Projects

Geometry of Seeing

How the primary visual cortex fills in contours that do not exist — a four-part mathematical investigation from Petitot's V1 model to an open problem on the cut locus beyond SE(2).

By Igor Moiseev · 26 April 2026
What this project is about
Your brain constructs edges and surfaces that are not physically present in the light hitting your retina. In Petitot's 2003 neurogeometry model, one idealisation of modal completion represents oriented cortical states by a contact bundle and selects connecting curves by a sub-Riemannian or elastica-type variational principle. The oriented double cover of that bundle is the Lie group SE(2), the group of rigid motions of the plane. One pendulum governs the extremals of two related problems: the smooth completion curves are Euler's elastica, parametrised by Jacobi elliptic functions, while the free geodesics project to their cuspidal siblings. This series develops the full theory from first principles, up to Sachkov's exact cut time and the question that remains open beyond SE(2).

The Series

Part Title Key objects Status
1 The Visual Cortex as a Contact Manifold SE(2), contact structure, Kanizsa PUBLISHED
2 Euler's Elastica and Jacobi Elliptic Functions sn, cn, dn, K(k²), elastica PUBLISHED
3 Maxwell Strata: When Optimal Paths Fork discrete symmetries, first Maxwell time PUBLISHED
4 The Exact Cut Time on SE(2) — and the Open Problem Beyond It cut locus, conjugate time, conjecture PUBLISHED

Appendices — Theory Background

The five appendices below build, from first principles, the mathematics the main parts use without proof. Each is self-contained, with derivations and 2–3 interactive figures, and re-uses code from the moiseevigor/elliptic package. Read in the order A1 → A2 → A3 → A4 → A5 — or as needed from the main parts.

Part Title Builds toward Status
A1 Lie Groups, Lie Algebras, and the Exponential Map of SE(2) $\mathfrak{se}(2)$, brackets, $\exp$ PUBLISHED
A2 Distributions, Frobenius, and Contact Geometry Chow, contact, V1 horizontality PUBLISHED
A3 Calculus of Variations and the Pontryagin Maximum Principle Lie–Poisson on $\mathfrak{se}(2)^*$ PUBLISHED
A4 Jacobi Elliptic Functions, Elliptic Integrals, and the AGM $\mathrm{sn}, \mathrm{cn}, \mathrm{dn}, K(m)$ PUBLISHED
A5 The Sub-Riemannian Exponential Map of SE(2) conjugate / cut / Maxwell PUBLISHED

Mathematical Setting

The model rests on three ingredients.

The state space is an orientation lift. An idealised orientation-selective cortical state records retinal position $(x,y)$ and a preferred line orientation $[\theta]$, with $\theta\sim\theta+\pi$. The biologically natural space is therefore $\mathbb{R}^2\times\mathbb{P}^1$. For calculation the series uses its oriented double cover $\mathbb{R}^2\times S^1\cong\mathrm{SE}(2)$, where heading is remembered modulo $2\pi$; endpoint statements modulo $\pi$ are explicitly projected back to the line bundle.

The metric is sub-Riemannian. Not all directions in $\mathrm{SE}(2)$ are allowed at unit cost. An idealised cortical state at orientation $\theta$ can move cheaply along its preferred direction $(\cos\theta, \sin\theta)$ and rotate cheaply by $d\theta$, but moving transversally is forbidden outright. This defines a rank-2 distribution (a contact structure) with the Pontryagin Hamiltonian

\[H = \frac{1}{2}(p_x \cos\theta + p_y \sin\theta)^2 + \frac{1}{2} p_\theta^2.\]

One pendulum, two curve families. Pinning the forward speed turns the problem into Euler’s elastica — the same curves Euler studied in 1744 when minimising the integral of squared curvature, with $\kappa(s) = 2k\,\mathrm{cn}(s \mid k^2)$ and spatial period $T = 4K(k^2)$. The free SE(2) geodesics share the pendulum but project to cuspidal curves; their cut time is $2K(k^2)$ — half the elastica clock (Sachkov 2010–2011).

Key Results Covered

Code and Data

The interactive figures use the elliptic library — Jacobi elliptic functions and complete/incomplete integrals implemented without Maple calls, accepting tensors as input.

Core References

  1. J. Petitot (2003). "The neurogeometry of pinwheels as a sub-Riemannian contact structure." J. Physiology–Paris 97(2–3): 265–309.
  2. I. Moiseev & Yu. L. Sachkov (2010). "Maxwell strata in sub-Riemannian problem on the group of motions of a plane." ESAIM: COCV 16(2): 380–399. arXiv:0807.4731
  3. Yu. L. Sachkov (2011). "Cut locus and optimal synthesis in the sub-Riemannian problem on the group of motions of a plane." ESAIM: COCV 17(2): 293–321. arXiv:0903.0727
  4. G. Citti & A. Sarti (2006). "A cortical based model of perceptual completion in the roto-translation space." J. Math. Imaging Vision 24(3): 307–326.