E Pluribusunum: Rethinking the Unity of Consciousness
R. Van Gulick
Rethinking the Unity of Consciousness
16
E pluribusunum: Rethinking the Unity of Consciousness
Robert Van Gulick
Etymology is not always a reliable guide to meaning and even less so to truth, but perhaps there is something to be learned from the fact that the word “conscious” derives from the Latin verb “conscio,” which literally translates as “know together” (con + scio). Indeed, in one archaic use, it could mean knowledge shared among different people. The Oxford English Dictionary (2nd edition, 2000) defines this obsolete use as “sharing knowledge with another” and cites Thomas Hobbes in Leviathan (1651, I. vii. 31) where he wrote, “When two, or more men, know one and the same fact, they are said to be conscious of it,” as well Robert South slightly later (1693, II.ii.88), “Nothing is to be conceal’'d from the other self. To be a friend and to be conscious are terms equivalent.” Being conscious in this sense of knowing together is a mutual or shared mental activity, just as one confides or conspires—literally “breathes together”- ”—with another. We no longer use “conscious” in that way, but perhaps the surviving concept of “conscious” we apply to single individuals retains some sense of being known together, a way in which the very word “conscious” implies some form of unity or integration. The relevant unity would be within one mind or self, but still involve some way in which features or states of mind are shared or integrated.
Consciousness is generally believed to be unified in some important respect, but in what specific ways and to what degree is not as clear. Nor is there agreement about the status of such unity: Is it essential to consciousness as a logical or empirical matter? And if so, how so and why? If not, might unity nonetheless be important to our understanding of consciousness, and how so?
Unity and integration might figure in two distinct but complementary ways in theories of consciousness: either as an explanandum or as an explanans, i.e.that is, as a real feature of consciousness that needs to be explained, or as something to which we might appeal in explaining consciousness and its properties. Indeed, given the complexity of the actual theoretical situation, it might serve as both.
As with any complex phenomenon, a theory of consciousness needs to describe, and perhaps model, its many important features and properties. We need a good sense of what consciousness is before we can explain how it can exist or be produced. Unities of various sorts seem likely candidates for inclusion on any adequate list of the properties of consciousness, including representational unity, object unity, and subject unity, as well as introspective, access, and phenomenal unity. Indeed, each of those unities subdivides into yet more specific types. Representational unity might be unity of content or of vehicle, and unity of content in turn can take make many forms and degrees. Unity of subjectmight concern a unified subject of thought or one of action, and each in turn can take many yet more specific forms and degrees of integration.
All these various possible unities need to be adequately described or modeled, and each serves as a possible explanandum, a property or feature whose existence and basis needs to be explained by a comprehensive theory of consciousness. Some forms of conscious unity might also serve as an explanans, in so far as we might appeal to one sort of conscious unity to explain another, e.g., explaining phenomenal unity in terms of the representational unity of consciousness. (Tye, 2003).
Unification and integration of various sorts can also occur at unconscious levels, and some theories try to explain consciousness or its properties in terms of such unconscious unities or integrations. Like conscious unity, unconscious unity comes in many forms both psychological and neural, including representational, spatial, and multi-modal unities, as well as many sorts of functional and causal integrations, both within and between modules or subsystems of the mind or brain.
In answering the “how question,” many theories of consciousness appeal to such nonconscious unities. Indeed, some explain the crucial transition from unconscious mental state to conscious state in terms of such integrative or unifying processes. For example, on Bernard Baars’s global workspace theory (1988, 1997), a specific unconscious mental state becomes conscious when it is brought into that workspace and thus globally “'broadcast”' for integration with other contentful states in a wide range of different subsystems or modules. Stan Deheane has further developed the global workspace theory and combined it with a proposed neural model of the brain regions involved in carrying out the relevant integrations (Dehaene Naccahe,2001[JMF1]).
Integration plays a more direct and essential role in Giulio Tononi’s (2008) integrated information theory of consciousness. On Tononi’s model, a state of a system is conscious just if it has the highest degree of integrated informational content, which Tononi defines in terms of an information-theory based measure he calls Φ, which depends in part on the degree of interdependence between the states of the system and thus on their integration. I myself have proposed a model, the Higher Order Global States model (or HOGS), that explains the transition from unconscious to conscious mental state as a matter of its being recruited into to the unified global state that constitutes the transient substrate of a subject’s conscious mental stream (Van Gulick, 2004, 2006). Though the HOGS model agrees with the workspace theories of Baars and Dehaene in treating the transition as a matter of increased global integration, it differs in the specific form of self-like unity it proposes.
