The Global Workspace Hypothesis
The studies we’ve looked at so far tell us something about where in the brain we can find activity crucial for consciousness; and as we’ve seen, there plainly is no single brain site that serves as the “seat of consciousness.” The studies also tell us something about when the brain changes that give rise to consciousness actually occur. However, we wantto go beyond merely specifying this where and when. We also want to learn how these brain regions might work together to produce the properties we associate with consciousness and to provide a biological base for the claims we made earlier about the function of consciousness.
In approaching these issues, we need to start with some basic facts. As we saw, each area of the brain has a highly specialized function. One brain region is specialized for controlling hand movements, another for processing auditory inputs, and still another for analyzing visual stimuli. Then, within each of these regions, we find fur-ther specialization. For example, the various aspects of visual perception depend on dis-tinct brain areas, one specialized for perceiving color, another for perceiving movement, and so on. Each of these areas does its own job, and the activity in each area is highly transient: The neurons respond to the current inputs and then, when their immediate task is done, they cease responding so they’re ready for the next input or the next task.
For many purposes, though, we need to sustain the activity in these various systems (e.g., for prolonged scrutiny). We also need to integrate the activity of the various systems into a coherent whole. What makes both of these steps possible is attention—probably implemented through mechanisms in the prefrontal cortex, but then influencing activity in many other brain sites (Maia & Cleeremans, 2005). The mechanisms of attention can amplify the activity within a specific neural system, and they can also sustain that activity.
Attention also seems to link the activity of different neural systems, binding them into a single representation. Thus, for example, a red moving object in front of our eyes will trigger a response in one brain area in which cells are sensitive to motion as well as in an area in which cells are sensitive to color. If we aren’t paying attention to this object, these two neural responses will be independent of each other. But if we are paying attention to it, the neurons in these two systems fire in synchrony . And when neurons fire in this coordinated way, the brain seems to register this as a linkage among the dif-ferent processing areas. As a result, these attributes are bound together so that we correctly perceive the stimulus as a unified whole
This coordination of separate neural systems requires communication among distinct brain areas, and this communication is made possible by “workspace neurons” that literally connect one area of the brain to another. However, the communication, and thus the information carried by the workspace neurons, is selective, and so it’s certainly not the case that every bit of neural activity gets linked to every other bit. Instead, it’s typi-cally the signals from more active areas that are transmitted to other sites. Bear in mind, though, that attention can be used to amplify activity and so can govern which brain areas are more activated and which are less. In this way, the information flow is controllable by virtue of what the person chooses to pay attention to.
These points are the backdrop for the global workspace hypothesis about consciousness. According to this hypothesis the integrated neural activity made possible by the workspace neurons provides the biological basis for consciousness. Activity in these neurons does not specify the content of consciousness; that content is instead represented in more specialized brain areas. Thus, when you’re aware of the blue sky overhead, that “blueness” is represented by the appropriate pattern of firing in your visual cortex; when you’re thinking about how cold your toes are, the “coldness” is represented by firing in somatosensory areas. What the workspace neurons do, however, is glue these bits together, creating a unified experience and allowing the exchange of information from one module to the next. (For several variants of this general hypothesis, see Baars, 2002; Baars & Franklin, 2003; Cooney & Gazzaniga, 2003; F. Crick & Koch, 2003; Dennett, 2001; Engel & Singer, 2000; Maia & Cleeremans, 2005; Roser & Gazzaniga, 2004; Tononi, 2004.)
The broad proposal, therefore, is that stimuli or ideas become conscious when they’re linked to each other in a dynamic, coherent representation made possible by the workspace neurons and supported by attention. This hypothesis helps us understand why our con-scious experience feels unitary and coherent. We aren’t, after all, separately aware of redness and movement and roundness. Instead, we’re aware of a single experience in which the red apple rolls slowly by. This coherence, of course, is precisely what the workspace allows—one representation, constructed from the coordinated activity of many processing components.
Likewise, we aren’t conscious of every aspect of our experience—we can focus on the rose’s color and fail to notice its thorns. But, of course, we can usually choose what we’re going to focus on (so that, when picking up the rose, we might decide to pay attention to those thorns). These observations are easily accommodated by the workspace hypothesis. As we’ve said, the information carried by the workspace neurons is selective (so might not carry the information, at least at the start, about the thorns) but is shaped by how some-one focuses their attention (and so, with a change in focus, would highlight the thorns).
Also bear in mind that the workspace neurons allow us to maintain a mental representation in an active state for an extended period of time, so that we can continue thinking about a stimulus or an idea after the specific trigger is removed. In this way, we can link the workspace to a form of memory known as working memory—the mem-ory that you keep ideas in while you’re working with them . We can also link it to the brain areas associated with this type of memory—specifically, the prefrontal cortex (Goldman-Rakic, 1987; also see McIntosh, Rajah, & Lobaugh, 1999).
Finally, the workspace hypothesis may also help with other puzzles—including the variations in consciousness that we sometimes experience. For example, when we’reasleep (and not dreaming), we aren’t conscious of time passing, of any ongoing stream of thought, or of many events taking place in our vicinity. Why is this? Evidence sug-gests that when we’re in sleep (without dreaming), the communication breaks down between different parts of the cortex so that, even though the sleeping brain is intensely active, the various activities aren’t coordinated with each other. The sugges-tion, of course, is that this communication (mediated by the neuronal workspace) is crucial for consciousness, so it makes sense that sleeping people, having temporarily lost this communication, aren’t conscious of their state or their circumstances (Massimini et al., 2005).
Likewise, when someone is anesthetized (e.g., before surgery), their brain remains active. Nonetheless, the person is not conscious—after all, that’s the point of anesthe-sia! So how should we think about the anesthetized brain? According to one recent review, the “loss of consciousness is associated with a breakdown of cortical connectiv-ity and thus of integration” (Alkire, Hudetz, & Tononi, 2008, pg. 879)—a proposal, again, in line with the idea that consciousness depends on communication and coordi-nation among distinct neural systems.
Overall, then, the global workspace hypothesis—rooted in psychology, philosophy, and neuroscience—links several lines of argument. These include claims about the subjective “coherence” of consciousness, research on the neural correlates of consciousness, and evidence from people who are sleeping or anesthetized. We should acknowledge that there is still controversy about some of these points (e.g., N. Block, 1997, 2001; Chalmers, 1995; Kinsbourne, 2000); but even so, the workspace hypothe-sis appears enormously promising as an account of how our brains make consciousness possible.