External and internal modes of inference.
Perception actively integrates new sensory information with internal predictions based on prior experiences. Our research explores how the brain dynamically shifts between two perceptual modes: an external mode (focused on immediate sensory input) and an internal mode (guided by past experiences). These shifts are key to understanding psychotic disorders such as schizophrenia.
Using large-scale behavioral data, we demonstrated systematic fluctuations in perception between an external mode (high accuracy, minimal reliance on previous experiences) and an internal mode (strongly biased by perceptual history)1. These fluctuations optimize perception, balancing novelty detection and stability.
False perceptions (“false alarms”) arise from internal predictions (“predictive templates”), especially during internal mode2. Computational modeling (HMM-GLM) revealed these templates fluctuate dynamically and linked them to the mechanisms underlying hallucinations.
We showed that pharmacologically-induced NMDAR hypofunction (using S-ketamine) prolongs external mode dominance, causing erratic perceptual inferences. This mirrors key aspects of perceptual disruptions seen in schizophrenia and offers a mechanistic explanation for episodic psychotic symptoms3.
Dynamic interplay between external and internal perceptual modes significantly influences psychosis. Over-reliance on external sensory input can produce unstable perceptions, leading to maladaptive internal predictions. When perception shifts to internal mode, these maladaptive predictions dominate perception and cognition, triggering hallucinations and delusions.
We combined multiple complementary methodologies:
Clarifying these perceptual dynamics offers important implications for psychiatric care. Dynamic perceptual mode shifts might serve as biomarkers for psychosis, informing novel neuromodulatory treatments targeting NMDAR function and perceptual stability.