Thus the unity of consciousness is not one issue or one question. It generates a variety of questions within a problem space defined by the many possible forms of conscious and unconscious integration and their possible explanatory connections. We must determine which types of conscious unity are real, and then describe and explain them. As to unconscious forms of unity and integration, they too must be modeled and described. And at least aAccording to many theorists at least, they are likely to play an important role in explaining the “how” of consciousness. Their guiding hypothesis is that consciousness, or at least some of its key features, is realized or produced by underlying nonconscious integrative processes. If so, nonconscious forms of unity and integration may figure as key explanantia in our understanding of consciousness.
Philosophical discussions of the unity of consciousness often concern whether unity of one sort or another is a necessary condition for consciousness, or alternatively whether it is sufficient for it. Both sorts of questions are open to logical as well as empirical readings. If phenomenal unity is a necessary feature of human consciousness, is that a matter of logical necessity, nomic necessity, or merely a contingent fact about the particular structure of human consciousness or its substrate? Some scientific theories of consciousness also assert or imply claims about the necessity or sufficiency of one or another sort of unity or integration. Tononi’s integrated information theory explicitly equates consciousness with having a high Φ value, and global workspace and HOGS models both regard integration into a larger unified state as a necessary element of the transition from unconscious to conscious state.
Unity may bear an important relation to consciousness even if it is not strictly necessary nor or sufficient. Scientific theories of a complex phenomenon Z often invoke explanatory properties that are in themselves neither necessary nor sufficient for Z, but nonetheless help us understand the nature of Z. The relevant property P, for example, might be a necessary part of some condition S that is sufficient for producing Z, but not uniquely so. Even though there may be other alternative ways to produce or realize Z, doing so in the S-way essentially involves P. For example, consciousness might berealized in one architecture that requires integration of content across modular subsystems; human consciousness may in fact do so. But there nonetheless be may be other ways to produce consciousness in systems with a different functional organization—e.g., some conditions S* [KH2]that suffice in systems without a modular structure. Thus, what might be necessary for consciousness in one systemic context might not be required in another.
Even if unity were not necessary for consciousness per se, it might nonetheless be necessary to understanding its function. Given any sort of unity one might initially think essential to consciousness, both clinical evidence and thought experiments may provide reason to believe that some limited cases of consciousness can occur without that form of unity, no matter how common it is in ordinary conscious experience. Nonetheless, consciousness may need to be unified in that way to carry out at least some of the functions that make it valuable and adaptive.
For example, our normal conscious life involves the unified experience of integrated objects and scenes, and having such experiences surely requires specific forms of representational integration at the conscious and underlying nonconscious levels. However, we know that patients suffering from perceptive visual agnosia have great difficulty integrating visual stimuli into coherent wholes, though there is no doubt that they have visual experiences. Patients with simultanagnosia, Bálint’sBalint’s syndrome, cannot see more than one object at a time, and thus are incapable of having a unified experience of a scene. Moreover, with unimpaired subjects, it seems possible to have some minimal experience which with no integration of object or scene. Imagine having just the experience of a dim flicker that passes so quickly that one cannot say just where it occurred or whether it was of any given color or shape, or the experience of brief, faint sound whose location and tone one cannot discern. Such stripped- down experience seems possible despite its lack of any parts to integrate or unify. It seems possible to have at least some conscious experiences that do not involve such unity. Thus if we think in terms of necessary conditions, we might conclude that such unities of object and scene are not essential or central to understanding consciousness.
However, that need not follow. Even if consciousness in some pathologically restricted cases lacks such unities, it may be the capacity of consciousness to support and enable such forms of unity and integration that explains why consciousness is important and useful. Enabling and supporting widespread integration in a dynamically unified representation may be one of consciousness’s central powers, even if it can be blocked from doing so in special cases. If so, understanding the nature of consciousness would require explaining how it comes to have that power and exercise it in normal conditions. If that is one of its key functions, then we need to understand what it is about consciousness and its underlying basis that enables it to play that role in normal contexts. The fact that the exercise of that power may be blocked in abnormal cases does not show that its capacity to support such integration is not central to its nature and value.
In introducing these issues, I have spoken interchangeably of “unity” and “integration,” and I will continue to do so below. The two notions are closely related, though they may have subtly different associations and convey somewhat different implications. Integration leads us to think in terms of a process, whereas unity may seem more like a basic fact or result. It is also natural to think of integration as admitting of degrees. Unity as well can be treated as a matter of degree, but there is also some pull toward thinking of it as all or none.
Once again, etymology is worth noting. The Latin root of “unity” literally invokes the idea of “one-ness” from the number “unum.” What is united is one thing; and that might seem like a simple and determinate fact, e.g.for example, is there one conscious subject or not? “Integration,” which shares its root with “integer,” turns on a slightly different metaphor, that of combining into an integer or whole (a whole as what is literally “untouched” --— from “in” meaning not + “tangere”). Especially when one is dealing with complex systems, what constitutes a whole may turn on many factors, and we are accustomed to the idea that new wholes may arise from suitably related or interacting parts. Though the idea of unity as one-ness may incline us more to think in terms of what is simple, and integration more in terms of what has an underlying complex basis, each of the two notions can be used to think about the way in which consciousness coheres and how it might result from the coherent interaction of underlying nonconscious processes. Indeed, having both notions may aid our theorizing by offering two slightly different conceptual perspectives on the same basic process.
