Informational processing of complex sound

Normal‐hearing listeners discriminated among complex, nonspeech sounds in which the difference to be discriminated was randomly varied from trial to trial. All other physical parameters of the sounds were fixed within a block of trials. On each trial, the listener heard two sounds (e.g., tone complexes). The values of the variable parameter (e.g., tone frequencies) were drawn from two normal distributions differing only in mean. The listener's task was to identify the sound having the higher mean value of the variable parameter. Discrimination performance was found to be largely independent of the particular physical dimensions along which the sounds varied. Rather, performance appeared to depend primarily on information content of the sounds. Information content was defined in terms of a stimulus equivocation factor that was derived from the data. Based on this model, transmitted information was estimated to be between 1.0–3.0 bits. [Work supported by AFOSR.]