The unity of consciousness: a conversation with Wolf Singer

Wolf Singer is professor of neurophysiology and director of the Department of Neurophysiology at the Max Planck Institute for Brain Research in Frankfurt, Germany. In 2004, he founded the Frankfurt Institute for Advanced Studies (FIAS), which conducts basic theoretical research in various areas of science, bringing together theorists from the disciplines of biology, chemistry, neuroscience, physics, and computer science. His main research interest lies in understanding the neuronal processes underlying higher cognitive functions, such as visual perception, memory, and attention. He is also dedicated to making the results of brain research known to the general public and is a recipient of the Max Planck Prize for Public Science.

Singer has been particularly active in the philosophical debate concerning free will. He is coeditor (with Christoph Engel) of Better Than Conscious? Decision Making, the Human Mind, and Implications for Institutions (2008).

Metzinger: Wolf, given the current state of the art, what is the relation between consciousness and feature-binding?

Singer: A unique property of consciousness is its coherence. The contents of consciousness change continuously, at the pace of the experienced present, but at any one moment all the contents of phenomenal awareness are related to one another, unless there is a pathological condition causing a disintegration of conscious experience. This suggests a close relation between consciousness and binding. It seems that only those results of the numerous computational processes that have been bound successfully will enter consciousness simultaneously. This notion also establishes a close link among consciousness, short-term memory, and attention. Evidence indicates that stimuli need to be attended to in order to be perceived consciously, and only then will they have access to short-term memory.

Metzinger: But why is there a binding problem to begin with?

Singer: The binding problem results from two distinct features of the brain: First, the brain is a highly distributed system, in which a very large number of operations are carried out in parallel; second, it lacks a single convergence center, in which the results of these parallel computations could be evaluated in a coherent way. The various processing modules are interconnected, in an exceedingly dense and complex network of reciprocal connections, and these appear to be generating globally ordered states, by means of powerful selforganizing mechanisms. It follows that representations of complex cognitive contents — perceptual objects, thoughts, action plans, reactivated memories — must have a distributed structure as well. This requires that neurons participating in a distributed representation of a particular type of content convey two messages in parallel: First, they have to signal whether the feature they’re tuned to is present; second, they have to indicate which of the many other neurons they’re cooperating with in forming a distributed representation. It is widely accepted that neurons signal the presence of the feature they encode by increasing their discharge frequency; however, there’s less consensus about how neurons signal with which other neurons they cooperate.

Metzinger: What are the constraints for such a signaling?

Singer: Because representations of cognitive contents can change rapidly, it needs to be decipherable with very high temporal resolution. We’ve proposed that the relation-defining signature is the precise synchronization of the discharges of the individual neurons.

Metzinger: But why synchronization?

Singer: Precise synchronization increases the impact of neuronal discharges, favoring further joint processing of the synchronized messages. Further evidence indicates that such synchronization is best achieved if neurons engage in rhythmic, oscillatory discharges, because oscillatory processes can be synchronized more easily than temporally unstructured activation sequences.

Metzinger: Then this isn’t just a hypothesis — there’s supportive experimental evidence.

Singer: Since the discovery of synchronized oscillatory discharges in the visual cortex more than a decade ago, more and more evidence has supported the hypothesis that synchronization of oscillatory activity may be the mechanism for the binding of distributed brain processes — whereas the relevant oscillation frequencies differ for different structures and in the cerebral cortex typically cover the range of beta- and gamma-oscillations: 20 to 80 Hz. What makes the synchronization phenomena particularly interesting in the present context is that they occur in association with a number of functions relevant for conscious experience.

Metzinger: Which functions are those?

Singer: These oscillations occur during the encoding of perceptual objects, when coherent representations of the various attributes of these objects have to be formed. The oscillations are consistently observed when subjects direct their attention toward an object and retain information about it in working memory. And finally, the oscillations are a distinctive correlate of conscious perception.

Metzinger: What is the evidence here?

Singer: In a test in which subjects are exposed to stimuli that are degraded by noise so that the stimuli are consciously perceived only half the time, you can study the brain activity selectively associated with conscious experience. Since the physical attributes of the stimuli are the same throughout, you can simply compare brain signals in cases where the subjects consciously perceive the stimuli with the signals in cases where they don’t. Investigations reveal that during conscious perception, widely distributed regions of the cerebral cortex transiently engage in precisely synchronized high-frequency oscillations. When the stimuli are not consciously perceived, the various processing regions still engage in high-frequency oscillations — indicating that some stimulus-processing is performed — but these are local processes and do not join into globally synchronized patterns. This suggests that access to consciousness requires that a sufficiently large number of processing areas — or in other words, a sufficient number of distributed computations — be bound by synchronization and that those coherent states be maintained over a sufficiently long period.

