How Matter becomes Imagination
Basic 2000




The Darwinian Perspective

Our brief  review of the vast field of neurological and neurophysiological evidence in this and the previous chapter leads to the following conclusions.

First, conscious processes are typically associated with distributed changes in activity in the thalamocortical system. 
Second, distributed changes in neural activity associated with conscious experience must be integrated through reentrant interactions that are both rapid and effective. 
Finally, these interactions are associated with conscious reports if they are highly differentiated but not if they are uniform or homogenous. 

These empirical observations suggest that underlying consciousness are distributed neural processes that, through reentrant interactions, are at once highly integrated but continually changing and thus are highly differentiated. 

This conclusion becomes particularly relevant when one realizes that integration and differentiation are also general properties of conscious experience, irrespective of its specific content.

To understand how these phenomenological properties relate to the actual neural mechanisms that are responsible for consciousness cannot be just a matter of accumulating additional facts.  It requires a robust theory - one that provides insights into the biological origins of pattern formation, perceptual categorization, memory, concepts, and values. 
What kind of brain theory is compatible with the distributed, integrated, but continnously changing patterns of neural activity that can give rise to the unitary yet immensely differentiated phenomenology of conscious experience? 

We turn to a brief description of one such theory, a theory that we believe provides the necessary basis for understanding the key principles underlying global brain function. 

Formulating such a theory required us to confront several challenging questions. What kind of svstem is the functioning brain? How can its properties lead to consciousness? How can we account for its function in the face of its enormous variability. In our effort to answer these questions, we take the position that the brain is a selective, or Darwinian, system, one whose rich functiorung actually requires variability.

pg 77

In his  theory of natural selection, Charles Darwin provided the chief foundation of modern biology. After his return from the voyage of the Beagle, he made continuing efforts to understand how the functions performed by the brain arose during evolution. His notebooks reveal his struggle to explain how perception, memory, and language could have arisen by what he called descent. 

We now have a rich evolutionary theory graced by the Darwinian perspective, but the problem of understanding mental processes is still with us. It remains for neuroscience to complete Darwin's program.

In this part, we show how Darwinian principles embedded in a theory of brain function provide insights into the processes of perception, memory, and the assignment of value, all of which are critical to an understanding of consciousness.

Once the reader grasps the nature of such processes, the stage will be set to consider the actual neural mechanisms by which consciousness arises during evolution and development. Our efforts here are focussed on consciousness, the ability to construct an integrated mental scene in the present that does not require language or a true sense of self. 

We believe that this integrated mental scene depends not only on the perceptual categorization of incoming sensory stimuli—the present—but, most important, on their interaction with categorical memories—the past. In other words, this integrated mental scene is a "remembered present." 

The main means by which the scene is constructed is through reentrant interactions among groups of neurons distributed in the thalamocortical sys-tem. As we show, these are just the kinds of interactions responsible for the integration and differentiation that we discussed in chapter 6.

The theory of neuronal group selection, or Neural Darwinism

In considering the origin of species, Charles Darwin made a great contribution that centered on population thinking: the idea that variation or diversity among individuals in a population provides a basis for competition during natural selec-tion. Natural selection is reflected in the differential reproduction of fitter indi-viduals in a species. 
In principle, selective events require the continual generation of diversity in repertoires of individual variants, the polling by environwental signals of these diverse repertoires, and the differential amplifcation or reproduc-tion of those repertoire elements or individuals that match such signals better than their competition. 

Could it be that the brain follows such principles? We believe it does, and in this chapter we briefly review some aspects of the theory of neuronal group selection, or Neural Darwinism. This theory embraces these selective principles and applies them to the functioning brain. Its main tenets are 

(1) the formation during brain development of a primary repertoire of highly variant neuronal groups tbat contribute to neuroanatomy (developmental selection), 
(2) the formation during experience of a secondary repertoire of facilitated neural circuits as a result of changes in the strength of connections or synapses (experiential selection), and 
(3) a process of reentrant signaling along reciprocal connections between and among distributed neuronal groups to assure the spa-tiotemporal correlation of selected neural events

Together, the three tenets of this global brain theory provide a powerfal means for understanding the key neural interactions that contribute to consciousness.


In his later vears, Darwin disagreed strongly with Alfred Russel Wallace, the codiscoverer of natural selection, who, as a spiritualist, insisted that the brain and mind of man could not have arisen by natural selection. 

