Stanislas Dehaene - Consciousness and the Brain: Deciphering How the Brain Codes Our Thoughts

Stanislas Dehaene: Consciousness and the Brain: Deciphering How the Brain Codes Our Thoughts (2014); $27.95 hardcover, $20.93 at Amazon.

Publisher's ad-copy:

A breathtaking look at the new science that can track consciousness deep in the brain

How does our brain generate a conscious thought? And why does so much of our knowledge remain unconscious? Thanks to clever psychological and brain-imaging experiments, scientists are closer to cracking this mystery than ever before.

In this lively book, Stanislas Dehaene describes the pioneering work his lab and the labs of other cognitive neuroscientists worldwide have accomplished in defining, testing, and explaining the brain events behind a conscious state. We can now pin down the neurons that fire when a person reports becoming aware of a piece of information and understand the crucial role unconscious computations play in how we make decisions. The emerging theory enables a test of consciousness in animals, babies, and those with severe brain injuries.

A joyous exploration of the mind and its thrilling complexities, Consciousness and the Brain will excite anyone interested in cutting-edge science and technology and the vast philosophical, personal, and ethical implications of finally quantifying consciousness.

Table of Contents:

Introduction: The Stuff of Thought
Ch. 1: Consciousness Enters the Lab
Ch. 2: Fathoming Unconscious Depths
Ch. 3: What Is Consciousness Good For?
Ch. 4: The Signatures of a Conscious Thought
Ch. 5: Theorizing Consciousness
Ch. 6: The Ultimate Test
Ch. 7: The Future of Consciousness

Global Workspace Theory

Global Workspace Theory (GWT) is a model of "cognitive architecture" proposed by Bernard Baars (1988, 1997, 2002) that seeks to account for a large set of matched pairs of conscious (declarative memory) and unconscious processes (procedural memory).

GWT expands the concept of Working Memory - a "momentarily active, subjectively experienced" event — the "inner domain in which we can rehearse telephone numbers to ourselves or in which we carry on the narrative of our lives. It is usually thought to include inner speech and visual imagery" (in Baars, 1997: In the Theater of Consciousness).

Global Neuronal Workspace Theory

Cognitive neuroscientists Stanislas Dehaene and Jean-Pierre Changeux developed this model beginning in 1986, and it was originally known by the acronym DCM. In essence, this a meta neural network (i.e. a network of neural networks) composed of a very large number of integrate-and-fire neurons.

Dehaene's model became a part of Baars' GWT, developing his broader idea, which operates at the level of neural circuits, down to the neuronal level.

In his new book, Consciousness and the Brain: Deciphering How the Brain Codes Our Thoughts (2014), Dehaene sets out his computational model of neural correlates of consciousness in a book friendly to non-experts and in prose that is engaging without simplifying the ideas.

Here is a brief encapsulation of his model:

Its main postulate is that conscious access is global information availability (see Baars 1989): what we subjectively experience as conscious access is the selection, amplification and global broadcasting, to many distant areas, of a single piece of information selected for its salience or relevance to current goals. (Dehaene, Changeux, and Naccache, 2001)

Dehaene offers a very simple (simplistic?) definition of consciousness: "Consciousness is brain-wide information sharing." Compare that definition with Dan Siegel's definition of mind: "The mind can be defined as an embodied and relational process that regulates the flow of energy and information." The key word here, which I have emphasized, is information.

When information enters the system (whether visual, auditory, tactile, and so on, including interoceptive information) it's all equal in terms of how the brain processes it, whether it becomes conscious or remains unconscious. However, if there is sufficient interference from earlier or newer stimuli, the response diminishes as it moves through the brain's modules, where it still can assert subliminal influence on behavior. If the information remains preconscious, the subject cannot offer a verbal report of the experience.

For Dehaene, that last point is crucial. His research into how and why some information becomes conscious (enters the global neuronal workspace) or remains unconscious depends on first-person, subjective accounts of whether they register a piece of information in conscious or do not.

