Storytelling as Inverse Inverse Planning

Area of Science:

• Cognitive Science
• Computational Linguistics
• Literary Theory

Background:

• Storytelling engages audiences through evolving beliefs about fictional worlds.
• Literary theorist Karin Kukkonen defines storytelling as "probability design."

Purpose of the Study:

• To computationally model storytelling as probability design.
• To explore inverse inverse planning as a framework for understanding narrative construction.

Main Methods:

• Utilized a Bayesian inverse planning model from human social cognition.
• Developed an inverse inverse planner to simulate storytelling processes.
• Conducted behavioral studies comparing inverse inverse planning with naïve planning.

Main Results:

• Inverse inverse planning demonstrated more expressive behavior than naïve planning in simulations.
• The model successfully accounts for core storytelling elements such as character, plot twists, and narrative arcs.
• Probability design naturally encodes narrative complexity through sequential information release.

Conclusions:

• Storytelling can be understood as the strategic manipulation of audience inferences via probability design.
• Inverse inverse planning provides a robust computational framework for analyzing narrative structure and effect.
• This approach offers new insights into the cognitive underpinnings of engaging fictional experiences.