Abstract:Recurrent excitatory circuits and the positive feedback they imply are assigned important roles in a variety of tasks in living organisms. Such networks obviously do not exhibit saturated behaviour in the sense of extremely fast rates and/or insensitivity to input variations, as artificial systems with positive feedback generally do. It is therefore important to identify how neural saturation is avoided. A single neuron that excites itself directly provides the simplest anatomical circuitry where this problem can be studied. Such a system was simulated experimentally by Diez-Martínez and Segundo using the pacemaker neuron in the crayfish stretch receptor organ. They showed that the feedback transmission time, called "delay", was strongly influential, and small changes led to markedly different outcomes. As the delay was increased the discharge patterns went from pacemaker spike trains to multiplets separated by silent intervals to still longer bursts and longer silent intervals. We hypothesized that neuronal sensitivity decreased along the rapid successive firings (adaptation) and prevented saturation and therefore played an important role in the observed behaviour. This hypothesis was tested using models of increasing complexity. The simplest model was an integrate and fire neuron without adaptation to repeated stimuli, in this case the dynamics were qualitatively different from the experimental data. The other models exhibited adaptation to repeated stimuli. Two were relatively simple: an integrate and fire and a leaky integrator. The last model was more complex, it included membrane conductances. Neither of these models exhibited saturation when recurrent excitation was introduced. Their dynamics were in fact similar to those in the crayfish preparation, both exhibiting pacemaker firing for short delays, and multiplets or burst for intermediate delays. Simulations were therefore compatible with the hypothesis that neuronal adaptation is important in preventing saturation. Copyright 1996 Elsevier Science Ltd

Activity of any neuron with delayed feedback stimulated with Poisson stream is non-Markov

Firing statistics of inhibitory neuron with delayed feedback. I. Output ISI probability density

On stationary distributions of stochastic neural networks

Leaky Integrate-and-Fire Neuron under Poisson Stimulation

Distribution of interspike intervals of a neuron with inhibitory autapse stimulated with a renewal process

Stochastic resonance in Hodgkin-Huxley neuron induced by unreliable synaptic transmission.

On the stationary state of a network of inhibitory spiking neurons

Superluminal propagation of light in gravitational field and non-causal signals: some comments

A stochastic-field description of finite-size spiking neural networks

A review of the integrate-and-fire neuron model: II. Inhomogeneous synaptic input and network properties

Neural Activity of Heterogeneous Inhibitory Spiking Networks with Delay

The firing of an excitable neuron in the presence of stochastic trains of strong synaptic inputs

On the Poisson equation for nonreversible Markov jump processes

Single Neuron with Recurrent Excitation: Effect of the Transmission Delay

Shifting Spike Times or Adding and Deleting Spikes—How Different Types of Noise Shape Signal Transmission in Neural Populations

Equivalence between synaptic current dynamics and heterogeneous propagation delays in spiking neuron networks

Metastability in networks of nonlinear stochastic integrate-and-fire neurons

A First-Order Non-Homogeneous Markov Model for the Response of Spiking Neurons Stimulated by Small Phase-Continuous Signals

Stimulus-Dependent noise facilitates tracking performances of neuronal networks

The neuron's response at extended timescales