The experiments developed and used to test the threshold of conscious awareness are ingenious and provide an intriguing and entertaining narrative.

The two serious issue I have with Dehaene are (1) his reliance on computational models of neural processing (the brain is so much more complex than anything that can be modeled on a computer), and (2) that his equivocations on free will is similar to those of Daniel Dennett:

“Our belief in free will expresses the idea that, under the right circumstances, we have the ability to guide our decisions by our higher-level thoughts, beliefs, values, and past experiences, and to exert control over our undesired lower-level impulses. Whenever we make an autonomous decision, we exercise our free will by considering all the available options, pondering them, and choosing the one that we favor. Some degree of chance may enter in a voluntary choice, but this is not an essential feature. Most of the time our willful acts are anything but random: they consist in a careful review of our options, followed by the deliberate selection of the one we favor.” 264

Like Dennett, he suggests that the only free will that matters is the free will we believe that we have.

These are small issues, really, for what is a great introduction to one of the leading theories of how our brains process information. I would highly recommend this book for masters level therapists who more than likely received little (if any) education on the brain and how consciousness becomes an emergent property of brain processes.

Reference:

Dehaene, S, Changeux, JP, and Naccache, L. (2011). The Global Neuronal Workspace Model of Conscious Access: From Neuronal Architectures to Clinical Applications. S. Dehaene and Y. Christen (eds.), Characterizing Consciousness: From Cognition to the Clinic? Research and Perspectives in Neurosciences. Berlin Heidelberg: Springer-Verlag.

The Brain's "Rich Club" Network Is the Foundation for Higher Order Tasks

Efforts to understand how the brain functions range from the molecular to the absurd, and everything in between. Increasingly, neuroscientists are looking at networks composed of various brain "modules" that, together, are responsible for many of the higher order functions researchers had previously sought in a single brain structure.

This article from QUANTA Magazine, via WIRED, offers an excellent overview of what researchers know about the "rich club" network, so named because it functions like a group of well-connected individuals. This is really interesting stuff in understand how the brain is networked in a complicated system of connections.

The Unlikely Network at the Core of Your Brain’s Internet

By Emily Singer, QUANTA Magazine
10.29.13


The rich club network, highlighted here in red, is more highly connected than the rest of the brain. (Image: Martijn van den Heuvel, Journal of Neuroscience.)
To successfully send a text message, your brain has to compose the wording and coordinate the deft tapping of your thumbs while checking for typos. Solving a simple algebra problem involves the visual cortex — to process the symbols — and the parts of the brain used for computation and short-term memory. And driving, best performed without texting, requires a combination of visual, tactile and spatial data. How do different areas of the brain communicate and coordinate their efforts to complete these tasks? Neuroscientists have long struggled to understand the brain’s ability to synthesize a dog’s breakfast of sensory inputs and cognitive processes.

But a growing body of evidence suggests that a network of highly interlinked brain regions exists that could prove essential in facilitating higher-order tasks. This collection of highly connected hubs has been dubbed the “rich club” network because it resembles groups of well-connected individuals, such as Ivy League alumni organizations, that help others meet and exchange information.

Within the brain, researchers “think that rich clubs have a key role to play in making global communication efficient and are also important for supporting integration of information,” said Olaf Sporns, a computational neuroscientist at Indiana University in Bloomington. Sporns first described the rich club in the brain, along with collaborator Martijn van den Heuvel, a neuroscientist at the Brain Center Rudolf Magnus in the Netherlands.

Sporns and others hope that studying the rich club will help solve one of the biggest challenges in neuroscience: linking the brain’s structure with its function. While each can be measured using different technologies, attempts to reconcile structural and functional information have proven elusive.

“It’s absolutely relevant to know as much as we can about the topology of the underlying structure to understand the function,” said Gustavo Deco, a neuroscientist at Universitat Pompeu Fabra in Barcelona, Spain.

