Abstract
Abstract Neural trajectories constitute propagating sequences of activity that are essential to cognitive representations. Understanding how neural trajectories overcome the disorganizing influence of neural noise in the awake state remains challenging. Theoretical and artificial network models have focused on optimizing the synaptic pathways underlying trajectory propagation. However, in a biophysical network with synaptic plasticity, solely altering synaptic connectivity induces a tight tradeoff between spurious trajectory interruptions and inductions by noise. Formal analysis predicts that separating the inhibitory fluctuations-driven spontaneous activity from the excitatory-driven trajectory propagation is key to reliable triggering and propagation. Amongst biophysical solutions, calcium-activated suprathreshold intrinsic conductances provided the best trajectory reliability by mitigating the reliance of propagation on connectivity. These mechanisms display parametric robustness and generalize to a variety of static and dynamic attractors with uni- and bidirectional propagation. Thus, we provide a unified framework for cognitive-associated neural trajectories robust to neural noise in the awake state.