Gładziejewski, P. (2025). Real patterns, the predictive mind, and the cognitive construction of the manifest image. Synthese 206(5): 225. doi:10.1007/s11229-025-05311-0.
Key takeaways
From the reading, NOTE-1056.
- Inference is compression. With optimal codes, maximising the posterior is minimising total description length, so “Bayesian inference is real-pattern extraction” (p. 8). Variational free energy’s complexity and accuracy terms are analogous to the model and data terms of minimum description length (p. 10). The first is a derivation, the second an analogy.
- The manifest image is structure in the senses. Generative models “specify probabilistic and causal structural descriptions of sensory patterns” (p. 12). They are constrained by the signal except in off-line simulation such as imagery or dreams (p. 13).
- Veridical in the vat. Matched sensory streams compress alike, so an envatted twin’s door is “as real as the door you open” (p. 13). What differs, the realizer, lies outside the manifest image (p. 14). This follows only if manifest content is wholly structural.
- The self is a compression. The self-model “achieves compression by positing a single, stable entity as the common cause of many endogenous patterns” (p. 16). Real as a pattern, unreal as a substance. “The conflict between those positions is apparent”, and it lies in the criteria of reality (p. 17).
- Dissociation as worse compression. Some dissociative states may be better compressed by a model positing two or more narrative selves (footnote 10, p. 17).
- Ordinary objects. On Bird’s (2023) compression criterion a cat composes and “whiskers + crust + dust” does not, so perception is evidence for ordinary objects (§4.2).
- Its limits. A sketch, by its own account, and conditional on a realist reading of predictive processing (footnote 2).
Why this is here
The owner’s exchange of 2026-10-21, “Experience and interpretation”, asked whether experience is constituted by interpretation or produced by it. The exchange cited this paper for joining real patterns to predictive processing and compression, and for its treatment of the self as a real pattern whose representation can exaggerate its unity. Bears on CLAIM-355.
Correction from the reading. The first half of that description is right. The second is not in the paper. It never says the self’s representation exaggerates its unity. It says the self-model compresses by positing one common cause, and that whether the result is a real self depends on which criterion of reality one adopts. For CLAIM-355 this cuts both ways. It supports the claim’s split between a real pattern and a constructed single owner. But it makes the single owner a lossy compression rather than a misdescription, which is the Drayson objection CLAIM-355 already names. Its one case of a single-self model going wrong is footnote 10’s dissociation. The paper does not mention consciousness, illusion or collectives.
Filed from this work on 2026-10-22: THEORY-872 (the self as a compression pattern; the realism dispute as one of criteria) and THEORY-877 (manifest content as structure in sensory data, veridical in matched streams).