Behavioral and neural dissociations of rhythmic temporal expectations from memory-based expectations

Background Sensory input with predictable dynamics can be used to create temporal expectations. It has been suggested that when the input is rhythmic, temporal expectations are created to in-phase time points due to oscillatory synchronization with the rhythm period. However, they can also be created by intentionally memorizing a fixed interval and applying it recursively. Using performance measures and EEG in 20 subjects, we dissociated these two processes. In one condition, targets appeared in-phase with a rhythmic sequence of stimuli in 80% of the trials. In another condition, they appeared in 80% of the trials at an interval that was recursively presented to allow memorization, but without creating a rhythmic sequence. In both conditions, a cue indicated explicitly that the next stimulus is the target. Results The behavioral validity effect of rhythmic stimulation was stronger than that of interval memorization, suggesting that the former is more temporally accurate. The CNV, an ERP which reflects temporal anticipation to the target, was more negative in the rhythmic condition, suggesting increased expectation. Furthermore, the effect of cue invalidity on the terminal CNV and on the latency of the P3, an ERP reflecting target evaluation, was more evident in the rhythmic condition, suggesting faster evaluation. Finally, in expected target times there was increased alpha desynchronization, previously shown to occur when directing attention. Conclusion Although the amount of temporal information was matched, the expectation created by rhythmic stimulation was superior over interval-based temporal expectation, both in behavior and in expectation-related neural activity. These findings are inconsistent with the idea that rhythmic expectation is no more than recursive applying a memorized interval, and instead suggest involvement of either an additional mechanism or a different one