Position Statement: The Predictive Mind (Predictive Processing / Bayesian Brain Framework)
The Predictive Processing (PP) framework, also known as the Predictive Mind or Bayesian Brain hypothesis (Friston 2010; Clark 2013; Hohwy 2013), proposes that the brain functions as a hierarchical prediction machine. It continuously generates top-down predictions about sensory input, compares them with bottom-up signals, and minimizes prediction error through active inference or perceptual updating. This unified model elegantly explains perception, action, attention, learning, emotion, and many aspects of psychopathology.
We regard Predictive Processing as one of the most powerful, parsimonious, and empirically supported frameworks in contemporary cognitive neuroscience and philosophy of mind. It has unified diverse phenomena under a single computational principle (free-energy minimization) and generated a wealth of testable predictions validated by neuroimaging, behavioral experiments, and clinical research. Our recent work (The Evolutionary Inevitability of Predictive Processing, Kriger 2026) further establishes PP as not merely empirically promising but physically and evolutionarily inevitable for adaptive systems facing realistic constraints such as finite neural conduction velocity, body size scaling, and environmental mediums (e.g., water, air).
Key strengths:
- Strong neurophysiological evidence (e.g., predictive coding in sensory cortices, mismatch negativity).
- Integration of perception and action into a single active inference loop.
- Successful applications in psychiatry (e.g., aberrant precision weighting in schizophrenia) and artificial intelligence (e.g., variational inference models, predictive coding networks).
- Evolutionary necessity: PP emerges as the only scalable architecture for overcoming latency in punishing, dynamic environments, as formalized by the Predictive Viability Law (mutual information I(X; Y_{t+τ} | Y_t) > 0 required for persistence).
While PP provides an outstanding mechanistic account of functional and access consciousness, it does not fully resolve the hard problem of phenomenal consciousness — why prediction-error minimization should be accompanied by subjective experience at all.
Our proposed approach
We fully endorse Predictive Processing as a leading computational and functional theory of cognition:
- Use it as a unifying framework for understanding perception, action, learning, and psychopathology, emphasizing its evolutionary inevitability.
- Apply operational terminology: describe PP mechanisms in terms of “prediction-error minimization,” “precision weighting,” and “active inference” without equating them directly with subjective experience.
- Integrate PP insights with our Functional Sufficiency Framework to assess the explanatory value of predictive mechanisms in cognition, including predictions for AI systems converging on PP architectures due to physical constraints.
Institute of Integrative and Interdisciplinary Research 