Pennartz’s Neurorepresentationalism

Neuroscientist Cyriel Pennartz’s Neurorepresentationalism holds that phenomenal consciousness is a multimodal, situational survey: a sensorily rich, spatially encompassing, best-guess representation of one’s body and environment, constructed by predictive inference and serving deliberate, goal-directed behavior rather than reflexes or habits (Pennartz, 2015, 2018, 2022). Its distinctive conjecture concerns qualia. Because cortex receives uniform spike trains and has no intrinsic knowledge of where its inputs originate, the qualitative identity of an experience—what makes it visual rather than auditory—must be fixed partly by feature-selective coding and partly by the correlational and predictive relations among sensory modalities, so that experience arises only at the level of a large-scale multimodal superinference assembled from simultaneous, multi-level representations (Pennartz, 2009, 2022). Pennartz characterizes the position as a non-reductionistic materialism in which experience stands in emergent correspondence to neural activity—real and supervenient, but exerting no top-down causation—and derives from it graded indicators of consciousness for animals, machines, and unresponsive patients, now under test in a structured adversarial collaboration with Integrated Information Theory and Active Inference.

In other words, neurorepresentationalism propose that phenomenal consciousness arises through the brain’s construction of a unified, spatially encompassing representation of the body and surrounding world. Developed in The Brain’s Representational Power, the account combines predictive processing with multiple levels of representation, culminating in a multimodal “situational survey” whose qualitative contents constitute experience (Pennartz, 2015, 2022). Its distinctive claim is that conscious qualities emerge through the organization and relationships of representational networks, without an internal spectator. This is a non-reductive physicalist account: experience is real and dependent on neural activity, while phenomenology and its neural basis constitute one functional entity rather than independently acting causes (Pennartz and Engeser, 2026).

Five Hallmarks and a Biological Function

Pennartz begins from phenomenology rather than mechanism, identifying five hallmarks of healthy human experience: multimodal richness, situatedness and immersion, unity and integration, dynamics combined with stability, and intentionality, the last treated as overarching because neural activity must be interpreted as referring to something other than itself (Pennartz, 2022). Perspectivalness is excluded as a necessary hallmark because it does not generalize across senses such as taste.

The function of consciousness is to present the subject with a survey of its situation that serves planning. Reflexes and habits run fast and largely without awareness; goal-directed behavior, which requires sustaining a representation of a goal and is tied to model-based learning, needs a rich, integrated picture of world and body. Pennartz argues that conscious contents do not depend on acute or long-term brain–environment interaction, and that the systems implementing conscious representation are densely connected to, but distinct from, those mediating goal-directed behavior and declarative memory (Pennartz, 2018). Consciousness is thus coupled to deliberation without being identical to it; daydreaming still counts.

Prediction without Action

Mechanistically, Neurorepresentationalism sits within predictive processing. Representations are hypotheses about the causes of sensory input, learned largely unsupervised and implementable in deep, Hebbian predictive-coding networks (Dora et al., 2021; Pennartz, 2022). Unlike Active Inference, it denies that overt or covert action is constitutive of experience: paralysis, locked-in syndrome, REM dreaming, and imagery show conscious content persisting without motor engagement. The vehicles of representation are confined to the brain, while their contents are experienced as located in the world. Pennartz rejects any literal projection of experience outward; what we inhabit is a brain-generated virtual reality that feels like direct contact because it reliably guides action.

Modality Identification and the Basis of Qualia

The theory’s signature problem is modality identification. Müller’s doctrine of labeled lines ties each modality to its receptors and nerves, but Pennartz argues this merely defers the question to the brain, which receives uniformly shaped action potentials from all afferents (Pennartz, 2009, 2022). Qualitative distinctness must therefore be constructed: partly by submodality-specific feature detectors, partly by the topology of cross-modal correlations and predictions, with an enriched multimodal experience corresponding to the attractor state of a dynamic associative network of unimodal and multimodal maps (Pennartz, 2009).

