We previously demonstrated a consistent relationship between visual spatial-frequency and auditory amplitude-modulation (AM) frequency, in which Gabors of 0.5a??8 cycles/degree (c/d) were linearly matched to auditory AM frequencies of 1a??12 Hz (Guzman et al., VSS 2009). Here, we investigated whether similar crossmodal associations occur for natural scenes, which are dominated by various spatial-frequency components. We asked whether people consistently match specific auditory AM frequencies to photographed scenes, and if so, how these crossmodal matches are associated with the dominant spatial-frequency component and subjective impression (dense, stimulating) of the scene. We found that eighteen observers matched specific auditory AM frequencies to 26 scenes from diverse categories (nature, urban, indoor) with surprising consistency. We applied a 2D Fourier transform to each scene to determine the contrast energy for 12 spatial-frequency bins ranging 0.05a??12.8 c/d. Interestingly, scenes with higher contrast energy in the mid-spatial-frequency range 0.5a??1.25 c/d were matched to faster AM frequencies, whereas other spatial-frequency components did not contribute to AM frequency matches. Analysis of our images suggests that scenes with stronger mid-spatial-frequency components appear to have numerous object boundaries. Thus, the results suggest a crossmodal association between the visual coding of multiple object boundaries and the auditory coding of AM frequency. Furthermore, a multiple regression of AM frequency matches to subjective scene ratings (obtained after the experiment) indicates that dense (vs. sparse) and stimulating (vs. calm) ratings independently contribute to faster AM frequency matches. Based on the spatial-frequency analysis and subjective ratings, our results demonstrate an association between visual object density and faster auditory AM frequencies in scene perception, and that visual features conveying stimulating content additionally contribute to faster AM frequency matches.
展开▼
