Unveiling Theory of Mind in Large Language Models: A Parallel to Single Neurons in the Human Brain

TL;DR

This study demonstrates that large language models exhibit Theory of Mind capabilities, with their internal embeddings responding to beliefs and perspectives similarly to neurons in the human brain, revealing an emergent cognitive property.

Contribution

The paper provides the first evidence of ToM-like responses in LLM embeddings, paralleling neural activity in the human dmPFC, and shows these responses depend on model size.

Findings

LLMs' embeddings respond to true- and false-belief trials

Embedding responses correlate with ToM task performance

Beliefs can be decoded from embeddings

Abstract

With their recent development, large language models (LLMs) have been found to exhibit a certain level of Theory of Mind (ToM), a complex cognitive capacity that is related to our conscious mind and that allows us to infer another's beliefs and perspective. While human ToM capabilities are believed to derive from the neural activity of a broadly interconnected brain network, including that of dorsal medial prefrontal cortex (dmPFC) neurons, the precise processes underlying LLM's capacity for ToM or their similarities with that of humans remains largely unknown. In this study, we drew inspiration from the dmPFC neurons subserving human ToM and employed a similar methodology to examine whether LLMs exhibit comparable characteristics. Surprisingly, our analysis revealed a striking resemblance between the two, as hidden embeddings (artificial neurons) within LLMs started to exhibit significant responsiveness to either true- or false-belief trials, suggesting their ability to represent another's perspective. These artificial embedding responses were closely correlated with the LLMs' performance during the ToM tasks, a property that was dependent on the size of the models. Further, the other's beliefs could be accurately decoded using the entire embeddings, indicating the presence of the embeddings' ToM capability at the population level. Together, our findings revealed an emergent property of LLMs' embeddings that modified their activities in response to ToM features, offering initial evidence of a parallel between the artificial model and neurons in the human brain.