Discussion

Experiment 1 investigated the influence of tonal relatedness on listeners’ judgments of tuning/mistuning of a target tone. This influence was studied with graduated pitch judgments, as previously used by Warrier and Zatorre (2002). Since the tonal function of the target tone was manipulated by changing only one (possibly repeated) tone at the beginning of the melodic context, the influence of context on listeners’ responses was linked to tonal, cognitive-based expectations, and not to features like pitch repetition, interval or contour.

Participants’ judgments showed an influence of tonal relatedness on pitch perception in terms of mean ratings, area scores, and hits/false alarms. The mean ratings and hits/false alarms analyses showed that the lower rank in tonal hierarchy of the subdominant targets increased listeners’ feeling that a mistuned target was out-of-tune and decreased the feeling that an in-tune target was in-tune (and the reverse for tonic targets). The lack of interaction between tonal relatedness and pitch deviation in mean ratings and hits/false alarms indicated the contribution of a response-bias shift: participants attributed more mistuning to less-related targets and were more tolerant of pitch deviations for related targets. However area scores, which provide an unbiased estimate of performance (Swets, 1973), indicated that the influence of tonal relatedness involved not only a shift in response bias, but also that discrimination performance between in- and out-of-tune targets was better for tonic targets than for subdominant targets.

The analyses thus revealed two influences in participants’ responses: increased discrimination of pitch tunings for tonics than for subdominants (as indicated by the area scores) and a response bias to attribute more mistuning to subdominants than tonics (as indicated by mean ratings and hits and false alarms). The differences in area scores suggest that a difference in tonal function affects pitch perception. The melodic context with its tonal structure (i.e., invoked key, implied tonal hierarchy) may give the tonic target the role of a cognitive reference point and facilitate the accuracy of mistuning judgments. This finding extends the observation by Warrier and Zatorre (2002): pitch perception is not only facilitated by a tonal context (in comparison to no context or a random tone context), but also influenced by the tonal function that a tone takes on inside this context (notably with a facilitation of the tonic).

Response bias might have been influenced by the instructions presenting concepts of “in-tune” and “out-of-tune” with the French words for “good” and “wrong” (“bonne,” “fausse”), which might have been ambiguous for the musically inexperienced participants. Listeners might have confounded a wrong note in pitch and a wrong note in terms of tonal relatedness, and this then biased them to judge tones as more out-of-tune when they were less related to the tonal context. In other words, the weaker tonal function of subdominants might have given the impression of a “tonal mistuning” that was confused with the pitch mistuning that listeners had to judge. A further indication of the ambiguity of instructions might be found in the mean ratings showing that in-tune tones were still heard as somewhat out of tune (0.65-0.7 in Figure 3). Since this is observed even for in-tune tonic targets, participants might have also have used other features than tonal relatedness in making their responses. Even if the to-be-judged pitch height is an objective property of a tone, this tendency might reflect the subjective noise in the perception and evaluation of an event. The task used in Experiment 3 aimed to focus on the influence of tonal relatedness on pitch discrimination by eliminating a possible source of response bias.

Finally, all three analyses revealed that the direction of pitch deviation influenced participants’ judgments: lowered pitches were perceived as more out-of-tune than pitches raised by the same amount. This asymmetry is in agreement with intonation used in musical performance: pitch perception is less accurate for raised pitch intonation than for lowered pitch intonation, leading to more pronounced pitch deviations in performance in the direction of raised pitch (Geringer & Madsen, 1984). Furthermore, a similar perceptual asymmetry has been observed in auditory non-musical contexts. In a study by Gottselig, Brandeis, Hofer-Tinguely, Borbély, and Achermann (2004), participants listened to tone sequences consisting of eight 50ms-long pure tones. The tones’ frequencies did not fall on a Western-tonal scale, and the intervals between tones did not correspond to Western-tonal intervals. The frequency of the sixth tone of the sequences was raised or lowered by 85 Hz, resulting in a deviant sequence. In a same/different discrimination task, participants listened to pairs of sequences where the first one was always the standard and the second could be a standard or a deviant. Discrimination performance was better for the low deviants than for the high deviants, and the MMN amplitude was larger for low deviants than for high deviants. This finding suggests that the asymmetry observed in our study and in intonation research might result from general (not specifically musical) auditory processes, as yet undocumented to our knowledge.