Wallace reasoned that savages had brains roughly the size of the brains of civilized Englishmen yet lacked mathematics and had no obvious need for abstract thought, so it was difficult for him to see how natural selection would have led to similar brain sizes in both cases. He was too thoroughgoing a natural selectionist and failed to recognize that during natural selection, there is correlative variation: A primary trait can be selected for and bring along changes that are used later for other selective events. For example, the selec-tion of enlarged brain structures for perception could be accompanied by enlargements of neighboring brain regions. At some later evolutionary epoch, these regions may become selectively advantageous for some other function, such as memory. In a letter to Wallace in the spring of 1869, Darwin said, "I hope that you have not murdered too completely your own and my child" - meaning, of course, natural selection.

Aside from the faultiness of Wallace's reasoning, the accumulation of evidence since that time strongly supports Darwin's conclusions: Whatever the specialness of the human brain, there is no need to invoke spiritual forces to account for its functions. 

Darwinian principles of variation in populations and natural selection are sufficient, and the elements invoked by spiritualism are not required for our being conscious. Being human in mind and brain appears clearly to be the result of an evolutionary process. 

The anthropolog-ical evidence emerging for the evolutionary origin of consciousness in humans further substantiates the notion that Darwin's is the most ideologi-cally significant of all grand scientific theories.

Darwinian principles turn out to be important even for a basic understanding of brain functions, especially given the enormous variation in the structure and function of individual vertebrate brains. As we have discussed, no two brains are alike, and each individual's brain is continually changing. Variations extend over all levels of brain organization, from biochemistry to gross morphology, and the strengths of myriad individual synapses are con-stantly altered by experience. The extent of this enormous variability argues strongly against the notion that the brain is organized like a computer with fixed codes and registers. Moreover, the environment or world from which signals are delivered to the brain is not organized to give an unambiguous message like a piece of computer tape. There is no judge in nature giving out specific pronouncements on the brain's potential or actual patterns and there is no homunculus inside the head deciding which pattern should be chosen and interpreted. 

These facts are incompatible with the notion that the brain operates according to an unambiguous set of algorithms or instructions, like a computer. Instructionism, the idea that the environment can reliably provide the kind of information required by a computer, fails as a principle of brain operation. 

Yet in a given species, individual animals show certain consistent behaviors within the broad spread of individual responses.

How does the brain give rise to such responses? What principles govern its global operations? To answer these questions, we need a global brain theory, one that sets out the principles governing the operation of vast and diverse neural networks. Of course, the principles of such a theory must be consistent with our observations of the neural processes necessary for con-sciousness.


We have repeatedly stressed that one of the most striking features of each brain is its individuality and variability. This variability occurs at all levels of brain organization, and it is so great that it cannot be dismissed as mere noise or ignored while pursuing a mechanical theory of brain action. As we shall see, this variability provides a key basis for the differentiation and diversity of conscious states. The existence of this enormous diversity and individuality, seen in the multilayered structures and dynamics of each brain, poses a major challenge to any theory that is proposed to account for global brain function. 

We believe that this challenge can be met by turning to population thinking, which Darwin invented. Population thinking centers on the idea that variations among individuals of a species provide the basis for natural selection in the struggle for existence that eventually leads to the ori-gin of other species.

Although Darwin did not have a correct picture of genetics, he understood that different individuals inherit different traits. Certain individuals would have greater fitness as the environment changed or new environwents were occupied. These individuals would, over several generations, leave more progeny than would be able to utilize resources in the competition for survival and reproduction. Natural selection would thus effectively lead to differential reproduction of those individuals who, on the average, had higher fitness. This population principle has deep ramifications: It not only provides the basis for the origin of species, but it governs processes of somatic selection occurring in individual lifetimes. 

When we say somatic selection, we mean what occurs in a single body in time frames ranging from fractions of seconds to years and, obviously, ending vith an animal's death. Thus, selection and variation can also occur in the cellular systems of ani-mals.

A well-analyzed example of somatic selection is provided by the immune system. In animals with backbones, there is an extraordinary cellular system capable of distinguishing foreign molecules or bacteria, viruses, and even another person's skin from the molecules of an individual body (or soma). The recognition is carried out by a set of remarkable proteins, called antibodies, that are made by circulatory blood cells. Antibodies have special sites that match or bind portions of other molecules, almost the way a cookie cutter matches a cookie of a given shape. What is remarkable is that practically any foreign molecule or antigen injected into the body vvill elicit the production of a complementary antibody that is essential for subsequent immune defense.

The original theory to account for the complementary fit between the antigen and the antibody was an "instructive" one: The antibody folded around the antigen's shape and kept the appropriately shaped fold. This theory turned out to be incorrect. 

