The Challenge of Controlled Negotiation Research
Traditional negotiation theory posits that successful outcomes require a delicate balance between competing interpersonal demands: empathy versus assertion, concern for others versus self-interest, and being 'soft on the people' while remaining 'hard on the problem.' Human subjects often struggle to maintain these balances consistently, making it difficult for researchers to isolate variables and test theoretical prescriptions under controlled conditions.
Personality Engineering as a Methodology
Vaccaro and Curhan introduce 'personality engineering' to overcome these human limitations. By using AI agents, researchers can precisely parameterize, manipulate, and evaluate specific personality traits. This methodology allows for:
- Precision: Fine-tuning agent behavior to specific personality profiles.
- Consistency: Ensuring agents maintain a stable persona throughout an entire negotiation.
- Scalability: Running thousands of controlled iterations to test theoretical outcomes.
The Interpersonal Circumplex Coordinate System
The authors propose the 'interpersonal circumplex' as the foundational coordinate system for this field. This model maps personality along two primary axes: warmth and dominance. By using these two dimensions, researchers can create a standardized grid for designing AI agents, allowing for rigorous, replicable experiments that test how varying levels of warmth and dominance impact negotiation success. This framework serves as both a scientific tool for testing existing theories and a practical design guide for developers building AI negotiation agents.