In 1998 Jana Iverson and Susan Goldin-Meadow videotaped twelve people who had been blind from birth, aged nine to eighteen, while they talked through a set of reasoning problems. All twelve gestured. They gestured at about the same rate as sighted speakers of the same age, with the same range of hand movements, and none of them had ever seen anyone do it. Iverson and Goldin-Meadow then sat four younger blind children down with a blind experimenter, a listener who could not see a single movement. The children gestured anyway, at about the same rate as sighted children talking to sighted listeners. Whatever the hands were doing, it could not have been only for the listener.
What embodied cognition is
Embodied cognition is the view that thinking is shaped by, and partly carried out through, the body and its dealings with the world, rather than being computation that happens in the brain and uses the body only to take things in and send instructions out. In 1991 Francisco Varela, Evan Thompson and Eleanor Rosch set out an early version of it in The Embodied Mind, drawing on Maurice Merleau-Ponty, who had argued in 1945 that perception is something a body does rather than something a mind does from inside one (you don’t think with your brain). The name now covers a family of claims that differ a great deal in strength.
The weak version says the body feeds into thought. Posture, movement and the signals coming up from the gut and heart (interoception) shape judgement the way lighting shapes what you see. Almost nobody disputes it. The strong version says the body is part of the machinery, so that understanding a word like kick draws on the systems that run kicking, and some thinking can’t happen without the movement that carries it. The strongest version, usually called the extended mind, says the machinery doesn’t stop at the skin. Cognitive scientists bundle these as 4E cognition: embodied, embedded, enacted and extended.
Your hands carry part of the thought
Gesture is where the strong claim has held up best. In 2001 Goldin-Meadow and colleagues asked children and adults to hold a list of letters or words in memory while explaining how they had solved a maths problem. When they were free to gesture during the explanation, they remembered more of the list than when they kept their hands still. The gesture was taking some of the weight of the explanation, and the memory task got the spare capacity.
Susan Wagner Cook, Terina Yip and Goldin-Meadow later tested whether any movement would do. Speakers explained maths solutions while gesturing, while making rhythmic movements that meant nothing or while keeping still. Only gesture freed up memory, and meaningless movement did no better than none. The benefit came from the hands moving in step with what the speech meant. Part of the explanation was being worked out in the hands.
It does not hold for every kind of thinking. When Acacia Overoye and Margaret Wilson ran the same design on verbal analogies in 2020, people remembered more when they were stopped from gesturing. Gesture seems to help with thinking that has a shape to it, quantities and positions and movement. On a word puzzle it may be one more thing to coordinate.
The body hacks that fell over
Most people meet embodied cognition through a different set of findings, the ones that promised a lever. Hold a pen in your teeth and the cartoons get funnier. Stand like a superhero for a couple of minutes and your hormones shift. Those results went into textbooks and TED talks, and they are the ones that came apart under replication.
The pen study came from Fritz Strack, Leonard Martin and Sabine Stepper in 1988. People held a pen between their teeth, which pulls the face toward a smile, or between their lips, which blocks one, and rated cartoons. The teeth group rated them funnier. In 2016 seventeen labs ran the study again on a shared protocol, with 1,894 participants between them. The original difference had been 0.82 points on a ten-point scale. The pooled difference was 0.03. Strack pointed out that the replicating labs had filmed their participants, and in 2018 a team at the Hebrew University of Jerusalem found that a camera in the room was enough to remove the effect. If that is right, the pen works only while nobody is watching, which is an awkward property for a classroom demonstration.
In 2022 the Many Smiles Collaboration, set up deliberately as an adversarial project with researchers from both sides of the argument, tested 3,878 people in 19 countries. Copying a smiling face and smiling on purpose both lifted feelings of happiness. The pen, the version that worked without people knowing what their faces were doing, gave no clear result. When people know what their faces are doing, the face feeds into feeling. The version that slips the feeling in unnoticed has not held up.
The 2010 study that launched power posing had 42 participants. In 2015 Eva Ranehill and colleagues ran it with 200 and held the poses for three minutes instead of one. People in the expansive poses said they felt more powerful. Their hormones and their appetite for risk did not move.
In 2002 Arthur Glenberg and Michael Kaschak reported that people judged a sentence like Close the drawer faster when their response meant moving the hand away from the body, the same direction as the sentence. The action-sentence compatibility effect became one of the most cited results in the field. Eighteen labs later ran it under a preregistered protocol, with Glenberg and Kaschak among the authors, and none of the eighteen found a reliable effect. The pooled estimate was essentially zero.
The two that collapsed loudest were the two that had been turned into techniques: do this with your body and get that in your mind. What held up, gesture and mimicry, describes a body that is already part of the thinking, which is much harder to sell as a two-minute routine before a job interview.
Is the body doing the thinking or feeding it?
Even where the results are solid, what they mean is disputed. In 2008 Bradford Mahon and Alfonso Caramazza accepted that sensory and motor systems become active when people process concepts, and disputed what followed from it. Activation spreading outward from an abstract concept would produce the same results, they argued, and the strong version did not fit the evidence from patients with brain damage. Their middle position, which they called grounding by interaction, keeps concepts abstract and lets the body colour them as they are used.
