Effect of stimulus-driven pruning on the detection of spatiotemporal patterns of activity in large neural networks

Abstract

Adult patterns of neuronal connectivity develop from a transient embryonic template characterized by exuberant projections to both appropriate and inappropriate target regions in a process known as synaptic pruning. Trigger signals able to induce synaptic pruning could be related to dynamic functions that depend on the timing of action potentials. We stimulated locally connected random networks of spiking neurons and observed the effect of a spike-timing-dependent synaptic plasticity (stdp)-driven pruning process on the emergence of cell assemblies. The spike trains of the simulated excitatory neurons were recorded. We searched for spatiotemporal firing patterns as potential markers of the build-up of functionally organized recurrent activity associated with spatially organized connectivity.

Introduction

There is experimental evidence that the cerebral cortex develops as a whole rather than regionally, as synaptogenesis proceeds concurrently in all cortical areas and layers. Simultaneous overproduction of a critical mass of synapses in each cortical area may be essential for their parallel emergence through competitive interactions between extrinsic afferent projections. Such competition has been observed between the projections of the two eyes during the formation of visual centers Hubel et al., 1977, Rakic, 1981.

Genetic programs are assumed to drive the primordial pattern of neuronal connectivity through the actions of a limited set of trophic factors and guidance cues, initially forming excessive axonal and dendritic branches and synapses, distributed somewhat diffusely (Innocenti, 1995). Then, refinement processes act to correct initial inaccuracies by pruning inappropriate connections while preserving appropriate ones. The embryonic nervous system is refined over the course of development as a result of the twin processes of cell death and selective axon pruning. It is generally agreed that changes in cortical function are associated with corresponding alterations in the density and arrangement of synaptic circuits. The density of synapses continues to increase during infancy and remains above adult levels. After a relatively short period of stable synaptic density, a pruning process begins: synapses are constantly removed, yielding a marked decrease in synaptic density. This process continues until puberty, when synaptic density stabilizes at adult levels which are maintained until old age (Huttenlocher, 1979). It was observed through widespread brain regions including cortical areas Bourgeois and Rakic, 1993, Huttenlocher et al., 1982 and the projection fibers between hemispheres (Innocenti, 1995). Pruning may also play a role in establishing topographic maps, as it can be seen in the retinotectal system (Nakamura and O’Leary, 1989). The relation between synaptic efficacy and synaptic pruning Chechik et al., 1999, Mimura et al., 2003 suggests that the weak synapses may be modified and removed through competitive “learning” rules. Trigger signals able to induce synaptic pruning could be related to dynamic functions that depend on the timing of action potentials.

In this article, we studied the emergence of cell assemblies out of a locally connected random network of integrate-and-fire units distributed on a 2D lattice (Iglesias et al., 2005) stimulated in both temporal and spatial dimensions. The originality of our study stands on the size of the network, 10,000 U, the duration of the experiment, $10^6$ time steps (1 time step = 1 ms), and the application of spike-timing-dependent synaptic plasticity (stdp) rule that is a change in the synaptic strength based on the ordering of pre- and post-synaptic spikes (Bell et al., 1997). Among several stdprules (Roberts and Bell, 2002) we selected a rather simple one compatible with a custom designed hardware implementation Eriksson et al., 2003, Torres et al., 2004. This hardware is the core of electronic devices, called POEtic, whose architecture include features derived from some of the properties present in living beings, like evolution, development, self-repair, self-replication and learning (Tyrrell et al., 2003). The combination of partial and total dynamic reconfiguration, as well as the self-configuration and dynamic routing capabilities make these devices an ideal candidate for the efficient implementation of large-scale spiking neural network models. In our study the synaptic modification rule was applied only to the excitatory–excitatory (exc, exc) connections. stdpis expected to strengthen the connections among neurons that belong to cell assemblies characterized by recurrent patterns of firing. Conversely, those connections that are not recurrently activated might decrease in efficiency and eventually be eliminated. The main goal of our study is to determine whether or not, and under which conditions, such cell assemblies may emerge from a large neural network receiving background noise and content-related input organized in both temporal and spatial dimensions.