Clinical Manifestations.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Psychiatry

Background:

• Anosognosia, or lack of illness awareness, is prevalent in neurodegenerative diseases.
• Implicit knowledge may influence behavior even without conscious awareness.
• The Emotional Stroop task can reveal preconscious processing of emotional stimuli.

Purpose of the Study:

• To assess the validity of a generative model based on Active Inference for simulating behavior in neurodegenerative diseases.
• To investigate implicit knowledge in Alzheimer's Disease (AD) patients using computational modeling.
• To explore potential for implicit metacognitive assessment of anosognosia.

Main Methods:

• Administered a computerized Emotional Stroop task to 37 healthy controls and 40 AD patients.
• Developed a Partially Observable Markov Decision Process (POMDP) model within the Active Inference framework.
• Incorporated hidden states for word meaning, color, action, and task sequence, with subject-specific parameters.

Main Results:

• Simulated reaction times mirrored the non-normal distribution of observed data.
• The model reproduced specific reaction time slowing for disease-related words when disease-related word salience was high and negative word salience was low.
• This pattern suggests implicit knowledge of the disease in AD patients.

Conclusions:

• The Active Inference model successfully reproduced behavioral patterns observed in AD patients.
• Model estimates show potential for integration with neuroimaging and neuropsychological data.
• This approach may enable implicit metacognitive assessment of anosognosia in neurodegenerative conditions.