Olaf Sporns of Indiana University. (Image: Indiana University)

One reason scientists are drawn to the neural rich club network is that it seems to violate one of the brain’s organizing principles — to minimize the cost of wiring. The brain appears to favor short connections that transfer messages quickly and take up little of the brain’s valuable volume and wiring material. The wiring diagram of the microscopic worm C. elegans follows this rule quite closely. And in mammalian brains, short connections are much more common than long ones. But in both worms and humans, the rich club network is a rare and expensive exception, suggesting an important role in neural function.

“This is a high-cost component of the brain that might be valuable in supporting high-level and more cognitively advanced forms of information processing,” said Ed Bullmore, a neuroscientist at the University of Cambridge in England who also studies the rich club. “It’s unlikely the brain would implement such a highly connected network without good reason.”

Form and Function

In the brain, the rich club is similar to the highway system, which accounts for only a small percentage of surface roads but a much larger percentage of all travel. The neural rich club makes up “a small piece of the overall wiring length of the brain, but it carries a massive amount of information flow,” van den Heuvel said.

The connections between rich club hubs are anatomically long but topologically short, meaning one node is connected to any other by only one or two intermediate nodes. Van den Heuvel said this could allow faster and less noisy communication among neurons.

Sporns and van den Heuvel originally defined the rich club by analyzing anatomical data gathered from images of the brain’s white matter. On a computer, they divided the brain into different regions and represented it as a network, with each region as a node and the neural connections as links between them. Using a mathematical approach called graph theory to analyze the properties of the network, they identified 12 highly interconnected hubs, which they dubbed the rich club. This central network was more interconnected than the rest of the brain and linked to almost every other brain region. Computer simulations revealed that damage to the rich club disrupted the overall network more than damage to other areas did.

Martijn van den Heuvel of the Brain Center Rudolf Magnus. 

(Image: Brain Center Rudolf Magnus, University Utrecht)

One big question that arose from the study is how the rich club network, based on anatomical data, corresponds to brain function. The scientists were particularly interested to find out whether the rich club corresponded to so-called functional networks, which have been mapped out using data from functional brain imaging. Scientists look for correlated patterns of brain activity when an individual is at rest or doing a specific cognitive task. Despite myriad studies on these functional networks, researchers have not yet defined the underlying physical structures.

Sporns and van den Heuvel set out to examine the link between structure and function with one of the biggest studies to date comparing both structural and functional imaging data from the same individuals. The results, published in the Journal of Neuroscience in September, show that the rich club network in the human brain overlaps many of the functional networks that have been mapped to date.

Van den Heuvel said the findings suggest there isn’t a simple linear relationship between structure and function. Instead, “the anatomy simply defines the boundaries in which functional configurations of the brain can take place,” he said. For example, when someone rapidly changes tasks, say from reading email to answering the phone, the brain rapidly switches functional states, which could be constrained by the rich club network.

The findings are important because they provide evidence that may “tell us places to look in the brain that are more likely to be sites of real interest, places where when something happens, it could have a more devastating consequence to brain function,” said Michael Hawrylycz, director of the Modeling, Analysis, and Theory Group at the Allen Institute for Brain Science in Seattle. Losing a major airport hub like O’Hare in Chicago, for example, would have a greater overall impact on the air traffic system than the closing of Hector International Airport in Fargo, N.D.

A second study, published earlier this year by Bullmore and collaborators, took a different tack at linking the rich club to functional networks. Researchers compiled brain activity data from more than 1,200 people performing a variety of tasks. Bullmore said the rich club nodes were more likely to be involved in higher-order cognitive functions such as action and cognition. “It adds another piece of data in support of the general argument that the rich club is expensive because it’s functionally valuable,” he said.

Interactive: This three-dimensional model of the network highlights its interconnectedness. Use your mouse to rotate the model and your scroll wheel to zoom. (Courtesy of Marcel de Reus and Martijn van den Heuvel, Brain Center Rudolf Magnus)

Cats and Worms

In April, Bullmore and collaborators published an analysis of a rich club in the microscopic roundworm C. elegans, the popular model organism whose entire set of neuronal connections, called a connectome, have been mapped out. “It’s as close to a canonical gold standard wiring diagram that we have in neuroscience so far,” Bullmore said.