Stephen Goldinger and colleagues went further in 2016, in a paper titled The Poverty of Embodied Cognition. Its principles, they argued, were either too vague to test or things cognitive science already knew. The gesture studies answer part of that charge. They don’t settle whether the hands are part of the thought or a very good notepad for it.
Past the skin
In 1998 Andy Clark and David Chalmers opened a short paper in Analysis by asking where the mind stops and the rest of the world begins. Their answer used two people walking to the Museum of Modern Art. Inga remembers that it is on 53rd Street. Otto has Alzheimer’s disease and carries a notebook with the things he needs written in it, and he looks it up and goes. If the notebook does for Otto what memory does for Inga, they argued, then the notebook is part of Otto’s memory, not a tool his memory uses.
Kim Sterelny, an Australian philosopher who had spent the previous decade dividing his time between the Australian National University and Victoria University of Wellington, accepted most of the argument and resisted the conclusion. In Minds: extended or scaffolded? he placed Otto’s notebook at one end of a much wider range of what he called scaffolding, the tools, records and other people that humans build around their thinking. A resource looks like part of a mind when it is reliable, fitted to one person and used by that person alone. The resources that matter most, on his account, are shared and handed from one generation to the next, which is the opposite of a private notebook.
For more than a decade John Sutton, Amanda Barnier, Celia Harris and colleagues in Sydney, at Macquarie University and later Western Sydney University, have asked long-married older couples to remember things, first separately and then together. The usual laboratory finding, called collaborative inhibition, is that a group remembers less than its members would have managed separately. The couples did not all follow it. In a 2011 study of twelve couples, some lost out by collaborating and some did better, and the difference tracked the strategies the pair used. In 2018, with 39 couples, the pairs recalled more together than apart, most clearly when listing the countries of Europe and the names of their mutual friends. The couples who collaborated well were still doing it two years later. When Harris and colleagues asked 58 couples how they made up for failing memory, the men reported leaning on their spouses more than the women did, and reliance on a spouse rose with age.
By Clark and Chalmers’ test a long-married spouse is a good candidate for part of a mind: reliable, fitted to one person, consulted without being checked.
When someone else built the notebook
If the notebook is part of the mind, so is whatever else is doing the remembering and the choosing, and not all of it was built by the person using it. Most writing on scaffolding is about what outside structure adds, and in 2023 Ryan Timms and David Spurrett, philosophers at the University of KwaZulu-Natal, set out the other case. They called it hostile scaffolding: outside structure that a person thinks with, which serves someone else’s interests at that person’s cost. Their worked example is the electronic gambling machine, built to produce near misses that keep a session going, inside casino systems that track players closely enough to predict when they are about to stop.
The phone that holds your calendar, your photographs and the shape of your week passes Sterelny’s test for part of a mind as well as Otto’s notebook does, and it was built by people with their own reasons for building it (mistaking the algorithm for your taste).
How much of it holds
Some of embodied cognition is settled. People think partly with their hands: gesture that matches the meaning of speech frees memory for other work, and people blind from birth gesture to listeners who cannot see them. Smiling on purpose and copying a smile lift mood, measurably, across 19 countries. The pen in the teeth did not survive large replications. Power posing changed what people said about themselves and nothing else that was measured. The action-sentence effect, which the field had treated as its best laboratory evidence, came out at essentially zero across eighteen labs. Long-married couples, in the Macquarie studies, remember some things better as a pair than either does alone.
Whether any of this makes the hands or the spouse or the phone part of the mind, rather than something the mind leans on very hard, is the question Clark and Chalmers asked in 1998, and the experiments have not answered it. The couples studies measured each pair while both people were alive to sit the test, and none of them follows the remembering after one of the two has gone.
FAQ
What is embodied cognition?
Embodied cognition is the view that thinking is shaped by, and partly carried out through, the body and its dealings with the world, rather than being computation that happens in the brain and uses the body only for input and output. It covers a family of claims of different strength, from the uncontroversial idea that bodily states influence judgement to the claim that some thinking is partly done by movement or by things outside the body.
What are the 4Es of cognition?
4E cognition is a label for four related claims: that cognition is embodied (shaped by the body), embedded (dependent on its surroundings), enacted (brought about through action) and extended (partly carried by things outside the body, such as notebooks or devices). Researchers accept the four claims to very different degrees.
Has embodied cognition been debunked?
Parts of it have failed to replicate and parts have held up. Large preregistered replications found no effect for holding a pen in the teeth to make cartoons funnier, no hormonal or behavioural effect of power posing and no action-sentence compatibility effect across 18 labs. Gesture freeing working memory and deliberate smiling or facial mimicry lifting mood have held up. Whether the body constitutes thought or only influences it is still argued.
What is the extended mind theory?
It is the claim, made by Andy Clark and David Chalmers in 1998, that if something outside the body does the job a mental process would do inside the head, it is part of that mental process. Their example was Otto, who has Alzheimer’s disease and uses a notebook in place of biological memory. Kim Sterelny and others accept that minds lean heavily on outside resources but argue that Sterelny’s term, scaffolding, fits most of them better than extension.
If this is the kind of thinking that doesn't leave when the tab closes.
Subscribe