Brain signal complexity during sleep
By Anis Zahedifard & Kelly Shen
This week marks our return in-person to the Society for Neuroscience meeting. Over the next few days, we’re giving you a sneak peek of what we’ll be presenting in San Diego.
In today’s sneak peek, we’ll be covering our poster titled “Signal complexity of local field potentials during sleep reflects both sleep stage and learning” which Randy will present on the morning of Tuesday November 15th (Poster #YY48).
Brain signal complexity is considered to be a measure of information processing capacity. Multiscale entropy (MSE), in particular, varies with cognitive performance and reflects accurate and stable behaviour. In this study, we were interested whether MSE could capture the effects of neuroplasticity that underlie learning and memory consolidation during sleep. Studies have shown a decrease in signal complexity with decreased wakefulness, but whether MSE varies with sleep stages or reflects the effects of training during sleep is unknown.
The Study
We answered these questions using data from a study by Lemke and colleagues. In their study, rats have undergone a motor skill learning task. We used the local field potentials (LFPs) from four male rats recorded from the forelimb area of M1. LFPs were recorded during sleep sessions that occurred before (Pre-task) and after (Post-task) training and later staged into rapid eye movement (REM) and non-REM periods. Rats were trained daily until they reached a performance criterion, ranging from 6 to 17 days. For analysis purposes, we used the first and last day of training for each rat regardless of how many days they underwent training, representing recordings when the rats were naïve to the task and when they had fully learned the task. We epoched the data and computed MSE on the epochs, grouped by training phase (Pre/Post-task), sleep stage (REM/non-REM) and learning phase (first/last day). We then performed a data-driven multivariate Partial Least Squares (PLS) analysis to compare multiscale entropy across the various conditions.
Our experimental workflow.
What did we find?
We found both an effect of sleep stage and an effect of training phase and the two effects were manifested in different timescales of MSE. MSE differentiated REM and non-REM sleep in both the finest and coarsest timescales, with a crossover of MSE curves occurring at mid-scales. MSE also differentiated the LFPs from the first day’s Pre-task condition (when rats were completely naïve to the task) and those from the last day of training in the non-REM conditions, with differences occurring in the middle timescales. It was interesting to us that the differentiation between the first and last day of training was detected in non-REM sleep, as these are the periods during which neuroplasticity mechanisms for learning and memory consolidation are thought to occur. This study was a first step towards understanding how MSE captures sleep-dependent effects on brain signals. More work is needed to understand how MSE varies in sleep with differences in learning/performance, as well as whether differences exist across brain regions and networks.