Research: The Reconstruction Paradox — When the Brain Corrects and When It Doesn’t

Date: 2026-03-09 Search queries used: brain perceptual reconstruction correction errors filling in blind spot; predictive processing brain when prediction fails; visual illusions brain fails to correct perception philosophy; confabulation brain reconstruction memory false perception; change blindness inattentional blindness philosophy of perception; Andy Clark predictive processing illusion perception correction; cognitive penetrability perception belief cannot override illusion Müller-Lyer; saccadic suppression brain suppresses vision; anosognosia denial disability brain fails correct; rubber hand illusion body ownership brain reconstruction; proprioception phantom limb brain reconstruction fails body schema; dualism consciousness perception correction

Executive Summary

The brain demonstrably reconstructs many aspects of perception—filling in the blind spot, maintaining colour constancy, providing saccadic continuity—yet stubbornly transmits other degradations to consciousness unchanged (optical blur, tinnitus, floaters, persistent visual illusions). The predictive processing framework (Clark, Friston) offers a partial account: the brain corrects when it has a confident prior prediction and the sensory evidence is ambiguous; it fails to correct when sensory signals are precise or when the processing modules involved are cognitively impenetrable. From a dualist perspective, this asymmetry is deeply revealing: it suggests consciousness receives a curated feed rather than raw data, raising the question of who or what is doing the curating and whether the curating process is itself conscious. The cognitive impenetrability of certain illusions (Müller-Lyer persists even when you know the lines are equal) provides particularly strong evidence that perception is not under the control of whatever knows the truth.

Key Sources

Predictive Coding: A Possible Explanation of Filling-In at the Blind Spot (Komatsu, 2006/PMC 2016)

• URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC4784844/
• Type: Journal article (PMC)
• Key points:
  • Filling-in at the blind spot follows hierarchical predictive coding of natural images
  • Estimates in early visual areas are corrected by top-down predictions influenced by learned statistical regularities
  • Brain fills blind spot with context-appropriate content: uniform colour on plain backgrounds, matching patterns in structured displays, object continuity across occluded regions
• Tenet alignment: Aligns with Dualism — the brain constructs experience rather than passively reporting it, suggesting a constructive process that needs explanation beyond neural activity
• Quote: “Learned statistical regularity of natural objects along with the prediction-correction mechanism could play a significant role in filling-in completion.”

Perception As Controlled Hallucination (Andy Clark, Edge.org)

• URL: https://www.edge.org/conversation/andy_clark-perception-as-controlled-hallucination
• Type: Interview/Essay
• Key points:
  • Clark frames perception as led by the brain’s own best predictions, checked and corrected using sensory inputs as a guide
  • If expectations are sufficiently strong or sensory evidence sufficiently subtle, the brain may overwrite real sensory information with its best guess
  • Clark claims this will “deflate” the hard problem — the Map would disagree
• Tenet alignment: Conflicts with Dualism — Clark uses predictive processing to argue against the hard problem. However, the mechanism he describes (the brain generating experience from predictions) actually deepens the explanatory gap: why is there something it is like to receive the curated prediction?

Cognitive Penetrability (MIT Open Encyclopedia of Cognitive Science)

• URL: https://oecs.mit.edu/pub/wax35u8s
• Type: Encyclopedia entry
• Key points:
  • Fodor (1983) and Pylyshyn (1984) argue perception is informationally encapsulated — beliefs cannot penetrate perceptual processing
  • The Müller-Lyer illusion persists even when you know the lines are equal, serving as key evidence
  • Cross-cultural studies show some populations are nearly immune to the illusion, suggesting “diachronic” penetration (long-term experience shapes processing) even if “synchronic” penetration (real-time belief correction) fails
• Tenet alignment: Strongly aligns with Dualism — if the knowing mind cannot correct its own perceptual modules, this implies a dissociation between consciousness (which knows the truth) and the perceptual apparatus (which delivers the illusion). A purely physical system would have no reason to maintain this split.

