Home / Materialism / Phylogenetic/Evolutionary / Pivovarov’s Evolving Self-Representations

Pivovarov’s Evolving Self-Representations

Pivovarov’s Theory of Evolving Self-Representations (TEVSER) grounds consciousness in the evolution of homeostatic regulation. It identifies the first feelings, or proto-qualia, with differentiated regulatory states organized within a shared internal domain. Following the Good Regulator Theorem, the models embodied in an organism’s regulators constitute its self-representation, which evolves through twelve idealized levels. Phenomenality begins when regulatory signals acquire distinct identities within one integrated internal space, yielding proto-qualia; spatial perspective (“here”), attention, and temporal presence (“now”) then assemble a fully structured conscious world, while intelligence, self-consciousness, and symbolic thought arise at higher levels. Qualia are felt configurations of whole situations. The core claim is an identity: self-representational dynamics are experience viewed from within. TEVSER offers falsifiable hypotheses and conditions for machine consciousness.

VerifiedVerified*
Photo of Igor Pivovarov

Igor Pivovarov

Physicist and AI researcher

Igor Pivovarov is a physicist and artificial-intelligence researcher who heads the Analytical Center of the Institute of Artificial Intelligence at the Moscow Institute of Physics and Technology (MIPT). Trained in theoretical physics at Moscow State University, with graduate study in biophysics, he spent decades as a deep-tech entrepreneur, founding several companies. His work examines the evolution of self-representation, consciousness, and intelligence, as well as biologically inspired architectures for autonomous agents. He co-develops the brain-inspired MARTI agent architectures and chairs the OpenTalks.AI conference.

*This summary was verified by Igor Pivovarov.

Pivovarov’s Evolving Self-Representations

Igor Pivovarov’s Theory of Evolving Self-Representations (TEVSER) identifies phenomenal consciousness with the organized dynamics of an organism’s self-regulation as lived from within, and it derives psyche, consciousness, and intelligence from the evolution of this homeostatic mechanism: the internal models embodied in an organism’s regulators constitute its self-representation (Ω), whose elaboration is ordered into twelve idealized levels, Ω⁰–Ω¹¹ (Pivovarov, 2026). The theory begins with homeostasis and proposes that minimal feeling appears when distinct regulatory signals are organized within a shared internal domain, giving rise to proto-qualia. Thus, phenomenality begins at Ω², when signals from different regulators acquire distinct identities within one integrated internal space; these proto-qualia are later enriched by spatial perspective (Ω³), attention (Ω⁵), and temporal presence (Ω⁶), where a fully structured conscious world first appears.

Evolutionary innovations situate these feelings in an embodied spatial perspective, select what occupies the experiential foreground, and integrate perception with memory into a temporally continuous present. Reflective self-awareness and symbolic thought arise later still. The sequence is an evolutionary hypothesis, and the identification of experience with regulatory dynamics is TEVSER’s central ontological commitment. The claim is an identity, not a generative mechanism: organized self-representational dynamics are experience as lived from within, and their neural description is the same process viewed from outside. Intelligence and reflective self-consciousness occupy higher levels, and the author claims that the framework yields falsifiable hypotheses and conditions for machine consciousness (Pivovarov, 2026).

Regulation as Self-Representation

TEVSER starts from homeostasis, not perception or introspection. A living system persists by monitoring and correcting variables essential to its existence, and by the Good Regulator Theorem, every effective regulator must embody a model of what it regulates (Conant and Ashby, 1970). The organization of these embodied models is the organism’s self-representation. Ω is not an observer’s construct but a functional organization participating in the system’s own regulation; at Ω⁰–Ω¹ it is implicit, nonconceptual, and without phenomenal content. Evolution integrates and differentiates these models; each level, favored because it offers new means of maintaining viability, answers a regulatory problem its predecessor cannot solve and receives approximate phylogenetic anchors (ganglionic nervous systems some 570 million years ago for Ω²), which Pivovarov treats as idealized transitions rather than sharply bounded epochs.

Proto-Qualia and the Onset of Phenomenality

At Ω², signals from different regulators are mapped into a shared, spatially organized internal space in which each retains a distinct identity. Sensations are mappings of individual regulator states; emotions are global states integrating many signals into a unified regulatory value; together they compose a primitive psyche. These differentiated states are proto-qualia: different regulatory conditions are felt differently from within. Because differentiation within an integrated space does the decisive work, phenomenality is withheld from mere regulation and from indiscriminate information integration. Nociception in simple nervous systems (C. elegans is the theory’s illustration) counts as minimal aversive feeling, though Pivovarov cautions that this implies neither consciousness nor reflective awareness. TEVSER thus distinguishes phenomenality, beginning at Ω², from consciousness in the full sense, assembled later.

