Tonal expectations

In musical pieces the tonal hierarchy is reflected in the statistical regularities of tone use: tones belonging to a key occur more frequently together than out-of-key, and those with central tonal functions (e.g., tonic, dominant) are used more often than the less central ones (Francès, 1958; Krumhansl, 1990). The tonal hierarchy is thus correlated with the tones’ probabilities of occurrence, and it has been hypothesized that listeners acquire implicit knowledge of tonal structure by internalizing the frequency of co-occurrence and occurrence of tonal events. Consequently sensory information in a musical context (i.e., which tones occur more frequently in the context) is entwined with tonal structure. The investigation of listeners’ implicit knowledge and of musical expectations linked to tonal hierarchy¹¹ thus gives rise to a methodological difficulty and requires the control of the sensory information in the material. In addition to sensory information linked to frequencies of occurrence (i.e., tone repetition), other features (namely intervals and melodic contour), which do not involve the tonal hierarchy, influence listeners’ expectations. These features elicit expectations according to Gestalt principles: a melodic interval generates strong expectations for successive tones to lie close in pitch height or to change contour direction (i.e., principles of pitch proximity and pitch reversal; Narmour, 1990; Schellenberg, Adachi, Purdy, & McKinnon, 2002). These Gestalt-based expectations are likely to influence tonal expectations, and have thus to be controlled in experimental material used to investigate cognitive expectations.

In focusing on cognitive expectations (i.e., top-down expectations linked to implicit knowledge of the tonal hierarchy), previous research has addressed the influence of sensory components with post-hoc analyses such as regression analyses (Bigand, 1997; Bigand & Pineau, 1997; Frankland & Cohen, 1990; Hébert, Peretz, & Gagnon, 1995), or has controlled for sensory features directly in the experimental material (e.g., Bigand, 1997; Krumhansl, 1979). The findings reveal that - beyond sensory factors such as tone repetition and pitch proximity - listeners’ knowledge of the tonal hierarchy influences melody perception. For example, when two melodies differ only in their initial few tones, musical tension judgments reveal that the remaining tones (which are the same in the two melodies) are perceived differently and according to their tonal functions in the key invoked at the beginning (Bigand, 1997).

Cognitive expectancies and their influence on processing speed have been investigated with the musical priming paradigm. This paradigm manipulates the tonal relatedness between a prime context (i.e., a chord, a chord sequence) and a to-be-processed target chord and compares tonally related targets to less-related targets. Related targets should be more expected in the priming context and thus should be processed faster than less-related targets. To focus on cognitive priming, the experimental material controls for acoustical overlap between prime and target. After single-chord primes, related targets are processed faster than unrelated targets, even when primes and targets do not share component tones (Bharucha & Stoeckig, 1987) or are separated by a white noise burst (Tekman & Bharucha, 1992). After 7-chord sequences, processing of related targets is facilitated even when less-related targets share more tones with the priming context (Bigand, Poulin, Tillmann, Madurell, & D'Adamo, 2003) or when they are immediately preceded by an identical chord (Bigand, Tillmann, Poulin-Charronnat, & Manderlier, 2005).

In sum, tonal expectations influence musical event processing, notably with faster response times for tonally related, more expected musical events than for less related, less expected ones. Our study aimed to investigate the levels of music processing influenced by tonal expectations and focused on the accuracy of pitch perception.