Difficulty: Intermediate | Prerequisites: None.
Altruism, helping others at a cost to yourself, is puzzling because selfish individuals should theoretically outcompete altruists. Reciprocity, reputation and parochial in-group favouritism help explain why altruism persists. Economic games (dictator, ultimatum) reveal that humans have robust fairness preferences. The bystander effect shows that people are less likely to help when others are present, due to failures at the notice, interpret and assume-responsibility stages.
Altruism
Actions or behaviour providing a benefit to others at a cost to oneself. Think of it as helping someone when there is nothing obvious in it for you (e.g. donating to a stranger).
Direct reciprocity
"I scratch your back, you scratch mine." Altruism can evolve if it is recognised and returned by partners.
Indirect reciprocity
"I saw you being nice, so I'll be nice to you." Keeping track of what others do for others builds a reputation that attracts future cooperation.
Dictator game
Two players: a dictator is given money and must offer some fraction (0 to 1.0) to a recipient. Any non-zero offer is thought to reflect altruistic preferences. The average offer is around 28%.
Ultimatum game
Two players: a proposer offers a split of money to a responder, who can accept (both get money) or reject (neither gets anything). Proposers are sensitive to fairness norms; responders engage in costly punishment by rejecting unfair offers.
Third-party punishment
A third player observes a dictator game and can pay from their own money to reduce the dictator's earnings. Any non-zero punishment reflects altruistic preferences in the third party.
Parochial altruism
People are more willing to engage in prosocial behaviour toward members of their own in-group than toward out-group members.
Bystander effect
A person is less likely to help someone in need when other bystanders are present.
Latane and Darley's decision tree
Three stages of bystander intervention: (1) notice the incident, (2) interpret it as an emergency, (3) assume responsibility. Failure at any stage means no help.
Altruism goes against the notion of rational self-interest that underlies behaviorism, Freudian theory, Darwinian fitness and mainstream economics. The evolutionary problem: imagine altruists and selfish individuals in equal numbers. Each round, individuals collect food (one to survive, two to reproduce). Altruists with two pieces share with a neighbour; selfish individuals do not. Over generations, selfish people dominate because they always keep their surplus. So how does altruism survive?
Direct reciprocity: mutual exchange between the same two individuals.
Indirect reciprocity: third parties observe your behaviour and reward or punish you based on your reputation. Keeping track of what others do for others promotes altruism because being seen as generous attracts future cooperation.
Dictator game: the dictator decides how to split money with a passive recipient. Average offers of about 28% show that people are not purely selfish.
Adding third-party punishment: a third player can pay to reduce the dictator's earnings ($1 to reduce $3). This reveals altruistic enforcement of fairness norms.
When normative expectations are manipulated (past participants described as selfish or fair), participants' empirical expectations (what they think others actually do) and normative expectations (what they think others should do) begin to diverge. Empirical expectations best predict a participant's own giving.
Ultimatum game: the responder's ability to reject unfair offers acts as costly punishment, enforcing fairness norms.
Cross-cultural evidence: behaviour in these games was examined across 15 different societies worldwide. All societies show altruistic preferences. The degree of third-party punishment in a society correlates with the average size of dictator game offers, suggesting cultural co-evolution of altruistic tendencies.
People are more willing to help in-group members. A study in Papua New Guinea (Bernhard, Fischbacher and Fehr) used a dictator game with third-party punishment across two groups (Ngenika and Wolimbka).
Third parties paid to punish inequitable transactions even when the victim was not from their own group.
Highest punishment occurred when the victim was in-group and the violator was out-group.
Third parties were more lenient when the norm violator was in their own in-group.
Supports parochial altruism: fairness enforcement is real but biased toward in-group members.
People are sensitive to reputation information and perform costlier altruistic punishment when reputation can be gained.
Study: participants played the ultimatum game for 20 rounds, each time with a new anonymous partner. In the reputation condition, the proposer learned the responder's last-round behaviour. In the no-reputation condition, this information was hidden.
