The contribution of pre-stimulus neural oscillatory activity to spontaneous response time variability

- We searched for oscillatory predictors of saccadic reaction times in humans.
- Pre-stimulus power at 5-45 Hz throughout dorsal pathway correlates with response time.
- These correlations mainly reflect spontaneous variance rather than deterministic factors.
- These correlations are dominated by different time scales across the brain.
- We found no evidence for a relationship between latency and phase of oscillations.

Large variability between individual response times, even in identical conditions, is a ubiquitous property of animal behavior. However, the origins of this stochasticity and its relation to action decisions remain unclear. Here we focus on the state of the perception-action network in the pre-stimulus period and its influence on subsequent saccadic response time and choice in humans. We employ magnetoencephalography (MEG) and a correlational source reconstruction approach to identify the brain areas where pre-stimulus oscillatory activity predicted saccadic response time to visual targets. We find a relationship between future response time and pre-stimulus power, but not phase, in occipital (including V1), parietal, posterior cingulate and superior frontal cortices, consistently across alpha, beta and low gamma frequencies, each accounting for between 1 and 4% of the RT variance. Importantly, these correlations were not explained by deterministic sources of variance, such as experimental factors and trial history. Our results further suggest that occipital areas mainly reflect short-term (trial to trial) stochastic fluctuations, while the frontal contribution largely reflects longer-term effects such as fatigue or practice. Parietal areas reflect fluctuations at both time scales. We found no evidence of lateralization: these effects were indistinguishable in both hemispheres and for both saccade directions, and non-predictive of choice - a finding with fundamental consequences for models of action decision, where independent, not coupled, noise is normally assumed.