Tononi’s Integrated Information Theory

Integrated Information Theory (IIT), developed by neuroscientist Giulio Tononi and supported by neuroscientist Christof Koch, is an original, indeed radical, model that states what experience is and what types of physical systems can have it (Tononi & Koch, 2015). IIT is grounded in experience, the phenomenology of consciousness, and it features mathematical description, quantitative measurement, scientific testability, broad applications, and nonpareil, intrinsic, cause-effect “structures.” In other words, “IIT addresses the problem of consciousness starting from phenomenology—the existence of my own experience, which is immediate and indubitable—rather than from the behavioral, functional, or neural correlates of experience” (Tononi et al., 2022). Controversial to be sure, IIT has become a leading theory of consciousness.^[1]

From Phenomenology to Mathematics

IIT accounts for consciousness in the following way. First, introspection and reason identify the essential properties of consciousness—the axioms of phenomenal existence. Then, each axiom is accounted for in terms of cause–effect power; that is, “translating” a “phenomenal property into an essential property of the physical substrate of consciousness” [PSC]—yielding the postulates of physical existence. In this way, IIT claims to “obtain a set of criteria that a physical substrate of consciousness (say, a set of cortical neurons) must satisfy” (Tononi et al., 2022).

IIT asserts that distinct conscious experiences are, in a literal sense, distinct kinds of conceptual structures in a radical and heretofore unknown kind of “qualia space.” IIT says (and introduced the idea) that for every conscious experience, there is a corresponding mathematical object such that the mathematical features of that object are isomorphic to the properties of the experience.

“Integrated information theory means that you need a very special kind of mechanism organized in a special kind of way to experience consciousness,” Tononi says. “A conscious experience is a maximally reduced conceptual structure in a space called ‘qualia space.’ Think of it as a shape. But not an ordinary shape—a shape seen from the inside.” Tononi stresses that simulation is “not the real thing.” To be truly conscious, he said, an entity must be “of a certain kind that can constrain its past and future—and certainly a simulation is not of that kind” (Tononi, 2014b).

Christof Koch envisions how IIT could explain experience—how consciousness arises out of matter. “The theory makes two fundamental axiomatic assumptions,” Koch explains. “First, conscious experiences are unique and there are a vast number of different conscious experiences. Just think of all the frames of all the movies you've ever seen or movies that will ever be made until the end of time. Each one is a unique visual experience and you can couple that with all the unique auditory experiences, pain experiences, etc. All possible conscious experiences are a gigantic number. Second, at the same time, each experience is integrated—what philosophers refer to as unitary. Whatever I am conscious of, I am conscious of as a whole. I apprehend as a whole. So, the idea is to take these two axioms seriously and to cast them into an information theory framework. Why information theory? Because information theory deals with different states and their interrelationships. We don't think the stuff the brain is made out of is really what's critical about consciousness. It's the interrelationship that's critical” (Koch, 2012b).

IIT starts from phenomenology itself—a point that Tononi stresses cannot be overstressed—with axioms that are deemed to be unequivocally and universally true for all instances of consciousness, such that whatever systems manifest these axioms will ipso facto manifest consciousness.

It is at this point that IIT seeks a mathematical expression of the fundamental properties of experience. It is not the reverse: IIT does not start from mathematics hoping to explain phenomenology; rather, it starts with phenomenology and ends with mathematics (Tononi, 2014a). Because IIT's consciousness is a purely information-theoretic property of systems, not limited to brains or even to biology, Tononi constructs a mathematical function φ (phi) to measure a system's informational integration, with levels of φ covarying with degrees of consciousness (Van Gulick, 2019).

In IIT, each experience, each conscious percept, has clear characteristics: it is specific: it is what it is by how it differs from alternative experiences; it is unified: irreducible to noninterdependent components; it is unique: it has its own one-off borders and a particular spatio-temporal grain (Oizumi et al., 2014; Haun & Tononi, 2019).

These pillar concepts, all grounded in experience, are expressed by five phenomenological axioms: intrinsic existence, composition, information, integration, and exclusion. These axioms are then formalized into postulates that prescribe how physical mechanisms, such as neurons or logic gates, must be configured to generate experience (phenomenology). The postulates are used to define integrated information as information specified by a whole that cannot be reduced to that specified by its parts (Tononi & Koch, 2015).

Each of IIT's five postulates defines and constrains the properties required of physical mechanisms to support consciousness (Tononi & Koch, 2015).

• (i) Intrinsic Existence. Consciousness exists of its own inherent nature: each experience is real, and it exists from its own inherent perspective; to account for experience, a system of mechanisms in a state must exist intrinsically, and it must have cause–effect power.

• (ii) Composition. Consciousness is structured: each experience is composed of phenomenological distinctions; the system must be structured: subsets of system elements (composed in various combinations) must have cause–effect power upon the system.

