Embodiment names incompatible claims, and the prototype proves a mechanism

Pick up a wooden block and the computer knows where your hand is. Embodied interaction turns that into a claim about mind. Meaning is not stored in a head and read out through the hands, but arises from a person's skilled practice in a physical and social setting. It is also a bundle.

This tenth of ten units argues that embodiment carries claims which disagree with each other. Its second claim concerns method: the prototype offered to demonstrate the stance shows that a mechanism works, not that it benefits anyone.

The stance treats meaning as made in practice, and the body does cognitive work

Dourish (2001) gives embodied interaction its vocabulary. On his account meaning arises from engaged, skilled practice in a physical and social setting, an idea taken from phenomenology rather than from experiment. His strongest claim is relational: tangible and social interaction are two views of one phenomenon.

Klemmer, Hartmann and Takayama (2006) convert that into design guidance, synthesising five themes: thinking through doing, performance, visibility, risk and thick practice. They offer them as inspiration, not as predictions with outcome measures attached.

Kirsh and Maglio (1994) studied people playing Tetris. They found that some cognitive and perceptual problems are solved more quickly and reliably by acting in the world than by computing in the head. They distinguish pragmatic actions, which reach a goal, from epistemic actions, which uncover hidden information. A standard information-processing model cannot explain the performance data.

That is a mechanism result, not a benefit result. It shows that rotations change the problem rather than implement a plan, and says nothing about whether a physical interface beats a graphical one, because the manipulation is the task.

The strongest version pushes the boundary past the skin

Clark and Chalmers (1998) move the boundary of the cognitive system itself, arguing for active externalism, in which the environment helps drive cognitive processes. An external resource then counts as part of one.

It carries no test in the paper, and Wilson (2002) treats that step as the weak point.

The argument runs on a case rather than a measurement. Otto has Alzheimer's disease and carries a notebook in which he writes what he will need later. Inga has an ordinary memory. Both hear of an exhibition at the Museum of Modern Art and each arrives at the address, Inga from memory and Otto from the notebook. Clark and Chalmers hold that the notebook plays for Otto the role memory plays for Inga. The information in it is therefore part of Otto's cognitive system by the same standard. The case is a thought experiment, so it supplies no data about whether any real interface is used that way.

Wilson's six claims show the term is not one position

Wilson (2002) performs the disambiguation the field needed. Six claims travel under the heading of embodied cognition: that cognition is situated; that it is time-pressured; and that we off-load work onto the environment. The other three are that the environment is part of the cognitive system, that cognition is for action, and that off-line cognition is body based.

A study supporting the third claim does not thereby support the fourth. The Tetris result of Kirsh and Maglio (1994) supports the third claim, that we off-load work onto the environment. It does not support the fourth, that the environment is part of the cognitive system. A demonstration that people act to simplify a problem is not a demonstration that an interface is part of their cognition. When a brief says a design is embodied, the first useful question is which of the six claims is being made.

Two different things are called embodiment, and only one is measured

A second meaning of the word has nothing to do with meaning-making. Kilteni, Groten and Slater (2012) define the sense of embodiment as three subcomponents: self-location, agency and body ownership. They argue that immersive virtual reality allows each to be manipulated experimentally. Their contribution is to turn an experience into a construct measurable part by part.

Slater, Spanlang, Sanchez-Vives and Blanke (2010) showed that a first-person perspective on a life-sized virtual female body, substituting the male participants' own bodies, generated a body transfer illusion. The evidence was a questionnaire and heart-rate deceleration under threat to the virtual body. Owning a body does not settle how meaning is made.

Design frameworks make the stance predictive

A stance becomes useful when it predicts a mapping. Hurtienne and colleagues (2010) derive predictions from primary metaphor theory, the idea that abstract domains are understood through basic sensorimotor experience. They test 12 metaphors relating spatial gestures to abstract content. In an empirical study, 65 participants from two age groups produced touch and free-form gestures in response to keywords. In 92 per cent of cases the gesture chosen confirmed the predicted mapping.

Bakker, Antle and van den Hoven (2011) apply the same idea to a tangible learning system. They map abstract sound concepts onto bodily actions, and build MoSo Tangibles, objects children use to manipulate pitch, volume and tempo. In a user study with 39 participants, all the children could reproduce sound samples.

Both papers stop short of the claim the stance is usually asked to support. Hurtienne and colleagues test whether a mapping is natural to the body, a prediction about gesture rather than learning. Bakker and colleagues test usability, and every child completed the task, a feasibility finding. Neither compares a tangible implementation with a graphical one.

The prototype proves a mechanism, and the proof stops there

Hartmann and colleagues (2006) treat prototyping as the activity that structures design, on the view that prototypes embody hypotheses and enable designers to test them. Their toolkit d.tools integrates design, test and analysis of information appliances, keeping the record of what was tried and what happened together. They evaluated it with 13 laboratory participants and by observing seven student teams.

A working prototype can establish that the sensor distinguishes the states the design requires, and that the behaviour is legible to someone who did not build it. It can also establish that the form fits the hand. That fit is a condition rather than a universal, since a prototype held by young hands does not show it suits an older user's grip. It cannot establish that the sensing survives a year of use, that the enclosure can be made at a workable cost, or that the interaction suits the task better than a switch. The gap between those two lists is where a physical computing project can overreach.

The contested question: what does embodiment name, and what would settle it?

The literature disagrees about the status of the term itself, and the disagreement runs in three directions. The first camp takes the stance as given. Dourish (2001) and Klemmer and colleagues (2006) treat embodiment as a paradigm whose evidence is illustrative. For them the test is whether it generates better design questions, not whether it wins a comparison.

The second holds that the stance is not one position. Wilson (2002) shows that most of what is claimed is carried by the weaker claims. The strongest, that the environment is part of the cognitive system, is the one she finds deeply problematic. The term then does persuasive work, letting a modest finding about off-loading be presented as a claim about the nature of mind. No shared criterion distinguishing functional coupling from ordinary tool use is agreed.

The third disputes the stance from inside its own tradition. Svanaes (2013) re-examines Dourish's use of Merleau-Ponty's phenomenology, and argues that the implications for design were drawn too narrowly. He sets out what a phenomenology of the lived body implies for whole-body systems. Hine (2025) argues that narratives of disembodiment overlooked that users were never disembodied but anonymised, and that this narrative still shapes contemporary claims.

Whether tangibility helps would be settled by a controlled comparison of a tangible and a graphical implementation of the same task, with learning or performance pre-registered as the outcome. Whether the sense of embodiment is one thing with three parts would be settled by convergent evidence. Self-location, agency and ownership would have to move together under manipulation, while remaining separable by independent behavioural measures.

The unit's thesis has a counter-argument. It says embodiment names claims that disagree, yet one of the two senses is a settled construct: Kilteni, Groten and Slater (2012) define the sense of embodiment as three subcomponents that can be manipulated and reported separately, and Slater and colleagues (2010) manipulate them experimentally. The incompatibility therefore sits between the philosophical stance and the measurable construct, not inside either. The strongest remaining challenge is to the extended-mind claim, which Wilson (2002) judges deeply problematic because no shared criterion distinguishes functional coupling from ordinary tool use.

What a designer should do while the dispute is unresolved. State which of Wilson's six claims a design is making, and report self-location, agency and ownership separately rather than as one score. Where the extended-mind claim is invoked, state the coupling criterion and test it. Treat a working prototype as a proof of mechanism, and evaluate usability and cost separately rather than folding them into the claim.

Where this is heading

The direction of travel is away from the metaphysical version of embodiment and towards claims that can be measured and claims about whose bodies are counted. Hine (2025) states the first half, arguing that the history of the debate still shapes present claims and calling for an inclusive approach to virtual and mixed-reality design. Kilteni, Groten and Slater (2012) proposed reporting self-location, agency and ownership separately. That is the standard the next generation of headset studies will be judged against.

