Spotify / Apple Podcasts / YouTube
The boundary of the crowd is not automatically the boundary of the action.
A group of people is marching. They wear blue shirts. They move along the same route and publicly support a cause.
Then one of them attacks someone outside the march.
What happened? Did a person attack? Did a subgroup attack? Did the marchers attack? Was the attack the purpose of the march all along?
Those descriptions may refer to very different events, even if a photograph shows roughly the same people in roughly the same place. To choose among them, we need to understand the relationships between the people, their purposes, and the action.
In Who Is “We”?, we examined the tendency to claim collective achievements while disowning collective wrongdoing. This follow-up asks a more immediate question: when is an action properly described as something people did together?
First, identify the kind of connection
Ontology asks what kind of thing we are describing. A collection of people sharing a characteristic is not necessarily the same kind of entity as a team engaged in a common undertaking or an organization capable of making decisions.
People wearing blue share a visible characteristic. People standing on the same street share a location. People knowingly participating in a march for a cause share more: their conduct is connected to a public undertaking.
Even then, the connection has a scope. Agreement to march does not establish agreement to every action another marcher might perform.
Michael Bratman’s account of shared agency helps explain why proximity alone is insufficient. His planning approach examines how people’s intentions and plans support an activity they undertake together. He distinguishes such activity from people merely adjusting their movements around one another on a crowded sidewalk.[1]
For this article, we do not need to settle every philosophical dispute about shared intention. We need to ask a practical question: what connects these people to this particular action?
Scenario one: a shared march
Imagine that the blue-shirted participants have agreed to march for a cause. They have a common route, an announced purpose, and coordinated participation.
We can describe them as marching together. That does not imply that they hold identical beliefs about every issue, possess the same motives, or have authorized one another to do anything that might advance the cause.
One participant might attend from conviction; another might attend to support a friend while endorsing the stated aim. Their private motivations can differ while they knowingly participate in the same public activity.
The important point is the level of the claim. “They marched together” identifies a shared undertaking. “They agreed to attack someone” would introduce another undertaking and require additional grounds.
Scenario two: one marcher attacks
Now one marcher attacks a non-marcher. In this version of the hypothetical, nobody else directs, encourages, assists, or joins the attack. It was not part of a shared plan.
We have an individual attacker within a larger group. The person’s blue shirt identifies a connection to the march; it does not, by itself, make the attack everyone’s action.
Notice how much work the hypothetical’s additional facts do. We have stipulated the absence of other involvement. In an actual event, seeing only one person strike someone would not establish those facts. Someone else might have directed the attack or knowingly helped it happen.
Nor does a conclusion about who attacked settle every question about everyone nearby. Whether a bystander should have sought help, opposed the conduct, or taken some other reasonable step is a separate inquiry. A person’s possible failure to respond is not automatically the same action as the attack itself.
Precision therefore neither spreads responsibility indiscriminately nor declares everyone else beyond criticism. It identifies the conduct we are evaluating.
Scenario three: a subgroup joins
Change the facts. Several marchers knowingly join the attack and act together against the non-marcher.
Now there is a participating subgroup within the larger march. Its members are connected by more than blue clothing or support for the cause: they are connected through the attack.
This subgroup need not have existed before the incident. People can form a joint undertaking in the moment. Conversely, several people acting at the same time would not alone prove that they were acting together; the question remains how their conduct and intentions relate.
For our hypothetical, participation in the shared attack is stipulated. That lets us make the relevant distinction: the attacking subgroup and the full set of marchers are not necessarily the same set of people.
Others might withdraw, object, assist the person being attacked, or remain elsewhere without knowing what is happening. Those differences matter. We should describe them rather than allow the outline of the crowd to substitute for an account of each person’s involvement.
Scenario four: the attack is the shared undertaking
Change the facts again. The organizers planned the march as an attack, and the participants knowingly carry out that plan.
Now the violence is part of the organized undertaking. Calling it an organized collective attack is supported by the shared plan and the participants’ knowing, coordinated conduct.
This does not require everyone to perform the same physical act. An undertaking can involve different roles. Describing it collectively does not make those roles identical or eliminate the need to distinguish knowledge, intention, and contribution.
It also does not automatically settle whether the group is a continuing collective agent in the stronger philosophical sense. A temporary joint action and an organization capable of sustaining decisions and commitments are related but distinct ideas.
Christian List and Philip Pettit defend the possibility of suitably organized collectives acting as intentional agents in their own right. Pettit’s further discussion of recognizing group agents emphasizes the significance of their interactions and capacity for commitments.[2][3] Those accounts caution us to look at organization and relationships rather than infer a unified agent from a crowd’s appearance.
The four scenarios are variations, not a prediction that a march inevitably progresses toward violence.
| Scenario | What connects the relevant participants? | What can we describe? |
|---|---|---|
| A shared march | Knowing participation in the public undertaking | Marching together |
| An individual attack | One person acts; other involvement is expressly absent | An attack by a marcher |
| A subgroup attack | Several participants knowingly act together in the attack | An attack by a participating subgroup |
| A planned collective attack | A shared attack plan knowingly carried out | An organized collective attack, with distinguishable individual roles |
Now suppose the cause is white supremacy
The blue shirts are an arbitrary visual marker. Now give the march an explicit ideological purpose: promoting white supremacy.
