Integrated Information Theory

From P2P Foundation
Jump to navigation Jump to search

= "IIT is a mathematically-grounded theory showing how the whole of experiential consciousness emerges from its merely material parts". [1]


Brendan Graham Dempsey:

"IT was pioneered by neuroscientist Giulio Tononi and has becoming one of the leading contenders for a comprehensive theory of consciousness. A complete explanation would require rather extensive mathematical exposition beyond the scope of this book, but a few works written for the layman offer a decent overview.

In The Feeling of Life Itself: Why Consciousness is Widespread But Can’t Be Computed, Christof Koch explains some of the fundamental assumption of IIT this way:

- “According to integrated information theory (IIT), consciousness is determined by the causal properties of any physical system acting upon itself. That is, consciousness is a fundamental property of any mechanism that has cause-effect power upon itself. Intrinsic causal power is the extent to which the current state of, say, an electronic circuit or a neural network, causally constrains its past and future states. The more the system’s elements constrain one another, the more causal power. …The causal powers can be represented as a constellation of points (distinctions) linked by lines (relations). According to IIT, these causal powers are identical to conscious experience, with every aspect of any possible conscious experience mapping one-to-one onto aspects of this causal structure” (p. 79).

The existence of parts with causal power with regard to the whole mean one can speak of an “irreducible” whole—or, as Koch puts it, “If partitioning some entity makes no difference to its cause-effect structure, then it is fully reducible to those parts, without losing anything” (p. 85). If something is lost, then the whole is genuinely irreducible to the parts. “The Whole is the most irreducible part of any system, the one that makes the most difference to itself. Per IIT, only the Whole has experience” (p. 87).

All of this is to say that consciousness is defined by irreducible complexity, where complexity is understood as the integrated relationship of parts in a whole. With this insight, one can then measure the complexity of the system—in this case, the neuronal/brain network—by an approximate measure known as “algorithmic complexity,” which reveal how irreducible something is.

Such measures are actually quite familiar to us. Any time you “zip” a file or folder on your computer, you have used algorithmic complexity to “compress” data to its irreducible form. This function essentially identifies those parts of the whole that can be removed without losing any essential information. A file with binary code can be compressed by finding the patterns in it, and reducing the length of the information needing to be conveyed to a shorter algorithm.


Between pattern and randomness, total order and total chaos, lies the domain of the sort of complexity which leads to the dynamic living structures we see around us. This is because maximal information is found not in total regularity, nor in total incoherence, but somewhere in-between.

IIT researchers use approximations like algorithmic complexity to measure the degree of integration in the brain/nervous system. This is what Φ is. Conscious complexity (Φ) drops when we enter sleep or are put under through anesthesia; it rises when we awaken.

Consciousness, in short, is a whole-part relationship, one maximally integrated; the degree of complexity and the degree of consciousness are directly correlated. Complexity is the exterior, structural aspect of the interior, subjective experience. Or, as Koch puts it: “IIT posits two sides to every Whole: an exterior aspect, known to the world and interacting with other objects, including other Wholes; and an interior aspect, what it feels like, its experience” (p. 166).

More than that, consciousness is emergent. As Anil Seth puts it in Being You: The New Science of Consciousness, “The easiest way to think about Φ is that it measures how much a system is ‘more than the sum’ of its parts, in terms of information. …In IIT, Φ measures the amount of information a system generated ‘as a whole,’ over and above the amount of information generated by its parts independently. This underpins the main claim of the theory, which is that a system is conscious to the extent that its whole generates more information than its parts” (p. 64). This, we have seen, is the very definition of emergence. IIT is a mathematically-grounded theory showing how the whole of experiential consciousness emerges from its merely material parts."



"Integrated information theory (IIT) is a theory of consciousness that directly correlates the increase in nervous system complexity with deepening internal mental experience and computational intelligence.

In essence, it posits that the relational complexity of a system—that is, the way its parts relate to form an irreducible whole with unique causal power—provides a measure of the information integrated by the system, and it is this metric that correlates directly with the depth, intensity, and resolution of mental life.

A greater number of parts in a greater number of relationships gives a greater degree of “integrated information.” In a nervous system, those parts are neurons, and their relationships create an irreducibly complex structural whole—a measure that can be computed using a quantifiable metric, Φ (phi). As brains grow in size and complexity, Φ increases, such that the Mind not only processes more information but also awakens to a more conscious subjective depth.

In his 2019 book on integrated information theory, The Feeling of Life Itself: Why Consciousness is Widespread But Can’t Be Computed, neuroscientist Christof Koch writes:

How does brain size affect consciousness? Bigger networks have combinatorially more potential states than smaller ones. …[I]ntegrated information in nervous systems shaped by the fierce forces of natural selection over eons will increase with brain size. As a consequence, the ability of the current state of such a network to constrain trillions of its past and future states becomes more refined as network size grows. That is, the bigger the brain, the more complex its maximally irreducible cause-effect structure may be, the bigger its Φmax and the more conscious it becomes.

The rise of more complex brains, and thus higher Φ, is driven by evolutionary pressures; consciousness is evolutionarily advantageous. Koch observes, “As adaptation increases, so do the animat’s integrated information and the number of distinctions the system is capable of supporting. Thus, evolution selects organisms with high Φmax.”

So has the evolutionary process driven the deepening of Mind, from simpler animals to those with rich internal models of reality."