Contextual pitch perception

Psychoacoustic research on pitch perception has investigated the influence of listeners’ expectations in nonmusical contexts. These expectations were mostly based on features of the experimental session such as the probability of occurrence of a specific pitch (Tanner & Norman, 1954) or the use of cues indicating the pitch of the to-be-detected signal (Greenberg & Larkin, 1968; Hafter, Schlauch & Tang, 1993; Howard, O’Toole, Parasuraman & Bennet, 1984; Watson & Foyle, 1985; Watson, Kelly & Wroton, 1976). In Greenberg and Larkin (1968) listeners’ expectations were manipulated by presenting a cue tone whose frequency corresponded to the frequency of the to-be-detected signal in 70-80% of the trials. Results of this probe-signal method revealed best detection performance for the signal at the cued (expected) frequency, and performance decreased for unexpected frequencies as a function of the distance with the cue tone. Facilitated pitch processing has been observed also with indirect cues that consisted of either one tone (Hafter, Schlauch & Tang, 1993) or a tone sequence (Howard, O’Toole, Parasuraman & Bennet, 1984). In Hafter et al. (1993) pitch processing was facilitated by cues that were a musical fifth (a 3/2 frequency ratio) away from the targets, showing that the cue need not rely on direct sensory priming of the target pitch. For longer tone contexts, pitch processing is facilitated when participants can develop expectations for target pitches based on pattern structure in the context (Howard et al., 1984). For example, in continuously rising (or falling) 12-tone sequences, the eleventh tone (the to-be-detected target) was either expected (or not) in terms of the homogeneous intervallic succession in the sequence and detection performance was better for expected targets (i.e., tones in the continuity of the pattern) than for unexpected targets violating the pattern (Howard et al., 1984).

Pitch processing has also been investigated in musical contexts. Comparing native and nonnative musical contexts, Lynch and Eilers (1992) reported that Western nonmusicians detected mistuned notes more accurately with melodies based on a Western tonal scale than on an unfamiliar scale (in this case, a Javanese pelog scale). Facilitated pitch processing by Western listeners in Western tonal scale contexts (vs. non-Western ones) has also been reflected in electrophysiological data showing larger amplitudes of Mismatch-Negativity (MMN- Brattico, Näätänen, & Tervaniemi, 2002). Focusing on Western tonal contexts, Francès (1958, Experiment 2) asked musician listeners to detect mistuned tones that were anchored to lower or higher tones. Results showed that mistunings were better detected when the direction of mistuning conflicted with the expected direction of tone motion than when it was coherent with it.

When pitch perception is rendered more difficult due to timbral changes of the involved tones, a tonal context can increase pitch perception accuracy. In Warrier and Zatorre (2002, Experiment 1) pairs of tones that differed in pitch (by 17, 35, or 52 cents) and/or timbre (more or less bright) were presented in isolation or preceded by a melodic context in which targets functioned as the tonic. Listeners rated the degree of difference between the tones (for the pairs) or the degree of mistuning (for the melodies). Melodic contexts improved pitch extraction in comparison to single tone-contexts (see also Brattico et al., 2002). Comparisons with a non-melodic context (with the same tone repeated five times) and with random, atonal contexts revealed that the facilitation in tonal contexts was not only due to the presence of other tones or to higher structural complexity, but to the tonal structure of the context (Warrier, Belin, Merlet & Zatorre, 1999; Warrier & Zatorre, 2002). These findings suggest that the tonal context creates tonal reference points allowing for more efficient pitch extraction in spite of the timbral changes.

Instead of comparing melodic contexts tono context or to a random context, our study investigated the influence of degrees of tonal relatedness on pitch processing by using only tonal melodies that were similar in structure complexity. More specifically, our aim was to investigate the influence of listeners’ tonal expectations on pitch perception for target tones differing in tonal function. The target (i.e., the last tone of the melodies) functioned either as the tonic, which is the cognitive reference point of a tonality (Krumhansl, 1990), or as the subdominant, which is less-related but still important in the tonal hierarchy. The melodic contexts were constructed in pairs to control for sensory differences (such as tone repetition and melodic contour) and to focus on the cognitive components of tonal expectations and their influence on pitch perception. In the related condition with the expected tonic target, pitch perception should be facilitated in comparison to the less-related condition with the less stable, less-expected subdominant target. Experiment 1 used ratings of degree of mistuning (proposed by Warrier and Zatorre, 2002). Experiment 2 used a speeded reaction time task requiring binary judgments of mistuning. Experiment 3 asked participants to make same/different-judgments on tone pairs that defined the end of the melodies (with one tone being mistuned, or not). In contrast to psychoacoustic studies, expectations for the target pitch depended on the key invoked by the melodic context and required listeners’ tonal knowledge.