Multiplexing neurons and multiple overlapping cell assemblies active during motor behavior

The cell assembly hypothesis [1] postulates that neurons coordinate their activity through the formation of repetitive co-activation of groups, called cellassemblies. We assume that spatio-temporal spike patterns (STPs) occur as an expression of active neuronal assemblies, at the resolution of a fewmilliseconds.In order to test this hypothesis, we used the SPADE method [2,3,4,5], which detects significant STPs in parallel spike trains. We analyzed experimental datarecorded by a 10x10 electrode Utah array in the pre-/motor cortex of macaque monkeys performing a reach-to-grasp task [6,7]. The task comprised four differenttrial types of grasping and pulling an object by combining two grip types and two force levels.We find significant STPs in 19/20 recording sessions (of 15min) from different days. They occur in all phases of the behavior and across all trial types. Their sizeranges between 2 and 6 neurons, with a maximal temporal extent of 60ms. The STPs show a high behavioral specificity, suggesting that different cellassemblies are active in the context of different behaviors. Moreover, we observed that pattern spikes are only a small fraction of the total recorded spikingactivity, which may be explained by downsampling due to the recording. A surprising finding is that STPs overlap on different levels: 1) the same neuron may beinvolved in a different STP during another behavioral epoch during an individual session, which may indicate overlapping assemblies; 2) in 85% of the sessionswith patterns at least one neuron participates in many patterns, which may be interpreted as a hub neuron linking assemblies; 3) even individual spikes take partin more than one STP.Concluding, our results indicate that STPs occur frequently in parallel spike trains. Quantitative analysis of their properties suggests that STPs are functionallyrelated to behavior and specific to it, and may be an indication of the presence of assemblies being activated during the task. The assemblies may include tensor even hundreds of neurons, however, given the sub-sampling of our experimental setting, we may capture their activation in the form of patterns composed ofa few neurons.References:[1] Hebb, D. O. (1949). John Wiley & Sons[2] Torre et al (2016) J Neurosci.[3] Quaglio et al. (2017). Front Comp Neurosci.[4] Stella, Quaglio et al.(2019). Biosystems[5] Stella, Bouss et al. (2022). eNeuro[6] Brochier et al. (2018). Scientific data[7] Riehle et al. (2013). Front. Neural Circuit