Wilson-Cowan Model

Wilson-Cowan Model

Audience

Difficulty	Intermediate

Notation on this page

The Wilson-Cowan model (1972) is the foundational mean-field model of neuronal population dynamics. It describes the interaction of an excitatory (E) and an inhibitory (I) population of neurons through two coupled ordinary differential equations.

It is the simplest neural-population model that exhibits oscillations, multistability, and pattern formation — and it serves as the conceptual core of nearly all later population-level models, including Amari fields, Jansen-Rit neural masses, and the framework's ψ-coupled extension.

Statement

The Wilson-Cowan equations for two interacting populations E(t) and I(t):

$ \tau _{E}\,{\frac {dE}{dt}}=-E+(1-r_{E}\,E)\,S_{E}\!{\bigl [}\,w_{EE}\,E-w_{EI}\,I+P(t)\,{\bigr ]} $

$ \tau _{I}\,{\frac {dI}{dt}}=-I+(1-r_{I}\,I)\,S_{I}\!{\bigl [}\,w_{IE}\,E-w_{II}\,I+Q(t)\,{\bigr ]} $

Components

Symbol	Meaning	Units / range
E(t)	Excitatory population activity	dimensionless ∈ [0,1]
I(t)	Inhibitory population activity	dimensionless ∈ [0,1]
τE, τI	Population time constants	~ 5–20 ms
rE, rI	Absolute-refractory periods	~ 1 ms
wEE, wEI, wIE, wII	Synaptic-weight matrix	dimensionless
SE, SI	Sigmoid activation, e.g. S(u) = 1/(1+exp(−(u−θ)/σ))	dimensionless
P(t), Q(t)	External drives (sensory / top-down)	dimensionless

Derivation sketch

Start with a population of neurons whose firing depends on aggregated synaptic input. Coarse-grain over individual neurons:

1. Each neuron fires when its summed synaptic input exceeds a threshold; the population fraction firing is a sigmoid function S of mean input.
2. The mean input to an excitatory neuron is (positive contributions from other E neurons) minus (inhibition from I neurons) plus external drive.
3. After firing, a refractory period (1 − r·E) reduces the fraction of cells available to fire.
4. The membrane dynamics imposes a time constant τ.

Combining these gives the Wilson-Cowan form.

Dynamics

Linearise about a fixed point (E*, I*); the Jacobian determines:

• Stable fixed point (both eigenvalues with negative real part) — resting / baseline state.
• Hopf bifurcation (complex-conjugate pair crosses imaginary axis) — emergence of α/β/γ rhythmic oscillations.
• Saddle-node (one eigenvalue passes through zero) — multistability; basis for perceptual bistability and working-memory models.

Wilson-Cowan with appropriate parameters reproduces:

• α (8–13 Hz) rhythms in cortex.
• β (13–30 Hz) sensorimotor oscillations.
• γ (30–80 Hz) high-frequency synchrony.
• Slow oscillations and up/down states.

Sanity-check limits

• No coupling (w_EE= w_EI= w_IE= w_II= 0): pure exponential relaxation E(t) → 0; I(t) → 0. ✓
• Linear regime (σ → ∞ in sigmoid): linear coupled ODEs; classical damped oscillator. ✓
• r → 0: no refractoriness; population activity unconstrained above. (Limit not biologically realistic; mathematically clean.)
• Single population, w_EE > 1: bistability between low and high firing.

Connection to ψ

The framework's ψ-coupled extension adds:

• Brain → ψ source: J_ψ(x,t) = κ_J · f(u(x,t)).
• ψ → brain feedback: extra term + β · ψ(x,t) in the activation argument.
• ψ field equation: □ψ − m²ψ − λψ³ = α F_μνF^μν + J_ψ.

Wilson-Cowan provides the baseline neural dynamics; ψ coupling enters as small additive terms with parameters κ_J, β.

Experimental status

Wilson-Cowan is a pillar of mainstream computational neuroscience. It is taught in every computational-neuroscience curriculum, used as the population-level component of most large-scale brain models (TVB, Spaun, others), and validated by:

• Direct fit to electroencephalographic (EEG) and magnetoencephalographic (MEG) spectra.
• Match to local-field-potential recordings in cortex.
• Reproduction of bistability, hysteresis, and oscillatory regimes in cortical slice preparations.

The Wilson-Cowan equations are not in dispute. The framework adds ψ-coupling on top of standard Wilson-Cowan.

See Also

• Amari_Neural_Field
• Hodgkin-Huxley_Equations
• FitzHugh-Nagumo_Equations
• Jansen-Rit_Neural_Mass
• Neural_Field_Equations
• Wilson-Cowan_Coupled_to_Psi
• Effective_Field_Theory_of_Consciousness

References

• Wilson, H. R., Cowan, J. D. (1972). "Excitatory and inhibitory interactions in localized populations of model neurons." Biophysical Journal 12: 1–24.
• Wilson, H. R., Cowan, J. D. (1973). "A mathematical theory of the functional dynamics of cortical and thalamic nervous tissue." Kybernetik 13: 55–80.
• Coombes, S., Beim Graben, P., Potthast, R., Wright, J. (eds.) (2014). Neural Fields: Theory and Applications. Springer.