A Brief Overview of Integrated Information Theories of Consciousness

I have posted before on the proposed relationship between information theory and conscious, phenomenal states. For a brief background, consider the following: Information states have two fundamental attributes, one being intrinsic and the other extrinsic, the latter of which can also be called “relational.” Take, for example, one bit of information, say 11001101. In this bit, a sequence of 1s and 0s stands to mean something when it is called upon. The individual 1s and 0s can be labelled as the intrinsic elements. The extrinsic aspect, then, refers to the internal structure of the bit, which is where the term relational comes in. Each element has a definite position within the bit—there is a 1 in the first position, and a 0 in the third position—which marks where it is relative to all of the other elements. We can apply this to consciousness research, some say, by thinking of the intrinsic elements as the subjective side of an issue, or what it is like to be something. On the other hand, the relational parts represent the third-person perspective that we take when we study physics (that is, when the study the relations between fundamental things. This is why, as I have commented before, physics is utterly hopeless when it comes to understanding phenomenal consciousness).

Now, the view of Integrated Information theorists takes this a step further. (If we weren’t to clarify the relationship between information and consciousness, then it might seem like we were saying that anything and everything that has an information state—e.g., a thermostat—is therefore conscious, in some, perhaps limited, fashion. Some, the panpsychists, do say this, but this is not necessarily the view from Integrated Information.) They claim that the phenomena of consciousness, while in some ways fundamental to information states, also depends on the integration and differentiation of those information states. Our brains, along with those of many mammals and “lower” species, do an excellent job of fulfilling these requirements. Through less-than-clear mechanisms, our brains are able to both synchronize activity at a global level, but also keep information very well stratified throughout the layers and structures contained therein. For a counterexample, think of the brain during an epileptic seizure: information is everywhere, with electrical signals firing at multiple locations simultaneously. It could be said that this represents a form of integration, but this situation also clearly does away with any sort of differentiation. As predicted, seizures are generally accompanied by a loss of consciousness, or a diminished conscious state at most.

It is still hard to see how exactly the tenets of this theory might explain the “why” of consciousness, but it presents, at the least, some interesting ways to think about the “how.”

Chalmers on Physics and Phenomenology

“Physics requires information states but cares only about their relations, not their intrinsic nature; phenomenology requires information states, but cares only about their intrinsic nature. This view postulates a single basic set of information states unifying the two. We might say that internal aspects of these states are phenomenal, and the external aspects are physical. Or as a slogan: Experience is information from the inside; physics is information from the outside.

The above comes from David Chalmer’s The Conscious Mind and provides a brief account of his personal attempt to reconcile phenomenal and physical aspects of the most basic of entities (to clarify that he is speaking to basic entities, his assertion taken to an extreme postulates nothing more than information states as actually existing, at a fundamental level)—though the view could be translated up to macroscopic structures with careful consideration (he notes the difficulty of such a task in the surrounding text, though I think that he overstates the problem). I’ll take up this problem below (note, not all of my ideas follow directly from Chalmers thesis, I have incorporated ideas from elsewhere, notably Damasio):

On scaling up, from, say, a cell to a full-fledged brain, we start to get successively larger functional units—units with their own informational states—forming a sort of nested hierarchy of phenomenology all the way to the uppermost level: one full self, in the ordinary sense of the word. A problem in this process that he remarks upon is the associated “jaggedness” that would seemingly result from summing up smaller “phenomenal” (or proto-phenomenal, if you prefer) sub-units into one coherent whole. In my estimation, it seems that this is not a necessarily a problem, much in the way that upon summing up individual atoms, or even molecules to give a better sense of the problem, into physical objects, we do not experience macroscopic objects as being “jagged” in any way, but rather as continuous, complete objects. Upon investigation below the level of every day experience with modern tools of magnification, we are able to peek into the jagged quality of physical objects, but our natural tools (i.e., eyes) for observation of such entities lack the resolution to pick out the underlying jaggedness. In other words, the jaggedness is there, but we do not notice it due to the limited resolution of our perceptual systems. It may be that conscious experience is similar to this: it, too, possesses a level of jaggedness, but this eludes our introspective observation due to the high-level nature of introspection itself. On this view, implied jaggedness does not detract from Chalmers related assertions.

As for the strength of Chalmers’ overall argument, I cannot say. On the surface, it seems plausible, though many would disagree with me on that. At the very least, he has advanced thinking on the matter in a fundamental way. I’ll post a fuller critique of the theory later on.