CALLIOPE · BCOM

This work investigates the 2D Ising model as a theoretical framework for understanding criticality, algorithmic complexity, and information dynamics, with direct implications for neuroscience and neuromodulation. Through computational simulation of classical and hyperlink-augmented Ising lattices, the study demonstrates that phase transitions leave characteristic signatures across multiple observables—including Lempel-Ziv-Welch compression ratios, autocorrelation length, magnetic susceptibility, and heat capacity—establishing algorithmic entropy as a practical marker of criticality. A central finding is that sparse long-range couplings (h-links) systematically shift the critical temperature: ferromagnetic h-links raise T_c while antiferromagnetic h-links lower it, yet the temperature at which information transmission peaks remains invariant at T_c regardless of h-link configuration, suggesting that maximal information flow is a universal property of criticality rather than a topological artifact. The framework further proposes that neural systems operating near criticality exhibit maximal susceptibility to weak external perturbations, offering a mechanistic account of why sub-threshold transcranial electrical stimulation can produce measurable effects in coupled neural networks. A roadmap is outlined for constructing personalized Ising models from individual connectome data to estimate subject-specific critical temperatures and predict neuromodulation responsiveness, with potential applications to psychiatric and neurodegenerative conditions characterized by departures from optimal criticality.

Phase transitions in spin lattices turn out to be a surprisingly good lens for understanding why weak brain stimulation works at all.

The 2D Ising model is one of the simplest systems in physics that undergoes a phase transition: a grid of binary "spins" (think: neurons that are either on or off) flips from a globally ordered state to a disordered one as temperature rises past a critical point T_c. This paper asks a pointed question — what happens to information and algorithmic complexity at that boundary? The answer is that criticality leaves fingerprints everywhere you look: compression ratios of the magnetization time series dip in a characteristic way, autocorrelation length drops sharply, susceptibility peaks, and information transmission between distant spins hits its maximum exactly at T_c. None of this is new physics, but grounding it in Lempel-Ziv-Welch compression (a practical proxy for Kolmogorov complexity — roughly, "how hard is it to describe this system?") is the useful move, because LZW is something you can actually run on EEG or fMRI data.

The more novel result involves "h-links" — a sparse sprinkling of long-range couplings added on top of the standard nearest-neighbor lattice. Ferromagnetic h-links (same-sign, reinforcing) push T_c upward; antiferromagnetic ones pull it down. That part is intuitive. What's striking is that the peak of information transmission stays pinned at T_c regardless of which h-links you add. The topology changes the phase boundary, but it doesn't change the fact that maximal information flow lives at that boundary. The paper interprets this as evidence that peak information transmission is a universal property of criticality itself, not an artifact of any particular wiring pattern.

The neuroscience motivation is this: transcranial electrical stimulation (tES) uses fields so weak that a single neuron can't detect them, yet measurable effects show up in behavior and physiology. The paper argues that a network operating near its critical point has maximal susceptibility — it is, by definition, most responsive to tiny external nudges. Criticality is the amplifier. The paper cites stochastic resonance and coupled-oscillator models as supporting evidence, though it doesn't rigorously derive the amplification within the Ising framework itself; that connection remains somewhat informal.

The most forward-looking section sketches a roadmap: take a person's structural brain connectivity (from diffusion MRI), build a personalized Ising model with h-links representing long-range electrical coupling between cortical patches, compute that individual's T_c, and use it to predict how responsive they'll be to stimulation. Pathological states — depression, epilepsy, neurodegeneration — would then correspond to brains operating too far from their critical point. This is a compelling vision, but the paper is honest that it's a roadmap, not a result: no real patient data is analyzed here. The formal link between Kolmogorov complexity and thermodynamic quantities also remains a conjecture rather than a theorem. What the paper does deliver is a clean computational demonstration that algorithmic complexity metrics and classical statistical-physics observables converge on the same story at the phase boundary — which is exactly the theoretical scaffolding needed before applying these ideas to messy biological data.