This Day in Philosophy — 8 September: Helmholtz, and the discovery that perception is inference

Hermann von Helmholtz died at Charlottenburg on 8 September 1894. Physiologist, physicist, and the most consequential philosopher of perception of the nineteenth century — a combination that would now require three careers.

Measuring the speed of thought#

In 1850, aged twenty-nine, Helmholtz did something his own teacher had declared impossible. Johannes Müller had held that nerve conduction was effectively instantaneous, or at any rate too fast to measure.

Helmholtz stimulated a frog's motor nerve at two points at different distances from the muscle and timed the difference. He obtained a speed of roughly 27 metres per second — slower than sound. He then measured reaction times in human beings.

The result is easy to state and hard to absorb. Nervous transmission takes time. Therefore the experience you are having is not of the present moment but of a moment slightly past, reconstructed. Mind is not outside the physical order looking in; it is an event in that order, with a measurable duration.

Unconscious inference#

Helmholtz spent the following decades on the physiology of vision and hearing — the three-volume Handbook of Physiological Optics, and On the Sensations of Tone, which grounded musical consonance in the physics of overtones.

Out of that work came his philosophical thesis: unconscious inference (unbewusster Schluss).

The problem that forces it is straightforward once stated. What arrives at the retina is a two-dimensional, inverted, constantly shifting distribution of light. What you see is a stable three-dimensional world of objects at determinate distances. The second is not contained in the first. Something must supply the difference.

Helmholtz's answer: perception is a conclusion, drawn from the sensory premises together with everything past experience has taught about how such premises are usually produced. The conclusion is reached without awareness, arrives already formed, and cannot be undone by knowing better — which is why an optical illusion persists after it has been explained to you.

He paired this with a sign theory of sensation. A sensation is not a copy or picture of its cause. It is a sign of it. The relation between the two need be no more resembling than the relation between a word and what it means — the same broken line that Ogden and Richards would later draw. What sensations preserve is not likeness but lawful correlation: same cause, same sign.

The quarrel with Kant#

Helmholtz took himself to be completing Kant by empirical means, and the claim was contested in his own lifetime.

He agreed with Kant that the mind is not a passive receiver, and that the perceived world is constituted rather than copied. But he turned several of Kant's a priori forms into empirical questions. Where Kant held Euclidean geometry to be the necessary form of outer intuition, Helmholtz — writing after Riemann — argued that which geometry holds of physical space is a matter for measurement, and constructed thought experiments about beings inhabiting a curved world to show that such experience is imaginable.

That move opened the road to the geometrical conventionalism of Poincaré and, eventually, to a physics in which the geometry of space is a physical variable.

The neo-Kantian movement that dominated German philosophy in the following decades — and against which the young Heidegger later defined himself — took Helmholtz as a principal point of departure.

Why it reads as current#

The Helmholtzian picture has returned, and not as an antiquarian revival.

Contemporary predictive processing theories of the brain hold that perception is the brain's best hypothesis about the causes of its sensory input, continuously corrected by error signals. That is unconscious inference in the vocabulary of Bayesian statistics; the literature says so explicitly, and Helmholtz is cited by name.

Andy Clark, whose work on the extended mind and on prediction this site covers in its directory, belongs to that lineage. So does the question of what a perceptual system has to do before it can be said to encounter a world at all — the question underneath current arguments about whether an artificial system that processes inputs is thereby perceiving anything.

Helmholtz's answer would have been that the question is not settled by the presence of inputs, but by whether the system is drawing conclusions it cannot inspect, and being corrected by the world when they fail.

Also on this date#

August Wilhelm Schlegel was born in Hanover on 8 September 1767. With his brother Friedrich he edited the Athenaeum, the journal of the Jena Romantics, and gave Romantic criticism its central idea: that a work of art is to be judged by the law it sets for itself rather than by rules imported from outside it. His Shakespeare translations — the Schlegel–Tieck edition — are the reason Shakespeare is a German classic. He held the first chair of Sanskrit in Germany, at Bonn, and his edition of the Bhagavad Gita opened Indian philosophy to European readers, Schopenhauer among them.

Frédéric Ozanam died on 8 September 1853. A professor of literature at the Sorbonne who founded the Society of Saint Vincent de Paul at twenty, he argued for what he called the passage "to the barbarians" — that Catholic thought should side with the industrial poor rather than with the established order. The tradition of Catholic social teaching, and of Maritain and Mounier after him, runs through that decision.

---

Each day in this column we take up an event from the history of philosophy that falls on today's date.