5.
Dynamics in the auditory representation of scene size. (a) Stimulus set comprised nine auditory scenes, generated by convolving each of three brief impact sounds with each of three room impulse responses. (b) MEG classification analysis revealed dissociable trajectories of source identity (blue) and space size (orange) decoding, peaking approximately 250 ms apart. Cross-classification analysis revealed that source identity and space size decoding were robust across different spaces and sources, respectively. Chance level of classification was 50%. Bold traces: main classification decoding curves. Thin traces: cross-classification decoding curves. Horizontal grey line indicates chance-level decoding. Vertical dotted line indicates stimulus onset. (c) Temporal generalization analysis of space size decoding reveals dynamics of evolving representations, tending towards greater stability (width relative to diagonal) with increasing time. (d) Model RDMs hypothesizing a categorical (i) versus ordinally progressive (ii) representation of space size. (e) Correlation with MEG data reveals stronger relationship with the progressive model. Figure adapted from [68].