Instead, the immune system works by somatic selection. The basis for molecular recognition of an enormous number of dif-ferent foreign molecules is somatic variation in the antibody genes of each individual that leads to the production of a vast repertoire of antibodies, each with a different binding site. Exposure of the enormous repertoire of different antibodies to a foreign molecule is followed bv the selection and growth of the cells bearing just those antibodies that fit the foreign chemical structure of a given antigen sufficiently well, even a structure that never occurred before in the history of the Earth. Although the mechanisms and timing of selective events obviously differ in evolution and immunity, the principles are the same—the Darwinian processes of variation and selection.

Over two decades ago, one of us began to think about how the mind could arise in evolution and development. It seemed that the mind must have arisen as result of two processes of selection: natural selection and somatic selection. 

The first process is hardly doubted except perhaps by some philosophers and theologians. Thinking about the second led to the proposal of a theory based on selective principles and concerned with the evolution, development, structure, and function of the brain. 

It is worth reviewing here not only because one of its main tenets (reentry) is central to the origin of consciousness, but because its way of dealing with variability in the brain is essential to understanding the complexity of conscious processes.

This theory of neuronal group selection (TNGS), or Neural Darwinism, has three main tenets that are illustrated in figure 7.2:
(1 ) Developmental selection leads to a highly diverse set of circuits, one of which is shown. 
2) Experiential selection leads to changes in the connection strengths of synapses favoring some pathways over others 1see the black lines). 
3) Reentrant mapping. Brain maps are coordinated in space and time through ongoing signaling across reciprocal connections. The black dots in the maps indicate strengthened synapses.

1. Developmental selection. During the early development of individuals in a species, formation of the initial anatomy of the brain is certainly constrained by genes and inheritance. But from early embryonic stages onward, the con-nectivity at the level of synapses is established, to a large extent, by somatic selection during each individual's ongoing development. For example, dur-ing development, neurons extend myriads of branching processes in many directions. This branching generates extensive variability in the connection patterns of that individual and creates an immense and diverse repertoire of neural circuits. Then, neurons strengthen and weaken their connections according to their individual patterns of electrical activity: Neurons that fire together, wire together. As a result, neurons in a group are more closely con-nected to each other than to neurons in other groups.

2. Experiential selection. Overlapping this early period and extending throughout life, a process of synaptic selection occurs within the repertoires of neuronal groups as a result of behavioral experience. It is known, for example, that maps of the brain corresponding to tactile inputs from the fin-gers can change their boundaries, depending on how much different fingers are used. These changes occur because certain synapses within and between groups of locally coupled neurons are strengthened and others are weakened without changes in the anatomv. This selectional process is constrained by brain signals that arise as a result of the activity of diffusely projecting value systems, a constraint that is continually modified by successful output.

3. Reentry. The correlation of selective events across the various maps of the brain occurs as a result of the dynamic process of reentry. Reentry allows an animal with a variable and uniquely individual nervous system to partition an unlabeled world into objects and events in the absence of a homunculus or computer program. As we have already discussed, reentry leads to the synchronization of the activity of neuronal groups in different brain maps, binding them into circuits capable of temporally coherent output. Reentry is thus the central mechanism by which the spatiotemporal coordination of diverse sensory and motor events takes place.

The first two tenets, developmental and experiential selection, provide the bases for the great variability and differentiation of distributed neural states that accompany consciousness. 

The third tenet, reentry, allows for the integration of those states. It is particularlv important to understand the central role played by reentry in our efforts to build a consciousness model, and it therefore requires some further elaboration. 
A key anatomical precon-dition for reentry is the remarkable massively parallel reciprocal connectiv-ity of brain areas. Although reciprocity between two different maps across multiple parallel fibers is common (think, for example, of the corpus callo-sum—the huge bundle of reciprocal fibers linking the two cortical hemi-spheres; see figure 6.1), many more complicated arrangements exist. 

The number of possible geometric and topological patterns possible in such a system is enormous. If we consider the combinatorial possibilities for reen-trant selection across such patterns, even after allowing a number of neuro-anatomical constraints to operate, we begin to glimpse the remarkable power of neuroanatomy in a selectional system. 

A jungle or food web, like the brain, has many levels and routes for the passage of signals but has noth-ing corresponding to reentrant neuroanatomv. Indeed, if asked, What char-acteristic uniquely differentiates higher brains from all other known objects or systems, we would sav "reentrant organization." Note that while complex wide-area computer networks are beginning to share some properties with reentrant systems, such networks rely fundamentally on codes and, unlike brain networks, they are instructional, not selectional.

