The Ravenous Brain: How the New Science of Consciousness Explains Our Insatiable Search for Meaning

by Bor, Daniel

  But for Tammet, it certainly wasn’t as if absolutely no structure was imposed on these sequences. He might not have noticed any additional structure for the mathematically chunkable sequences, but for him, because of his rich form of synesthesia, in a sense every trial was highly structured—it’s just that the structure resided in his head, in the multisensory mountain range of his mind’s eye that appears whenever he thinks of a stream of numbers. Reflecting this, for all trials, on average, whether they were structured or not, he had markedly increased activity in the prefrontal part of his prefrontal parietal network compared to the normal volunteers. This is because, for him, every trial, not just the structured trials, were being chunked at least to some extent. So, in two unexpected ways, this experiment reconfirmed the links between chunking, consciousness, and the prefrontal parietal network.

  This collective evidence, then, shows that consciousness is most closely connected with the prefrontal parietal network, which supports not only attention and working memory processes but also any kind of novel or complex task. But if you want to activate this network the most powerfully, and by extension engage your consciousness to its most intense heights, you need to detect some useful pattern. These chunking processes are an embedded part of our advanced conscious cognitive machinery. They are also perhaps the essence of what it means to be conscious. They are the mechanisms by which we convert awkward obstacles into innovative solutions and initial, error-prone fumblings into adept automatic habits.

  HARMONIOUS EXPERIENCES

  Admittedly, though, these are broad strokes in painting the details of how the brain creates consciousness. One way of looking at the problem in a deeper way is to ask how neurons communicate with each other in order to generate our experiences. Although this question was partially answered in the previous chapter, by my description of how attention equates to coalitions of neurons competing for dominance, another feature of neuronal communication is the waves of activity used to connect brain cells together. The main tool for examining this neural chatter is EEG, which lacks the fine-grained spatial resolution of fMRI but can collect data every millisecond, compared to the second or two it takes for an fMRI scanner to grab a picture of the brain’s activity.

  Francis Crick was the main early champion in this area. One of the most famous scientists of the twentieth century, Crick codiscovered the structure of DNA in the early 1950s, and proceeded to contribute revolutionary findings in genetics for a generation before turning in the last two decades of his life to the science of consciousness, which he believed was the greatest unsolved mystery in biology.

  Crick popularized the idea that when neurons act in harmony in a certain way, then consciousness ensues. The particular frequency bandied about for this love-in between neurons was originally the gamma band, roughly averaging to 40 cycles per second, one of the fastest frequencies that neuronal communication is capable of (and a frequency previously linked with attention).

  Although there is solid support for the link between this gamma band and awareness, this thesis requires a couple of tweaks. For instance, in rats at least, these fast waves are observed, along with slow delta waves, when the animals are under general anesthesia. Rats can also generate these swift types of neuronal synchrony when in a deep sleep.

  The resultant updates to this suggestion that the gamma band reflects consciousness suggest that local communication between neighboring neurons using these frequencies is not sufficient for consciousness to arise. What’s required is a gamma rhythm binding together the information between neurons that may be on separate neural continents—for instance, one region in the prefrontal cortex toward the front, and another in the posterior parietal cortex toward the back.

  And perhaps gamma isn’t quite fast enough for consciousness—instead, what’s called “high gamma,” with frequencies from 50 cycles a second, possibly even up to 250 cycles a second, is currently a hot topic in consciousness research. This frequency band is too swift to study using conventional EEG because of the interference of the signal by the scalp. But some epileptic patients soon to undergo a brain operation to remove the locus of their epilepsy have EEG arrays implanted directly onto their cortex, under their scalp, in order to investigate where they are having their seizures. When the electrodes are immediately over neurons, then you can pick up such high frequencies with ease. Two independent labs, Stanislas Dehaene’s near Paris and Bob Knight’s at the University of California at Berkeley, have both shown, using this technique, that when the patients are free from seizures and normally awake, these ultrahigh waves of neuronal activity are exquisitely connected with consciousness, and might, in concert with and locked to lower frequencies, be the main neuronal signature of awareness. Why such high frequencies?

  The answer follows from the purpose of awareness. In order for consciousness to carry out such complex tasks and generate insights, it needs to make connections in two ways: First, it needs attention, not just to select the most pertinent mental objects to place in working memory, but for all aspects of that object to be knitted together into a single, coherent whole. So when I see Angelina Jolie in that red dress on the silver screen, I don’t see her eyes as one object, her nose as a separate, disconnected object, her hands and name and voice as others. Instead, she is just a single, unified, though complex object, with many different components all connected together. Very importantly, I really can’t help consciously recognizing her as Angelina Jolie, as opposed to all these distinct parts.

  The color of her dress may nevertheless be represented in my visual cortex at the back of my brain, her face in my fusiform face area at the bottom of my cortex, her name and various other facts about her in my semantic store, at the front of my temporal lobes, and so on. When I recognize Angelina Jolie, many specialist cortical areas, all over my brain, are involved in that recognition, as well as my prefrontal parietal network, which acts as a conscious, temporary manager of the information in working memory. Now, if all of these spatially disparate regions need to be joined together to represent the conscious mental object, Angelina Jolie, a slow neuronal rhythm, say of a few cycles a second, simply isn’t up to the task, since there is too much data to hold together in the time. This slow kind of rhythm, the delta band, is instead commonly found when someone is under general anesthesia. The conscious frequency, in contrast, is as fast as our brains can allow. High gamma waves of activity are initiated by the thalamus, the Grand Central Station of brain regions in the center, and then these ultrafast waves perpetuate through the relevant parts of the cortex and bind together all components of an object in consciousness.

  The second type of connection required for consciousness is between those items in working memory in order to spot patterns or chunks, or simply to maintain a sequence of items. A fast rhythm between neurons throughout the cortex can maintain the links between distinct objects and allows us to analyze and manipulate them in working memory.

  EEG, with its millisecond accuracy, is also excellent at providing a time frame for consciousness. These widespread high gamma rhythms don’t immediately follow the presentation of a stimulus. Instead, they are relatively protracted, as you might expect when much of the brain has to coordinate its activity, and they take at least 300 milliseconds to form. This is a very similar value to the time it takes attention to filter sensory input according to our current goals.

  CONSCIOUSNESS, IN THEORY

  I can now paint an overall empirical picture of consciousness. If a vivid red rose comes into view, my experience of it is built up over a third of a second, as an initially brutal neuronal competition leads to the shaping of brain activity around my attention toward the rose. An ultrafast, harmonious neuronal rhythm spreads outward from the thalamus and merges my collective neural information of the rose, which is stored in specialist areas throughout my cortex. This high-frequency, long-range, unified mental chunk will also broadcast itself into the prefrontal parietal network, where the experience will come to life.

&n
bsp; But if I were faced with a more novel or complex task, my consciousness would show its true potential. My prefrontal parietal network activity would reflect an engaged working memory, a focused attention, and a ravenous search for patterns in order to conquer whatever mental obstacle was in my way. Meanwhile, my specialist regions of cortex—for example, areas that store knowledge about objects at the front of the temporal lobes—would take turns to support my consciousness by providing the specific contents to my experiences.

  The next challenge for consciousness researchers, in the decades to come, is to discover exactly how neurons collectively represent the information they do and what forms of neuronal interaction generate consciousness. For instance, just how is information transmitted between the fusiform face area and the prefrontal parietal network, via these high gamma waves, to generate my experience of my daughter’s face? And what is the precise code that neurons use to represent information? Such questions may be answered by simultaneously recording the activity of each of thousands of neurons in multiple regions. At present, the state of the art is limited to dozens of simultaneous electrodes (in the macaque monkey, the closest species to us where these studies are routinely carried out), so the technology falls considerably short of the kind of data collection required. But there’s every reason to assume that in the next decade or two the methods will be sufficiently advanced for us to discover and extract the precise neural signature of consciousness.

  In the meantime, many scientists have created detailed theories based on the existing empirical picture. Admittedly, there was a rather wild crop of early theories in the final decade or two of the twentieth century—for instance, one seemed to rely on the impeccably argued syllogism that because consciousness is mysterious and quantum mechanics is mysterious, then quantum mechanics must explain consciousness. But now the story is very different, with theories linking closely with the latest empirical findings. What is striking about the most prominent current crop of theories is how they are all broadly converging on the same overall position.

  The three most popular serious theories of the day all, at their heart, see consciousness as a particular flavor of dense information transmission across a large cortical network. But each theory differs subtly in its particular perspective on this general view.

  Victor Lamme’s recurrent processing model starts with the stark assumption that we may think we know when we are conscious of something, but we couldn’t be more wrong. According to Lamme, there are many times when we are actually conscious but we don’t even realize it. So he abandons talking about psychology, and what experiences we can or cannot report, and so on, and instead centers entirely on what is happening in the brain. Sometimes one brain region will feed information to another, but the second brain region won’t talk back to the first. Other times, there will be “recurrent processing,” where both brain regions are entering into a proper back-and-forth, two-way dialogue as they exchange information. Lamme believes that it is only when this second kind of neuronal chatter is taking place, with information bouncing between regions, that consciousness occurs. If this back-and-forth talk happens only between specialist areas, such as different visual cortical regions, then there will be some level of consciousness, but it will not be strong enough that we could say, for instance, “Ah, there’s a red rose in front of me.” But if this two-way communication stretches into the prefrontal parietal network, then we have a full, deep consciousness and can report on what we’re seeing.

