Harming Many People is Easier than Harming One
This study by behavioral economists shows that people who behave quite generously in experimental games where they have the opportunity to give money to a single individual, often behave very selfishly in a game that allows them to take money from each person in a group of 16 people.
The abstract reads:
The seemingly rampant economic selfishness suggested by many recent corporate scandals is at odds with empirical results from behavioural economics, which demonstrate high levels of prosocial behaviour in bilateral interactions and low levels of dishonest behaviour. We design an experimental setting, the ‘Big Robber’ game, where a ‘robber’ can obtain a large personal gain by appropriating the earnings of a large group of ‘victims’. In a large laboratory experiment (N = 640), more than half of all robbers took as much as possible and almost nobody declined to rob. However, the same participants simultaneously displayed standard, predominantly prosocial behaviour in Dictator, Ultimatum and Trust games. Thus, we provide direct empirical evidence showing that individual selfishness in high-impact decisions affecting a large group is compatible with prosociality in bilateral low-stakes interactions. That is, human beings can simultaneously be generous with others and selfish with large groups.
The researchers note the consistency of these findings with psychic numbing and compassion fade, concluding that harming many individuals might be easier than harming just one, in line with received evidence that people are more willing to help one individual than many (Shelling, 1968; Fetherstonhaugh et al., 1997; Butts et al., 2019).
Schelling, T. C. in Problems in Public Expenditure Analysis (ed. Chase, S. B.) 127–162 (The Brookings Institute, 1968).
Fetherstonhaugh, D., Slovic, P., Johnson, S. & Friedrich, J. Insensitivity to thevalue of human life: a study of psychophysical numbing. J. Risk Uncertain. 14, 283–300 (1997).
Butts, M. M., Lunt, D. C., Freling, T. L. & Gabriel, A. S. Helping one or helping many? A theoretical integration and meta-analytic review of the compassion fade literature. Organ. Behav. Hum. Decis. Process. 151, 16–33 (2019).
The study's authors also note that their results are conceptually aligned with previous research showing that behavior is crucially linked not only to underlying preferences and values but also the specific characteristics and properties of the situation at hand. This is central to the arithmetic of compassion where the value of protecting a life is not fixed within each of us but rather depends on the number of lives at risk and other contextual factors.
Photograph by abbythepup.com CC0 1.0