Functional connectivity among spikes in low dimensional space during working memory task in rat

Abstract

Working memory (WM) is critically important in cognitive tasks. The functional connectivity has been a powerful tool for understanding the mechanism underlying the information processing during WM tasks. The aim of this study is to investigate how to effectively characterize the dynamic variations of the functional connectivity in low dimensional space among the principal components (PCs) which were extracted from the instantaneous firing rate series. Spikes were obtained from medial prefrontal cortex (mPFC) of rats with implanted microelectrode array and then transformed into continuous series via instantaneous firing rate method. Granger causality method is proposed to study the functional connectivity. Then three scalar metrics were applied to identify the changes of the reduced dimensionality functional network during working memory tasks: functional connectivity (GC), global efficiency (E) and casual density (CD). As a comparison, GC, E and CD were also calculated to describe the functional connectivity in the original space. The results showed that these network characteristics dynamically changed during the correct WM tasks. The measure values increased to maximum, and then decreased both in the original and in the reduced dimensionality. Besides, the feature values of the reduced dimensionality were significantly higher during the WM tasks than they were in the original space. These findings suggested that functional connectivity among the spikes varied dynamically during the WM tasks and could be described effectively in the low dimensional space.

Figure 1. The Y maze working memory task and preprocessing of the data.

(A) The Y maze working memory task. For the Y maze, dashed lines represent the removable guillotine doors. The occurrence of behavioral event is detected by an infrared sensor. Food rewards are located at the ends of goal arms. Arrow shows in the right plot possible correct path, dotted line in the right plot shows possible incorrect path. (B) Processes of spike sorting. (C) Rastergram of the spike trains recorded during the Y-maze task (3 s pre and 1 s post the tripping time) and the converted continuous series. The red triangle denotes the tripping time of the ‘choice run’ behavioral event in the Y-maze. (D) Plot of the principal components obtained from the continuous time series. The first 10 principal components (PCs) account for over 90% energy of the total variables. (E) Granger causality matrices in the original (left) and in the reduced dimensionality (right).

Figure 2. Dynamic variations of granger causality during working memory tasks.

The data are divided into six 1(4 s pre and 2 s post the tripping time). The red triangle indicates the tripping time of the infrared sensor in the Y-maze. (A) Dynamic variations of the granger causality matrixes during a working memory task of rat 1. (B) Variations of the GC values in the original dimensionality during the working memory tasks of each rat (mean±SEM). (C) Variations of the GC_PC values in the reduced dimensionality during the working memory tasks of each rat (mean±SEM). (D) Comparisons of granger causality (mean±SEM). The granger causality values of the original and the reduced dimensionality are both significantly higher at the working memory state (WMS) than at the beginning state (WMBS). Besides, the granger causality levels in the original dimensionality at the WMS and the WMBS are significantly lower than those in the reduced dimensionality (80 trials for 6 rat, paired sample t-test, ^* P<0.05, ^** P<0.01).

Figure 3. Dynamic variations of global efficiency and causal density both in the original and in the reduced dimensionality during working memory tasks, 6 rats respectively.

(A) The dynamic variations of global efficiency in the original dimensionality (dashed lines) and the reduced dimensionality (solid lines) during the WM tasks, 6 rats respectively (mean±SEM). (B) The dynamic variations of causal density in the original dimensionality (dashed lines) and the reduced dimensionality (solid lines) during WM tasks, 6 rats respectively (mean±SEM). The red triangle indicates the tripping time of the infrared sensor in the Y-maze. (C) Comparisons of global efficiency (mean±SEM). The values of the global efficiency in both the original and the reduced dimensionality are significantly higher at the WMS than at the WMBS. The E values in the original dimensionality at the WMS are significantly lower than the E_PC values in the reduced dimensionality (80 trials for 6 rat, paired sample t-test, ^** P<0.01). No significant difference is found at the WMBS (80 trials for 6 rat, paired sample t-test, P>0.05). (D) Comparisons of causal density (mean±SEM). The values of the causal density in both the original and the reduced dimensionality are significantly higher at the WMS than at the WMBS. The CD values in the original dimensionality at the WMS are significantly lower than the CD_PC values in the reduced dimensionality (80 trials for 6 rat, paired sample t-test, ^** P<0.01). No significant difference is found at the WMBS (80 trials for 6 rat, paired sample t-test, P>0.05).

Figure 4. Dynamic variations of granger causality, global efficiency and causal density during the incorrect tasks in both the original and the reduced dimensionality (20 trials for 6 rats).

The data are divided into six 1(4 s pre and 2 s post the tripping time). The red triangle indicates the tripping time of the infrared sensor in the Y-maze. (A) Dynamic variations of the granger causality matrixes during an incorrect trial of rat 1. (B) The variations of granger causality (left), global efficiency (middle) and causal density (right) during the incorrect tasks (20 trials for 6 rats) and the correct trials (80 trials for 6 rats) in the original dimensionality. The feature values in the correct trials are significantly higher 2 s (GC, E, CD) and 1 s (E, CD) pre the tripping time than those in the incorrect trials (t test, ^* P<0.05, ^** P<0.01). No statistical difference is found at the WMBS between the incorrect and the correct trials (t test, P>0.05). (C) The variations of granger causality (left), global efficiency (middle) and causal density (right) during the incorrect tasks (20 trials for 6 rats) and the correct trials (80 trials for 6 rats) in the reduced dimensionality. The feature values in the correct trials are significantly higher 2 s (GC_PC, E_PC, CD_PC) and 1 s (GC_PC, E_PC, CD_PC) pre the tripping time than those in the incorrect trials (t test, ^** P<0.01). In addition, the feature values were significantly higher at the WMBS in the incorrect trials (t test, ^* P<0.05).