Comparing content and timing predictions in music and speech: How domain and musical training shape our predictions

Prediction is a fundamental aspect of cognition that allows us to anticipate what is about tooccur, and when. However, little is known about how these predictions are related, andwhether they differ depending on stimulus domain (i.e., music, speech). To investigate thesequestions, content and timing predictability were comparably manipulated within melodiesand sentences to investigate effects on completion judgements for good and bad note/wordendings as a measure of prediction sensitivity. Musicians (n=26) and non-musicians (n=26)were recruited to investigate effects of music training on prediction in music, and possiblerelationships between prediction sensitivity in music and speech. Different result patternswere found for content and timing manipulations depending on domain and musical training.For melodies, timing predictability influenced sensitivity independently of contentpredictability and musical training. For sentences, timing predictability only influencedmusicians when speech content was predictable, suggesting an interaction between contentand timing related to musical training. Content predictability influenced sensitivity more forsentences than melodies, whereas timing predictability influenced melodies more thansentences. Musicians were more sensitive than non-musicians for melodies, and for sentenceswith unpredictable content. These differences were not accounted for by working memorycapacity. Overall, these results suggest that content and timing affect music and speechsensitivity in different ways, perhaps depending on the relative importance of these cues foreach domain, and also the musical expertise of the listener. These findings have implicationsfor understanding different types of prediction in the brain, and effects of expertise onpredictive processing.