Consciousness as a Thermostat Glitch express
A thermostat keeps a room at 20 degrees. Open a window in winter, and it kicks on the heat to restore equilibrium. The brain does something similar: it constantly predicts what will happen a moment ahead and corrects itself when wrong. That's how it learns. But where in this machinery does consciousness fit — the taste of an apple or the sting of a burn?
Two theories have approached this from different angles. One, the free energy principle, says any living system strives to minimize surprise, much like that thermostat. The other, integrated information theory, tries to measure consciousness through a system's ability to influence itself. If you break the brain into parts, it loses the holistic picture. Until now, there was no clear mathematical link between these views.
The authors of the new study found that link. They proposed counting information as the difference between what a system actually did and what it "should" have done if it merely followed the most likely trajectory. It turns out that key concepts in consciousness theory, so-called cause-effect repertoires, follow naturally from this definition. In other words, a measure of awareness turns out to be a special case of how a system copes with surprises.
This doesn't mean a thermostat gains a soul. It's about one and the same mathematical language being able to describe both learning and subjective experience. The bridge between them rests on a simple idea: consciousness is not a special substance but the degree to which a system must deviate from the familiar to remain itself.
Original abstract
as written by the authors arXiv:2605.12536
The Free Energy Principle (FEP) is a leading framework for mathematically modeling self-organization and learning, while Integrated Information Theory (IIT) is a computational ontology of consciousness oriented around irreducible cause and effect. While conceptual unifications have been proposed and appear to be supported by empirical findings, the absence of a rigorous mathematical mapping places upper bounds on their precision and testability. This work proposes that information can be defined as the deviation $$\psi$$ of realized dynamics from a constrained maximum-caliber (MaxCal) path ensemble over a finite time horizon. Under this definition, each of the cause/effect repertoires central to IIT 3.0 emerge directly from MaxCal variational principles, allowing IIT's phenomenological calculus to be re-derived from constrained entropy-maximization (CMEP). This framework supplies a theoretical bridge to active inference, which is mathematically dual to CMEP under Langevin dynamics, and offers a principled route for extending IIT to new dynamical regimes. When the approach is applied under the Central Limit Theorem (CLT) for Markov chains and via large deviations theory (LDT) to Ising models, information $$\psi$$ is shown to be equivalent to prediction error under accompanying predictive coding models. This may hold relevance to the ``hill-shaped trajectory'' of $$\Phi$$ observed in neuronal cultures adapting to sensory inputs. Together, these results provide a physically and mathematically grounded rationale for studying the convergence of FEP, IIT, and thermodynamic frameworks of cognition such as recent work grounding consciousness in violations of the Fluctuation-Dissipation Theorem (FDT).
Neuroscience knows a "hill-shaped" curve: when neurons in a test tube learn to respond to stimuli, the integrated information measure first rises, peaks, then falls, as if consciousness burns brightest at the moment of mastering something new.
In Peter Watts's novel Blindsight, an alien intelligence turns out to lack consciousness precisely because its behavior is always optimal and never deviates from the predicted.