Emergent Social Transmission of Model-Based Representations Without Inference
Summary: arXiv:2604.05777v1 Announce Type: new
Abstract
How do people acquire rich, flexible knowledge about their environment from others despite limited cognitive capacity? Humans are often thought to rely on computationally costly mentalizing, such as inferring others’ beliefs. In contrast, cultural evolution emphasizes that behavioral transmission can be supported by simple social cues. Using reinforcement learning simulations, we show how minimal social learning can indirectly transmit higher-level representations.
Key Findings
This research explores a model in which a naive agent learns to navigate a reconfigurable environment. The agent can learn either alone or by observing an expert, crucially without the need to infer mental states. The following points summarize the key findings:
- Naive Agent’s Learning Process: The naive agent selects actions or boosts value representations based on observed actions from an expert, rather than attempting to understand the expert’s thoughts or beliefs.
- Bias in Learning Experience: The observed actions from the expert serve as social cues that bias the learner’s experience, leading to a convergence of the learner’s representations toward those of the expert.
- Benefits of Social Exposure: Model-based learners, who utilize higher-level cognitive processes, benefit significantly from this type of social exposure, resulting in faster learning rates and more expert-like representations.
- Cultural Transmission via Simple Processes: This study provides evidence that cultural transmission can emerge from simple, non-mentalizing processes that leverage asocial learning mechanisms.
Implications of the Study
The implications of this research are significant in understanding how knowledge and skills are transmitted in social contexts. The findings suggest that:
- Social learning can be effective even when complex mentalizing processes are not utilized, making learning more accessible.
- Educational strategies may be enhanced by focusing on observational learning and social cues, rather than purely cognitive approaches.
- Future models of cultural evolution should consider the role of simple social interactions in the transmission of knowledge.
Conclusion
This study sheds light on the mechanisms of social learning, suggesting that individuals can acquire complex knowledge through minimal social cues without the need for deep inference. The results highlight the potential of reinforcement learning simulations in exploring cognitive processes and offer a new perspective on cultural transmission. As research in this area continues, it may lead to the development of more effective learning strategies that leverage social interactions.