# Neural mechanisms in resolving prior and likelihood uncertainty in scene recognition

## TL;DR

This study explores how the brain processes sensory and prior information during scene recognition using AI-generated images and brain scans.

## Contribution

The study introduces a novel method combining AI-generated images and fMRI to separately manipulate and identify neural representations of likelihood and prior uncertainty.

## Key findings

- Higher visual areas were more active with low likelihood uncertainty.
- The default mode network showed increased activity with higher prior information.
- AI-generated images enabled precise control of uncertainty in a neuroscience experiment.

## Abstract

Recognizing real-world scenes requires integrating sensory (likelihood) and prior information, yet how the brain represents these components remains unclear. To investigate this, we employed deep image transformation to generate images with parametrically controlled naturalness, enabling precise manipulation of likelihood uncertainty. Concurrently, we designed a sequential image-scene recognition task that quantitatively modulates prior information. By combining these AI-generated images with the task, we conducted a functional magnetic resonance imaging (fMRI) experiment enabling systematic control of both likelihood and prior information. The results revealed that higher visual areas were activated when viewing images with low likelihood uncertainty. In contrast, the default mode network, which includes the medial prefrontal gyrus, inferior parietal lobule, and middle temporal gyrus, exhibited higher activation when more prior information was available. This approach highlights how applying AI technology to neuroscience questions can enhance our understanding of neural mechanisms underlying scene recognition.

•Images with varying levels of naturalness were generated using deep learning•Tasks involving these images manipulated the uncertainty of both likelihood and prior•Human scene recognition performance was modeled within a Bayesian framework•Brain regions representing likelihood and prior information were identified

Images with varying levels of naturalness were generated using deep learning

Tasks involving these images manipulated the uncertainty of both likelihood and prior

Human scene recognition performance was modeled within a Bayesian framework

Brain regions representing likelihood and prior information were identified

Neuroscience; Sensory neuroscience; Cognitive neuroscience