标题： 因果尖峰时间依赖可塑性防止循环网络中的集合融合 显示英文标题

标题： Causal Spike Timing Dependent Plasticity Prevents Assembly Fusion in Recurrent Networks

摘要： 神经元组织成功能相关的集合是皮层网络的一个基本特征，然而我们对这些集合在共享成员的同时如何保持独特身份的理解仍然有限。 在这里，我们分析了突触时间依赖可塑性（STDP）如何塑造反复耦合的脉冲神经元模型网络中重叠神经元集合的形成和稳定性。 通过数值模拟和相应的平均场理论，我们证明了STDP规则的时间结构，特别是其因果性程度，关键地决定了在训练完成后共享神经元的集合是否保持分离或合并。 我们发现，因果STDP规则，其中增强/抑制严格发生在前突触脉冲先于/后于后突触脉冲时，即使成员有大量重叠，也能使集合保持独立。 这种稳定性是因为因果STDP有效地抵消了由共享神经元的共同输入引入的对称相关性。 相反，非因果STDP规则在重叠超过临界阈值时会导致集合融合，这是由于共同输入相关性的不受控制的增长。 我们的结果为突触时间依赖学习规则如何支持分布式表示提供了理论见解，其中单个神经元参与多个集合的同时保持功能特异性。 显示英文摘要

摘要： The organization of neurons into functionally related assemblies is a fundamental feature of cortical networks, yet our understanding of how these assemblies maintain distinct identities while sharing members remains limited. Here we analyze how spike-timing-dependent plasticity (STDP) shapes the formation and stability of overlapping neuronal assemblies in recurrently coupled networks of spiking neuron models. Using numerical simulations and an associated mean-field theory, we demonstrate that the temporal structure of the STDP rule, specifically its degree of causality, critically determines whether assemblies that share neurons maintain segregation or merge together after training is completed. We find that causal STDP rules, where potentiation/depression occurs strictly when presynaptic spikes precede/proceed postsynaptic spikes, allow assemblies to remain distinct even with substantial overlap in membership. This stability arises because causal STDP effectively cancels the symmetric correlations introduced by common inputs from shared neurons. In contrast, acausal STDP rules lead to assembly fusion when overlap exceeds a critical threshold, due to unchecked growth of common input correlations. Our results provide theoretical insight into how spike-timing-dependent learning rules can support distributed representation where individual neurons participate in multiple assemblies while maintaining functional specificity.