Double training reveals an interval-invariant subsecond temporal structure in the brain

Subsecond temporal perception is critical for understanding time-varying events. Many studies suggest that subsecond timing is an intrinsic property of neural dynamics, distributed across sensory modalities and brain areas. Furthermore, we hypothesize the existence of a more abstract and conceptual representation of subsecond time, which may guide the temporal processing of distributed mechanisms. However, one major challenge to this hypothesis is that learning in temporal interval discrimination (TID) consistently fails to transfer from trained intervals to untrained intervals. To address this issue, here we examined whether this interval specificity can be removed with double training, a procedure we originally created to eliminate various specificities in visual perceptual learning. Specifically, participants practiced the primary TID task, the learning of which per se was specific to the trained interval (e.g., 100 ms). In addition, they also received exposure to a new interval (e.g., 200 ms) through a secondary and functionally independent tone-frequency discrimination (FD) task. This double training successfully enabled complete transfer of TID learning to the new interval, indicating that training improved an interval-invariant component of temporal interval perception, which supports our general proposal of an abstract and conceptual representation of subsecond time in the brain.