Before moving on to consider some more specific questions, let me recap the general structure of the problem space. Unity may occur in many conscious and nonconscious forms as shown in table 16.1. Some questions concern the reality of those varying sorts of unity. Which are true of consciousness in general, or of human consciousness? Other questions concern relations between the various sorts of unity, both conscious and nonconscious. Which sorts of unities might be explained fully, or at least partly, in terms of others? Which forms of unity might be necessary or sufficient for consciousness, or human consciousness, or at least important to our understanding of its nature, function, and substrate?
[Table 16.1 near here]
Table 16.1 aims to display the general problem space, with column three having a special structure that includes both various types of conscious unity as well as consciousness itself. The table can be read either across the columns or up and down within column 3 (and perhaps within column 1). Reading across, one set of questions can be generated by selecting specific items from each of the three columns: Is unconscious multimodal integration necessary for multimodal conscious integration? Is the unconscious unity of thought and action sufficient for the conscious unity of subject? Is the unconscious representational unity of content sufficient for consciousness itself? Other questions can be generated by applying one of the linking relations from column 2 with various pairings within column 3, either between various specific forms of conscious unity, or between such unities and consciousness itself: Is conscious representational unity sufficient for phenomenal unity? Is conscious object unity necessary for conscious subject unity? Is phenomenal unity necessary for consciousness? Does the unity of the experienced world explain the functional value of consciousness?
Some cross pairings generate more interesting and plausible linkages than others, but it is useful to have an overview of the full range of possible connections. Understanding the unity of consciousness requires understanding how its various forms relate to each other and to consciousness itself, as well as to the various sorts of nonconscious unity that may provide their underlying substrate. A comprehensive exam of the full problem space is beyond the scope of the present chapter, and I will instead focus for the remainder of this chapter on a few specific questions about the relations between representational unity, phenomenal unity, and consciousness.
As noted above, the neuroscientist Giulio Tononi has developed an influential theory of consciousness that identifies it with a form of integrated information that his theory defines in purely information theoretic terms (Tononi, 2008). Tononi’s proposal is thus a reductive theory that aims to fully explain consciousness in terms of nonconscious integration. He writes, “The integrated information theory (IIT)[KH3]of consciousness claims that, at the fundamental level, consciousness is integrated information, and that its quality is given by the informational relationships generated by a complex of elements.” (2008). Since the supposed relation is one of identity, relative to figure 16.1 Tononi’s theory should be understood as asserting that a type of nonconscious informational unity from column 1 provides both a necessary and sufficient condition (link from column 2) for consciousness itself in column 3. The key idea in Tononi’s IIT is that of integrated information for which he proposes a mathematical measure he terms “Φ” defined in purely information theoretic terms (with the symbol “Φ” itself composed of two components “I” for information and the circular “Ο” for integration within a whole).
For present purposes, we need not go into the precise mathematical definition of Φ used by IIT. What matters is that Φ concerns the information within a complex or system that results from the interactions and causal dependencies among its parts as oppposed to the information in the parts themselves. As Tononi puts it, “” In short, integrated information captures the information generated by causal interactions in the whole, over and above the information generated by the parts.” (2008, p.221).
To illustrate his point, Tononi uses the example of the detector in a digital camera as an example of non-integrated information. The camera’s detector may have five million pixel elements, each with its own information value, but that information is not integrated; each is an independent unit simply signaling the light value for its small portion of the scene in isolation. By contrast, when one has a conscious visual experience—as when I look at the cluttered desk in front of me—the information about all the parts of the scene is integrated into a unified awareness of the overall environment from a single subjective viewpoint.This is with an understanding of how the parts fit together as well as their connections with all sorts of other stored information, including my knowledge and memory of the various items on my desk.[KH4]
According to Tononi, a complex that embodies such integrated information literally has a point of view, or at least does so if it is not embedded within a yet more integrated complex with a higher Φ value. He writes, “Specifically, a complex X is a set of elements that generate integrated information (> 0) that is not fully contained in some larger set of higher Φ.” (2008, p.221). A complex, then, can be properly considered to form a single entity having its own, intrinsic “point of view” (as opposed to being treated as a single entity from an outside, extrinsic point of view). The restriction on not being contained within a set of elements with a higher Φ is relevant to the case of the conscious mind or brain. A human brain will contain many subsystems with some significant measure of integrated information such as the visual cortex or auditory cortex, but they do not each have their own separate consciousness or subjective point of view. Only the larger corticoal-thalamic complex of globally integrated elements is conscious and has such a view point, or at least that is the supposed implication of Tononi’s theory.