What is it about?

Most people experience inner speech - that voice in your head narrating thoughts and working through problems. But why do we need it? This paper proposes that inner speech helps our brains navigate uncertainty. It acts like a translator in two directions: converting overwhelming sensory experiences into compact linguistic forms (like reducing a complex painting to "Mona Lisa"), and unpacking abstract goals into concrete actions (transforming "stay healthy" into "choose salad, not chips"). When you talk yourself through an anxious situation, you're reducing uncertainty by converting fuzzy feelings into structured sequences your brain can predict. This explains why inner speech becomes more frequent and extensive when you're confused, then fades and condenses when things become clear.

Featured Image

Why is it important?

Most research describes inner speech as "silent talking" without explaining why we actually need it. This framework offers the first unified explanation for why inner speech helps with some tasks but not others, why its forms varies dramatically between people and situations, and how it connects to working memory and mental health. This changes our view from "inner speech is just thinking in words" to "inner speech is a uncertainty control mechanism for making the world easier to predict and control."

Perspectives

I've been fascinated by inner speech for years, and seeing how a computational perspective helps connect previously scattered findings has been incredibly exciting. What really drives my enthusiasm is how viewing inner speech through the lens of uncertainty reduction makes everything click - from its various forms to its different functions. I hope this framework not only bridges gaps in our understanding but also inspires new ways to study this fascinating conscious experience!

Bo Yao
Lancaster University

Read the Original

This page is a summary of: Rethinking inner speech through linguistic active inference., Psychological Review, December 2025, American Psychological Association (APA),
DOI: 10.1037/rev0000607.
You can read the full text:

Read

Contributors

The following have contributed to this page