Sources

• Dourish, P. (2001). Where the Action Is: The Foundations of Embodied Interaction. The MIT Press. https://mitpress.mit.edu/9780262541787/where-the-action-is/ Argues that embodied interaction treats meaning as arising from engaged, skilled practice in a physical and social setting. Its claim that tangible and social interaction are two views of one phenomenon is the stance this unit examines.
• Kirsh, D. and Maglio, P. (1994). On distinguishing epistemic from pragmatic action. doi:10.1207/s15516709cog1804_1 Studies Tetris play and finds that some problems are solved more quickly and reliably by acting in the world than by computing in the head. It separates pragmatic actions, which reach a goal, from epistemic actions, which uncover hidden information. It is the empirical core of the claim that the body does cognitive work.
• Clark, A. and Chalmers, D. (1998). The extended mind. doi:10.1093/analys/58.1.7 Argues for active externalism, on which the environment helps drive cognitive processes, so an external resource counts as part of one. It is the strongest form of the claim that an interface is part of thought, and it is argued conceptually rather than tested.
• Wilson, M. (2002). Six views of embodied cognition. doi:10.3758/bf03196322 Distinguishes six claims carried by the term and assesses them separately, judging the first three and the fifth at least partially true, the fourth deeply problematic, the sixth possibly the best documented. It shows the term names incompatible positions, so it lets a reader ask which of the six any design claim is making.
• Klemmer, S. R., Hartmann, B. and Takayama, L. (2006). How bodies matter: five themes for interaction design. doi:10.1145/1142405.1142429 Draws on psychology, sociology and philosophy to synthesise five design themes, among them thinking through doing and thick practice. It converts the stance into guidance a designer can follow, themes offered as inspiration rather than predictions.
• Hartmann, B., Klemmer, S. R., Bernstein, M., Abdulla, L., Burr, B., Robinson-Mosher, A. and Gee, J. (2006). Reflective physical prototyping through integrated design, test and analysis. doi:10.1145/1166253.1166300 Presents d.tools, a toolkit integrating design, test and analysis of information appliances on the view that prototyping is thinking by doing. It was evaluated with 13 laboratory participants and by observing seven student teams, showing prototyping produces knowledge about the prototype.
• Hurtienne, J., Stossel, C., Sturm, C., Maus, A., Rotting, M., Langdon, P. and Clarkson, J. (2010). Physical gestures for abstract concepts: inclusive design with primary metaphors. doi:10.1016/j.intcom.2010.08.009 Tests 12 primary metaphors relating spatial gestures to abstract content, with 65 participants producing touch and free-form gestures. In 92 per cent of cases the chosen gesture matched the predicted mapping, so the stance yields falsifiable predictions, though what is tested is the mapping.
• Slater, M., Spanlang, B., Sanchez-Vives, M. V. and Blanke, O. (2010). First person experience of body transfer in virtual reality. doi:10.1371/journal.pone.0010564 Induced a body transfer illusion in which a first-person view of a life-sized virtual female body substituted the male participants' own bodies, measured by questionnaire and heart-rate deceleration under threat to that body. The design crossed perspective, mirror synchrony and touch synchrony as independent binary factors, so bodily experience can be manipulated experimentally.
• Bakker, S., Antle, A. N. and van den Hoven, E. (2011). Embodied metaphors in tangible interaction design. doi:10.1007/s00779-011-0410-4 Develops a people-centred, iterative method for mapping abstract concepts onto bodily actions, and applies it to MoSo Tangibles, objects children use to manipulate pitch, volume and tempo. In a user study with 39 participants all the children could reproduce sound samples, a feasibility finding rather than a comparison.
• Kilteni, K., Groten, R. and Slater, M. (2012). The sense of embodiment in virtual reality. doi:10.1162/pres_a_00124 Defines the sense of embodiment as three subcomponents, self-location, agency and body ownership, reviewing the measures and manipulations reported in the literature. It turns embodiment into a construct measurable part by part, which is the unit's alternative to arguing about the word.
• Svanaes, D. (2013). Interaction design for and with the lived body. doi:10.1145/2442106.2442114 Re-examines Dourish's use of Merleau-Ponty's phenomenology and argues that the implications for design were drawn too narrowly, setting out what the lived body implies for whole-body systems. It is a critique from inside the phenomenological tradition, which makes it harder to dismiss than an outside objection.
• Hine, E. (2025). Virtual reality, cyberspace, and embodiment: a historical debate with contemporary resonance. doi:10.1007/s10055-025-01130-3 Re-reads the history of digital embodiment in cyberspace and virtual reality, arguing that narratives of disembodiment overlooked that users were never disembodied but anonymised, usually into a white male default. It argues the dualist narrative still shapes contemporary claims, turning the debate into a question about whose bodies and which access.