Identifying neuronal oscillations using rhythmicity.

Neuronal oscillations are a characteristic feature of neuronal activity and are typically investigated through measures of power and coherence. However, neither of these measures directly reflects the distinctive feature of oscillations: their rhythmicity. Rhythmicity is the extent to which future phases can be predicted from the present one. Here, we present lagged coherence, a frequency-indexed measure that quantifies the rhythmicity of neuronal activity. We use this method to identify the sensorimotor alpha and beta rhythms in ongoing magnetoencephalographic (MEG) data, and to study their attentional modulation. Using lagged coherence, the sensorimotor rhythms become visible in ongoing activity as local rhythmicity peaks that are separated from the strong posterior activity in the same frequency bands. In contrast, using conventional power analyses, the sensorimotor rhythms cannot be identified in ongoing data, nor can they be separated from the posterior activity. We go on to show that the attentional modulation of these rhythms is also evident in lagged coherence and moreover, that in contrast to power, it can be visualised even without an experimental contrast. These findings suggest that the rhythmicity of neuronal activity is better suited to identify neuronal oscillations than the power in the same frequency band.