That changes what the march is advocating. It does not erase the distinctions among the four scenarios.
We can condemn a racist purpose that asserts racial superiority or domination. We can evaluate people’s knowing support for that purpose through their statements and participation. We can also investigate who performed, directed, encouraged, or assisted a particular attack.
Those inquiries concern related conduct, but they are not interchangeable.
Someone could knowingly support the racist cause while opposing a particular assault. Their opposition to the assault would not make their ideological support disappear. Equally, their ideological support would not, without more, establish that they planned or participated in that assault.
Shared ideology does not establish identical responsibility for every act. Distinguishing individual roles does not make the shared racist purpose disappear.
There is another boundary to preserve. A white-supremacist organization, the participants at one of its events, and white people as a demographic category are not the same entity. The organization’s claim to speak for a race does not establish that everyone placed in that racial category has authorized it.
This follows the same discipline as the earlier discussion of historical “we.” A speaker cannot create everyone else’s participation simply by announcing that they belong.
Two opposite mistakes
The first mistake is attributing the whole event to everyone in sight. Shared clothing, location, or category membership becomes a shortcut around the questions of knowledge and participation.
The second is reducing the event to the person who made physical contact. That can overlook those whose directions, encouragement, or assistance were part of the undertaking.
Both mistakes lose relationships. The first invents a connection that has not been established. The second ignores connections that may matter.
“I did not throw a punch” answers one question about a person’s conduct. It does not answer every question about their role. “I was at the march” establishes presence and perhaps participation in the march, depending on the facts. It does not by itself establish participation in the attack.
Our descriptions should be specific enough to accommodate both points.
What evidence would change the description?
In a hypothetical, we can state everyone’s intentions. In an actual event, intentions and relationships need to be established through evidence.
Relevant questions include:
- What purpose was announced, and what did the participants understand?
- Was the attack planned, encouraged, rejected, or improvised?
- Who coordinated with whom before and during the incident?
- What did each person contribute, and what did they know about the undertaking?
- Did anyone oppose the attack, withdraw, or assist the person being attacked?
- What evidence supports the claimed connection between organizers and participants?
A later statement of approval can establish approval at that later time. It does not, by itself, prove prior planning. A later denial does not, by itself, erase earlier involvement. Both should be considered alongside the rest of the evidence.
Similarly, silence can be ambiguous. Someone might be unaware, afraid, supportive, or occupied with something else. Identical outward behavior can have different explanations. We should distinguish an inference supported by evidence from an assumption that merely completes a preferred story.
Why the labels can wait
Terms such as “lone wolf,” “mob,” “gang,” or “organized crime” can bring additional assumptions into the discussion before we have established the underlying relationships.
“Lone wolf,” for example, can suggest independence beyond the observation that one person physically attacked. “Mob” can obscure the distinction between the attacking subgroup and everyone nearby. Legal classifications require their own definitions and factual analysis.
For this introduction, the narrower descriptions do more work: an individual attacker, a participating subgroup, and an organized collective attack. The article addresses philosophical attribution, not a test of criminal liability.
What makes the action ours?
For Intend to Relate, this is a question about the scope of relationships and agreements.
What did people undertake together? What did they knowingly support? What authority, if any, did they give someone to speak or act for them? Did the undertaking change, and did they knowingly participate in that change?
Agreeing to one activity is not a blank authorization for another. Shared purpose also need not mean agreement about every method. A broad cause can connect people while leaving their relationship to a particular act unsettled.
That is why the unit of analysis matters. We may be examining one person’s conduct, a subgroup’s joint action, an organization’s decision, or a broader movement’s public message. The same event can raise questions at several of these levels without making them interchangeable.
Accurate attribution lets us name racism clearly, describe collective violence when the facts support it, and preserve distinctions among individual roles. It asks us to establish the connection rather than let a shirt, a crowd, or a pronoun establish it for us.
The boundary of the crowd is not automatically the boundary of the action.
Ask who did what, who acted together, and what they knowingly supported.
Sources and further reading
The marching scenarios are hypothetical. Their application to racist ideology and the practical questions above are this article’s analysis. The sources introduce relevant philosophical approaches; they do not establish a single uncontested account of collective agency or responsibility.
- Michael E. Bratman, “A Planning Theory of Acting Together,” Journal of the American Philosophical Association 8, no. 3 (2022): 391–398. Discusses shared intentional activity through interconnected planning and distinguishes it from merely navigating alongside other people.
- Christian List and Philip Pettit, Group Agency: The Possibility, Design, and Status of Corporate Agents (Oxford University Press, 2011). See List’s author summary of the argument for suitably organized collectives as intentional agents.
- Philip Pettit, “How to Tell if a Group Is an Agent,” in Jennifer Lackey, ed., Essays in Collective Epistemology (Oxford University Press, 2014), 97–121. Author-affiliated repository abstract and publication record. The brief discussion here draws on the published abstract.
For broader orientation, see the Stanford Encyclopedia of Philosophy entries on Collective Intentionality, Shared Agency, and Collective Responsibility.
Part of Philosophical and Theoretical Foundations. Explore the project at Intend2Relate.com and the Intend to Relate publication.
Member discussion: