Biol Psychiatry
May 25, 2018
Philipp Sterzer, Rick A. Adams, Paul Fletcher et al.
901 citations
Psychosis may stem from altered predictive coding in the brain, where prior beliefs are combined with sensory signals to interpret the world. The canonical account suggests that weakened prior beliefs relative to sensory data lead to maladaptive inferences. However, evidence is mixed: some studies find weakened priors, others stronger priors. Hallucinations and delusions may involve distinct predictive coding alterations. Challenges include non-uniform deficits across sensory modalities, hierarchical processing where lower and higher levels interact non-linearly, and the role of active inference—anticipating how actions affect sensations. Considering these complexities could reconcile conflicting findings and yield a more comprehensive framework for psychosis.
Frontiers in Human Neuroscience
December 22, 2016
Philipp Sterzer, Aaron Mishara, Martin Voss et al.
78 citations
Thought insertion in schizophrenia may arise from altered Bayesian inference within the predictive coding framework. Early 20th-century phenomenological accounts by the Heidelberg School described thought insertion as a self-disturbance involving disrupted inner connectedness of thoughts, which become sensory and feel inserted. Mescaline was used as a model psychosis to explore these mechanisms. The authors propose that reduced precision of context-dependent predictions, relative to sensory precision, increases prediction-error signals for internal events like thoughts. This aberrant salience, analogous to that proposed for external events, leads individuals to interpret thoughts as inserted by an alien agent, similar to delusion formation from aberrant sensory salience.
Consciousness and cognition
August 1, 2021
Tim Julian Möller, Yasmin Kim Georgie, Guido Schillaci et al.
16 citations
Self-disorders are increasingly seen as the root cause of schizophrenia, not merely a symptom. This aligns with philosophical views of an enactive self, formed through action and interaction. The authors analyze definitions of the self and evaluate computational theories, particularly Bayesian and predictive processing approaches, for modeling the active self. They assess the implementation and challenges of these models in computational psychiatry and cognitive developmental robotics. Embodied robotic systems are described as valuable tools for assessing, validating, and simulating mechanisms of self-disorders, especially those involving sensorimotor learning, prediction, and self-other distinction. This link offers insights into self-formation and new avenues for treating psychiatric disorders.
Schizophrenia bulletin
October 21, 2021
Leonie Klock, Martin Voss, Markus Weichenberger et al.
Healthy participants experienced thoughts as self-generated when they believed a fake device was deactivated, and attributed thoughts externally when they believed the device was activated—an experience usually reported only by patients diagnosed with schizophrenia. Distinct activations in the medial prefrontal cortex were observed: ventral mPFC activation linked to a sense of thought authorship and dorsal mPFC activation to a diminished sense of thought authorship. This functional differentiation corresponds to research on self- and other-oriented reflection processes and on patients with schizophrenia who show abnormal mPFC activation. Results support the notion that the mPFC might be involved in thought authorship and anomalous self-experiences.