Figure 2.

Dynamical description of the neuronal model. (a) We use detailed digital neuron reconstructions from NeuroMorpho.org. The dendrites are partitioned into fine compartments. (b) The network of compartments forms an excitable tree to account for dendritic spikes. Each compartment undergoes the same type of dynamics. A temporal evolution of states of a compartment is shown here, corresponding to the susceptible (green), spiking (red), and refractory (yellow) stages of a dendritic spike. (c) Snapshot of neuronal dynamics with $h=0.5$ and $P=0.96$³⁰. For a full illustration of the dynamics, please refer to Supplementary Video 1. (d) The firing rate of each compartment can be plotted for a particular set of parameters h and P (here, $h=1.12$ and $P=0.9$ using Neuron B, see “Methods” and Table 1), which govern the external driving and the dendritic excitability (see “Methods” for details), to obtain a map of most active regions within the neuron. Warm colors correspond to high firing rates, and cold colors to low firing rates. (e) Plotting the average firing rate of a particular compartment against the input rate h yields the response curve. The range of h values over which the firing rate varies most (between 10% and 90% of the maximum firing rate, F₁₀ and F₉₀) is called the dynamic range Δ. This example uses Neuron B with $P=0.9$. (f) The energy consumption of a neuron is given by the number of total dendritic spikes per somatic spike, and is plotted over the parameter space. This example uses Neuron B. (g) The relative energy consumption represents the number of times a dendritic compartment fires, on average, per somatic spike. If it is less than 1, the neuron is energy efficient because dendrites effectively amplify the neuronal output (somatic spikes). If it is larger than 1, the neuron is considered energy inefficient. This example uses an interneuron from the Drosophila^30,31. (h) Based on the number of branches connected to the soma and its centrality, neurons can be classified into four functional types: energy efficient (Type 1), partially efficient/inefficient (Type 2), energy inefficient (Type 3), and a transitional type (Type T) which can exhibit a mixture of the other types. See reference³⁰ for more details on the functional classification of neurons based on energy consumption.