Social decision-making is a fundamental aspect of human behavior. Utilitarian philosophy, as articulated by Hume (2003), posits that social decision-makers should opt for choices that maximize the overall welfare of the group. This approach necessitates comparing individual preferences on a common scale—meaning that each person's welfare must be measured in a way that allows meaningful comparisons across individuals. However, the legitimacy and possibility of such interpersonal utility comparisons has been the subject of significant debate. Roberts (1997) rigorously shows that combining different individuals’ utility judgments into a single, objective scale is only feasible under very strong assumptions; otherwise, the process leads to extreme outcomes, such as relying solely on one person's perspective. Despite these theoretical difficulties, humans routinely engage in informal forms of such comparisons in everyday life. From choosing a restaurant with friends to formulating national policies, we constantly face the challenge of balancing diverse preferences to maximize collective welfare. This cognitive ability is crucial for preventing social discord.
The neural mechanisms underpinning our ability to maximize collective welfare have been a subject of increasing interest in neuroscience, as researchers seek to bridge the gap between abstract philosophical concepts and concrete biological mechanisms. A key component of this ability is the generation of neural representations of value associated with stimuli, actions, or decisions for each individual in the group. Previous research has identified several brain regions involved in value representation and social decision-making. The orbitofrontal cortex (OFC), ventromedial prefrontal cortex (VMPFC), and striatum can all encode subjective values (Padoa-Schioppa and Assad, 2006; Bartra et al., 2013; Clithero and Rangel, 2014). The VMPFC and striatum are also implicated in processing vicarious rewards and representing others’ experiences of value. Moreover, the VMPFC can represent subjective value when choosing on behalf of others (Nicolle et al., 2012) and encode differences between others’ and one's own preferences (Garvert et al., 2017). Thus the VMPFC may play a key role in social decision-making by allowing comparison of values across individuals.
To further understand how the brain represents others’ preferences and allocates resources based on those preferences, Soutschek et al. (2024) asked participants to perform a series of tasks while undergoing functional magnetic resonance imaging (fMRI). First, participants performed a food choice task in which they chose between different quantities of two snacks that they had previously rated positively. This allowed the researchers to determine the relative subjective value of two snacks for each participant. Activity in the VMPFC was correlated with the subjective value difference between the two options, consistent with previous work showing that the VMPFC encodes one’s own subjective values.
Next, participants were asked to learn the snack preferences of two agents who had completed the snack choice task during pilot studies. For each participant, one agent with a similar snack preference and one with a dissimilar snack preference was selected. After training, subjects were able to predict choices of both similar and dissimilar agents, demonstrating the ability to learn others’ preferences regardless of their similarity to one's own. Crucially, the participants’ VMPFC activities were correlated with the estimated subjective value differences between options of the two agents. Furthermore, a decoder trained to classify VMPFC activity patterns associated with the preferences of either agent (e.g., the similar agent) could also predict the preferences of the other agent (e.g., the dissimilar agent). This suggests the VMPFC uses a common neural code to encode the preferences of others, regardless of whether they have preferences similar or dissimilar to the subject's own preferences. However, attempts to cross-decode between self-value and other-value failed, indicating that VMPFC employs distinct neural codes for one's own preferences versus those of others.
Finally, subjects performed a snack allocation task, in which they were asked to choose between two options for allocating snacks to the two agents. Each option provided different quantities of two snacks to each agent. In most cases, there was a conflict of interest, with one option favoring one agent and the other option favoring the other agent. Subjects were not given any instructions about how they should make the choice; in particular, they were not told to maximize the total welfare of the agents. Nevertheless, participants often chose allocation options that maximized the value for the pair of the agents. As with individual subjective values, VMPFC activity during this task was correlated with the difference in total subjective value (combined subjective value for the two agents) between the two options. Notably, the activity patterns recorded in VMPFC during the preference learning task could be used to decode the total value differences between options in the snack allocation task. This means that the participants’ neural representation of how much each agent valued different options could predict how participants decided to allocate resources. Conversely, VMPFC activity during the snack allocation task could be used to decode the agents’ preferences in the preference learning task. This suggests that VMPFC uses the same neural code for representing both estimated preferences of others and preferences of them as a group.
Collectively, the authors provide compelling evidence that VMPFC encodes one's own subjective preferences during self-directed snack choices and, in a distinct neural code, encodes the inferred preferences of other agents—whether those tastes are similar to or diverge from the subject's own. When participants allocate snacks between two agents, VMPFC reuses that same “others’ preference” code to represent the summed utilitarian welfare of both agents. Thus, VMPFC maintains segregated encodings for self-referential preference versus others’ preferences but collapses all other-referential value processing onto a single unified scale whenever collective welfare must be calculated. Such an organization suggests a general computational principle: valuation regions can preserve separable codes for self and others while enabling efficient integration of multiple others into a coherent, utilitarian metric.
The work by Soutschek et al. (2024) also opens avenues for future research to expand our understanding of the biological basis of altruism and cooperation. Individuals who exhibit more altruistic behavior may do so because they assign higher value to other's rewards. Future research might ask whether this is reflected in different levels of VMPFC activity when choosing for oneself versus choosing for others during a reward allocation task including the subjects as well as multiple other agents. Research may also examine whether cooperation only happens when both subjects assigned a higher weight for other's reward value. Such insights might shed light on the neural underpinnings of human sociality and moral behavior (Hutcherson et al., 2015).
Despite making important advances, the study by Soutschek et al. (2024) has some limitations. In particular, the laboratory setting and specific tasks used does not fully capture the complexity of real-world social decision-making. Future studies could explore how these findings translate to more naturalistic settings and diverse populations. For example, researchers can explore how human subjects consider group welfare when the group include different genders, different races, different socioeconomic status, even people you like or dislike.
In conclusion, Soutschek et al. (2024) have advanced our understanding of the neural basis of utilitarian decision-making and welfare maximization. By demonstrating that the VMPFC encodes the preferences of other individuals and group welfare using the same neural code, the study provides empirical support for the possibility of interpersonal utility comparisons, a concept central to utilitarian philosophy. This research opens new avenues for investigating the biological foundations of moral reasoning and social behavior, with potential implications for addressing social decision-making challenges in various contexts, from small-group dynamics to larger-scale policy decisions.
Footnotes
Editor’s Note: These short reviews of recent JNeurosci articles, written exclusively by students or postdoctoral fellows, summarize the important findings of the paper and provide additional insight and commentary. If the authors of the highlighted article have written a response to the Journal Club, the response can be found by viewing the Journal Club at www.jneurosci.org. For more information on the format, review process, and purpose of Journal Club articles, please see http://jneurosci.org/content/jneurosci-journal-club.
Review of Soutschek et al.
I thank my journal club mentor Dr. Kate M. Wassum (Professor, Department of Neuroscience, University of California, Los Angeles) and Dr. Teresa Esch for their helpful comments in the writing process of this journal club article.
This Journal Club was mentored by Kate M. Wassum.
The author declares no competing financial interests.
- Correspondence should be addressed to Manning Zhang at manningzhang{at}g.ucla.edu.
