Comprehensive review: Computational modelling of schizophrenia.

Area of Science:

• Neuroscience
• Computational Psychiatry
• Psychopathology

Background:

• Schizophrenia presents diverse symptoms, prompting computational modeling approaches.
• Models range from abstract behavioral (Bayesian) to biologically realistic (neural networks).
• Existing models often support diverging hypotheses, leading to a fragmented research landscape.

Purpose of the Study:

• To synthesize computational modeling literature for schizophrenia and psychotic symptoms.
• To categorize studies supporting key hypotheses: dopamine, glutamate, GABA, dysconnection, and Bayesian inference.
• To compare model predictions with empirical evidence and identify under-investigated hypotheses.

Main Methods:

• Systematic review of computational modeling studies in schizophrenia.
• Categorization of models based on underlying hypotheses (dopamine, glutamate, GABA, dysconnection, Bayesian inference).
• Comparative analysis of model predictions against empirical findings.

Main Results:

• Synthesis of computational approaches for schizophrenia, linking behavioral and neurobiological models.
• Evaluation of evidence supporting major etiological hypotheses.
• Identification of specific hypotheses requiring further computational investigation.

Conclusions:

• Computational modeling provides a framework for understanding schizophrenia's complexity.
• A critical comparison of model predictions with empirical data is crucial for advancing the field.
• Further research should focus on under-explored computational hypotheses to refine our understanding of schizophrenia.