Metzinger: This could be interesting from a philosophical perspective. Wouldn’t this ideally account for the unity of consciousness?

Singer: Indeed, this would also account for the unity of consciousness — for the fact that the contents of phenomenal awareness, although they change from moment to moment, are always experienced as coherent. Admittedly, the argument is somewhat circular, but if it is a necessary prerequisite for access to consciousness that activity be sufficiently synchronized across a sufficient number of processing regions, and if synchronization is equivalent with semantic binding, with integrating the meaning, it follows that the contents of consciousness can only be coherent.

Metzinger: What remains to be shown, if what you describe here turns out to be the case?

Singer: Even if the proposed scenario turns out to be true, the question remains as to whether we have arrived at a satisfactory description of the neuronal correlates of consciousness. What do we gain by saying that the neuronal correlate of consciousness is a particular metastable state of a very complex, highly dynamic, nonstationary distributed system — a state characterized by sequences of ever-changing patterns of precisely synchronized oscillations? Further research will lead to more detailed descriptions of such states — but these will likely be abstract, mathematical descriptions of state vectors. Eventually, advanced analytic methods may reveal the semantic content, the actual meaning of such state vectors, and it may become possible to manipulate these states and thereby alter the contents of consciousness, thus providing causal evidence for the relation between neuronal activity and the contents of phenomenal awareness. However, this is probably about as close as we can come, in our attempts to identify the neuronal correlates of consciousness. How these neuronal activation patterns eventually give rise to subjective feelings, emotions, and so on, will probably remain a conundrum for quite some time even if we arrive at precise descriptions of neuronal states corresponding to consciousness.

Metzinger: In your field, what are the most urgent questions, and where is the field moving?

Singer: The most challenging questions are how information is encoded in distributed neuronal networks and how subjective feelings, the so-called qualia, emerge from distributed neuronal activity. It is commonly held that neurons convey information by modulating their discharge rate — that is, by signaling the presence of contents for which they are specialized through increases in their firing rate. However, accumulating evidence suggests that complex cognitive contents are encoded by the activity of distributed assemblies of neurons and that the information is contained in the relations between the amplitudes and in the duration of the discharges. The great challenge for future work is to extract the information encoded in these high-dimensional time series. This requires simultaneous recordings from a large number of neurons and identification of the relevant spatio-temporal patterns. It is still unclear which aspects of the large number of possible patterns the nervous system exploits to encode information, so searching for these patterns will require developing new and highly sophisticated mathematical search algorithms. Thus, we’ll need close collaboration between experimentalists and theoreticians to advance our understanding of the neuronal processes underlying higher cognitive functions.

Metzinger: Wolf, why are you so interested in philosophy, and what kind of philosophy would you like to see in the future? What relevant contributions from the humanities are you waiting for?

Singer: My interest in philosophy is nurtured by the evidence that progress in neurobiology will provide some answers to the classic questions treated in philosophy. This is the case for epistemology, philosophy of mind, and moral philosophy. Progress in cognitive neuroscience will tell us how we perceive and to what extent our perceptions are reconstructions rather than representations of absolute realities. As we learn more about the emergence of mental functions from complex neuronal interactions, we will gain insight into possible solutions of the mind-body problem, and as we learn to understand how our brains assign values and distinguish between appropriate and inappropriate conditions, we will learn more about the evolution and constitution of morality.

Conversely, cognitive neuroscience needs the humanities — for several reasons. First, progress in the neurosciences raises a large number of new ethical problems, and these need to be addressed not only by neurobiologists but also by representatives of the humanities. Second, as neuroscience progresses, more and more phenomena that have traditionally been the subject of humanities research can be investigated with neuroscientific methods; thus, the humanities will provide the taxonomy and description of phenomena awaiting investigation at the neuronal level. Brain research begins with the analysis of such phenomena as empathy, jealousy, altruism, shared attention, and social imprinting — phenomena that have traditionally been described and analyzed by psychologists, sociologists, economists, and philosophers. Classification and precise description of these phenomena are prerequisites for the neuroscientific attempts to identify the underlying neuronal processes. There will undoubtedly be close collaborations in the near future between the neurosciences and the humanities — a fortunate development, as it promises to overcome some of the dividing lines that have segregated the natural sciences from the humanities over the last centuries.

PART TWO

Дата добавления: 2015-10-31