• (iii) Information. Consciousness is specific: each experience is the particular way it is; the system must specify a cause–effect-enabling structure that is the particular way it is; the system has a set of specific cause–effect repertoires that distinguishes it from all other possible structures (differentiation).

• (iv) Integration. Consciousness is unified: each experience is irreducible to noninterdependent subsets of phenomenal distinctions; the cause–effect structure specified by the system must be unified: it must be intrinsically irreducible.

• (v) Exclusion. Consciousness is definite in content and spatio-temporal grain: each experience has the set of phenomenal distinctions it has, not less or more, and flows at the speed it does, not faster or slower; the cause–effect structure specified by the system must be definite and is maximally irreducible intrinsically (“conceptual structure”).

It is this conceptual structure that is especially intriguing. Maximally irreducible intrinsically, it is also known as a “quale” (plural: qualia). Its arguably infinite varieties are formed when higher-order mechanisms specify concepts, with the constellation of all concepts specifying the overall form or shape of the quale. On this basis, Tononi and Koch formulate the central identity of IIT quite simply: an experience is identical to a conceptual structure that is maximally irreducible intrinsically (Tononi & Koch, 2015).

The Ontology of Intrinsic Entities

Questions that IIT seeks to address: Why the cerebral cortex gives rise to consciousness but the cerebellum does not, though the latter has even more neurons and appears to be just as complex? Is consciousness present in coma patients, preterm infants, non-mammalian species? Can computers, artificial intelligence (e.g., large language models) become conscious as humans are conscious?

Most relevant to our Landscape is IIT's fundamental ontology. Put simply, it begins with “the ontological primacy of phenomenal existence.” The proper understanding of consciousness, IIT states, is “true existence, captured by its intrinsic powers ontology: what truly exists, in physical terms, are intrinsic entities, and only what truly exists can cause” (Tononi et al., 2022).

Seeking to embed its theory of consciousness within a coherent metaphysical framework, IIT introduces its “0th postulate” or “principle of being.” To exist physically, IIT states, “means to have cause–effect power—being able to take and make a difference. In other words, physical existence is defined purely operationally, from the extrinsic perspective of a conscious observer, with no residual ‘intrinsic’ properties (such as mass or charge). Furthermore, physical existence should be conceived of as cause–effect power all the way down—namely down to the finest, ‘atomic’ units that can take and make a difference” (Tononi et al., 2022).

IIT deep conclusion is that “only a substrate that unfolds into a maximum of intrinsic, structured, specific, irreducible cause–effect power—an intrinsic entity—can account for the essential properties of phenomenal existence in physical terms.” IIT goes on to claim that “only an intrinsic entity can be said to exist intrinsically—to exist for itself, in an absolute sense. By contrast, if something has cause–effect power but does not qualify as an intrinsic entity, it can only be said to exist extrinsically—to exist for something else—say, for an external observer—in a relative sense. And intrinsic, absolute existence is the only existence worth having—what we might call true existence. Said otherwise, an intrinsic entity is the only entity worth being.”

In a crucial move, according to Tononi and colleagues, “IIT asserts an explanatory identity: an experience is identical to a Φ-structure. In other words, the phenomenal properties of an experience—its quality or how it feels—correspond one-to-one to the physical properties of the cause–effect structure unfolded from the physical substrate of consciousness. Thus, all the contents of an experience here and now—including spatial extendedness; temporal flow; objects; colors and sounds; thoughts, intentions, decisions, and beliefs; doubts and convictions; hopes and fears; memories and expectations—correspond to sub-structures in a cause–effect structure (Ф-folds in a Ф-structure)” (Tononi et al., 2022).

This means that “all contents of experience correspond to sub-structures within a maximally irreducible cause–effect structure—to Φ-folds within a Φ-structure. This applies not only to the experience of space, time, and objects, but also to conscious thoughts and feelings of any kind … Conscious alternatives, too, are Φ-folds within the Φ-structure corresponding to an experience.

Fundamentally, then, it is IIT's claim that when one is conscious, “what actually exists is a large Ф-structure corresponding to my experience, and it exists at its particular grain. No subsets, supersets, or parasets of that Ф-structure also exist, just as no other grains also exist. Moreover, what actually exists is only the Ф-structure corresponding to my experience, not also an associated physical substrate. Crucially, any content of my experience, including alternatives, reasons, and decisions, corresponds to a sub-structure [i.e., Φ-folds] within my Ф-structure, not to a functional property emerging from my [neural] substrate (Tononi et al., 2022).

Because “IIT starts from phenomenal existence and defines physical existence operationally in terms of cause–effect power ‘all the way down,’ with no intrinsic residue, such as mass and charge … a physical substrate should not be thought of as an ontological or ‘substantial’ basis—an ontological substrate—constituted of elementary particles that would exist as such, endowed with intrinsic properties.”

