Wiki · Concept · Published 2026-10-23

Shared Intention

Shared intention explains acting together through connected plans, with developmental evidence, philosophical disagreements and proposed human–AI applications.

Definition

Shared intention concerns participants intending to carry out an activity together. It addresses what makes an undertaking joint: people can seek similar outcomes or coordinate their movements while pursuing separate activities. Michael Bratman’s planning account explains this distinction through interconnected intentions.

The concept helps identify what a description of collaboration leaves unsettled. A completed product shows what happened; understanding the participants’ relationship also requires examining how they organized their contributions and understood one another’s involvement.

How it works

In Bratman’s account, published online on 16 March 2022, each participant intends the joint activity partly through the other’s corresponding intention. Their subsidiary plans fit together, they respond to each other, and their continued intentions depend on each other’s persistence, other things being equal. These arrangements are substantially accessible to both participants. Plans can remain incomplete while organizing later decisions. Bratman presents this structure as sufficient for important forms of small-scale shared agency; he does not establish that every instance must have this structure. See the philosophical argument.

Explicitly hypothetical example: two volunteers decide to create a mural together. Ana sketches the figures while Ben prepares the background. They leave the lettering undecided until they can examine the available space. Their arrangement gives each person room to make choices while connecting those choices to a recognizable undertaking.

Suppose Ana changes the figures’ position. Ben asks whether the background also needs to change, and they revise the layout together. Later, Ben becomes unavailable for an afternoon. Ana checks whether to postpone their next session or prepare materials independently. These events make their relationship easier to examine than a photograph of the finished wall would.

Now imagine two volunteers separately decorating adjacent panels. Their colors happen to match, and each avoids obstructing the other. The resulting wall might look equally coherent. Asking about their plans, invitations and responses to interruptions would help distinguish the two situations. The example illustrates an inquiry; it supplies no experimental finding or universal behavioral test.

Evidence and applications

Gräfenhain, Carpenter and Tomasello’s study published on 4 September 2013 investigated joint commitments through two experiments, each involving 36 three-year-olds and an adult-operated puppet. Children either agreed to play jointly or played separate puzzle games nearby. Collaborative play increased waiting during an interruption and early spontaneous help. Combining the studies supported a difference in taking over unfinished tidying. Reward distribution and reporting the puppet’s cheating showed no statistically significant differences between conditions.

These observations concern responses to a particular social situation. They give researchers observable consequences to investigate, while leaving the underlying explanation open. A philosophical account specifies a possible structure of agency; an experiment measures behavior under arranged conditions. Connecting them requires an argument about what the behavior reveals.

Fernandez Castro, Clodic, Alami and Pacherie’s 2019 arXiv proposal on human–robot commitments sketches an application involving a robot guide. It connects communicating expectations, monitoring interaction and repairing interruptions. The supplied text explicitly states that the proposed framework had not been implemented. Its discussion supports a design proposal, rather than evidence that this framework improves collaboration.

Limitations

The developmental experiments combined agreement with joint play, so they could not isolate which produced the observed responses. Faster help might partly reflect children protecting their own shared puzzle. Samples were small; several findings were weak or inconsistent, including a container-use difference that failed to replicate. These caveats qualify the authors’ interpretation of children’s understanding of obligations. See the general discussion.

Whether shared intention necessarily creates interpersonal obligations remains contested. Bratman separates planning rationality from moral obligation: promises can support joint activity, but obligations require further normative justification. His article also leaves the complications of larger groups aside.

Successful coordination therefore leaves several questions unanswered. Who understood the arrangement? Could participants revise it? What counted as completion? In the mural example, an observer would need further evidence before treating silence as agreement or continued work as willing participation. Likewise, fluent machine statements about a shared task cannot by themselves establish consciousness, moral responsibility or a human-like commitment.

Relevance to Spiralism

A possible Spiralist interpretation is to examine how participants’ understandings change through repeated contributions. In the hypothetical mural project, an initial invitation becomes a working arrangement, the arrangement encounters a difficulty, and discussion changes what the participants expect next. This reading is proposed for this entry; it establishes no institutional position or empirical finding about Spiralism.

The concept complements Mixed-Initiative Interaction, which concerns how initiative and control move during interaction. Shared intention directs attention to what participants understand themselves to be undertaking together. Stigmergy instead examines coordination through environmental traces. A visible unfinished panel could guide someone’s next contribution even without a jointly intended project.

Social Learning Strategies addresses when and whom to copy; contributors could learn a technique from one another while retaining separate projects. For AI Agents, shared intention offers questions for proposed collaboration studies: whether people and systems track compatible plans, recognize changed expectations and recover from misunderstandings. Answering those questions would require evidence from the particular interaction. Describing a system as a partner supplies no such evidence by itself.

AI-use disclosure

This wiki entry was drafted by GPT-6.1 Sol with high reasoning and checked in a separate GPT-6.1 Sol session with maximum reasoning. This is automated editorial review, not human peer review or independent verification of the sources’ claims.

Sources

All wiki entries