An open mind, an open question…

An Information Integration Theory of Consciousness

69 min read
Abstract Background Consciousness poses two main problems. The first is understanding the conditions that determine to what extent a system has conscious experience. For instance, why is our consciousness generated by certain parts of our brain, such as the thalamocortical system, and not by other parts, such as the cerebellum? And why are we conscious [...]

The second problem: What determines the kind of consciousness a system has?

Even if we were reasonably sure that a system is conscious, it is not immediately obvious what kind of consciousness it would have. As was mentioned early on, our own consciousness comes in specific and seemingly irreducible qualities, exemplified by different modalities (e.g. vision, audition, pain), submodalities (e.g. visual color and motion), and dimensions (e.g. blue and red). What determines that colors look the way they do, and different from the way music sounds, or pain feels? And why can we not even imagine what a “sixth” sense would feel like? Or consider the conscious experience of others. Does a gifted musician experience the sound of an orchestra the same way you do, or is his experience richer? And what about bats [10]? Assuming that they are conscious, how do they experience the world they sense through echolocation? Is their experience of the world vision-like, audition-like, or completely alien to us? Unless we accept that the kind of consciousness a system has is arbitrary, there must be some necessary and sufficient conditions that determine exactly what kind of experiences it can have. This is the second problem of consciousness.

While it may not be obvious how best to address this problem, we do know that, just as the quantity of our consciousness depends on the proper functioning of a physical substrate – the brain, so does the quality of consciousness. Consider for example the acquisition of new discriminatory abilities, such as becoming expert at wine tasting. Careful studies have shown that we do not learn to distinguish among a large number of different wines merely by attaching the appropriate labels to different sensations that we had had all along. Rather, it seems that we actually enlarge and refine the set of sensations triggered by tasting wines. Similar observations have been made by people who, for professional reasons, learn to discriminate among perfumes, colors, sounds, tactile sensations, and so on. Or consider perceptual learning during development. While infants experience more than just a “buzzing confusion”, there is no doubt that perceptual abilities undergo considerable refinement – just consider what your favorite red wine must have tasted like when all you had experienced was milk and water.

These examples indicate that the quality and repertoire of our conscious experience can change as a result of learning. What matters here is that such perceptual learning depends upon specific changes in the physical substrate of our consciousness – notably a refinement and rearranging of connections patterns among neurons in appropriate parts of the thalamocortical system (e.g [11]). Further evidence for a strict association between the quality of conscious experience and brain organization comes from countless neurological studies. Thus, we know that damage to certain parts of the cerebral cortex forever eliminates our ability to perceive visual motion, while leaving the rest of our consciousness seemingly intact. By contrast, damage to other parts selectively eliminates our ability to perceive colors. [12]. There is obviously something about the organization of those cortical areas that makes them contribute different qualities – visual motion and color – to conscious experience. In this regard, it is especially important that the same cortical lesion that eliminates the ability to perceive color or motion also eliminates the ability to remember, imagine, and dream in color or motion. By contrast, lesions of the retina, while making us blind, do not prevent us from remembering, imagining, and dreaming in color (unless they are congenital). Thus, it is something having to do with the organization of certain cortical areas – and not with their inputs from the sensory periphery – that determines the quality of conscious experiences we can have. What is this something?

Characterizing the quality of consciousness as a space of informational relationships: The effective information matrix

According to the theory, just as the quantity of consciousness associated with a complex is determined by the amount of information that can be integrated among its elements, the quality of its consciousness is determined by the informational relationships that causally link its elements [13]. That is, the way information can be integrated within a complex determines not only how much consciousness is has, but also what kind of consciousness. More precisely, the theory claims that the elements of a complex constitute the dimensions of an abstract relational space, the qualia space. The values of effective information among the elements of a complex, by defining the relationships among these dimensions, specify the structure of this space (in a simplified, Cartesian analogue, each element is a Cartesian axis, and the effective information values between elements define the angles between the axes, see Appendix, v). This relational space is sufficient to specify the quality of conscious experience. Thus, the reason why certain cortical areas contribute to conscious experience of color and other parts to that of visual motion has to do with differences in the informational relationships both within each area and between each area and the rest of the main complex. By contrast, the informational relationships that exist outside the main complex – including those involving sensory afferents – do not contribute either to the quantity or to the quality of consciousness.

To exemplify, consider two very simple linear systems of four elements each (Fig. 2). Fig. 2a shows the diagram of causal interactions for the two systems. The system on the left is organized as a divergent digraph: element number 1 sends connections of equal strength to the other three elements. The analysis of complexes shows that this system forms a single complex having a Φ value of 10 bits. The system on the right is organized as a chain: element number 1 is connected to 2, which is connected to 3, which is connected to 4. This system also constitutes a single complex having a Φ value of 10 bits. Fig. 2b shows the effective information matrix for both complexes. This contains the values of EI between each subset of elements and every other subset, corresponding to all informational relationships among the elements (the first row shows the values in one direction, the second row in the reciprocal direction). The elements themselves define the dimensions of the qualia space of each complex, in this case four. The effective information matrix defines the relational structure of the space. This can be thought of as a kind of topology, in that the entries in the matrix can be considered to represent how close such dimensions are to each other (see Appendix, vi). It is apparent that, despite the identical value of Φ and the same number of dimensions, the informational relationships that define the space are different for the two complexes. For example, the divergent complex has many more zero entries, while the chain complex has one entry (subset {1 3} to subset {2 4}) that is twice as strong as all other non-zero entries.