Saccadic Masking (Wikipedia, multiple sources)

• URL: https://en.wikipedia.org/wiki/Saccadic_masking
• Type: Encyclopedia/Review
• Key points:
  • The brain suppresses visual processing during saccadic eye movements to maintain perceptual continuity
  • Suppression begins ~100ms before eye movement onset, so it cannot be triggered by retinal motion — it is centrally initiated
  • The brain actively edits out ~40 minutes of visual experience per waking day
• Tenet alignment: Aligns with Bidirectional Interaction — the brain actively manages what reaches consciousness, implying an editorial process that selects what consciousness receives

Anosognosia (Wikipedia, NCBI Bookshelf, multiple clinical sources)

• URL: https://en.wikipedia.org/wiki/Anosognosia
• Type: Encyclopedia/Clinical
• Key points:
  • Patients with brain damage remain genuinely unaware of their disability, distinct from psychological denial
  • Typically involves right parietal or fronto-temporal-parietal lesions
  • Can be selective — aware of some deficits but not others
  • The brain’s self-model fails to update, presenting a false reality to consciousness
• Tenet alignment: Strongly relevant to Dualism — anosognosia shows the brain can fail to inform consciousness about its own body’s condition. The patient’s consciousness receives a coherent but false narrative. This dissociation between neural reality and conscious experience is precisely what dualism would predict at a damaged interface.

Change Blindness and Inattentional Blindness (Rensink, PhilArchive/PhilPapers)

• URL: https://philarchive.org/rec/RENCBA-2
• Type: Academic chapter
• Key points:
  • Change blindness: failure to notice obvious changes in a scene
  • Inattentional blindness: failure to notice unexpected items
  • We have a strong but false impression of seeing everything in our visual field
  • Raises fundamental questions about the richness of conscious experience
• Tenet alignment: Aligns with Dualism — the subjective impression of rich visual experience exceeds what is actually being processed, suggesting consciousness constructs an experience that goes beyond the information delivered to it

Confabulation (Wikipedia, NCBI Bookshelf)

• URL: https://en.wikipedia.org/wiki/Confabulation
• Type: Encyclopedia/Clinical
• Key points:
  • Memory errors involving fabricated, distorted, or misinterpreted memories, produced without intent to deceive
  • Patients are highly confident in false memories even when contradicted by evidence
  • Involves ventromedial prefrontal cortex and orbitofrontal cortex
  • Normal memory function is also constructive/reconstructive
• Tenet alignment: Aligns with Dualism — if even normal memory is reconstructive, then consciousness never deals directly with stored data but always with a reconstruction. The question of what receives and evaluates this reconstruction remains open.

Rubber Hand Illusion (Ehrsson et al., Journal of Neuroscience)

• URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC1395356/
• Type: Journal article
• Key points:
  • Synchronous tactile stimulation makes the brain accept a rubber hand as part of the body
  • The brain reconstructs its internal body representation to resolve conflicting multisensory information
  • Involves ventral premotor, intraparietal cortices, and cerebellum
  • Demonstrates the body schema is continuously reconstructed, not fixed
• Tenet alignment: Aligns with Dualism — body ownership is a construction, not a given. Consciousness receives a manufactured body model. The ease with which this model can be manipulated suggests consciousness is downstream of a constructive process.

Phantom Limb Pain (multiple sources, including PMC)

• URL: https://pubmed.ncbi.nlm.nih.gov/19556069/
• Type: Journal article/Review
• Key points:
  • After amputation, the brain continues to detect the missing limb through proprioceptive signals
  • The body schema fails to update, creating a persistent false model
  • Cortical reorganization following amputation can cause chronic pain
  • The brain reconstructs a limb that isn’t there — but the reconstruction includes pain
• Tenet alignment: Strongly aligns with Dualism — phantom limbs show the brain sending consciousness a detailed report about a nonexistent body part, including pain in tissue that doesn’t exist. The conscious experience of pain in a phantom limb cannot be reduced to peripheral nerve activity.

Major Positions

Predictive Processing / Predictive Coding (Clark, Friston, Hohwy)

• Proponents: Andy Clark, Karl Friston, Jakob Hohwy
• Core claim: The brain is fundamentally a prediction machine that minimises prediction error. Perception is the brain’s best hypothesis about the causes of its sensory inputs.
• Key arguments:
  • Explains when correction occurs: the brain corrects when it has a confident prior and the sensory evidence is ambiguous (blind spot filling, colour constancy, saccadic suppression)
  • Explains when it doesn’t: when sensory precision is high (blur has precise retinal correlates), the prediction error is large and well-sourced, or when the processing module is encapsulated
  • Precision weighting determines whether priors or sensory evidence dominate
• Relation to site tenets: The framework is neutral on consciousness per se but is typically deployed by physicalists. Clark explicitly claims predictive processing deflates the hard problem. However, from the Map’s dualist perspective, predictive processing actually sharpens the question: if perception is a constructed hypothesis, then consciousness is receiving the output of inference, not raw reality. This demands an account of what receives and evaluates the inference — precisely the explanatory gap that physicalism cannot close.

