Discover how the brain anticipates movement according to Berthoz. A new perspective on perception and action that changes everything.
Published on June 2, 2026
Hello to you, the movement professional,
This week I listened again to Alain Berthoz's inaugural lecture at the Collège de France, "Physiology of Perception and Action," given in 1993. It lasts an hour, and with each listen, I come away with the same slightly dizzying sensation: that I've been looking at the problem of movement the wrong way around for years. We readily imagine that the body captures the world, the brain decides, and then the muscle executes, a neat chain in the right order. Berthoz turns this chain inside out. And when working with athletes or patients, this reversal is not just a laboratory curiosity: it shifts the very place where we intervene, and thus what we choose to train first. Here are the shifts I retain from this listening. The last one still makes me uncomfortable.
The first shift is summed up in one sentence: the brain functions primarily as a prediction machine, anticipating information long before receiving it. Berthoz puts it bluntly, "the senses are hypothesis verifiers." The sensory circuit is not primarily used to discover the world; it serves to confirm or correct a prediction the brain has already made.
Imagine a slalom skier at full speed. If he had to wait for visual information from each gate to react, the processing delay would have him down before the third one. What he does is simulate the descent in advance, then pick a few chosen signals on the track to adjust his map. You feel the body's weight shift a fraction of a second before the gate, not after. Perception becomes a constant bet between an expectation and a return. In the field, a poorly calibrated sensory sensor not only lies to you about the present: it also distorts the prediction of the next gesture. And if the prediction is wrong upstream, what is the value of the correction downstream?
Berthoz rejects the clear boundary between perception and action. He takes up Merleau-Ponty's phrase, "vision is palpation by the gaze." The gaze does not record an image; it explores, touches, questions. Perceiving is already a motor act.
The physiological proof is elegant. When the brain sends a movement order, it keeps a copy that it immediately sends to the sensory areas, the famous efference copy. As a result, the room appears still to you when you move your eyes, even though the image slides on the retina. Action informs perception before it even occurs. For us, it's a complete change of logic: we stop treating perception as an input and movement as an output. The two form a single loop, and it's this loop we seek to read, link by link, rather than each piece in isolation.
We learn five senses in school. Berthoz reminds us that one is missing, named kinesthesia by Goldscheider as early as 1898: the sense of movement itself. And its central sensor, the vestibular system of the inner ear, is a small gravito-inertial center that measures head accelerations.
Here lies a difficulty that Einstein had formalized: an acceleration sensor cannot distinguish a true acceleration from a simple effect of gravity when the head tilts. The vestibular system is therefore ambiguous by nature. To resolve this ambiguity, the brain constantly cross-references three inputs: the vestibular, vision, and somesthesia, that is, the sensation of the body. This is exactly the multi-modal reading of Pillar C in Neuro-Postural Reprogramming. When one of the three inputs becomes deaf or offset, the entire integration is disrupted along with it, and with it balance, tone, sometimes vertigo. Reading movement as a sense means accepting to calibrate it as seriously as correcting blurred vision.
If the brain predicts, then the correct movement begins before the visible movement. Berthoz demonstrates this with anticipated postural adjustments, described by Babinski and later worked on by Massion. Even before bending the trunk forward, your body slightly moves back to preemptively compensate for the fall that gravity will cause. Posture anticipates imbalance instead of suffering it.
This anticipation has a huge computational cost, and nature has found shortcuts. Bernstein spoke of reducing the number of degrees of freedom: the brain controls already wired synergies, groups of joints, rather than each segment one by one. And to choose, it inhibits as much as it commands. Berthoz has this image that I really like, "the brain is a spirited horse directed by inhibition." The freedom of movement comes from the ability not to respond to everything. Practically, postural anticipation is trained like a circuit. The question remains with which stimuli it is truly engaged.
Spatial memory, too, is a matter of movement. O'Keefe and Nadel discovered in 1978 hippocampal neurons that light up when the animal passes through a specific place, place cells. Berthoz draws an interpretation from this that should be presented for what it is, an interpretation and not a settled fact: remembering a place would mean simulating the movements needed to get there, rather than consulting a fixed map.
He relies on an intuition of Poincaré, "to locate an object means to represent the movements needed to reach it." Space is constructed by moving: a perfectly immobile being would have no notion of it. A telling detail, these cells only discharge when the animal actively explores its environment. Movement is not the backdrop of spatial memory; it is its substance. This strongly resonates with what we observe in rehabilitation: we learn a space by physically engaging with it, and the body retains what the gaze alone forgets.
An honest reading note must name its gray areas. Berthoz recounts the beautiful experiment by Melvill Jones and Gonshor in 1965: prisms invert vision, and the brain eventually reverses a reflex millions of years old, the vestibulo-ocular reflex. This adaptation fails if the cerebellum is damaged, making it a magnificent model of plasticity. But making the cerebellum the sole teacher of learning is now contested, and Berthoz himself leans towards a plasticity distributed across several structures. So I keep the demonstration, and leave the strong conclusion aside.
Second caution. Berthoz holds together two families of ideas that science has since separated: on one side internal simulation, the predictive brain; on the other Gibson's affordances, the idea that we directly perceive what the environment allows us to do. I refrain from merging these two roots into a comfortable super-model, because they are distinct frameworks that do not summarize one another. Berthoz remains valuable precisely because he inhabited both shores, at a time when physiology still dared to write "in the beginning was action."
If I had to reduce this lesson to field levers, I would retain three. Stabilization of the gaze, because the gaze guides the loop much higher than we think. Integration of the three vestibular, visual, and somesthetic systems, because it is their intersection that calibrates balance, not each input taken alone. And postural anticipation, because a body that prepares anticipates the gesture where a body that reacts lags by a fraction of a second.
Berthoz provides a framework for reading rather than a protocol, and this framework shifts attention from muscular output to the sensory preparation of the gesture. It's a shift in perspective before being a change in exercises. As long as you correct what the body produces, you always arrive too late; the lever is in what it predicts.
This is exactly the field of Neuro-Postural Reprogramming.
👉 I want to learn to read the loop upstream: labo-rnp.com/fr/pros
Alain Berthoz's inaugural lecture, "Physiology of Perception and Action" (Collège de France, 1993), is available here:
Romain LabO-RNP
