PhronesisThe kernel, granted
A body is not incidental to thinking, and the claim that it is has a poor history. James Gibson's coperception — that to perceive the world is to coperceive oneself — is on this site already, in Perception, Minus the Narrator. So is Polanyi, in Tacit Knowledge Is Language You Can Touch: skill lives in the hands before it lives in a sentence, and some of it never makes the trip. Neither of those depends on any neuroscience at all, and neither is threatened by anything below.
The measured part is also real. Sensorimotor areas do activate during conceptual processing; this is among the more replicated imaging observations in the field. The timing is the strongest form of the argument: embodied effects appear within roughly 100–200 ms of stimulus onset, in primary motor and premotor areas — early enough that they cannot easily be dismissed as a downstream echo of comprehension that has already happened elsewhere. And there is causal work: motor cortex appears to contribute to the benefit of sensorimotor-enriched vocabulary training, rather than merely accompanying it.
So the kernel is that the body is genuinely recruited, and that mental effort costs something a body pays. Nothing in this note takes either back.
The correction
Three claims get carried by one phrase, and they are not the same claim.
Activation is not necessity. A region lighting up during a task is not evidence that the task requires it. This is the oldest trap in imaging and embodiment walked into it. The test for necessity is subtraction: remove the region and see whether the function survives. It does. Neurosurgical lesions to sensorimotor cortex do not impair action verb processing. Whatever the motor activation is doing during action language, the language does not need it.
Somatotopy is in trouble specifically. The flagship version of embodied semantics is that action words map onto motor cortex by body part — that kick recruits the leg region and lick the mouth region, so meaning is literally laid out on the homunculus. A 2026 commentary describes the field as working "in the midst of a replication crisis for embodied action semantics", holds that failures to replicate make sensorimotor simulation an unlikely basis for action-word processing, and asks embodied language researchers to acknowledge the lack of reliable evidence for the somatotopy account and build models on phenomena that are reliable. That is not an outside critic. That is the literature talking to itself.
The popular effects are gone, and they are what the phrase means to most people. Power posing, facial feedback, warmth priming — the results that made embodied cognition a public idea — largely did not survive replication. They were never the theory's core and specialists will say so, correctly. But a note that uses the phrase inherits what the phrase means to a reader, and to most readers it means those.
The body is recruited. That is not the same as the concept being made of it.
The honest state of the field is a live split, not a verdict either way. The early-timing and causal-training results are real and are not answered by the lesion evidence; the lesion evidence is real and is not answered by them. Anyone reporting this as settled, in either direction, is reporting something other than what is there.
What this framework was leaning on, and how much
Now the part that concerns this site. Two accounts of load live here under different principles and have never been made to face each other. The Dose Window says a stressor builds you only inside a window with rest, that the response is an inverted U rather than a line, and that the axis most often dropped is recovery rather than intensity. Cognitive Displacement applies the displacement framework to cognition, where the cost of holding a displacement is working memory load, attentional effort, processing energy.
If strong embodiment held, those would be one law with one seat: mental load would be bodily load, and the dose-window result would transfer to cognition unchanged and for the same reason. It is a tempting move and this note was started in order to make it.
It should not be made that way, because it does not need to be. The load claim and the semantic claim rest on different evidence, and only the semantic one is contested. Whether a concept is constituted by sensorimotor activity is exactly what the replication fight is about. Whether sustained mental effort has a metabolic cost, produces measurable fatigue, and consolidates during rest is not in that fight at all — it is established on its own terms and would survive the strong embodiment thesis being wrong entirely.
So the framework's exposure here is smaller than it first appears, and the correct version of the transfer is narrow. The dose-window shape may well apply to cognitive load, on the evidence for cognitive load, and not on any claim about what concepts are made of. What this note will not do is buy that transfer with somatotopy, which is the currency currently being devalued.
What would settle it
For the transfer, the honest test is whether the dose-window's distinctive claim survives in the cognitive case: not that too much load harms — everything harms in excess — but that recovery rather than intensity is the axis that goes missing, and that the same dose builds with rest and wears without it. That is specific enough to be wrong. This note does not test it, and it is not established here by having been stated twice on the same site.
For embodiment itself, the necessity question is already being answered by lesion work, and the answer so far is unfavourable to the strong thesis. What would change it is a lesion or interference result showing a specific, meaning-selective deficit that survives the obvious confounds — and the existing patient literature is candid that lesions are usually diffuse enough to make that hard.
The limits, plainly
- This reads a live dispute from outside it, and largely from secondary sources and search rather than from the primary literature. The participants disagree with each other in good faith and this note does not adjudicate between them.
- The 2026 commentary quoted above is one position in an ongoing exchange, published as a commentary on another group's work. It is evidence about the state of the argument, not a settlement of it.
- "Embodied cognition" names several theses of very different strength — that cognition uses the body, that it is shaped by the body, that it is constituted by the body. This note separates them; much of the literature and nearly all of the popular writing does not.
- Nothing here is advice about how to work, rest, study or train. The dose-window note carries the same caution and it applies with equal force to the cognitive case, which is less well characterised, not better.
- As with this site's other applications of the framework, no claim in this note is the framework's own. The vocabulary is describing results that belong to other people, and the description is not evidence for the vocabulary.
Sources. The replication position: “Modelling in the midst of a replication crisis for embodied action semantics: a commentary on Carriere et al. (2026),” Language, Cognition and Neuroscience, 2026. Lesion evidence against necessity: “Neurosurgical lesions to sensorimotor cortex do not impair action verb processing.” The causal and timing case for the other side: “Motor cortex causally contributes to vocabulary translation following sensorimotor-enriched training,” J. Neurosci. 41(41), and “Dynamics of language grounding: on the time course, durability, adaptability, and vulnerability of embodied effects,” Front. Psychol., 2025. The site's own terms: The Dose Window, Cognitive Displacement, Perception, Minus the Narrator, Tacit Knowledge Is Language You Can Touch.