After identifying 11 neurons that belonged to the rich club in C. elegans, the researchers discovered remarkable similarities in the rich club of the human brain, Bullmore said. Each neuron in C. elegans has a catalogued time of birth and function, enabling researchers to explore both how the rich club develops and what it does. Like other network studies of the brain, Bullmore’s team used graph theory to map out the rich club. “It turned out that the neurons we identified by this purely mathematical analysis are known as command interneurons, which play a role in coordinated movement of the nervous system,” Bullmore said. The worms can’t move until the rich club is fully developed.

Though not as complete as the C. elegans diagram, a wiring map of the much more complex cat cortex was created in 1995. One of the benefits of studying the cat brain, said van den Heuvel, is that scientists have extensive data from physiological experiments on the function of specific brain regions. In a new analysis of the cat’s connectome data, van den Heuvel and his graduate student Marcel de Reus found that the rich club network helps link the major functional networks — visual, auditory, somatomotor and frontolimbic — that had been defined in the cat brain. Indeed, 90 percent of the messages passed among these regions appear to go through the rich club. Their results were published in the Journal of Neuroscience in August.

In the cat, scientists also know the direction in which neurons in one node send information to another. “That allows you to map the theoretical information flow between regions in the cat cortex,” van den Heuvel said. The rich club has more reciprocal connections, meaning nodes that send information in both directions, than other networks. “Maybe that creates an infrastructure with a high level of flexibility rather than a network of segregated modules,” van den Heuvel said. More research is needed to explore whether this structure is indeed more flexible.

The fact that scientists have identified rich club networks in people, cats, worms and other species suggests that it’s a basic feature of the brain. “Maybe the rich club is a general feature of how neural systems grow,” van den Heuvel said.

The variety of animals with rich club networks also suggests that the network isn’t an artifact of human brain imaging — a concern raised by critics. Although scientists can use brain imaging data to create detailed images of the human brain, these pictures are indirect — the structural maps used to analyze the rich club use the diffusion of water molecules as a proxy for neural wiring, for example — and therefore subject to experimental biases. (Diffusion maps may be better at detecting some kinds of wiring than others.) “We don’t have a perfect measure of brain structure or function in humans, so there is always a question in my mind: To what extent are the network properties we see in human imaging studies really there?” Bullmore said. The use of more direct anatomical tracing techniques in studies of other animals, however, has been consistent with the findings in humans.

Rich Club in Disarray

Rich clubs have been discovered in many interconnected systems, including the Internet, social networks and flight patterns. One recent study found that a group of college students who started interacting intensely during the first week of class and continued to exchange information throughout the semester — mostly with each other — performed better than others in the class. These students had, without realizing it, formed a rich club and reaped its benefits.

But what happens when the brain’s rich club network is defective? Although the answer is still unclear, preliminary evidence suggests a connection between abnormalities in the rich club and neurodevelopmental disorders like schizophrenia, for which reproducible flaws in the brain have been difficult to pinpoint using standard brain imaging. “Despite the fact that it’s so devastating, what’s actually different in the brain of a person with schizophrenia is a much more subtle question,” Sporns said.

Looking for defects in the rich club and examining other network measures, such as the number and structure of connections, is showing early promise in illuminating the disorder. According to research published in JAMA Psychiatry in August, Sporns and van den Heuvel found that people with schizophrenia tend to have thinner and less well-developed connections between hub regions than people without the disorder, but their connections linking peripheral nodes were less defective or unimpaired. In a separate study, by comparison, people with the motor disorder amyotrophic lateral sclerosis showed differences in connectivity of the motor network. The researchers are now looking at the rich club network in people with Alzheimer’s disease.