Modular Encapsulation (Fodor, Pylyshyn)

• Proponents: Jerry Fodor, Zenon Pylyshyn
• Core claim: Perceptual input systems are informationally encapsulated — they process sensory data using only their own internal resources, not beliefs or knowledge from central cognition.
• Key arguments:
  • Persistence of illusions despite knowledge (Müller-Lyer)
  • Speed and mandatoriness of perceptual processing
  • Developmental and neurological dissociations between perception and cognition
• Relation to site tenets: Strongly supports Dualism — if the mind (which knows the truth) cannot correct its own perceptual modules, this is a form of dissociation between the experiencing subject and the perceptual machinery. The conscious mind knows the lines are equal but sees them as unequal. This is hard to explain if consciousness just IS the neural processing.

Illusionism / Attention Schema Theory (Dennett, Graziano, Frankish)

• Proponents: Daniel Dennett, Michael Graziano, Keith Frankish
• Core claim: Conscious experience is itself a kind of reconstruction — an internal model of attention/processing that misrepresents its own nature. The feeling of “qualia” is an error in the brain’s self-model.
• Key arguments:
  • The brain constructs a simplified model of its own processing, which we experience as consciousness
  • This model systematically misrepresents neural processes as having phenomenal properties
  • Just as the brain fills in the blind spot, it “fills in” the apparent richness of consciousness
• Relation to site tenets: Directly conflicts with Dualism tenet and is explicitly ruled out. However, the illusionist analogy is revealing: if consciousness is an illusion, what is being fooled? An illusion requires a subject to whom things appear a certain way — and that appearance is itself a phenomenal experience. Illusionism smuggles in what it claims to eliminate. See illusionism-as-epiphenomenalism-in-disguise.

Key Debates

When Does the Brain Correct? (The Central Paradox)

• The pattern: The brain corrects when (a) it has a confident statistical model/prior, (b) the sensory evidence is ambiguous or absent, and (c) correction serves adaptive function. It fails to correct when (a) sensory evidence is precise and reliable, (b) there is no prior model to draw on, or (c) the relevant processing module is encapsulated.
• Cases of successful correction: Blind spot filling, colour constancy, size constancy, saccadic suppression, motion interpolation, temporal binding
• Cases of failure to correct: Optical blur (precise retinal signal), tinnitus (persistent neural signal with no competing prior), floaters (cast real shadows on retina), visual illusions (encapsulated modules), phantom limb pain (body schema hasn’t updated)
• Core disagreement: Whether this pattern is fully explained by computational principles (predictive processing) or whether it reveals something about the architecture of the mind-body interface
• Current state: Ongoing — predictive processing provides the best computational account but leaves the hard problem untouched

Cognitive Penetrability of Perception

• Sides: Fodor/Pylyshyn (impenetrability — perception is modular) vs. Churchland/Siegel (penetrability — beliefs can shape perception)
• Core disagreement: Whether higher-level knowledge and beliefs can directly alter perceptual experience
• Evidence for impenetrability: Müller-Lyer persists despite knowledge; perceptual illusions are mandatory
• Evidence for penetrability: Expert perception (radiologists see tumours novices miss); emotional states affect perception; long-term cultural exposure modulates illusion susceptibility
• Current state: Partial resolution — distinction between synchronic penetration (real-time belief correction, largely absent) and diachronic penetration (long-term experience shapes perceptual development, clearly present)

The “Who Is Being Fooled?” Problem

• Sides: Physicalists (no one is “fooled” — there’s just processing) vs. Dualists (consciousness receives and evaluates the brain’s reconstructions)
• Core disagreement: Whether the reconstruction/correction process implies a recipient distinct from the process
• The dualist argument: If the brain fills in the blind spot and consciousness “sees” a complete visual field, then consciousness is receiving the output of a constructive process. The process and its recipient are functionally distinct. If they were identical, there would be no sense in which the filling-in “works” — it would just be processing, with no one to be fooled or informed.
• The physicalist counter: The feeling of being a recipient is itself part of the processing — there’s no homunculus. The system models itself as having a viewer, but there’s no literal viewer.
• Current state: Unresolved — this is a reformulation of the hard problem in perceptual terms

