Volume 8 Supplement 2

Sixteenth Annual Computational Neuroscience Meeting: CNS*2007

Open Access

Bumps and waves in a two-dimensional multilayer neural field model

BMC Neuroscience20078(Suppl 2):P68

DOI: 10.1186/1471-2202-8-S2-P68

Published: 6 July 2007

Background

Neural field models of firing rate activity have played a major role in developing an understanding of the dynamics of neural tissues [1]. They can be used to model extrinsic optical imaging signals and understand how different neuronal layers contribute to them. A two-dimensional, multi-population approach is therefore required. At a higher level of detail, biological data on horizontal cortical connectivity must be well taken into account. Finally, the spatial resolution of extrinsic optical imaging and biological connectivity studies involving patches of neurons [2, 3] suggest a mesoscopic neural mass approach.

Methods

We model a cortical area as a two-dimensional neural field composed of one excitatory and one inhibitory layer of neural masses. It is governed by a four-dimensional integro-differential system that we write as the sum of two terms. The first term is linear and describes the synaptic integration made by the neural masses. The second term is the input feeding a neural mass at a given point of the field. It sums up the contributions of all neural masses in the field by a weighted integral of their instantaneous firing rates. This is done through kernels that include both quantitative ("In which proportion do different types of neurons connect to each other?") and spatial ("How are these connections distributed on the cortical surface?") information between each pair of neuronal types. Neural masses are described by two average variables: the average membrane potential (dendritic compartment) and the average firing rate (axonal compartment), which is obtained from the potential by a Heaviside transformation.

Results

We have considered translation invariant, rotationally symmetric connectivity kernels and looked for rotationally symmetric bumps and pulses solutions [1, 4]. For both problems, the analysis falls in two parts: find solutions and check their stability. In the case of stationary bump solutions, expressing connectivities in terms of Bessel functions leads to closed forms depending on the parameters of the model. But not all these solutions are actual bumps and we need sufficient conditions to characterize acceptable bump radii according to other parameters values. A first step is made by writing several local necessary conditions, e.g., the solution must be equal to the threshold of its Heaviside voltage-to-rate transformation on the boundaries of the bump. The same problem arises for traveling pulses solutions. We then check the stability of bumps and pulses solutions to a family of perturbations with separated polar coordinates by reducing the analysis to an eigenvalue problem. Technical computations lead to implicit formulas for the eigenvalues. In the case of bumps it takes the simple form of a second order polynomial.

Conclusion

The model we have proposed extends previous related work [1, 4] in two directions: we can deal with several populations of neural masses and perform the model analysis in the framework of a 2D continuum as opposed to 1D. Our analysis raises several interesting biological questions related to connectivity functions that may be answered using optical imaging.

Authors’ Affiliations

(1)
Odyssee Laboratory, INRIA/ENPC/ENS, INRIA Sophia Antipolis

References

  1. Coombes S: Waves, bumps, and patterns in neural fields theories. Biol Cybern. 2005, 93: 91-108. 10.1007/s00422-005-0574-y.PubMedView ArticleGoogle Scholar
  2. Thomson A, Bannister A: Interlaminar Connections in the Neocortex. Cerebral Cortex. 2003, 13: 5-14. 10.1093/cercor/13.1.5.PubMedView ArticleGoogle Scholar
  3. Kisvrday ZF, Toth E, Rausch M, Eysel U: Orientation-specific relationship between populations of excitatory and inhibitory lateral connections in the visual cortex of the cat. Cerebral Cortex. 1997, 7: 605-618. 10.1093/cercor/7.7.605.View ArticleGoogle Scholar
  4. Pinto D, Ermentrout G: Spatially structured activity in synaptically coupled neuronal networks: 1. traveling fronts and pulses. SIAM J Appl Math. 2001, 62: 206-225. 10.1137/S0036139900346453.View ArticleGoogle Scholar

Copyright

© Faugeras and Grimbert; licensee BioMed Central Ltd. 2007

This article is published under license to BioMed Central Ltd.

Advertisement