Abstract
Two sustained notes, A4 and E5, and a repetition of C#5, which was between the two sustained notes and given by eight notes interrupted by eight rests, made up ambiguous music; at each eight rest, one either heard a continuous melody of repetition of E5-C#5 or of A4-C#5 although the sustained notes were never repeated. Sometimes one heard only the intermittently presented C#5 just as the actual stimulus. MEG measurements during presentation of the ambiguous music showed different (although not always significantly different) responses to the last note of the stimulus depending on which melody the subject was induced to perceive. Behavioral experiments yielded the distribution time among the three inner states of the subject corresponding to the three phrases the subject heard. The obtained transition time distribution was used to estimate the state transition probability based on a model.