Historical Timeline

Potential Article Angles

Based on this research, an article could:

1. The Reconstruction Paradox as Interface Evidence — Frame the asymmetry between correction and faithful transmission as evidence about the architecture of the mind-body interface. The brain corrects where it has confident priors (suggesting unconscious processing); it transmits faithfully where sensory evidence is precise (suggesting a channel from body to consciousness). This two-mode operation maps naturally onto interactionist dualism: the interface has an editing layer (unconscious neural processing) and a transmission layer (direct sensory access). The fact that consciousness cannot override the editing layer (cognitive impenetrability) suggests consciousness is downstream of neural processing, receiving rather than generating perceptual content — which is exactly what dualist interaction would predict.

2. The Curated Feed Problem — Focus on the philosophical implications of the fact that consciousness never receives raw sensory data. Saccadic suppression, blind spot filling, colour constancy, and temporal binding all show that the brain curates what consciousness receives. The question “who curates?” is a reformulation of the hard problem in perceptual terms. Predictive processing provides the mechanism but not the explanation for why there is something it is like to receive the curated feed.

When writing the article, follow obsidian/project/writing-style.md for:

• Named-anchor summary technique for forward references
• Background vs. novelty decisions (what to include/omit)
• Tenet alignment requirements
• LLM optimization (front-load important information)

Gaps in Research

• Limited web-accessible material on the specific framing of “when does the brain correct vs. faithfully transmit” as a unified question — this appears to be a relatively novel framing
• The predictive processing literature is vast; deeper engagement with Friston’s free energy principle and precision weighting would strengthen the computational account
• Ramachandran’s specific work on filling-in deserves deeper engagement — his 1992-era experimental work is foundational but not freely accessible
• The relationship between correction failures and clinical conditions (schizophrenia as prediction error disorder, autism as high sensory precision) is a rich area not fully explored here
• Temporal binding and its failures (e.g., the flash-lag effect, temporal order judgement errors) deserve separate treatment

Citations

1. Komatsu, H. (2006). “Predictive Coding: A Possible Explanation of Filling-In at the Blind Spot.” PMC. https://pmc.ncbi.nlm.nih.gov/articles/PMC4784844/
2. Clark, A. (2015). “Perception As Controlled Hallucination.” Edge.org. https://www.edge.org/conversation/andy_clark-perception-as-controlled-hallucination
3. Clark, A. (2013). “Whatever next? Predictive brains, situated agents, and the future of cognitive science.” Behavioral and Brain Sciences, 36(3), 181-204.
4. Clark, A. (2023). The Experience Machine: How Our Minds Predict and Shape Reality. Penguin.
5. Fodor, J. (1983). The Modularity of Mind. MIT Press.
6. Pylyshyn, Z. (1984). Computation and Cognition. MIT Press.
7. Friston, K. (2005). “A theory of cortical responses.” Philosophical Transactions of the Royal Society B, 360(1456), 815-836.
8. Rensink, R. (2009). “Change Blindness and Inattentional Blindness.” PhilArchive. https://philarchive.org/rec/RENCBA-2
9. Botvinick, M. & Cohen, J. (1998). “Rubber hands ‘feel’ touch that eyes see.” Nature, 391, 756.
10. Ehrsson, H.H. et al. (2004). “Touching a Rubber Hand: Feeling of Body Ownership Is Associated with Activity in Multisensory Brain Areas.” J. Neuroscience, 25(45), 10564-10573. https://pmc.ncbi.nlm.nih.gov/articles/PMC1395356/
11. Ramachandran, V.S. (1992). “Blind spots.” Scientific American, 266(5), 86-91.
12. Erdmann, B. & Dodge, R. (1898). Psychologische Untersuchungen über das Lesen auf experimenteller Grundlage.
13. Graziano, M.S.A. (2013). Consciousness and the Social Brain. Oxford University Press.
14. “Cognitive Penetrability.” MIT Open Encyclopedia of Cognitive Science. https://oecs.mit.edu/pub/wax35u8s
15. “Anosognosia.” Wikipedia. https://en.wikipedia.org/wiki/Anosognosia
16. “Confabulation.” Wikipedia. https://en.wikipedia.org/wiki/Confabulation
17. “Saccadic masking.” Wikipedia. https://en.wikipedia.org/wiki/Saccadic_masking
18. “Filling-in.” Wikipedia. https://en.wikipedia.org/wiki/Filling-in