This post is a deep-dive analysis of a cognitive emergence theory based on information topology, summarized under the slogan "CYCLE IS ALL YOU NEED: MORE IS DIFFERENT."
1. Core Framework: From Information to Cognitive Cycles
The Cycle as the Basic Unit of Cognition
The theory's central claim, "CYCLE IS ALL YOU NEED," holds that the basic unit of cognitive activity is not the static data point (bit) but the dynamic, closed information loop. Closure is essential: an information fragment that fails to connect with others into a closed cycle behaves like an open boundary and eventually dissolves into noise. Once a fragment closes into a loop, it acquires a form of topological invariance, remaining stable under the system's dynamics—becoming a carrier of memory, meaning, and consciousness. The essence of intelligence, on this view, is the ability to organize information into cycles, not merely to process it.
Mathematical Basis: Homology and ∂² = 0
The mathematical foundation is homology theory, specifically the identity "the boundary of a boundary is zero" (∂² = 0). Interpreted cognitively:
- Points/bits are 0-simplices—isolated information fragments.
- Cycles are 1-cycles—closed chains of relations with zero boundary.
- Chains that do not close (open boundaries) leave no lasting trace in the homology group; they are filtered out as noise or forgotten.
- Only chains that close persist as stable invariants—the basis of memory, knowledge, and meaning.
- Cross-frequency coupling: high-frequency local cycles (gamma-band) nested inside low-frequency global cycles (theta-band), creating multi-scale spatiotemporal structure.
- Memory replay: during rest/sleep, micro-cycles from experience are re-activated, compressed, and recombined, exposing hidden statistical regularities and causal relations that consolidate into high-order, generalizing cycles.
- Oscillatory phase coding: neural rhythms (theta, alpha, gamma) parametrize time; firing at specific phases positions events within a cycle, converting linear time flow into a traversable loop.
- Coincidence detection: when the end of an open chain (A→B→C) synchronously activates with its start, the chain closes (A→B→C→A), eliminating the open boundary; synchrony triggers plasticity (e.g., LTP), strengthening the loop—both the trigger and engine of cycle consolidation.
- Polychronous Neural Groups (PNGs) (Izhikevich): neurons firing with precise millisecond delays form repeatable, self-organized temporal patterns—the most natural biological carriers of cycles, with vast capacity for encoding fine-grained information.
- Beyond the Turing paradigm: shift from operating on points (symbols, features, weights) to discovering, closing, and recombining cycles—naturally handling ambiguity, context, and common-sense reasoning.
- Homological information carriers: a "homology engine" that continuously forms loops via coincidence detection and plasticity; reasoning becomes traversal, combination, and transformation of cycles.
- AGI implications: cycles map many possible input orderings onto a finite set of stable structures, mitigating combinatorial explosion; perception–action cycles ground symbols in the world, addressing the grounding problem—yielding embodied agents rather than "brains in vats."
- Perception, memory, and action unify as different kinds of cycles (input-to-representation closure; re-entry of stable neural trajectories; intention-to-feedback closure).
- The framework resonates with Grothendieck's structuralism: mathematical (and cognitive) focus shifts from isolated objects to relations and structures.
- Classical perception–action loops are deepened with higher-order invariance: high-order cycles are insensitive to the order of their local steps, explaining flexibility and generalization.
- Consciousness is interpreted as the phenomenological correlate of the most stable, persistent, high-order cycles integrating multisensory, cross-temporal information; qualia are topological properties of body-coupled perception–action cycles.
- Information ontology: cycles, not bits, are cognition's natural atoms; a fragment's existence in a cognitive system is tied to its functional participation in a loop.
- Sheaf–cosheaf duality unifies the two directions of cognition: sheaves organize local data (points) into global structure (cycles), while cosheaves decompose global structure back into local data—suggesting points and cycles are complementary sides of one whole.
Thus ∂² = 0 serves as a first principle by which cognitive systems extract order from chaos and build durable structure from transient experience.
"More Is Different": Emergence
Echoing Philip Anderson, the theory describes how simple local loops (few neurons, reflex arcs, short-term memories) nest, couple, and recursively combine into global cognitive functions:
From "It from Bit" to "Cognition from Cycle"
The theory revises John Wheeler's "It from Bit": isolated bits are meaningless in cognition. The proposal replaces the bit with the cycle as the cognitive "natural atom," emphasizing the dynamism and relationality of information, the coherence and stability of closed structures, and a non-reductive route to consciousness as the phenomenological correlate of the most stable, persistent high-order cycles.
The "Four No's for Cognition"
| Principle | Core idea | Implication for AI design | | :--- | :--- | :--- | | No isolated information | Meaning arises from relational closure within cycles | Build systems that construct and exploit context, not isolated data | | No privileged order | Cognition is insensitive to the ordering of local steps that close a cycle | Robust, parallel, fault-tolerant architectures | | No static storage | Memory is the dynamic ability to re-enter a latent cycle, not stored data | Plastic, lifelong-learning memory systems | | No prediction without invariance | Prediction depends on stable invariants extracted from cycles | Systems that identify stable environmental patterns |
Non-Ergodicity
Intelligent systems are explicitly non-ergodic: unlike ergodic physical systems, they do not explore all states uniformly but inhabit learned, path-dependent subspaces. Measure preservation is replaced by cycle preservation—the probability distribution may drift, but topological invariants (the cycles encoding memories and behaviors) are retained. Cycles act as attractors guiding the system's dynamics, enabling efficient learning and generalization.
2. Neurobiological Grounding
The theory maps cycles onto known neural mechanisms:
Multi-scale structure arises from cross-frequency nesting (theta cycles as macro-loops containing gamma micro-loops) and replay (recursive recombination of micro-cycles into higher-order ones, e.g., from "red light → brake" to "obey traffic rules"), enabling the emergence of global cognitive structures like concepts from networks of interconnected local cycles.