Encoding Workig Memory Via Coordination Of Multi-mode Neural Signals

Objective:Working memory (WM) refers to the short-term maintenance and manipulation of information necessary for higher cognitive functions. There is a growing interest in the investigation of working memory mechanism. Recent researches show that different frequency components (eg. theta and gamma) of local field potentials (LFPs) and spikes, as different modes of neural signals, encode WM respectively. A question arises here is how these signals encode WM in coordination. The aim of this study is to investigate the mechanism of encoding WM in coordination of multi-mode neural signals, including:(1) encoding WM in coordination of theta (4-12Hz)-gamma (30-100Hz),(2) encoding WM in coordination of spikes-LFPs.Methods:(1) Acquisition of different mode neuronal signals during WM:The experimental data were multi-channel LFPs and spikes obtained from SD rat prefrontal cortex (PFC) through the implanted microelectrode array during the WM tasks in Y-maze. Then the short-time Fourier transform (STFT) was applied to analyze the power changes of WM related frequency bands (theta:4-12Hz, gamma:30-100Hz) in the LFPs.(2) Theta-gamma encoding WM in coordination:The coupling strength between theta and gamma was measured by modulation index (MI) via cross frequency coupling (CFC). The distribution of coupling strength was obtained by applying MI measure to several narrowed-filtered frequency pairs (4Hz bandwidths and2Hz steps). Dynamic coupling strength between theta and gamma during WM were then calculated.(3) Spikes-LFPs encoding WM in coordination:The joint entropy indexes (JEIs) between spikes and the principle components (theta, gamma) of LFPs were calculated for each pair of the spike and the LFP series during WM to measure the coordination strength. Dynamic changes of the spikes-theta, spikes-low gamma and spikes-high gamma JEI values was analyzed to characterize the spikes-LFPs joint encoding pattern.Results:(1) Power changes of LFPs during the Y-maze taskThe low frequency (<20Hz) power was higher during WM tasks. The power of theta and gamma both peaked in Is pre the reference point (RP) in Rat1to Rat5and around1s in Rat6.(2) Dynamic theta-gamma coupling levels during the WM tasks1) The distribution of coupling strength between narrowed-filtered frequency pairs The MI peak values of the narrowed-filtered frequency pairs appeared between the phase of low frequencies (4-10Hz) and the amplitude of high frequencies (around45Hz and80Hz). Accordingly, gamma was divided into low gamma band (LG,30-60Hz) and high gamma band (HG,60-100Hz). There was no interaction between the phase of higher frequencies (20-100Hz) and the amplitude of4-100Hz LFPs signals.2) Dynamic changes of theta-LG coupling levels during the WM tasksTheta-LG modulation strength significantly increased during the accurate trials of WM task (4rats,63trials in all). Meanwhile, the theta phase was more strongly coupled with LG during the accurate trials than in the error trials.3) Dynamic changes of theta-HG coupling levels during the WM tasksThere was no significantly change of the theta-HG modulation strength during the accurate trials of WM task (4rats,63trials in all). Meanwhile, difference between the theta-HG coupling strength in the accurate trials and the incorrect trials was not statistically significant.4) Dynamic changes of theta-LG coupling levels of each channel during the WM tasksChanging tendency of theta-LG MI values for each channel was not consistent. The theta-LG MI values significantly increased during the correct trials in most of the channels.(3) Dynamic spikes-LFPs coupling levels during the correct WM tasks 1) Dynamic spikes-theta coupling levels during the correct WM tasksThe normalized spikes-theta JEI values significantly increased during the correctly performed working memory tasks (6rats,10trials for each rat). The mean values of the normalized spikes-theta JEI increased from0.345±0.0383s pre the RP to the peak value of0.721±0.033ls pre the RP.2) Dynamic spikes-LG coupling levels during the correct WM tasksThe mean values of the normalized spikes-LG JEI values increased from0.408±0.0403s pre the RP to the peak value of0.696±0.040ls pre the RP in the correct trials (6rats,10trials for each rat). The differences between these two group JEI values were statistically significant.3) Dynamic spikes-HG coupling levels during the correct WM tasksThe normalized spikes-HG JEI values significantly increased during the correctly performed working memory tasks (6rats,10trials for each rat). The mean values of the normalized spikes-HG JEI significantly increased from0.393±0.0413s pre the RP to the peak value of0.690±0.038ls pre the RP.Conclusion:(1) The power of LFPs WM principle components theta and gamma were both increased.(2) The coupling strength between theta and LG significantly increased during the correct trials but not in the incorrect trials (4rats). The coupling strength of theta and low gamma encode WM in coordination.(3) The JEIs between spikes and LFPs components (theta, LG and HG) in PFC significantly increased and peaked before the reference point during the correct trials. The coordination between spikes and LFPs components (theta. LG and HG) could encode WM effectively.