Navigation requires a gradient. A landscape with structure, where different directions lead to measurably different states. Entropy provides both the gradient and the arrow.
You know the difference between a good day and a bad one. On a good day, the world feels open. Options seem available. Your mind moves freely between thoughts, plans, memories. On a bad day, the world narrows. Thoughts loop. The same anxious scenario plays on repeat. You can measure this.
Robin Carhart-Harris and colleagues established something striking: entropy levels in neural dynamics track conscious states quantitatively. Higher entropy corresponds to richer, more expansive conscious experience. Think of the psychedelic state, where the mind feels vast and unbound. Lower entropy corresponds to more constrained, rigid experience. Think of depression, or disorders of consciousness, where the world narrows. There are many ways to measure entropy in brain signals, and every single one of them tracks consciousness level.
The evidence is not subtle. Put someone under anesthesia and the complexity of their brain signals drops. Wake them up and it rises. The relationship holds whether you measure it with EEG complexity, fMRI entropy, or perturbational response. It distinguishes minimally conscious patients from unresponsive ones. It can track perceptual transitions without the person reporting anything at all. Multiple research groups, using different methods on different populations, converge on the same finding.
The Perturbational Complexity Index (PCI), developed by Casali and colleagues, provides the most direct connection between these entropy measures and NFT's claims. PCI measures how the brain responds to perturbation, combining TMS stimulation with algorithmic complexity of the evoked EEG response. It reliably tracks conscious level across wakefulness, sleep, anesthesia, and disorders of consciousness without requiring behavioral report. A system at criticality should show maximal PCI; departures from criticality should reduce it. This maps directly onto NFT: at criticality, the navigational system has maximal flexibility; away from criticality, navigation is impaired.
The Massimini group's complementary finding tells the same story from the other side. Loss of consciousness is characterized by breakdown of cortical effective connectivity, the brain's inability to sustain complex responses to perturbation, exactly what NFT would predict from loss of criticality-mediated amplification. The underlying physics has been measured. Tagliazucchi and colleagues provide direct evidence that the brain shifts away from criticality during loss of consciousness in sleep and anesthesia, grounding the claim in empirical data rather than theoretical arguments.
Entropy characterizes the landscape through which consciousness moves.
The word "entropy" does different work in different parts of this argument.
Signal entropy is what the empirical literature measures. Shannon complexity, Lempel-Ziv complexity, the information content of recorded neural time series. It describes what the brain is doing right now. Trajectory entropy is what consciousness reduces. It measures uncertainty over distributions of future states, conditioned on the organism's goals and available actions. Signal entropy tracks consciousness state. Trajectory entropy is what navigation changes.
The two are correlated. Suppressing consciousness reduces both. But they are not the same quantity. A classical Bayesian agent can reduce trajectory entropy through prediction and planning without any quantum resources. NFT's Level B claim requires something stronger: that the efficiency of trajectory-entropy reduction in conscious systems exceeds what classical stochastic processes can achieve under matched biological constraints.
As the candle burns down, the barn roof weathers, and your hot cup of tea cools to warm, entropy increases. Every complex system that persists does so by actively managing that increase, importing energy, shedding waste, rebuilding what burns. Jha argues that consciousness is what this management looks like from the inside. When a system gets complex enough that steering against entropy becomes a full-time job, the steering itself becomes experience. Consciousness is thermodynamic management, felt.
Hemmo and Shenker come at the same territory from the other side. Why does time feel like it flows? Physics has no flow. Einstein's equations run identically in both directions. But entropy does not. Entropy increases, and increase has a direction. Hemmo and Shenker argue that the psychological arrow of time, the felt sense that this moment gives way to the next, is thermodynamic asymmetry experienced from within. You feel time flowing because entropy is increasing. The feeling and the physics point the same way.
NFT synthesizes both. Consciousness navigates along the entropy gradient, and the arrow of time is the direction of navigation. This is the directionality that DeLancey's complexity framework (Chapter 2) cannot provide [A].
A conscious navigator is not a god moving freely through a block universe. It is an entity embedded in a thermodynamic gradient with limited degrees of freedom. It can swim and steer its course, but it cannot reverse the current or step outside it.
This constrained-agency picture has independent support from multiple directions. Prigogine's dissipative structures demonstrated that open systems far from equilibrium maintain coherent organization precisely by dissipating entropy; they navigate along the gradient, not against it. England showed that driven systems spontaneously restructure to dissipate energy more efficiently. And Friston's free energy principle, which keeps reappearing in this story for good reason, says essentially the same thing from another angle: biological agents minimize surprise. Which is to say they steer within the entropic current toward states compatible with their continued existence.
NFT does not compete with Friston's framework here; it reinterprets it. Minimizing free energy is navigating possibility space along the entropy gradient. The distinction is that NFT grounds the navigation in quantum substrate coupling rather than in Bayesian inference on classical computation. The reinterpretation becomes substantive at Level B.