  Lamme’s notion that recurrent processing is required for conscious levels of information transmission to take place is a very plausible suggestion. But I find his insistence that we are still conscious even when we are quite convinced that we are not to be a deeply unpalatable stance. In order to build a coherent theory of consciousness, it’s fine to be suspicious of the edges of what we report about our experiences, but it is not sensible completely to ignore the very event you are trying to explain. Partly because of Lamme’s rejection of the experiential intricacies that make up how we are aware of the world, his model fails to capture much detail about the nature or purpose of consciousness.

  The model most closely aligned with the existing data, and the view of consciousness I’ve been describing throughout this book, is the global neuronal workspace model proposed by Stanislas Dehaene and Jean-Pierre Changeux. This model is largely the neuronal extension of the global workspace theory put forward by Bernard Baars. In Baars’ theory, consciousness is roughly equated with working memory. It’s a spotlight on a stage, or scribbles on a general-purpose cognitive white board, which lasts only a second or two, but which can contain and manipulate working memory items by drawing them from our vast unconscious reservoirs of knowledge in specialist nonconscious systems.

  In the global neuronal workspace model, the brain also divides along specialist and generalist lines. First there are the specialist, content-dedicated areas at the edge of the collective brain network, which store our memories, crunch data from our senses, and so on. The neurons here have short to medium connections with each other and are all that’s needed when we perform an effortless, automatic, largely unconscious task. The inferotemporal cortex, processing visual objects, is one example of such a region. Then there are the general-purpose regions at the densest center of the network, comprising the prefrontal parietal network and thalamus, which “ignite” in dramatic activity whenever an effortful task is required, so that this entire core can become fully activated simultaneously. This central core includes lots of long-range connections between neurons, enabling this neuronal workspace to draw in specialist knowledge from the content-dedicated regions at the thinner outer edges of the network. If necessary, the prefrontal parietal network and thalamus can also control and modify the activity of these subordinate distant areas, so that complex information processing can occur and difficult tasks can be achieved. Activity in this core set of regions, particularly involving the prefrontal parietal network, is the locus of consciousness.

  Anatomical studies of how the brain is wired put the lateral prefrontal cortex, one of the main regions of the prefrontal parietal network, at the top of the league in terms of how many other regions it is connected to, although the posterior parietal cortex and thalamus are not far behind. So in terms of brain wiring, the prefrontal parietal network, in concert with the thalamus, constitutes an “inner core” of regions that are ideally suited to collect information from the rest of the brain, carry out the most complex tasks we are capable of performing, and generate our sense of experience from this central hub of information processing.

  But because Dehaene’s model is so closely wedded to the empirical neural details that are associated with awareness, he has opened himself up to the charge of not more ambitiously capturing the mechanistic essence of consciousness.

  The third and final theory, Giulio Tononi’s information integration theory, travels in the opposite direction, only discussing mechanisms while refusing to get its hands dirty with too many tawdry details about the brain. Information integration theory is also by far the most abstract and ambitious of the current crop of consciousness models.24 An entirely mathematical theory, it tries to distill consciousness into its informational essence. Whereas the previous two models were rooted in the real cortical networks of the human brain, Tononi’s theory applies to any network of nodes whatsoever, be they a connected series of neurons, computer transistors, or any other information-carrying object one would care to imagine. For Tononi, a network’s capacity for consciousness is directly related to how many different kinds of information it can represent and how well those pieces of information can be combined. The more nodes there are in a network—as long as they are sufficiently connected to each other—the more varied the possibilities for combined forms of information and the greater that network’s capacity for consciousness.25

  This simple yet powerful recipe for awareness can map quite neatly onto concepts such as attention, with its drive to combine information about
an object into a unified whole, or even the previously mentioned global neuronal workspace model, which also cares about a dense central network to carry combined forms of information.

  Under the information integration theory, regions like the cerebellum can never support much consciousness, as they have few connecting internal wires. Specialist brain regions, such as the primary visual cortex, can play only a minimal role in consciousness, again being at the edges of the main network. In contrast, the prefrontal parietal network, being so densely interconnected and also linked to many specialist regions, is just the kind of network shape that can support high levels of consciousness.

  Without the prefrontal parietal network, we are really just processing an independent collection of facts in parallel—for instance, the color of Angelina Jolie’s dress as one datum, the sound of her voice as a completely separate feature, and so on. But when there is some highly interconnected network involved, such as the prefrontal parietal network, with attention combining those facts within it, the richness of that information far exceeds the sum of each individual piece of data, and consciousness ensues.

  And just how much information of this merged form a network can contain is the same as how many different states of activity it can be in. In practical terms, this means how many different kinds of experiences we are ever capable of. I might have inferior senses to a dog, and therefore at best a matched level of information input, but because I can combine the data from my senses in so many different ways, due to the powerful analytical machine of my prefrontal parietal network, the range of experiences I can have far exceeds that of a dog.