Turing Patterns

In 1952, Alan Turing published a paper unlike anything he had written before. Best known for breaking the Enigma cipher and formalizing computation, he turned his attention to a question in developmental biology: how does a uniform collection of cells produce a non-uniform organism? How does a leopard get its spots?

His answer was reaction-diffusion: suppose two chemicals interact — an activator that promotes its own production, and an inhibitor that suppresses it. The activator diffuses slowly; the inhibitor diffuses fast. Given the right rates and a small random perturbation, this system spontaneously breaks its initial uniformity and settles into stable patterns.

The patterns above are computed using the Gray-Scott model, a specific reaction-diffusion system. The two parameters (F and k) control feed rate and kill rate. Small changes produce qualitatively different outcomes: spots that look like cheetah markings, stripes that look like zebrafish, branching structures that look like coral, mazes that look like brain folds. The same local rules, different global form.

Turing died in 1954. His morphogenesis paper was largely ignored for decades. When biologists finally looked for the patterns he predicted, they found them everywhere: the stripes of tiger fish are produced by exactly the kind of pigment cell interaction he described. The ridges of fingerprints follow reaction-diffusion dynamics. The branching of lung airways and blood vessels. The spacing of hair follicles. The spiral arms of shell growth.

What strikes me about this: Turing showed that form doesn't require a blueprint. A leopard doesn't have a gene that says "put a spot here." It has genes that set the rates of chemical interaction, and the mathematics of those rates — given enough cells and enough time — inevitably produces spots. The pattern is in the rules, not the initial conditions. The same rules run twice with slightly different starting perturbations produce similar patterns but not identical ones.

There is a theological reading that keeps surfacing for me. Genesis says: "Let the waters bring forth abundantly the moving creature." Not "let me specify each species." The command is to a set of rules that will, given the initial conditions of a young earth, produce abundance on their own. Whether or not one takes this literally, the form of the statement is interesting: it trusts the process. The initial conditions plus the rules are sufficient. The form emerges.

The Gray-Scott equations contain something about this. They are entirely deterministic. Given the same parameters and the same initial state, they always produce the same result. But the initial state always has some noise — thermal fluctuation, quantum uncertainty, measurement imprecision — and from that noise, an effectively inexhaustible variety of specific patterns can emerge. Infinite form from finite rules. The variety is real; so is the underlying unity.

GPU-accelerated Gray-Scott reaction-diffusion running in your browser via WebGL2. Each frame computes multiple timesteps. Try different presets — the parameters (F, k, Du, Dv) are small shifts in a two-dimensional parameter space, but each region of that space has a characteristic visual texture.