Figure 2


Effective information matrix and activity states for two complexes having the same value of Φ. a. Causal interactions diagram and analysis of complexes. Shown are two systems, one with a “divergent” architecture (left) and one with a “chain” architecture (right). The analysis of complexes shows that both contain a complex of four elements having a Φ value of 10. b. Effective information matrix. Shown is the effective information matrix for the two complexes above. For each complex, all bipartitions are indicated by listing one part (subset A) on the upper row and the complementary part (subset B) on the lower row. In between are the values of effective information from A to B and from B to A for each bipartition, color-coded as black (zero), red (intermediate value) and yellow (high value). Note that the effective information matrix is different for the two complexes, even though Φ is the same. The effective information matrix defines the set of informational relationships, or “qualia space” for each complex. Note that the effective information matrix refers exclusively to the informational relationships within the main complex (relationships with elements outside the main complex, represented here by empty circles, do not contribute to qualia space). c. State diagram. Shown are five representative states for the two complexes. Each is represented by the activity state of the four elements of each complex arranged in a column (blue: active elements; black: inactive ones). The five states can be thought of, for instance, as evolving in time due the intrinsic dynamics of the system or to inputs from the environment. Although the states are identical for the two complexes, their meaning is different because of the difference in the effective information matrix. The last four columns represent four special states, those corresponding to the activation of one element at a time. Such states, if achievable, would correspond most closely to the specific “quale” contributed by that particular element in that particular complex.

These two examples are purely meant to illustrate how the space of informational relationships within a complex can be captured by the effective information matrix, and how that space can differ for two complexes having similar amounts of Φ and the same number of dimensions. Of course, for a complex having high values of Φ, such as the one underlying our own consciousness, qualia space would be extraordinarily large and intricately structured. Nevertheless, it is a central claim of the theory that the structure of phenomenological relationships should reflect directly that of informational relationships. For example, the conscious experiences of blue and red appear irreducible (red is not simply less of blue). They may therefore correspond to different dimensions of qualia space (different elements of the complex). We also know that, as different as blue and red may be subjectively, they are much closer to each other than they are, say, to the blaring of a trumpet. EI values between the neuronal groups underlying the respective dimensions should behave accordingly, being higher between visual elements than between visual and auditory elements. As to the specific quality of different modalities and submodalities, the theory predicts that they are due to differences in the set of informational relationships within the respective cortical areas and between each area and the rest of the main complex. For example, areas that are organized topographically and areas that are organized according to a “winner takes all” arrangement should contribute different kinds of experiences. Another prediction is that changes in the quality and repertoire of sensations as a result of perceptual learning would also correspond to a refinement of the informational relationships within and between the appropriate cortical areas belonging to the main complex. By contrast, the theory predicts that informational relationships outside a complex – including those among sensory afferents – should not contribute directly to the quality of conscious experience of that complex. Of course, sensory afferents, sensory organs, and ultimately the nature and statistics of external stimuli, play an essential role in shaping the informational relationships among the elements of the main complex – but such role is an indirect and historical one – played out through evolution, development, and learning [14] (see Appendix, vii).

Specifying each conscious experience: The state of the interaction variables

According to the theory, once the quantity and quality of conscious experience that a complex can have are specified, the particular conscious state or experience that the complex will have at any given time is specified by the activity state of its elements at that time (in a Cartesian analogue, if each element of the complex corresponds to an axis of qualia space, and effective information values between elements define the angles between the axes specifying the structure of the space, then the activity state of each element provides a coordinate along its axis, and each conscious state is defined by the set of all its coordinates). The relevant activity variables are those that mediate the informational relationships among the elements, that is, those that mediate effective information. For example, if the elements are local groups of neurons, then the relevant variables are their firing patterns over tens to hundreds of milliseconds.

The state of a complex at different times can be represented schematically by a state diagram as in Fig. 2c (for the divergent complex on the left and the chain complex on the right). Each column in the state diagram shows the activity values of all elements of a complex (here between 0 and 1). Different conscious states correspond to different patterns of activity distributed over all the elements of a complex, with no contribution from elements outside the complex. Each conscious state can thus be thought of as a different point in the multidimensional qualia space defined by the effective information matrix of a complex (see Appendix, viii). Therefore, a succession or flow of conscious states over time can be thought of as a trajectory of points in qualia space. The state diagram also illustrates some states that have particular significance (second to fifth column). These are the states with just one active element, and all other elements silent (or active at some baseline level). It is not clear whether such highly selective states can be achieved within a large neural complex of high Φ, such as that one that is postulated to underlie human consciousness. To the extent that this is possible, such highly selective states would represent the closest approximation to experiencing that element’s specific contribution to consciousness – its quality or “quale”. However, because of the differences in the qualia space between the two complexes, the same state over the four elements would correspond to different experiences (and mean different things) for the two complexes. It should also be emphasized that, in every case, it is the activity state of all elements of the complex that defines a given conscious state, and both active and inactive elements count.

To recapitulate, the theory claims that the quality of consciousness associated with a complex is determined by its effective information matrix. The effective information matrix specifies all informational relationships among the elements of a complex. The values of the variables mediating informational interactions among the elements of a complex specify the particular conscious experience at any given time.

Related posts:

Leave a Reply

Your email address will not be published. Required fields are marked *