Key distance effects on perceived harmonic structure in music.

Two experiments provide convergent evidence for the hypothesis that listeners interpret chords in terms of their harmonic functions in a system of interrelated keys. The perceived associations between chords undergo significant changes depending on the broader tonal framework in which they are embedded. Three independent context effects are identified, and their magnitude is found to be a systematic function of the distance between the context key and the key (or keys) of which the chords are members. In the first experiment, listeners rate how musically related one chord is to a second chord; the chords are those that function within two maximally distant major keys (Cmajor and F# major). All possible chord pairs from this set are presented in each of three context keys: G major (whichis close to C major and distant from F# major), A major (which is moderately distant from both C major and F# major), and B major (whichis close to F# major and distant from C major).The second experiment measures recognition memory for the same chords embedded in tonal sequences in C, G, A, or B major keys, or random sequences. In the two experiments, the distance between the context key and the key of the chords on the circle-of-fifths affects: (1) the probability that a repeated chord is correctly recognized, (2) the strength of association between chords from the same key measured in terms of both confusion errors and direct relatedness judgments, and (3) asymmetries in confusion errors and relatedness judgments when the chords are in different keys. Perceived harmonic relations are thus found to be strongly context dependent, but the context effects are lawful functions of interkey distance. We conclude that listeners possess a highly articulated system of knowledge about the harmonic functions of chords in musical keys and an appreciation of musical structure at the level of abstract tonal centers.