Dictator Game Decisions with Robot Recipients | Avantika Dev et al. | ResearchPod

Transcript: Dictator Game Decisions with Robot Recipients

Alex: People offered money to robots, even when keeping it all was allowed. But their generosity shrank when money became real, and shrank again when refusing became an explicit choice.

Sam: That sounds like a finding about robots and about how we ask people to share. Whose research is this?

Alex: It’s a series of experiments by Avantika Dev, Roy de Kleijn, and Sumitava Mukherjee, published in the International Journal of Social Robotics. Students in India and the Netherlands made allocation decisions involving robot recipients.

Sam: For researchers studying social machines, the distinction matters. Giving a robot money might reflect a social response, or an expectation created by the experiment.

Alex: And this study helps separate those possibilities, without fully resolving them. We’ll look at what the broader robot sample adds, how the payment rules changed decisions, and what these offers can actually tell us.

Sam: Start with the task. What does offering money to a robot mean here?

Alex: The authors used the dictator game, where one person decides how much money to share with a passive recipient. The recipient cannot reject the offer or retaliate. Keeping everything is permitted, so sharing does not require a strategic payoff.

Sam: That removes bargaining, but it doesn’t automatically make every offer altruistic. How do the authors interpret it?

Alex: As a possible proxy for economic social preferences, with qualifications. They acknowledge that giving can reflect social norms or expectations about the experiment. That uncertainty becomes central when they change the task.

Sam: What was missing from earlier research on robot recipients?

Alex: Much of it examined just one or two robot models, or individual aspects of perception. That leaves open whether a result belongs to robots generally or to a particular design. Across four studies, these authors collected decisions from 451 participants.

Sam: So breadth is part of the contribution. How different were the recipients?

Alex: They used 18 commercially available robots spanning humanoids, consumer devices, industrial machines, medical robots, and educational robots. Participants saw static images, not moving robots or live interactions. Each participant evaluated the whole set, presented in random order.

Sam: With that much variety, a human-resemblance scale alone could miss why someone likes a machine. How did they capture those impressions?

Alex: They deliberately chose a broad set of perception questions rather than one existing scale. Participants rated friendliness, creepiness, apparent emotional capacity, and willingness to let the robot perform particular tasks. For example, would they let it cook or care for their family?

Dictator Game Decisions with Robot Recipients

Avantika Dev, Roy de Kleijn, Sumitava Mukherjee

Mar 1, 2026·8:53·user_upload·Original Paper

Dictator GameHuman-Robot InteractionAnthropomorphismSocial RoboticsProsocial PreferencesEconomic Decision Making

Listen

9 min

Summary

Research Question and Approach

This paper investigates whether economic prosocial preferences, typically studied in human-human interactions, extend to human-robot interactions. The researchers employed the dictator game—a standard behavioral economic paradigm for measuring altruism and social norms—to observe how participants allocate money to a diverse set of 18 different commercial robots. Across four studies involving participants from India and the Netherlands, the researchers measured how participants' offers changed based on whether the stakes were hypothetical or real, and whether the game structure was a one-step or two-step process.

Key Findings

The study confirmed that participants consistently make monetary offers to robots, with behavior mirroring patterns observed in the human-human dictator game literature. Notably, likeability emerged as the strongest predictor of generosity toward a robot, surpassing both perceived anthropomorphism and social functionality. The findings show that while participants are generally less generous when real money is involved compared to hypothetical stakes, the three identified robot characteristics remain robust predictors of giving across all contexts. Interestingly, the study found no significant cross-country differences between Indian and Dutch participants in the proportion of the stake offered.

Why It Matters

As socially interactive robots increasingly enter public and private spaces, understanding the psychological and economic mechanisms of human-robot interaction becomes vital. This research moves beyond the common focus on single-robot models, providing a data-driven look at how diverse designs influence human social preferences. The results suggest that humans naturally apply social heuristics to robots, and that designers can influence these economic interactions by manipulating specific perceptual traits like likeability and social utility.

Key Terms and Definitions

Dictator Game — A behavioral economic experiment where a 'dictator' is given a stake and can share any portion with a passive recipient, serving as a measure of prosocial economic preferences.
Likeability Index — An emergent perceptual factor capturing positive affective responses, friendliness, and the desire to possess or interact with a robot.
Anthropomorphism Index — An emergent perceptual factor reflecting the attribution of human-like physical and cognitive qualities, such as the ability to think or feel.
Social Functionality Index — An emergent perceptual factor related to the robot's perceived utility and its suitability for assisting with household or social tasks.
Two-Step Dictator Game — A variation of the game where participants first make a binary decision of whether to give at all, followed by a decision on how much to give, intended to isolate the initial decision-making process.

Abstract

Abstract While extensive research has examined human-human interactions in dictator games, studies exploring human-robot dictator games remain scarce. Here, we conducted multiple studies in India and The Netherlands to explore how participants allocate money to robot recipients. Using various versions of the dictator game paradigm with both hypothetical and real stakes, we measured offers toward 18 different robots while also assessing participants’ perceptions of these robots. Overall, our findings reveal that participants allocated money to robot recipients in dictator games, with offers varying significantly for hypothetical versus real monetary outcomes, mirroring patterns observed in the existing human-human dictator game literature. Factor analyses identified three key characteristics – perceived likeability, anthropomorphism, and social functionality of the robots – as consistent predictors of offers made in the dictator games. Results were broadly consistent across participants from the two countries. We discuss similarities and differences observed across dictator game variations, key predictors of offers made to robots, and cross-country differences in robot perception, along with study limitations. Our results offer empirical insights into the dynamics of human-robot economic interactions that could inform the design of socially interactive robots.

Dictator Game Decisions with Robot Recipients: Research Paper Podcast