Results: responders had higher acceptance thresholds (required more even splits) when reputation was allowed. Behaviour changed immediately in response to the presence of reputation tracking.
Latane and Darley's decision tree: three stages where helping can break down.
Notice the incident. In a crowd, you may not even see it.
Interpret it as an emergency. In a crowd, if nobody else is reacting, you may decide it is not serious (pluralistic ignorance).
Assume responsibility. In a crowd, responsibility is diffused ("someone else will help").
Smoke-filled room study
Columbia University men filled out a questionnaire in a room, either alone or with two strangers. Smoke began pouring into the room.
Results: 80% of people working alone reported the smoke within four minutes. Only 10% of three-person groups reported it within four minutes. People alone were also faster to notice the smoke.
The presence of passive bystanders reduced both the likelihood and the speed of helping.
Students often think altruism requires zero self-interest. In practice, altruism is defined as benefiting others at a cost to oneself. The motive does not need to be purely selfless; the cost is what matters.
Students often think the bystander effect means people in groups do not care. The mechanism is more subtle: failures at the notice, interpret and assume-responsibility stages, not indifference.
Students often think the dictator game average offer of 28% shows people are generous. Compared to a 50-50 split, 28% is below fair. It shows altruistic preferences exist but are modest.
⚠️ Be able to explain why altruism is an evolutionary puzzle and describe the food-sharing thought experiment.
⚠️ Know the rules of the dictator game, ultimatum game and third-party punishment, and be able to explain why each is relevant for studying altruism.
⚠️ Be prepared to describe the Papua New Guinea parochial altruism study and its results.
⚠️ Know the three stages of Latane and Darley's decision tree and be ready to describe the smoke-filled room study.
⚠️ Understand how reputation promotes altruism (ultimatum game reputation study).
Fill in the blank: In the dictator game, the average offer is approximately _____%. (28)
True or False: In the smoke-filled room study, people in groups reported the smoke faster than people alone. (False. People alone reported it much faster.)
Fill in the blank: The three stages of Latane and Darley's bystander decision tree are notice, interpret and _____. (assume responsibility)
True or False: Parochial altruism means people punish norm violations only when the victim is an in-group member. (False. Third parties punished inequitable transactions even when the victim was out-group, but punishment was highest when the victim was in-group and the violator was out-group.)
Q: Describe the rules of the dictator game and the ultimatum game. Explain why they are relevant for understanding altruism.
A: In the dictator game, one player divides money and the recipient passively receives whatever is offered. Any non-zero offer suggests altruistic preferences. In the ultimatum game, the responder can reject the offer, in which case neither player gets anything, making rejection a form of costly punishment. The games are relevant because purely self-interested players would offer nothing (dictator game) or accept any positive offer (ultimatum game). The fact that people offer roughly 28% and reject unfair offers shows genuine fairness preferences and willingness to pay costs to enforce norms.
Q: Explain the bystander effect and describe the smoke-filled room study.
A: The bystander effect is the finding that people are less likely to help when other bystanders are present. In the smoke-filled room study, individuals filling out questionnaires alone reported smoke 80% of the time within four minutes; in three-person groups, only 10% reported it. The presence of passive others inhibited helping at all three decision stages: noticing (slower in groups), interpreting (if others are calm, it must not be serious) and assuming responsibility (diffusion).
Reciprocity and reputation connect to the cooperation mechanisms in the Conflict and Conflict Resolution notes (tit-for-tat, punishment in public goods games). Parochial altruism connects to in-group bias from the Stereotypes and Prejudice notes. The bystander effect connects to diffusion of responsibility, which parallels social loafing from the Group Influence notes.
Altruism, helping behaviour, prosocial behaviour, evolutionary puzzle, reciprocity, direct reciprocity, indirect reciprocity, reputation, dictator game, ultimatum game, third-party punishment, parochial altruism, Bernhard Fischbacher Fehr, Papua New Guinea, bystander effect, Latane, Darley, decision tree, smoke-filled room, pluralistic ignorance, diffusion of responsibility, fairness norms, costly punishment, NYU social psychology