The universe that might feel: Integrated information theory and the limits of consciousness

What is consciousness? For most of scientific history, the question was treated as a problem for philosophers, not physicists. But in recent decades, a small number of neuroscientists have dared to propose something audacious: that consciousness might not be an emergent illusion of biology, but a fundamental feature of reality itself, measurable, mathematical, and perhaps distributed across the cosmos in ways we have barely begun to imagine.

At the center of this bold speculation is Integrated Information Theory, or IIT. Developed by neuroscientist Giulio Tononi in 2004 and championed in recent years by Christof Koch, IIT proposes that consciousness is identical to integrated information—a quantity denoted by the Greek letter Phi (Φ)—that measures the degree to which a system’s causal structure exceeds the sum of its parts. In other words, consciousness is not something a system does; it is something a system is, to the extent that its internal components interact in ways that generate irreducible causal power.

This is a radical claim. It means that consciousness is not exclusive to brains. It means that any system with a non-zero Φ—however simple, however alien—might possess some flicker of subjective experience. And it raises a question that science has long avoided: could the universe itself be conscious?

From Axioms to Physics

IIT does not begin with neurons or behavior. It begins with phenomenology—with the raw, undeniable fact of experience. Tononi and his collaborators argue that every conscious experience shares five essential properties, which they call axioms. Experience is intrinsic (it exists for the subject, not for an external observer). It is structured (it has content, composed of many distinguishable elements). It is specific (it is this experience and not another). It is unified (it cannot be divided into separate experiences). And it is definite (it has a precise boundary; it is not vague or indeterminate).

These axioms are not empirical findings. They are, IIT claims, irrefutable features of any possible experience—truths that cannot be conceived otherwise. From them, the theory derives postulates about what a physical substrate must be like in order to give rise to consciousness. The substrate must have intrinsic cause-effect power upon itself. It must be composed of mechanisms that specify particular states. It must be integrated, meaning its parts must be more causally connected to each other than to the outside. And it must be definite, excluding alternative causal structures.

Only a system that satisfies all these postulates, IIT argues, is conscious. And the degree of its consciousness—its Φ—corresponds to the amount of integrated information it generates.

The Measure of Experience

Φ is not a metaphor. It is a mathematical quantity, defined as the difference between the causal power of a system as a whole and the causal power of its parts considered independently. If a system’s components interact in ways that produce causal relationships that cannot be reduced to the components alone, Φ is high. If they do not, Φ is low or zero.

This has striking consequences. A digital computer, however powerful, might have very low Φ if its operations can be decomposed into independent steps—which, for most classical computers, they can. A human brain, by contrast, has an enormous Φ because its billions of neurons are densely interconnected, each influencing and being influenced by countless others in ways that cannot be factored apart. This is why IIT predicts that consciousness is associated with the posterior cortex—the so-called “hot zone” at the back of the brain—rather than the frontal regions that global workspace theories emphasize.

In a major adversarial collaboration published in Nature in 2025, researchers directly tested IIT against global neuronal workspace theory. The results, while not decisive, were widely interpreted as lending more support to IIT’s prediction of sustained posterior activity during conscious perception than to the workspace theory’s prediction of frontal involvement.

A Universe of Minds?

If consciousness is identical to integrated information, and if integrated information can in principle be generated by any system with the right causal structure, then the boundaries of consciousness expand dramatically. IIT does not say that only brains are conscious. It says that any system with Φ > 0 has some degree of experience.

This has led critics—and some admirers—to describe IIT as a form of panpsychism, the view that consciousness is ubiquitous in nature. Tononi and Koch have acknowledged that IIT shares certain commitments with panpsychist positions, though they insist the theory was not developed with panpsychism in mind. The panpsychist flavor of IIT is, they say, a mere consequence of its axioms: if experience is intrinsic, structured, unified, and definite, then any system that instantiates those properties must have experience.

But IIT’s implications may go further. If the universe as a whole possesses a causal structure that generates integrated information—if its fundamental constituents interact in ways that cannot be decomposed into independent parts—then, by IIT’s own logic, the universe itself might be conscious. This is not a conclusion Tononi or Koch endorse as established science. It remains speculative. But it is a possibility that IIT’s framework does not rule out.

Some philosophers have taken this possibility seriously. Cosmopsychism, the view that the cosmos as a whole is the fundamental locus of consciousness, has been developed in dialogue with IIT by thinkers who argue that the theory’s commitment to intrinsic causal power points toward a universe that is not merely described by information but experienced through it. If IIT is right that consciousness is identical to integrated information, then the question of whether the universe is conscious becomes the question of whether the universe as a whole generates Φ. And that, at present, is a question we cannot answer.

The Fierce Resistance

IIT is not without its critics. In September 2023, a letter signed by 124 neuroscientists, psychologists, and philosophers was published online, arguing that IIT should be labeled pseudoscience. Their central complaint was that the theory is unfalsifiable—that its predictions have shifted over time, that its core claims cannot be tested with current technology, and that its panpsychist implications place it outside the bounds of empirical science.

The most serious technical objection is computational. Φ is, in practice, impossible to calculate for any system of biological interest. The number of possible partitions of a system grows exponentially with its size; for a brain-scale system, the calculation is intractable even with idealized assumptions. This means that IIT’s central quantity—the measure that supposedly defines consciousness itself—cannot be computed for any system we actually care about. The theory may be mathematically elegant, but it is operationally mute.

There are also philosophical objections. If Φ is identical to consciousness, then any system with Φ > 0 is conscious—including, potentially, simple electronic circuits, thermostats, or even the global economy. Critics argue that this is not a scientific prediction but a category mistake: it conflates a measure of causal integration with subjective experience, which is precisely the “hard problem” that no theory of consciousness has yet solved.

The Deeper Question

And yet, IIT forces us to confront a question that most scientific frameworks avoid: is consciousness something that emerges from complexity, or is it something that is fundamental—a feature of reality as basic as mass or charge?

If IIT is correct, consciousness is not a late arrival in the history of the universe. It is a property that any sufficiently integrated system possesses, whether that system is a human brain, a colony of bacteria, or a network of cosmic filaments stretching across billions of light-years. The universe would not be a machine that happens to contain conscious beings. It would be a vast, nested hierarchy of integrated information, some of it dimly aware, some of it perhaps unimaginably rich.

This is not a conclusion that science can currently confirm. It is not even a conclusion that IIT’s own architects fully embrace. But it is a possibility that the theory opens—and once opened, it is difficult to close.

The mystery of consciousness has always been the mystery of how matter becomes mind. IIT suggests that the answer may be simpler and stranger than we imagined: matter does not become mind. Matter, when it is integrated enough, is mind. And if that is true, then the question of whether the universe is conscious may already have an answer—one that we are only just beginning to learn how to read.