State dependence of network output: modeling and experiments

Abstract

Emerging experimental evidence suggests that both networks and their component neurons respond to similar inputs differently, depending on the state of network activity. The network state is determined by the intrinsic dynamical structure of the network and may change as a function of neuromodulation, the balance or stochasticity of synaptic inputs to the network, and the history of network activity. Much of the knowledge on state-dependent effects comes from comparisons of awake and sleep states of the mammalian brain. Yet, the mechanisms underlying these states are difficult to unravel. Several vertebrate and invertebrate studies have elucidated cellular and synaptic mechanisms of state dependence resulting from neuromodulation, sensory input, and experience. Recent studies have combined modeling and experiments to examine the computational principles that emerge when network state is taken into account; these studies are highlighted in this article. We discuss these principles in a variety of systems (mammalian, crustacean, and mollusk) to demonstrate the unifying theme of state dependence of network output.

Figure 1.

A, B, Neuromodulation of the crustacean gastric mill CPG network states. Computational modeling shows that neuromodulation (by the peptide PK) of the gastric mill rhythm elicited by the projection neuron MCN1 can result in little qualitative change in network output (top traces). Yet, in contrast to the unmodulated or weakly modulated state (A), the generation of network oscillations in the strongly modulated state (B) does not depend on presynaptic inhibition of the axon terminals of the projection neuron MCN1 (bottom traces). Models modified from Kintos et al. (2008).

Figure 2.

Large scale modeling of olfactory bulb network state. A, Schematic illustration of the model. Olfactory sensory neurons (OSN), responding to odor input in a distributed manner, each project to one glomerulus (Glom), in which they make excitatory synapses with olfactory bulb output neurons [mitral cells (Mi)] and diverse local interneurons (not shown). Each mitral cell receives sensory input from a single class of OSNs within a given glomerulus. Mitral cells have extensive lateral dendrites through which they form reciprocal synaptic connections with inhibitory interneurons [granule cells (Gr)]. Mitral cell axons project onto a large number of secondary olfactory structures. Based on data from brain slice experiments, synaptic plasticity was implemented on excitatory synapses between OSNs and mitral cells as well as between mitral and granule cells (arrows). B, Simulation results showing the membrane potential and action potentials of a subset of mitral cells in response to stimulation with two odors A and B, before and after simulated exposure to odor A. Note the sparsening of mitral cell responses to odorants accompanied by increased oscillatory dynamics and synchrony among individual mitral cells. C, Synaptic plasticity in the model leads to stronger inputs to mitral cells (red arrow) because of increased synaptic efficacy as well as a stronger feedback loop (blue arrow) between mitral and granule cells. Together, these changes enhance the oscillatory power (gray arrow) of the excitatory-inhibitory feedback loop formed by the mitral/granule cell reciprocal synapse.

Figure 3.

Feeding CPG dynamics and behavior in Aplysia. A, Aplysia in a feeding posture. The ingestive and egestive arrows indicate movement of the seaweed strip into and out of the mouth of the animal, respectively. Adapted from Lum et al. (2005). B, Functional performance of a computational model of the feeding CPG dynamics shown in C in a simulated feeding task modeled on the behavior in A, plotted over a range of two basic parameters of a simulated feeding environment: the length scale of the environment (the average length of the seaweed strips present in the environment) and the certainty with which the feeding stimuli in the environment can actually be detected. Warm colors indicate good performance, cool colors, poor performance. C, Diagram of the dynamics of the feeding CPG as quantified by Proekt et al. (2004). Discrete motor programs are not represented; rather, the underlying evolution of their ingestive-egestive character is indicated by a time-continuous variable, the “feeding behavior.” The red rectangle indicates a period during which, in the behavioral simulations described in the text, the model would perform goal-driven egestion, that is, egestion despite an ingestive stimulus.