Collective responsibility
A Socratic walk-through of collective responsibility — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why can a crowd be blamed for a harm that no single person in it did enough to be blamed for?
A company poisons a river. The board voted for a cost-cutting programme, the plant manager applied it, the engineer flagged a concern in a memo nobody read, the operator followed the procedure he was given. Go round the room afterwards and every individual has a defence that is more or less good. Go and stand by the river, and something is plainly to blame.
The instinct is to say the accounting must be wrong — that if you looked harder you would find the person whose share was large enough. Sometimes that is true. The interesting cases are the ones where it is not, and no individual share reaches the threshold. Can blame belong to a group when it does not distribute to its members?
Reasoning it through
REASONING #Start with the sceptical position, because it is the strong one. H. D. Lewis argued in 1948 that responsibility is individual, full stop — value belongs to persons, and collective guilt is a barbarous idea whose historical record speaks for itself. Anyone tempted to wave that away should remember what group blame has been used for. So the burden is on the other side: what would a group have to be, for blame to attach to it and not merely to the people in it?
The observation that makes the case is technical. Christian List and Philip Pettit built their account of group agency on the discursive dilemma. Picture a three-member panel deciding whether a firm is liable, which requires both that a duty existed and that it was breached. Two members think the duty existed; two think it was breached; no single member thinks both. Vote by premise and the panel is committed to liability. Vote on the conclusion and it rejects liability. Both are honest readings of the same three sets of views.
Sit with that. The group's position cannot be read off its members' positions, because there is more than one way to aggregate and the results conflict. An organisation wanting a consistent position over time therefore has to adopt a procedure and be bound by its output — including outputs no member endorses. Once something has commitments of its own, revisable in the light of reasons, the ordinary grounds for treating an entity as answerable are in place. This is why Peter French argued that a corporation with an internal decision structure — a way of turning individual acts into corporate intentions — is a moral agent rather than a metaphor for one.
That gets us the organised case. What about the crowd, which has no procedure at all? Virginia Held asked in 1970 whether a random collection of individuals can be morally responsible, and gave a careful yes for a narrow case: where the group could have organised itself and what needed doing was obvious, the failure to constitute a decision procedure is itself the fault. Nobody in the crowd around a drowning swimmer did the wrong thing individually, and the crowd failed anyway — not for lacking a leader, but for not becoming the kind of thing that could act.
The analogy
THE ANALOGY #Think of a rowing eight losing a race. No single rower can be shown to have cost the boat the margin, and the crew's failure is not the sum of eight individual failures — it is a failure of timing, which exists only at the level of the boat. Nothing mystical is present, and the coach's remedy is directed at the crew as a unit rather than at any oar.
a crew has a shared aim and trains as one, so it already has most of what makes a group agent, whereas the hard cases are collections that never agreed to anything — a mob, a market, a nation — where there may be no unit for blame to be directed at.
Clarifying the model
THE MODEL #Three refinements keep this from sliding into either mysticism or denial.
First, group and individual responsibility are not a fixed quantity being divided. Both can be full at once: the firm bears responsibility for the river and the executive who suppressed the memo bears his own. Pettit's point is stronger still — there can be a residue at the group level that no distribution accounts for, which is exactly what the sceptic denies is possible.
Second, the group must have a structure that can carry attitudes, or the claim is empty. A decision procedure, standing roles, a way of revising commitments: these are the whole basis for treating a group as answerable rather than as a convenient collective noun — which is why "the West" or "this generation" as a bearer of blame deserves suspicion, and why Karl Jaspers, writing about German guilt in 1946, had to separate criminal from political, moral and metaphysical guilt before the question could be discussed at all.
Third, this is a different question from the imperceptible-harms puzzle in this collection, and confusing the two wastes a lot of argument. That one is about the victim side and the size of a contribution: real people are made worse off, and each contribution is too small to detect. This one is about the agent side: the harm may be perfectly visible, and the question is whether anything other than individuals can be answerable for it. One asks how small a contribution can be and still count, the other what kind of thing can be a defendant. That is the fixed point of difference.
Whether irreducible group responsibility is real is a conceptual disagreement, not an empirical one — no measurement adjudicates it. But it is not idle, because our institutions have taken sides: we fine corporations, prosecute them, and treat their obligations as surviving everyone who incurred them. Progress would look like an account of group answerability explaining why those practices are not convenient fictions, or a demonstration that convenience is all they ever were.
A picture of it
THE PICTURE #How to readEach point is a case, placed by the two features that decide where blame can land — how far the collection has a standing procedure for acting (left to right) and how large any one member's contribution is (bottom to top). Top-left is the mob, where the group label adds nothing because each member's act is blameworthy alone. Bottom-right is the case this walk-through is about: a board whose members each cast one defensible vote, and a decision that is nonetheless the firm's. Bottom-left is the uncomfortable corner where neither route works and blame has nowhere to go — where Held tries to put a floor under the drowning-swimmer case.
What became clearer
WHAT CLEARED #Blaming a group is not shorthand for blaming everyone in it. Where a collection has a procedure for forming and acting on commitments, it can hold positions no member holds — the discursive dilemma shows this is a structural feature of aggregation, not an oddity — and having commitments of its own is the ordinary ground for being answerable. That justifies group blame where the structure exists and withholds it where the group is only a label: a corporation is a far better candidate than a crowd, and a crowd than a nation. The sceptic's warning survives as a constraint on the loose cases, which is where the damage has always been done.
Where to go next
ONWARD #- Whether a member who dissented can still owe something for what the group did.
- The responsibility gap in automated systems, where harm follows from a process no person designed end to end.
Key terms
TERMS #| Term | What it means |
|---|---|
| Discursive dilemma | the result that majority voting on premises and on the conclusion they entail can commit a group to opposite verdicts, so group attitudes are not a simple function of members' attitudes. |
| Non-distributive responsibility | responsibility borne by a group that does not divide into shares carried by its members. |
Every term the collection defines is gathered in the glossary.