The Interplay of Interval Models and Entrainment Models in Duration Perception

Despite extensive research demonstrating the effect of temporal context on time perception, its underlying mechanisms remain poorly understood. One influential proposal to explain the temporal context effect is McAuley and Jones' (2003) framework that incorporates 2 classic timing models, interval and entrainment models. They demonstrated that listeners' duration estimates were shifted from reality in opposite directions when to-be-judged durations occurred earlier versus later than an expected beat, which is predicted by their entrainment models. However, it is unclear about how long the entrainment lasts after the cessation of external stimulation. Here, we investigated the persistence of the entrainment effect in 2 experiments. In Experiment 1, we found that entrainment models predict the behaviors better after short delays (2 beats), while interval models predict better after long delays (4 beats). In Experiment 2, we extended the finding to a faster tempo and added 1 more delay length. Again, we found that entrainment was strongest after short delays (2 beats), while disappeared after medium (4 beats) and long delays (8 beats). Our findings suggest an interplay between entrainment and interval timings as a function of delays between successive events. Public Significance Statement This was the first study to test how long listeners expect sounds to occur "on the beat" after the end of a sequence of beats. Using a duration-judgment test, we found that beat sensitivity is present for about 2 cycles, but is mostly gone by 4 cycles after the end of a sequence of beats. We found that duration judgments are still accurate even when the beat is gone, which suggests two different processes may help listeners with making time judgments-one based on beats, the other based on long-term duration knowledge.