“Since the rich club is high cost and high value, it’s likely to be particularly vulnerable to disease,” said Bullmore, whose team is now studying the growth of the rich club in nervous systems and the pathological implications when it’s disturbed.

However, more work is needed to understand the link between damage to the rich club and disease. For example, it’s not clear whether the thinner connections cause the disorder or are a product of having schizophrenia. Eventually, scientists should be able to disrupt the rich club network in model organisms, such as rodents, flies or worms, and measure whether the animals suffer schizophrenia-like impairments.

Beyond studying the rich club’s role in disease, Sporns and collaborators hope to better define how the network varies among people. Part of their effort will focus on data from the Human Connectome Project, an effort to map neural connections and activity from more than 1,000 people and make that data publicly available. Sporns said the researchers are trying to answer questions such as whether there are heritable differences in the rich club structure and whether these structures change over a person’s lifetime.

Ultimately, the rich club is only one part of a complex network, so scientists also hope to understand how the rich club fits into the broader brain network, which will be essential to understanding its role in integration. “What we really need to understand,” said Bullmore, “is the structure of brain networks as a whole.”

Sidebar: Concerns About Graph Theory

Despite the growing popularity of graph theory and other approaches for analyzing networks in the brain, it has its detractors. One of the major criticisms is its simplicity. Graph theory methods “are merely descriptive statistics that ignore our prior biological knowledge, which is substantial,” Partha Mitra, a neuroscientist at Cold Spring Harbor Laboratory on Long Island, said in an email. The analysis also depends highly on the quality of the data. Incomplete or faulty data can lead to faulty network structures.

“There’s obviously a lot of detail excluded from the description of the brain as 300 to 400 nodes, such as the differences in neurons and synapses, which are lost in that translation,” Bullmore said. “I try to make that clear early on when I talk about graphs that we are trying to strip the brain down to simple description of nodes and edges.”

Bullmore pointed out that graph theory has been useful in analyzing other complex systems, such as social networks, where connections represent friendships. Identifying highly connected people has helped illuminate how information or infection travels through a community. “Which agents in social network should be targeted to prevent spread of infection or to spread a message?” Bullmore said. “That’s a concrete example where thinking about a complex system in simple terms of a graph is useful.”

Original story reprinted with permission from Quanta Magazine, an editorially independent division of SimonsFoundation.org whose mission is to enhance public understanding of science by covering research developments and trends in mathematics and the physical and life sciences.

Original story reprinted with permission from Quanta Magazine, an editorially independent division of SimonsFoundation.org whose mission is to enhance public understanding of science by covering research developments and trends in mathematics and the physical and life sciences.

A bow tie representation of the network of connections between cortical areas in the brain. (Credit: University of Notre Dame)

The image above represents the authors' conception of how the brain organizes itself for the most efficient processing of information. Here are a couple of passages from the research summary, explaining the architecture and how it functions (with comparisons to the internet and other complex information processing systems):

Using brain-wide and consistent tracer data, the researchers describe the cortex as a network of connections with a "bow tie" structure characterized by a high-efficiency, dense core connecting with "wings" of feed-forward and feedback pathways to the rest of the cortex (periphery). The local circuits, reaching to within 2.5 millimeters and taking up more than 70 percent of all the connections in the macaque cortex, are integrated across areas with different functional modalities (somatosensory, motor, cognitive) with medium- to long-range projections.
___

This bow tie arrangement is a typical feature of self-organizing information processing systems. The paper notes that the cortex has some analogies with information-processing networks such as the World Wide Web, as well as metabolism, the immune system and cell signaling. The core-periphery bow tie structure, they say, is "an evolutionarily favored structure for a wide variety of complex networks" because "these systems are not in thermodynamic equilibrium and are required to maintain energy and matter flow through the system." The brain, however, also shows important differences from such systems. For example, destination addresses are encoded in information packets sent along the Internet, apparently unlike in the brain, and location and timing of activity are critical factors of information processing in the brain, unlike in the Internet.

This is another piece of info in the growing awareness of how the brain works as a collection of networks.

