Neurons and Cognition (q-bio.NC)

Abstract: Recently, interest has grown in exploring the hypothesis that neural activity conveys information through precise spiking motifs. To investigate this phenomenon, various algorithms have been proposed to detect such motifs in Single Unit Activity (SUA) recorded from populations of neurons. In this study, we present a novel detection model based on the inversion of a generative model of raster plot synthesis. Using this generative model, we derive an optimal detection procedure that takes the form of logistic regression combined with temporal convolution. A key advantage of this model is its differentiability, which allows us to formulate a supervised learning approach using a gradient descent on the binary cross-entropy loss. To assess the model's ability to detect spiking motifs in synthetic data, we first perform numerical evaluations. This analysis highlights the advantages of using spiking motifs over traditional firing rate based population codes. We then successfully demonstrate that our learning method can recover synthetically generated spiking motifs, indicating its potential for further applications. In the future, we aim to extend this method to real neurobiological data, where the ground truth is unknown, to explore and detect spiking motifs in a more natural and biologically relevant context.

Abstract: In recent years, recurrent quantification analysis (RQA) and its multi-dimensional version (MdRQA) have emerged as a popular tool for assessing interpersonal behavioral or physiological synchrony in groups of two or more individuals. While experimental data in such studies are typically collected for a fixed, pre-determined duration, naturally occurring phenomena may often reach a state of transition after an unpredictable or varying duration of time. The resulting recurrence plots(RPs) across groups cannot be compared directly via linear scaling because the sensitivity of RQA variables to local dynamics would vary. We propose to address this by using the sliding window technique on individual RPs and using the summary statistics of the different RQA variable distributions computed across the sliding windows to differentiate the dynamics of the original time series of unequal durations. We tested our approach in two models: 1) the Rossler attractor and 2) the Kuramoto model. We compared the ability of different summary statistics of RQA variable distributions to accurately predict the dynamic states of the system across varying levels of noise, unequal lengths of time series, and, in the case of the Kuramoto model, different numbers of oscillators across samples. We found that while the mean, compared to other measures of central tendency, was a more accurate predictor of the underlying dynamic state of the system at high noise conditions, the mode was the most robust to the degree of noise in the signals, performing better than RQA variables from the whole RP, in general. To our knowledge, this is the first systematic attempt to validate the use of MdRQA in computing and comparing synchrony between systems of non-uniform composition and unequal time series data, paving the way for future work that examines interpersonal synchrony in more naturalistic, ecologically valid contexts.

Abstract: It is known that, during learning, modifications in synaptic transmission and, eventually, structural changes of the connectivity take place in our brain. This can be achieved through a mechanism known as structural plasticity. In this work, starting from a simple phenomenological model, we exploit a mean-field approach to develop a modular theoretical framework of learning through this kind of plasticity, capable of taking into account several features of the connectivity and pattern of activity of biological neural networks, including probability distributions of neuron firing rates, selectivity of the responses of single neurons to multiple stimuli, probabilistic connection rules and noisy stimuli. More importantly, it describes the effects of consolidation, pruning and reorganization of synaptic connections. This framework will be used to compute the values of some relevant quantities used to characterize the learning and memory capabilities of the neuronal network in a training and validation procedure as the number of training patterns and other model parameters vary. The results will then be compared with those obtained through simulations with firing-rate-based neuronal network models.