Discussion

The auditory short-term memory model (Leman, 2000) simulated successfully the tonal priming effect when the melodies were played with piano tones (as in Experiment 1). For piano tones, the outcome of the simulations challenges the interpretation of the priming effect of Experiment 1 in terms of cognitive processes: The model simulates the behavioral data pattern on the basis of periodicity pitch information, without postulating cognitive processes. When running simulations on the same melodies played by pure tones, no difference between related and less-related conditions was predicted by the model. This failure of the model results from the loss of acoustic information provided by the melodies played with pure tones. In contrast to piano tones, pure tones do not carry any overtones and thus provide reduced periodicity pitch information, which is insufficient for the model to differentiate related and unrelated melodies.

For pure tones, the simulations lead to two alternative hypotheses that oppose sensory and cognitive interpretations and were tested in Experiment 2. If the cognitive interpretation is correct, listeners should also show a priming effect for melodies played with pure tones. Since a pure tone of frequency F0 is sufficient to elicit the perception of the musical tone whose fundamental frequency is F0, the lack of overtones should not prevent a tonal framework to be abstracted from the melodies and the implicit tonal knowledge to drive tonal expectations. This should then result in a tonal priming effect. If, however, the priming effect observed in Experiment 1 resulted from sensory-driven expectations, no priming effect should be observed for melodies played with pure tones in Experiment 2.