Full Citation:
N. T. Markov, M. Ercsey-Ravasz, D. C. Van Essen, K. Knoblauch, Z. Toroczkai, H. Kennedy. (2013, Nov 1). Cortical High-Density Counterstream Architectures. Science, 342 (6158): 1238406 DOI: 10.1126/science.1238406

New Insights Into Brain Neuronal Networks

Nov. 4, 2013 — A paper published in a special edition of the journal Science proposes a novel understanding of brain architecture using a network representation of connections within the primate cortex. Zoltán Toroczkai, professor of physics at the University of Notre Dame and co-director of the Interdisciplinary Center for Network Science and Applications, is a co-author of the paper "Cortical High-Density Counterstream Architectures."

Using brain-wide and consistent tracer data, the researchers describe the cortex as a network of connections with a "bow tie" structure characterized by a high-efficiency, dense core connecting with "wings" of feed-forward and feedback pathways to the rest of the cortex (periphery). The local circuits, reaching to within 2.5 millimeters and taking up more than 70 percent of all the connections in the macaque cortex, are integrated across areas with different functional modalities (somatosensory, motor, cognitive) with medium- to long-range projections.

The authors also report on a simple network model that incorporates the physical principle of entropic cost to long wiring and the spatial positioning of the functional areas in the cortex. They show that this model reproduces the properties of the connectivity data in the experiments, including the structure of the bow tie. The wings of the bow tie emerge from the counterstream organization of the feed-forward and feedback nature of the pathways. They also demonstrate that, contrary to previous beliefs, such high-density cortical graphs can achieve simultaneously strong connectivity (almost direct between any two areas), communication efficiency, and economy of connections (shown via optimizing total wire cost) via weight-distance correlations that are also consequences of this simple network model.

This bow tie arrangement is a typical feature of self-organizing information processing systems. The paper notes that the cortex has some analogies with information-processing networks such as the World Wide Web, as well as metabolism, the immune system and cell signaling. The core-periphery bow tie structure, they say, is "an evolutionarily favored structure for a wide variety of complex networks" because "these systems are not in thermodynamic equilibrium and are required to maintain energy and matter flow through the system." The brain, however, also shows important differences from such systems. For example, destination addresses are encoded in information packets sent along the Internet, apparently unlike in the brain, and location and timing of activity are critical factors of information processing in the brain, unlike in the Internet.

"Biological data is extremely complex and diverse," Toroczkai said. "However, as a physicist, I am interested in what is common or invariant in the data, because it may reveal a fundamental organizational principle behind a complex system. A minimal theory that incorporates such principle should reproduce the observations, if not in great detail, but in extent. I believe that with additional consistent data, as those obtained by the Kennedy team, the fundamental principles of massive information processing in brain neuronal networks are within reach."

Here is the full abstract from Science (the full article is behind a paywall, but I will have a paper copy in the next few days, thanks to a PT client with a subscription):

Background

The cerebral cortex is divisible into many individual areas, each exhibiting distinct connectivity profiles, architecture, and physiological characteristics. Interactions among cortical areas underlie higher sensory, motor, and cognitive functions. Graph theory provides an important framework for understanding network properties of the interareal weighted and directed connectivity matrix reported in recent studies. 

Density and topology of the cortical graph. (Left) The 66% density of the cortical matrix (black triangle) is considerably greater than in previous reports (colored points) and is inconsistent with a small-world network. (Right) A bow-tie representation of the high-density cortical matrix. The high-efficiency cortical core has defined relations with the cortical periphery in the two fans.

Advances

We derive an exponential distance rule that predicts many binary and weighted features of the cortical network, including efficiency of information transfer, the high specificity of long-distance compared to short-distance connections, wire length minimization, and the existence of a highly interconnected cortical core. We propose a bow-tie representation of the cortex, which combines these features with hierarchical processing.

Outlook

The exponential distance rule has important implications for understanding scaling properties of the cortex and developing future large-scale dynamic models of the cortex.