Not all-or-nothing: Intracellular action potential waveform varies systematically with extracellular gamma oscillatory state

Individual neurons’ action potentials (APs) have a dynamic waveform, the shape of which differs by neuron subtype and species. In analyses of single-unit data, however, this waveform is reduced to a binary “all-or-nothing” event, with systems and cognitive neuroscience focusing on the number and timing of spike events. This analytical convenience has, in turn, resulted in computational theories of neural coding that assume binarity.Despite this assumption, a considerable body of evidence shows that AP waveforms exhibit systematic, within-neuron variation that may contribute additional information beyond the rate or timing of discharge that is ignored in the interpretation of classical results in systems, cognitive, and computational neuroscience.To address this concern, we have developed an AP waveform parameterization approach that quantifies the fine-scale features of extracellular and intracellular spike waveforms. Using this parameterization on patch-clamp recordings of APs, we find within-neuron correlations between features such that, for example, individual APs that decay faster result in subsequent spikes occurring more quickly.Leveraging this parameterization, we analyzed a dataset of simultaneously recorded patch-clamp APs and local-field potentials (LFPs) in rats. This allows us to examine, for the first time, the temporally causal influence of the LFP on individual AP waveforms. We observed within-neuron differences in spike waveform features in relation to LFP gamma oscillatory states that have been overlooked using traditional spike analyses. Our results provide preliminary evidence for the importance of studying within-neuron variance in AP waveform alongside traditional spiking metrics to enhance our understanding of neural coding and cognition.