New neuronal model shows adaptive thresholds amplify firing noise

In a recent paper on arXiv (2607.18428), researchers Oliver Gambrell and Abhyudai Singh from the University of Delaware analyze the statistics of inter-spike intervals (ISIs) in a classical integrate-and-fire model of a postsynaptic neuron receiving independent excitatory and inhibitory inputs (EI circuit). The key innovation is the study of adaptive threshold potentials—both depolarizing (threshold increases with membrane potential) and hyperpolarizing (threshold decreases as membrane potential hyperpolarizes)—compared to the traditional fixed threshold.

Their analysis reveals that a depolarizing adaptive threshold increases ISI noise, measured by the coefficient of variation (CV), relative to a fixed threshold at the same mean ISI. Simulations further show that the ISI noise can be hypo-exponential (CV<1) or hyper-exponential (CV>1) depending on the balance of excitation and inhibition. More strikingly, the hyperpolarizing adaptive threshold model can generate action potentials even when driven solely by inhibitory inputs, a phenomenon not seen in standard models. This work provides a systematic stochastic framework for understanding how threshold adaptation shapes interneuronal communication and could inform both biological modeling and neuromorphic computing designs.