Population dynamics of interacting spiking neurons

Physical Review E - Statistical Physics, Plasmas, Fluids, and Related Interdisciplinary Topics

Volumn 66, Issue 5, 2002, Pages 19-

  Mattia, Maurizio ^a   Del Giudice, Paolo ^a  

^a INFN SEZIONE DI ROMA   (Italy)

Author keywords

[No Author keywords available]

Indexed keywords

BIFURCATION (MATHEMATICS); COMPUTER SIMULATION; PERTURBATION TECHNIQUES; RANDOM PROCESSES; SPURIOUS SIGNAL NOISE; TRANSFER FUNCTIONS;

DEPOLARIZATION; FINITE-SIZE EFFECTS; NEURONS; POWER SPECTRUM;

NEURAL NETWORKS;

EID: 85036133906     PISSN: 1063651X     EISSN: None     Source Type: Journal    
DOI: 10.1103/PhysRevE.66.051917     Document Type: Article

References (50)

1. A more complete historical account of the subject can be found, for example, in 7.
2. B. W. Knight, J. Gen. Physiol. 59, 734 (1972).
3. H. R. Wilson and J. D. Cowan, Biophys. J. 12, 1 (1972).
4. D. J. Amit and M. Tsodyks, Network 2, 259 (1991).
5. A. Treves, Network 4, 259 (1993).
6. L. F. Abbott and C. van Vreeswijk, Phys. Rev. E 48, 1483 (1993).
7. W. Gerstner, Phys. Rev. E 51, 738 (1995).
8. D. J. Amit and N. Brunel, Cereb. Cortex 7, 237 (1997a).
9. N. Brunel and V. Hakim, Neural Comput. 11, 1621 (1999).
10. M. Spiridon and W. Gerstner, Network 10, 257 (1999).
11. B. W. Knight, D. Manin, and L. Sirovich, in Proceedings of Symposium on Robotics and Cybernetics, Lille-France, July 9-12, edited by E. C. Gerf (1996).
12. S. Fusi and M. Mattia, Neural Comput. 11, 633 (1999).
13. N. Brunel, J. Comput. Neurosci. 8, 183 (2000).
14. M. Tsodyks and T. Sejnowsky, Network 6, 111 (1995).
15. W. Gerstner, Neural Comput. 12, 43 (2000).
16. S. Thorpe, D. Fize, and C. Marlot, Nature (London) 381, 520 (1996).
17. D. Q. Nykamp and D. Tranchina, J. Comput. Neurosci. 8, 19 (2000).
18. Conditions favoring independence are sparse connectivity, a random distribution of synaptic efficacies which can compensate for sparsness when connectivity is high or, in general, a suitable source of quenched randomness affecting the communication among the neurons.
19. H. Risken, The Fokker-Planck Equation: Methods of Solution and Applications (Springer-Verlag, Berlin, 1984).
20. We remark that in deriving the above Fokker-Planck equation in the general case (2.2) of a multiplicative noise, the condition (Formula presented) plays a crucial role, since if the latter condition is not fulfilled, a further, “noise-induced” drift appears in the A coefficient of the Fokker-Planck equation.
21. L. Sirovich, A. Omurtag, and B. W. Knight, SIAM (Soc. Ind. Appl. Math.) J. Appl. Math. 60, 2009 (2000).
22. B. W. Knight, Neural Comput. 12, 473 (2000).
23. C. van Vreeswijk and H. Sompolinsky, Science 274, 1724 (1996).
24. The ratio between the standard deviation and the mean of the interspike intervals.
25. The irrelevance of the detailed statistics of the interspike intervals for the collective dynamics of the network has been recognized in a different context in 50.
26. In Ref. 19 it is suggested (for a slightly different expansion strategy) that this is indeed the case when the terms A and B in the Fokker-Planck equation are rational functions of (Formula presented) (see 11 for an explicit example).
27. Not to be individually considered as probability currents, since the (Formula presented)’s are not individually probability density functions.
28. As noted by several authors, the above definition of (Formula presented) is free from the difficulties that arise in the definition based on taking averages of single spike trains over small time intervals.
29. Indeed, the (Formula presented) term is finite in the potentially dangerous limit (Formula presented) and the singularities possibly arising from the term in the numerator involving (Formula presented) are eliminated by the analogous term in the denominator.
30. While the stability condition (Formula presented) appears superficially consistent with general and standard arguments, it applies to a special case (see below for other examples). Besides, we argue in Sec. IV that simple arguments appropriate for a first-order dynamics lead to wrong conclusions in the case under study.
31. G. Mongillo and D. J. Amit, J. Comput. Neurosci. 11, 249 (2001).
32. A pole near one of the eigenvalues of L, driving the instability of the network for small (Formula presented) (which our analysis proves to be a special, large (Formula presented) effect) had been predicted in Ref. 22 to be a general property, on the basis of what appears as an unwarranted condition imposed on the poles near (Formula presented)
33. H. C. Tuckwell, Introduction to Theoretical Neurobiology (Cambridge University Press, Cambridge, England, 1988), Vol. 2.
34. C. A. Erickson and R. Desimone, J. Neurosci. 19, 10404 (1999).
35. We mention for completeness that in the small-(Formula presented) limit, in the drift-dominated regime the eigenvalues (therefore, in most conditions, the diffusion poles) have a real part (Formula presented) such that as (Formula presented) the diffusion poles bring about a marginal stability. See also 6 and 22 for a discussion of the eigenvalues of L in the limit of small (Formula presented)
36. For example, for the extreme case of full connectivity, with all synapses chosen as random variables drawn from the same probability distribution, the afferent currents to two generic neurons are determined by the convolution of the same realization of a stochastic point process (the common sequence of afferent spikes) with two independent realizations of the process generating the synaptic couplings.
37. D. J. Amit and N. Brunel, Network 8, 373 (1997b).
38. D. Q. Nykamp and D. Tranchina, Neural Comput. 13, 511 (2001).
39. P. Dayan and L. F. Abbott, Theoretical Neuroscience (MIT Press, Cambridge, MA, 2001).
40. W. Gerstner and W. Kistler, Spiking Neuron Models (Cambridge University Press, Cambridge, England, 2002).
41. D. J. Mar, Proc. Natl. Acad. Sci. U.S.A. 96, 10450 (1999).
42. It is clear from the above discussion that the characteristic times of the population are not directly determined by single neuron properties such as the membrane time constant.
43. D. J. Amit, Behav. Brain Sci. 18, 617 (1995).
44. P. Del Giudice and M. Mattia, in Neurocomputing–Computational Neuroscience: Trends in Research 2001, edited by J. Bower (Elsevier Science, Amsterdam, 2001), Vols. 38–40, pp. 1175–1180.
45. B. W. Knight, J. Gen. Physiol. 59, 767 (1972).
46. A. Destexhe, Z. F. Mainen, and T. J. Sejnowski, in Methods in Neuronal Modeling, edited by C. Koch and I. Segev (MIT Press, Cambridge, MA, 1998), pp. 1–25.
47. E. Haskell, D. Q. Nykamp, and D. Tranchina, Network 12, 141 (2001).
48. N. Brunel, F. S. Chance, N. Fourcaud, and L. F. Abbott, Phys. Rev. Lett. 86, 2186 (2001).
49. Previous attempts within the diffusion approximation are reported in 9. 47 takes into account the higher-order contributions in the Kramers-Moyal expansion.
50. W. Gerstner and J. L. van Hemmen, Biol. Cybern. 67, 195 (1992).