Dendritic excitability controls overdispersion

Abstract

A neuron's input-output function is commonly understood in terms of the fluctuation-driven and mean-driven regimes. Here we investigate how active dendrites falling in either regime shape the input-output function and find that dendritic input primarily controls interspike interval dispersion, a prediction that is validated with dual patch clamp recordings. This phenomenon can be understood by considering that neurons display not two, but three fundamental operating regimes, depending on whether dendritic spikes or the somatic input reach threshold.

A Burst-Dependent Algorithm for Neuromorphic On-Chip Learning of Spiking Neural Networks

“…[29,30,31,32]. For spiking neural networks, burst-dependent algorithms exploit the target-dependence of short-term plasticity [33,34,35], dendrite-dependent bursting [36,37] and burst-dependent plasticity [13,12,14,15] to perform credit assignment in a way that approximates backpropagation [10,11,38]. These burst-dependent learning methods, referred to as 'Burstprop' algorithms, can in principle allow for multilayer local learning in SNNs.…”

Silences, spikes and bursts: Three‐part knot of the neural code

“…Deep layer, thick tufted neurons with projection to the pyramidal tract neurons can generate burst of action potentials mainly when input to the apical dendrite is combined with an action potential (Larkum, et al., 1999; Larkum et al., 2001, 2004). Other neuron types and dendritic compartments are also expected to show burst‐like coding whenever dendritic spikes are potent (Leinweber et al., 2017; Friedenberger & Naud, 2022). For example, bursts of high‐frequency or spatio‐temporally co‐ordinated inputs onto basal dendrites produce NMDA spikes, increasing the probability of producing a somatic burst (Polsky et al., 2009).…”

“…The absence of potent calcium spikes (for L2–3 cells (Larkum et al., 2007)) would indicate that these bursts do not arise from the same mechanism as in deep layer pyramidal cells. NMDA‐spikes and dendritic sodium‐ion channels can produce burstiness and have been observed in these cells (Brumberg et al., 2000; Palmer et al., 2014; Smith et al., 2013; Friedenberger & Naud, 2022), but the contrast between known mechanisms and the phenotype remains puzzling.…”

“…GABAergic interneurons of the cortex, on the contrary, may show burst‐dependent transmission (Campagnola et al., 2022), but have not been shown to have burst‐dependent plasticity (Hennequin et al., 2017) or particular burst‐generation mechanisms. Recent work has revealed that these cells are producing NMDAR‐based dendritic spikes, which can alter the ISI dispersion as in burst coding cells (Friedenberger & Naud, 2022). Putative GABAergic cells have also been shown to modulate bursting in an attention task (Voloh & Womelsdorf, 2018) in vivo .…”

“…Focusing on independent response features is more likely to reveal independent streams of information. We have provided here one example of such independent features in the event rate and the burst fraction, but other features such as the coefficient of variation (Doron et al., 2014; Friedenberger & Naud, 2022) or the timing of the burst (Herzfeld et al., 2018; Kayser et al., 2009; Bittner et al., 2017) are plausible choices.…”