Interpolative Decoding: Exploring the Spectrum of Personality Traits in LLMs

TL;DR

This paper introduces interpolative decoding to efficiently simulate and explore the spectrum of human personality traits in large language models, enabling more nuanced behavioral emulation and reducing experimental overhead.

Contribution

The paper presents a novel interpolative decoding method that modulates LLM behavior along personality dimensions, improving simulation fidelity and experimental efficiency.

Findings

Interpolative decoding reliably modulates Big Five personality scores.

LLMs can mimic human decision-making in economic games.

Preliminary results show potential for twin-like replication of individual human behaviors.

Abstract

Recent research has explored using very large language models (LLMs) as proxies for humans in tasks such as simulation, surveys, and studies. While LLMs do not possess a human psychology, they often can emulate human behaviors with sufficiently high fidelity to drive simulations to test human behavioral hypotheses, exhibiting more nuance and range than the rule-based agents often employed in behavioral economics. One key area of interest is the effect of personality on decision making, but the requirement that a prompt must be created for every tested personality profile introduces experimental overhead and degrades replicability. To address this issue, we leverage interpolative decoding, representing each dimension of personality as a pair of opposed prompts and employing an interpolation parameter to simulate behavior along the dimension. We show that interpolative decoding reliably modulates scores along each of the Big Five dimensions. We then show how interpolative decoding causes LLMs to mimic human decision-making behavior in economic games, replicating results from human psychological research. Finally, we present preliminary results of our efforts to ``twin'' individual human players in a collaborative game through systematic search for points in interpolation space that cause the system to replicate actions taken by the human subject.