A thought experiment sharpens the point. A V1–MT loop isolated from the rest of cortex, recurrent feedback intact, could not sustain conscious motion vision, because it has no means of identifying its signals as visual motion. This consciousness-in-a-bottle argument implies that a minimal substrate must span several submodalities and requires lateral connectivity across nominally unimodal areas—a claim that distinguishes the view from Lamme’s Recurrent Processing Theory, on which local recurrence within sensory cortex suffices.

Multi-Level Representation and Superinference

Pennartz distinguishes four representational levels: single neurons coding features, local ensembles coding within a submodality, unimodal metanetworks binding object hypotheses, and multimodal metanetworks integrating across senses, at which phenomenal experience arises (Pennartz, 2022). These are functional rather than anatomical levels, unfolding simultaneously rather than as serial way stations; V1 and inferotemporal cortex both contribute. Conscious experience is the phenomenological result of a superinference spanning low- and high-level hypotheses across modalities. The levels are not higher-order in the metarepresentational sense: no state becomes conscious by being represented by another.

Emergent Correspondence and the Hard Problem

Pennartz defends a non-reductive representationalism. Experience depends on neural activity but not conversely, and both levels are held to be real; phenomenal experience stands on an equal ontological footing with its electrophysiological basis, emergent and supervenient yet not an epiphenomenon (Pennartz, 2022). Invoking Kim’s overdetermination worry, he denies mental top-down causation.

On the Hard Problem, Pennartz locates the felt impossibility of deriving experience from neurons in the limits of imagination: shaped by prior experience, we never learned to “look under the hood” of our own imagery (Pennartz, 2022). He disclaims a tangible solution, reframing instead: qualities such as color and shape are the best available representations for an agent planning action, and a multimodal survey is functionally superior to an undifferentiated stream of information. Pennartz distances the view from classical functionalism, which he reads as neglecting qualitative character, while not ruling out machine consciousness in principle.

Evidence and Explanatory Limits

The framework encourages converging indicators of consciousness in nonverbal beings, including goal-directed behavior, episodic-like memory, susceptibility to illusions, and relevant neurophysiology. None independently proves experience; their convergence strengthens an inference while leaving room for alternative explanations (Pennartz, Farisco, and Evers, 2019).

A recent methodological development further qualifies the earlier language of essential hallmarks: Bassler and Pennartz (2026) argue that theories should be evaluated through testable lawful regularities rather than definitions claiming to capture consciousness’s essence. This shifts evaluative emphasis without itself supplying a new explanation of phenomenality.

The outstanding question is whether the proposed organization explains why experience exists, beyond explaining its structured contents and adaptive usefulness. Emergent correspondence states a substantive brain–experience relationship, but does not independently demonstrate why multimodal superinference must feel like anything. The framework thus combines a developed account of phenomenal organization with an unresolved explanatory bridge from representation to subjectivity.

Indicators, Rival Theories, and Testing

The theory yields six graded indicators of consciousness: goal-directed behavior and model-based learning; neural substrates for integrative multimodal representation; psychometrics and metacognition; episodic memory; susceptibility to illusions and multistable perception; and specific visuospatial behaviors (Pennartz et al., 2019). Consistency across indicators, not any single criterion, carries evidential weight; rodents and birds score positively. Responding to theory-derived indicators for AI systems (Butlin et al., 2026), Pennartz argues that internal-architecture assessments must be validated by long-term behavioral-cognitive study, since no theory of consciousness commands consensus (Pennartz, 2026).

Pennartz contrasts his view with Global Workspace Theory, which he reads as addressing the logistics of information distribution and hence access rather than phenomenal content, and with Integrated Information Theory, whose panpsychist implications and lack of a world-matching constraint he contests (Pennartz, 2022). Its neural core is sensory corticothalamic hierarchies with extensive cross-modal connectivity; frontal executive and motor systems are not considered essential.

In the INTREPID adversarial collaboration, which Pennartz directs, Neurorepresentationalism, Integrated Information Theory, and Active Inference are tested against one another through jointly designed experiments, with replications, targeting their incompatible predictions (Corcoran et al., 2025). A 2026 protocol examines perceived spatial extent around the visual blind spot. Neurorepresentationalism predicts no major disruption when contextual information permits adequate spatial inference; the protocol presents experimental designs and simulated outcomes, not completed empirical confirmation (Abbatecola et al., 2026).

