René Thom

The Topologist

It is easy to meet René Thom only through catastrophe theory and miss that he was already one of the most decorated topologists alive before he wrote a word of it. In his 1954 thesis he built cobordism theory — a way of classifying smooth manifolds not by isomorphism but by whether two of them together bound a higher-dimensional manifold — and showed how to compute the resulting classification using homotopy theory, via what are now called Thom spaces and the Thom construction. It became a template for how algebraic topology is done: turn a geometric classification problem into a computation in a cohomology theory. The 1958 Fields Medal, awarded at the Edinburgh ICM alongside Klaus Roth's, cited this work explicitly. Thom was, before anything else, a mathematician whose instinct was classification: not "does this system have property X" but "what are all the ways a system in this class can look, up to the equivalence that actually matters."

Seven Ways to Fall Off a Smooth Curve

Catastrophe theory, developed through the 1960s and published as Stabilité structurelle et morphogénèse in 1972 (English translation, Structural Stability and Morphogenesis, 1975), asks the same kind of question about a different object: not manifolds, but families of smooth functions — potentials — depending on a handful of slowly varying control parameters. As the controls move, the function's critical points (its equilibria) can appear, merge, and vanish. Thom's classification theorem says that if there are at most four independent control parameters and at most two state variables, only seven qualitatively distinct types of these transitions are structurally stable — robust to small perturbation of the function itself, not just the parameters. Every other way a critical point could degenerate is a knife-edge case that a small nudge destroys.

The proof is not casual: it leans on Whitney's theory of singularities of smooth maps, the Malgrange preparation theorem, and Mather's classification of stable map-germs. Thom's own contribution was recognizing that this machinery, built for pure singularity theory, exhaustively classifies the local qualitative behavior of any smooth potential family within the stated parameter bounds — a genuine, hard, load-bearing theorem.

Zeeman, and the Controversy

Thom's own applications, in the 1972 book, stayed close to the parts of biology he knew best: embryology, morphogenesis, the qualitative shape of developmental transitions — offered tentatively, as a language for classifying the discontinuities, not as a machine for deriving their thresholds from first principles. The popularization ran further. E.C. Zeeman, over the following decade, applied catastrophe-theoretic models to an enormous range of subjects — prison riots, dog aggression, the collapse of a bridge, stock market crashes, the beating heart — with genuine mathematical craft but, increasingly, with numbers attached that the underlying theorem had no way of certifying.

The correction came from within mathematics itself. Raphael Zahler and Hector Sussmann's "Claims and Accomplishments of Applied Catastrophe Theory" (Nature, October 1977) went through the best-known applications in close technical detail and reported that they were, in their words, "characterised by incorrect reasoning, far-fetched assumptions, erroneous consequences, and exaggerated claims" — concluding that catastrophe theory had, at that point, "made no significant contributions to biology and the social sciences" beyond what better-established tools already provided. Sussmann followed with a companion piece in The Sciences, "Catastrophe Theory: Mathematics Misused." A public exchange in Nature followed — a "catastrophe controversy" — that settled, more or less, into the position this chapter states plainly: the classification theorem is real and Thom's caution was justified; a great many of the applications built on top of it were not.

Place in the Series

This corpus ran its own, small-scale version of exactly that pattern this week, and it is worth naming plainly rather than passing over. Two toy dynamical systems were built to justify the constants $\mu=-2$ and $\mu=-3$ for an autophagy switch: a pitchfork saturation $\dot x = x - x^3$, and a cusp-family potential $V(q) = q^3 - 3q$ — both, structurally, Thom normal forms, the fold and the cusp from the table above, wearing biological variable names. Both produced the target eigenvalues. Neither derivation started from a measured rate constant; both were reverse-engineered to land on a pre-chosen number, and the second one additionally used a non-standard formula to get there. The correction, worked out over the following exchange and written up as WP-31C and WP-31D, is the Zahler – Sussmann move exactly: not "catastrophe theory is wrong," but "a real classification theorem was quietly asked to deliver a numerical claim it was never built to make," followed by going and finding the actual published model, the actual missing feedback edge, and the one real rate constant ($u_0 \approx 0.01\ \text{s}^{-1}$, traced to Szymańska et al. 2015) that could be defended.

Thom's classification theorem is not diminished by any of this — if anything, this episode is a small, current demonstration of exactly the caution he built into the 1972 book and that his own popularizers too often dropped: a fold or a cusp tells you the generic shape a bifurcation is allowed to take. It was never going to tell you, for free, what a real cell's kinase does.

References