Work

Macaque motor cortex resting state spiking activity is characterized by low correlations and high dimensionality

Year: 2018

Type: conference-paper

Abstract: Resting state activity is typically studied in large scale measurements such as fMRI or M/EEG with the aim to understand spontaneous brain activity as opposed to task- or stimulus-evoked. To investigate network activity on the level of single units, we recorded the spiking activity and LFP from the hand-movement area of macaque (pre-)motor cortex during rest, using a chronically implanted 4x4 mm^2 100 electrode Utah Array (Blackrock Microsystems). Based on video recordings of the monkeys, we defined epochs of resting (RS), sleepiness (RSS), spontaneous body and contralateral arm movements (M, AM). Each recording session lasted 15-20 min and yielded approx. 140 simultaneously recorded single units, which we separated into putative inhibitory (INH) and excitatory (EXC) based on their wave shape (Dehghani et al. 2016). We characterized the data by their firing rates (FR), global (CV) and local (LV) regularity of firing, pairwise fine temporal correlations (~2 ms), rate covariances (~200 ms), and the power spectra of the LFP.We find that INH units fire faster compared to EXC in all states, with a tendency towards lower LV and higher CV values. The CV is generally highest during RSS and AM. Since it does not account for nonstationarities, its high values indicate that firing rates vary over time. We find that the change of FR with respect to behavior is neuron-specific and more prominent in INH. INH-INH fine temporal correlations are higher than EXC-EXC, independently of the behavior. During RS the distributions of correlations and rate covariances are narrower and show lower absolute values compared to other states. Moreover, the principal component analysis of the pairwise covariances reveals high dimensionality of RS activity while it is significantly lower during RSS and AM. LFP spectra show different dominant frequencies with respect to the behavioral state: the power in alpha-beta range (7-30 Hz) is enhanced during RS and RSS, and in high-gamma (60-200 Hz) during M and AM. In summary, resting state activity is characterized by less correlated spike trains, less non-stationarity, and a high dimensionality compared to other states. However, a clear classification of states based on the neuronal activity is not yet possible, due to a high overlap of state-resolved distributions. Support: DFG DE2175/2-1 & GR1753/4-2 Priority Program (SPP 1665); EU Grants 720270 & 785907 (HBP); LIA “Vision for Action” of CNRS, Marseille & Research Centre Jülich.

Source: JuSER (Forschungszentrum Jülich)