Trolley Variants: A Taxonomy and Annotated Exemplars

Which moral axis do you want to isolate?

Classification prompt — pick an axis

Decide which conceptual axis you want to examine. Read the short definitions and follow the section anchors below for the canonical variants that isolate that distinction.

Canonical variant Axes emphasized Two-line summary
Classic Switch Act/Omission · Means/Side-effect A lever diverts a runaway trolley from five people to one on a side track; highlights causal distance and whether switching counts as acting.
Footbridge (Push) Personal force · Intention A bystander must push a large person off a bridge to stop a trolley, thereby killing them to save five; isolates physical contact and intended means.
Loop / Sacrifice Collective trade-offs · Identity framing A diversion route requires sacrificing a passenger connected to the trolley to save multiple outsiders; probes agent-relative permissions and passenger identity.
Bystander/Omission Omission · Role framing A bystander can refrain from flipping a switch, letting five die; explores whether omissions are morally distinct from acts.
Probabilistic Switch Risk · Expected outcomes Choosing a diversion with probabilistic outcomes (e.g., chance to save or kill) tests how uncertainty alters permissibility judgments.
MEANS
Means vs. side-effect asks whether harm that is used as the means to an end is morally different from harm that is merely foreseen.

When to consult — Use loop/sacrifice or push variants to sharpen whether killing someone as the means to stop a trolley differs from letting them die as a side-effect.

Avoid conflating foreknowledge with instrumental use: a close causal chain that uses the person's body as a stopping-mechanism is prototypical for testing means.

ACT/OMIT
Act vs. omission examines whether doing harm is morally distinct from passively allowing harm to occur.

When to consult — The classic switch and bystander omission variants contrast an active diversion with standing by; frame the same outcomes both ways to test the symmetry.

Be explicit about role and obligations: labeling someone as a 'driver', 'passenger', or 'bystander' changes the moral salience of omission.

PERSONAL
Personal force probes how direct bodily contact and proximity affect judgments about permissibility.

When to consult — The footbridge/push example is canonical: it isolates emotional salience and whether 'pushing' differs from flipping a remote lever.

Distinguish physical contact from emotional proximity: both can increase perceived wrongness but for different explanatory reasons.

UNCERTAINTY
Probability and uncertainty test how risk and epistemic limitations change permissibility.

When to consult — Use probabilistic-switch variants to contrast deterministic outcomes with risky choices; useful for arguments about expected outcomes vs. rights-based constraints.

Keep the background epistemic state explicit: what the agent knows about probabilities matters for the conceptual question being isolated.

Compact comparison matrix

Classic Switch
Tests: act/omission and causal distance. Pitfall: conflating remote intervention with impersonal means. Prefer when you want a clear act that remains impersonal.
Footbridge / Push
Tests: personal force and intention. Pitfall: emotional salience obscuring conceptual claims about means. Prefer when isolating proximity and bodily contact.
Loop / Sacrifice
Tests: means vs. side-effect and passenger identity. Pitfall: role-framing—passenger consent or status changes the analysis. Prefer to discuss agent-relative permissions.
Probabilistic Switch
Tests: risk, expected outcomes, and moral uncertainty. Pitfall: conflating chance with moral permissibility without specifying epistemic stance. Prefer for arguments about expected-value reasoning.

Reader’s guide: choosing and adapting variants

Start by naming precisely which conceptual contrast you intend to isolate (for example, 'intention vs foreseen side-effect'). Then pick the canonical variant that most directly foregrounds that contrast. If you combine features, state which axis is primary to avoid category-mixing.

Framing note: Describing agent roles (passenger, driver, bystander) changes the argument frame; prefer neutral labels when the role should not be the focus.

Limitations & ethics

Thought experiments isolate conceptual distinctions; they do not establish empirical claims about how people decide. Avoid overgeneralizing from a single variant: different variants test different principles and cognitive contrasts. Be explicit about which distinction you treat as primary.

Further reading: seek canonical discussions under the labels used in this page (classic trolley, footbridge/push, loop/sacrifice, probabilistic-switch) and primary sources that introduce the doctrine of double effect and act/omission debates.