Moreover, functional activation and gray-matter volume in post- and precentral regions before training predicted individual learning abilities as indexed after training. We showed that learning of time is associated with a series of functional and structural changes within several nodes of a sensory-motor circuit. A general issue with activations associated with time processing is whether these reflect modifications of the representation of time

per se or, rather, they reflect some changes at higher stages of the discrimination process, like attentional or decision-making levels. In our fMRI experiment we compared trials that were different in terms of the duration encoded (i.e., trained versus untrained) but were otherwise identical with respect to other cognitive aspects BMN 673 purchase (i.e., check details attention, working memory, and decision components). Therefore, the activations

observed here are ought to genuinely reflect a change in the representation of the trained duration. An alternative possibility is that learning has changed the ability to temporarily store a 200 ms template rather than changing the representation of the duration itself. However, the finding that training-related changes were duration specific and were associated with the activation of visual cortices, where the encoding of time information in the millisecond range has been previously hypothesized (Bueti et al., 2010; Heron et al., 2012; Shuler and Bear, 2006), suggests that memory processes are unlikely to fully explain our results. Nonetheless, our findings cannot exclude that training may affect both the representation of time, as well as the capacity to store specific durations (i.e., here, the trained 200 ms interval). For instance, visual cortices may play a direct role in the representation of time, providing a “low level” sensory-specific

substrate for time representation; while the insula, activating here irrespective of sensory modality, may be involved in “higher level” storage-related operations of temporal information. The behavioral results showed that learning in the visual modality next generalized to the auditory modality in 11 out of the 13 “visual learners.” The generalization of learning across sensory modalities has been often interpreted as suggesting the existence of a central “amodal” timing mechanism, as opposed to the proposal of distributed modality-specific clocks (Rousseau et al., 1983). This view implies that the same mechanisms of time processing mediate both “intermodal generalization” and temporal learning. Here we found that not all subjects generalized learning from vision to audition and that there was no significant subject-by-subject correlation between learning in the two modalities.