Altruistic traits are predicted by neural responses to monetary outcomes for self vs charity

Abstract

Human altruism is often expressed through charitable donation-supporting a cause that benefits others in society, at cost to oneself. The underlying mechanisms of this other-regarding behavior remain imperfectly understood. By recording event-related-potential (ERP) measures of brain activity from human participants during a social gambling task, we identified markers of differential responses to receipt of monetary outcomes for oneself vs for a charitable cause. We focused our ERP analyses on the frontocentral feedback-related negativity (FRN) and three subcomponents of the attention-related P300 (P3) brain wave: the frontocentral P2 and P3a and the parietal P3b. The FRN distinguished between gains and losses for both self and charity outcomes. Importantly, this effect of outcome valence was greater for self than charity for both groups and was independent of two altruism-related measures: participants' pre-declared intended donations and the actual donations resulting from their choices. In contrast, differences in P3 subcomponents for outcomes for self vs charity strongly predicted both of our laboratory measures of altruism-as well as self-reported engagement in real-life altruistic behaviors. These results indicate that individual differences in altruism are linked to individual differences in the relative deployment of attention (as indexed by the P3) toward outcomes affecting other people.

Keywords: P300; altruism; event-related potentials; feedback-related negativity; monetary outcomes.

Fig. 1.

Experimental design. (A) Participants chose between four decks, each one labeled with a different symbol. The position of these decks was randomized across trials, but the symbol label for each deck remaining constant from trial to trial. Feedback concerning outcomes for self and for the charity was sequentially presented as a number of colored stars and circles. The specific symbol-recipient association (stars-self/circles-charity vs stars-charity/circles-self) was randomized across participants (although consistent within a participant), and the specific sequence (self-first/charity-second vs charity-first/self-second) was randomized across trials for each participant. Red-colored shapes indicated the number of points lost and green-colored shapes indicated the number of points won. (B) There were four possible outcome valence combinations on each trial: Self-win\Charity-win, Self-win\Charity-loss, Self-loss\Charity-win and Self-loss\Charity-loss. The outcome on each trial was randomly determined according to the weight that the chosen deck assigned to each of these outcome valence combinations. For example, deck S ++\C- was associated with a 30% chance of observing a self-win\charity-win outcome, 45% for self-win\charity-lose, 15% chance for self-lose\charity-win and 10% chances for self-lose\charity-lose. Outcome magnitudes (1–4) were equiprobable (i.e. 25% each), resulting in specific EVs for self and charity associated with each deck.

Fig. 2.

Distribution of intended-donation scores. Participants were asked about the proportion of the money earned during the task they would be willing to donate to the charity. Based on the distribution of their responses we defined a selfish-group and an altruistic-group of participants. This group definition was used as a between-subjects factor in several of our subsequent behavioral and ERP data analyses.

Fig. 3.

Behavioral adjustment related to gains and losses for self and for the charity. For participants in the selfish-group the probability of persevering on a particular choice across trials was affected by outcomes for self but not outcomes for the charity. For participants in the altruistic-group, outcomes for self and outcomes for the charity had similar effects on behavioral adjustment. Error bars correspond to the standard error of the means (SE).

Fig 4.

Effect of outcome valence and outcome magnitude on ERPs and difference waves. (A) Left, Grand average ERPs for each outcome from an anterior ROI of seven electrodes centered on FCz and located over frontocentral cortex. Cream-colored rectangles show the time windows defined for each of the ERP components. The thin rectangle in light red, immediately preceding the FRN time window, represents the anchoring reference latency window for quantifying the FRN amplitude. Right, Grand average ERPs from each outcome from a more posterior ROI of seven electrodes centered on Pz and located over the parietal cortex. (B) ERP difference waves obtained for the fronto-central ROI (left panel) and for the parietal ROI (right panel). Brown traces show the difference waves obtained by subtracting the ERP response elicited by a gain outcome from that elicited by a loss outcome, separately for each outcome magnitude. Blue traces show the difference waves obtained by subtracting the ERP response elicited by outcome magnitudes 2, 3 and 4 from that elicited by outcome magnitudes 1, regardless of outcome valence.

Fig. 5.

ERP traces and loss-minus-win difference waves in the fronto-central ROI showing effects involving the factors ‘group’, ‘recipient’ and ‘outcome valence’, shown separately for the selfish group (left panels) and the altruistic group (right panels). Upper panels: Grand ERP averages for each outcome valence (gain outcome and loss outcomes) separately for each recipient (self and charity). Lower panels: Black traces show difference waves obtained by subtracting the ERP responses to gains from the ERP responses to losses, separately for each outcome recipient (self and charity). Blue and green traces show difference waves obtained by subtracting the ERP responses to outcomes for self from the ERP responses to outcomes for charity.

Fig. 6.

The across-participants association between the self-minus-charity-P3a contrasts and both actual-donations (left) and self-reported real-life altruism (right). Participants included in the selfish group for previous analyses are represented as blue rings. Participants included in the altruistic group for previous analyses are represented as green rings.