This means, according to IIT, “because I actually exist—as a large intrinsic entity—the neurons of my substrate as such but the Ф-structure expressing its causal powers … Moreover, because my alternatives, reasons, and decisions exist within my experience—as sub-structures within an intrinsic entity—the neuronal substrates of alternatives, reasons, and decisions cannot also exist.” If this picture is correct, IIT claims controversially, “it leaves no room for emergence or dualism of any sort” (Tononi et al., 2022).

As a defining corollary to its radical theory of consciousness, IIT claims that true free will exists, based on “the proper understanding of experience as true existence and on the intrinsic powers view: what truly exists, in physical terms, are intrinsic entities, and only what truly exists can cause.” In contrast, in materialistic theories, with ontological and causal micro-determination, much of the debate about free will has revolved not around existence but around determinism/indeterminism, so that true free will is incompatible (Tononi et al., 2022).

Testing and Koch’s Perspective

In the same set of “adversarial collaboration” experiments that tested Global Workspace Theory, IIT was also subjected to the putatively rigorous protocols (Templeton World Charity Foundation, n.d.). “The specific IIT prediction examined was that consciousness is a kind of “structure” in the brain formed by a particular type of neuronal connectivity that is active for as long as a given experience, say, seeing an image, is occurring. This structure is said to be in the posterior cortex (the occipital, parietal, and temporal cortices in the back part of the brain). Preliminary results indicate that while “areas in the posterior cortex do contain information in a sustained manner”—which could be taken as evidence that the “structure” postulated by the theory is being observed—the independent “theory-neutral” researchers didn't find sustained synchronization between different areas of the brain, as had been predicted.

Preliminary brain-scanning data to calculate φ for simplified models of specific neural networks within the human brain, such as the visual cortex, seem to correlate with states of consciousness (Lenharo, 2023a, 2023b, 2024). Scanning the brain as people “slip into anesthesia” is said to offer support for IIT by calculating phi “for simplified models of specific neural networks within the human brain that have known functions, such as the visual cortex” (Wilson, 2023)—though, by all accounts, the empirical neuroscience of IIT is still rudimentary.

More recently, Koch defines IIT’s consciousness as “unfolded intrinsic causal power, the ability to effect change, a property associated with any system of interacting components, be they neurons or transistors. Consciousness is a structure, not a function, a process, or a computation.” He calls out “the theory’s insistence that consciousness must be incorporated into the basic description of what exists, at the rock-bottom level of reality”—a claim that “has also drawn considerable fire from opponents.” He explains that IIT “quantifies the amount of consciousness of any system by its integrated information, characterizing the system’s irreducibility. The more integrated information a system possesses, the more it is conscious. Systems with a lot of integration, such as the adult human brain, have the freedom to choose; they possess free will” (Koch, 2024, p. 16).

IIT in the Landscape

Personally, I see IIT operating in three dimensions. First, measurement: IIT is a test of consciousness, assessing what things are conscious, and in those things that are, quantifying the degree of consciousness (e.g., coma patients). Second, mechanism: IIT can predict brain structures and functions involved in consciousness. Third, ontology (the most controversial): IIT speculates that the conceptual structures of qualia are “located” in some kind of “qualia space.”

The first two dimensions, IIT's measurement and mechanism, could sit comfortably in the Materialism Theories area of the Landscape. The third, IIT's ontology of qualia, is radically distinct, its classification unclear—which is part of the reason why I have given IIT its own category on the Landscape.^[2]

IIT claims that integrated information is both necessary and sufficient for consciousness: necessary seems uncontroversial; sufficient is the rub to many. But what I especially like about IIT's “conceptual structures” in “qualia space” is that IIT makes a stake-in-the-ground commitment to what consciousness per se may literally be—an appreciated rarity on the Landscape of consciousness (which does not mean that I [RLK] subscribe to it).

Footnotes

[1]. In my original Landscape of Consciousness paper, I had Integrated Information Theory as its own category, but I did not do so because I thought IIT was the leading theory of consciousness. I did so because IIT is (i) a leading theory; (ii) original in premises and approach; (iii) controversial; and (iv) it would be misleading if classified in any of the other categories. Now, in this Landscape of Consciousness website, I broaden the category to "Information" broadly. I do give IIT pride of place as the first theory, because it is thoroughly and meticulously developed, and because it is unique.

[2]. In attempting to classify IIT's ontology of “conceptual structures in qualia spaces,” one could make a case that IIT could be a form of Panpsychism, a kind of Dualism, or part of a much-enhanced Materialism. IIT leaders reject Dualism, distance themselves from Panpsychism, and probably would argue that, to subsume IIT, Materialism as currently practiced would need to be stretched beyond the snapping point.