Abstract
Spike timing-dependent plasticity (STDP) is a fundamental mechanism that modulates synaptic strength in response to the precise temporal relationships between input and output spikes. However, the exact role of transmission delays in this mechanism, particularly at the network level, remains unclear. Delays introduce an additional layer of complexity, as they directly influence the timing of spikes and consequently the direction and magnitude of synaptic modifications through STDP. Understanding how these delays shape and are shaped through plasticity rules is critical to reveal their role in neural computations. In this study, we explored the effects of repetitive presentations of synchronous spike volleys on the distribution of transmission delays in a feedforward (FF) network through STDP. Our findings showed that STDP preferentially selected connections with shorter transmission delays, whereas connections with longer delays were weakened and pruned from the network. We showed that "learning of delays" by the impact of synchronous pulse packets propagated forward; that is, the modification of the delays began from the connections between upstream layers and continued to downstream ones. This forward propagation of the delay selection process was accompanied by the successful transmission of synchronous signals along the network. The interaction between transmission delays and synaptic modification established a dynamic feedback loop. The transmission of synchronous signals drove the selective strengthening or weakening of connections, while the evolving synaptic structure, in turn, shaped the pathways for signal propagation. This bidirectional interplay highlighted the fundamental role of delays in orchestrating both network dynamics and the plasticity rule.