Taxonomic Placement

Pennartz calls Neurorepresentationalism a non-reductionistic form of materialism, in which experience and its neural basis are equally real (Pennartz, 2022), a description that might point to Landscape’s Non-Reductive Physicalism category. However, Landscape places it in Materialism, under First-Order, because classification follows metaphysics, and the decisive driver here is causal. Non-Reductive Physicalism is often marked by top-down causation, in which conscious content does causal work of its own; Pennartz rejects exactly this. Experience is emergent and supervenient, corresponds to neural activity simultaneously rather than standing in a causal sequence with it, and exerts no top-down influence on the neurons that realize it. Following Kim’s overdetermination argument, causal efficacy belongs to those neural processes, so consciousness is not an epiphenomenon yet has no independent causal powers, and physical causal closure remains intact. On Landscape’s reading, his irreducibility concerns levels of organization rather than an autonomous mental order, as with Bunge’s Emergent Materialism and Searle’s Biological Naturalism. Within Materialism, First-Order fits: phenomenal character is the content of world-directed sensory representation, and no higher-order state is required.

References

Abbatecola, C., ’t Hart, B. M., Montabes de la Cruz, B. M., Petro, L. S., Pennartz, C. M. A., et al. (2026). Protocol for investigating the warping of spatial experience across the blind spot to contrast predictions of the Integrated Information Theory and Predictive Processing accounts of consciousness. PLOS One, 21(1), e0340593. doi:10.1371/journal.pone.0340593.

Bassler, M., and Pennartz, C. M. A. (2026). Overcoming essentialism: Evaluation of theories of consciousness by the lawful regularities they predict. Neuroscience of Consciousness, 2026(1), niag069. doi:10.1093/nc/niag069.

Butlin, P., Long, R., Bayne, T., Bengio, Y., Birch, J., Chalmers, D., Constant, A., Deane, G., Elmoznino, E., Fleming, S. M., Ji, X., Kanai, R., Klein, C., Lindsay, G., Michel, M., Mudrik, L., Peters, M. A. K., Schwitzgebel, E., Simon, J., & VanRullen, R. (2026). Identifying indicators of consciousness in AI systems. Trends in Cognitive Sciences, 30(6), 488–501. https://doi.org/10.1016/j.tics.2025.10.011

Corcoran, A. W., Haun, A. M., Dorman, R., Tononi, G., Friston, K. J., Pennartz, C. M. A., & TWCF INTREPID Consortium. (2025). Integrated information and predictive processing theories of consciousness: An adversarial collaborative review. arXiv:2509.00555. https://doi.org/10.48550/arXiv.2509.00555

Dora, S., Bohte, S. M., & Pennartz, C. M. A. (2021). Deep gated Hebbian predictive coding accounts for emergence of complex neural response properties along the visual cortical hierarchy. Frontiers in Computational Neuroscience, 15, 666131.

Pennartz, C. M. A. (2009). Identification and integration of sensory modalities: Neural basis and relation to consciousness. Consciousness and Cognition, 18(3), 718–739.

Pennartz, C. M. A. (2015). The Brain’s Representational Power: On Consciousness and the Integration of Modalities. MIT Press.

Pennartz, C. M. A. (2018). Consciousness, representation, action: The importance of being goal-directed. Trends in Cognitive Sciences, 22(2), 137–153. https://doi.org/10.1016/j.tics.2017.10.006

Pennartz, C. M. A. (2022). What is neurorepresentationalism? From neural activity and predictive processing to multi-level representations and consciousness. Behavioural Brain Research, 432, 113969. https://doi.org/10.1016/j.bbr.2022.113969

Pennartz, C. M. A. (2026). How can we validate theory-derived indicators of consciousness in artificial intelligence? Trends in Cognitive Sciences, 30(7), 573–574.

Pennartz, C. M. A., Farisco, M., & Evers, K. (2019). Indicators and criteria of consciousness in animals and intelligent machines: An inside-out approach. Frontiers in Systems Neuroscience, 13, 25. https://doi.org/10.3389/fnsys.2019.00025