Chains-of-Thought, Inner Speech, and Artificial Epistemic Agency

Cameron Buckner (University of Florida)

We are pleased to announce our monthly online talk series on "Inferences & Capacities."

Our next speaker is:

Cameron Buckner (University of Florida)
"Chains-of-Thought, Inner Speech, and Artificial Epistemic Agency"

July 20: 11am (Buenos Aires), 10am (New York), 4pm (Berlin).

Abstract: The frontier of AI is being pushed forward now by “Large Reasoning Models” (LRMs) that self-generate long textual “Chains-of-Thought” (CoTs) before answering user queries. The role of these CoTs in generating final answers was inspired by and generates obvious allusions to the roles played by inner speech in human reasoning and metacognition. In both cases, we might wonder whether access to causally relevant streams of linguistic representations might reveal the structure of the agent’s rational inferences or the way they construe their evidence as supporting their conclusions. I argue that there is a degree of negative epistemic parity in both cases: inner linguistic representations require interpretation in both cases, which limits the role such representations might play in rational explanations of inferences or luminous access to inferential grounds. However, in both cases inner linguistic representations might play a role in more forward-directed metacognitive control—though there are still important disanalogies in the epistemic architecture of humans and current artificial agents, especially involving epistemic feelings and the stable adjustment of inferential policies over time. These disanalogies limit the sense in which even frontier AI models possess the kind of individual perspective on the world through which such notions obtain their distinctive explanatory import, though this suggests less in-principle limitations than ambitious targets for near-term AI research.

How to participate: Please, send an email to Alfredo Vernazzani at: 

alfredo-vernazzani AT protonmail.com

_______

• About Inferences & Capacities:

The series brings together work on inferential capacities, rationality, normativity, and cognition — across both human and non-human animals — with the aim of fostering discussion on the nature and limits of the cognitive sphere.  

2026 line-up:  

April 27: Angelica Kaufmann (University of Milan): "Mind Blanking as Mental Imagery." 

May 18: Federico Burdman (Universidad Alberto Hurtado) "Constrained choices: addiction, attention, and reasons-responsiveness." 

June 22: Susanna Schellenberg (Rutgers): TBA 

July 20: Cameron Buckner (University of Florida): "Chains-of-Thought, Inner Speech, and Artificial Epistemic Agency."

September 7: Ulf Hlobil (Concordia University): TBA  

October 19: Eva Schmidt (TU Dortmund): TBA 

November 16: Hans-Johann Glock (University of Zürich): "Is Ascribing Inferences to Brains or Non-human Animals a Fallacy?" 

December 14: Emma Borg (SAS, University of London): "Twitches, Fidgets, Habits, Skills: The Scope of Common-Sense Psychology." 

Each talk lasts c. 40 minutes followed by 40 minutes open Q&A.  

The series is co-organized by:

Mariela Aguilera (University of Córdoba)

Matías Osta-Vélez (Universidad de la República)

Alfredo Vernazzani (TU Dortmund; Hanse-Wissenschaftskolleg).

Visit our website: