Segmentation of continuous stimulus response patterns in a cultured neuronal network

In this study, we developed a cultured neuronal circuit as a simplified model of the brain. Our objective was to investigate the emergence of language acquisition elements within this minimal living model, particularly focusing on the segmentation of input series stimulation. Our hypothesis posited that within this cultured neuronal circuit, the segmentation of series stimuli would manifest through an emphasized response pattern to a specific repeated sequence of input patterns. Initially, we observed a notable augmentation in the reproducibility of response patterns following repeated series stimulation from a single electrode. Specifically, the reproducibility of responses to continuous stimulation intervals of 1 s appeared to significantly increase. Moreover, we confirmed a link between the increased reproducibility of response patterns to series stimulation and the stability of background activity. Additionally, we analyzed the mutual similarity of invoked response to triple stimuli (triplets) from three independent electrodes, $\mathbf{A}, \mathrm{B}$, and $\mathbf{C}$, in various permutations (e.g., ABC, BCA, etc.). By applying 10 series-stimuli with equalized triplet occurrence frequencies, the reproducibility of response patterns invoked by each triplet stimulus was standardized, suggesting that when triplet stimuli are input with equal frequencies, their responses exhibit comparable reproducibility. If the input frequencies of triplet stimuli are biased, and if the reproducibility of responses to specific triplet stimuli with higher input frequencies increases, it implies the segmentation of these triplet stimulus patterns as cohesive units akin to words. In future analyses, we will investigate the reproducibility of responses to sequences with biased triplet occurrence frequencies.