Data sonification requires inference-preserving mappings: conclusions drawable from the sound must correspond to conclusions about the source data

Ballantine (ch21) defines good data sonification as requiring ‘inference-preserving’ cross-domain mapping: when a listener draws a conclusion from hearing the sonification, that conclusion should correspond to something true about the original data. This is what distinguishes useful scientific sonification from arbitrary sound design. The chapter distinguishes order-zero mappings (playing data directly as audio), first-order mappings (using data to modulate synthesis parameters), and higher-order mappings (using data to control synthesis structure). Mapping to a process or model rather than directly to parameters allows richer inference but requires more careful design.

Examples

Seismic data played at 10x speed: listeners can hear the acoustic properties of rock through which the earthquake traveled — an inference-preserving mapping because acoustic and seismic physics are analogous.

Assessment

Explain what ‘inference-preserving mapping’ means in the context of data sonification. Give one example of a good inference-preserving mapping and one example of a sonification that would NOT be inference-preserving, explaining why.