Building a Conscious World

At Ω³, spatial self-localization establishes a nonreflective, embodied perspective: the subject “here.” Elementary associative learning at Ω⁴ extends regulation beyond fixed reflexes. Attention at Ω⁵ prioritizes internal representations by need and relevance, but is not equated with consciousness. At Ω⁶, object differentiation and memory bind the current scene to recent events, yielding temporal presence, the subject “now.” The conjunction of psyche, spatial perspective, attention, and temporal presence constitutes basic consciousness, reached by progression rather than across a threshold.

A quale, accordingly, is not an isolated property such as redness but the immediate felt configuration of a whole situation, integrating perception, regulatory concerns, remembered events, and possibilities for action. Proto-qualia supply differentiated feeling; full qualia at Ω⁶ add spatial placement, temporal depth, and memory-dependent context.

Identity Rather than Production

Self-representational dynamics do not generate a second, immaterial phenomenon called experience. In Pivovarov’s formulation, “phenomenology is not the outcome of function but its inner side” (Pivovarov, 2026); the neural and functional description is the third-person view of the same process. The “meta-space” (Ψ) of dynamically organized control patterns is a higher level of physical organization, not an additional substance. Offered as a constructive response to the hard problem (Chalmers, 1995), the identification is, Pivovarov acknowledges, a philosophical commitment rather than an experimentally established bridge from function to phenomenality, and one dualists will reject.

Relations to Neighboring Theories

TEVSER’s nearest neighbors are both homeostatic and affective theories, and phylogenetic & evolutionary theories. It differs from homeostatic and affective theories over what experience is first identified with. Panksepp, Damasio, and Solms ground phenomenality in affect; TEVSER makes differentiation of regulatory states the minimal phenomenal fact, with emotion one of its products and explicit valuation (reward-prediction error) arriving only at Ω⁷. It differs from phylogenetic & evolutionary theories in terms of the order of Ω⁰–Ω¹¹. Ginsburg and Jablonka take unlimited associative learning as the marker of minimal consciousness; TEVSER places proto-qualia before elementary associative learning. Where Feinberg and Mallatt specify neurobiological requirements, TEVSER specifies functional-regulatory ones. Unlike Birch’s deliberately theory-light gradualism, TEVSER commits to specific transitions and accepts sharper falsification risk. Against Kriegel’s self-representationalism, its foundational self-representation is the organism’s regulatory model, not a state’s representation of itself. TEVSER recasts rival theories as accounts of particular levels: Baars’s Global Workspace Theory access at Ω⁵, Higher-Order theories at Ω¹⁰, and Anokhin’s acceptor of action results as the goal-setting “virtual regulator” of Ω⁸ (Anokhin, 1974). Unlike Tononi’s Integrated Information Theory, TEVSER confines consciousness to self-regulating systems and supplies an evolutionary mechanism; predictive processing (Friston) is embedded within its hierarchy.

Predictions, Intelligence, and Machines

TEVSER advances seven falsifiable hypotheses. For example, conscious level should track integration between regulatory processes more than overall neural activity across sleep, anesthesia, and disorders of consciousness, and higher-level regulation should be selectively disruptable while elementary regulation is spared. Pivovarov concedes that the theory introduces no new data, that its neural mappings are approximate, and that attributing proto-qualia to simple organisms remains open to challenge. Higher levels carry intelligence and selfhood: reinforcement learning (Ω⁷), hierarchical prediction and goals (Ω⁸), modeling of other agents (Ω⁹), socially mediated self-consciousness, the subject “who” (Ω¹⁰), and symbolic reasoning (Ω¹¹). Large language models approximate fragments of Ω¹¹ detached from the levels beneath, so linguistic sophistication alone does not establish experience. Machine consciousness is possible in principle but would require an embodied, intrinsically regulating hierarchy; partial implementations of lower and middle levels exist (Pivovarov and Shumsky, 2022, 2025), and no current AI system is claimed to qualify.

References

Anokhin, P. K. (1974). Biology and Neurophysiology of the Conditioned Reflex and Its Role in Adaptive Behavior (S. A. Corson, Ed.). Pergamon Press.

Chalmers, D. J. (1995). Facing up to the problem of consciousness. Journal of Consciousness Studies, 2(3), 200–219.

Conant, R. C., and Ashby, W. R. (1970). Every good regulator of a system must be a model of that system. International Journal of Systems Science, 1(2), 89–97.

Pivovarov, I. (2026). TEVSER: A theory of evolving self-representations. Frontiers in Human Neuroscience, 20, 1858621. https://doi.org/10.3389/fnhum.2026.1858621

Pivovarov, I., and Shumsky, S. (2022). MARTI-4: New model of human brain, considering neocortex and basal ganglia—learns to play Atari game by reinforcement learning on a single CPU. In B. Goertzel, M. Iklé, A. Potapov, and D. Ponomaryov (Eds.), Artificial General Intelligence: AGI 2022 (Lecture Notes in Computer Science, Vol. 13539, pp. 62–74). Springer.

Pivovarov, I., and Shumsky, S. (2025). MARTI-5: A mathematical model of “self in the world” as a first step toward self-awareness. arXiv:2512.10985.

Tools

Share
El quote rightCite
Download PDF