It is important to emphasize that reentry is not feedback. Feedback occurs along a single fixed loop made of reciprocal connections using previous instructionally derived information for control and correction, such as an error signal. In contrast, reentry occurs in selectional systems across multiple paral-lel paths where information is not prespecified. Like feedback, however, reen-try can be local (within a map) or global (among maps and whole regions).

Reentry carries out several major functions.  For example, it can account for our ability to discern a shape in a display of moving dots, based on inter-actions between brain areas for visual movement and shape.5 Thus, reentry can lead to the construction of new response properties. It can also mediate the synthesis of brain functions by connecting one submodality, such as color, to another, such as motion. It can also resolve conflicts among com-peting neural signals. Reentry also ensures that changes in the effficacy of synapses in one area are affected by the activation patterns of distant areas, thereby making local synaptic changes context-dependent. Finally, by assuring the spatiotemporal correlation of neuronal firing, reentry is the main mechanism of neural integration.

Since the initial formulation of the TN'GS, considerable evidence to support the theory has accumulated. Moreover, certain aspects of the theory have been greatly expanded. 

One of these aspects relates to the issue of degeneracy—the ability of structurally different variants of brain elements to produce the same function. 

Another important aspect of the theory is related to the notion of value, which we touched on briefly in chapter 4 in our discussion of diffusely projecting value systems. We consider each of these aspects in turn.


All selectional systems share a remarkable property that is as unique as it is essential to their functioning: In such systems, there are typically many dif-ferent ways, not necessarily strurturally identical, by which a particular output occurs. We call this property degeneracy.

Degeneracy is seen in quantum mechanics in certain solutions of the Schrödinger equation and in the genetic code, where, because of the degenerate third position in triplet code words, many different DNA sequences can specify the same protein.

Put briefly, degeneracy is reflected in the capacity of structurally different components to yield similar outputs or results. In a selectional nervous system, with its enormous repertoire of variant neural circuits even within one brain area, degeneracy is inevitable. Without it, a selectional system, no matter how rich its diversity would rapidly fail - in a species, almost all mutations would be lethal; in an immune system, too few antibody variants would work; and in the brain, if only one network path was available, signal trafffic would fail. Degeneracy can operate at one level of organization or across many. It is seen in gene networks, in the immune system, in the brain, and in evolution itself. For example, combinations of different genes can lead to the same structure, antibodies with different structures can recognize the same foreign molecule equally well, and different living forms can evolve to be equally well adapted to a specific environment.

There are countless examples of degeneracy in the brain. The complex meshwork of connections in the thalamocortical system assures that a large number of different neuronal groups can similarly affect, in one way or another, the output of a given subset of neurons. For example, a large num-ber of different brain circuits can lead to the same motor output or action. Localized brain lesions often reveal alternative pathways that are capable of generating similar behaviors. Therefore, a manifest consequence of degeneracy in the nervous system is that certain neurological lesions may often appear to have little effect, at least within a familiar environment. 
Degeneracy also appears at the cellular level. Neural signaling mechanisms utilize a great variety of transmitters, receptors, enzymes, and so-called sec-ond messengers. The same changes in gene expression can be brought about by different combinations of these biochemical elements. Degeneracy is not just a useful feature of selectional systems; it is also an unavoidable consequence of selectional mechanisms. 

Evolutionary selective pressure is typically applied to individuals at the end of a long series of com-plex events. These events involve many interacting elements at multiple temporal and spatial scales, and it is unlikely that well-defined functions can be neatly assigned to independent subsets of elements or processes in bio-logical networks. For example, if selection occurs for our ability to walk in a particular way, connections within and among many different brain structures and to the muscoloskeletal apparatus are all likely to be modified over time. While locomotion will be affected, many other functions, including our ability to stand or jump, will also be influenced as a result of the degen-eracy of neural circuits. 

The ability of natural selection to give rise to a large number of nonidentical structures yielding similar functions increases both the robustness of biological networks and their adaptability to unforeseen environments.


As powerful as it is in providing alternative pathways for a given function, degeneracy cannot provide constraints for a selectional system; indeed, it is a relaxation of constraints. Given that this is so, how can a selectional system achieve its goals without specific instructions? It turns out that the necessarv constraints or values are provided bv a series of diverse phenotypic struc-tures and neural circuits that have been selected during evolutionary time. 

We define values as phenotypic aspects of an organism that were selected during evolution and constrain somatic selective events, such as the synaptic changes that occur during brain development and experience. For example, the mere fact of having a hand with a certain shape and a certain propensity to grasp in one way and not in another enormously enhances the selection of synapses and of neural patterns of activity that lead to appropriate actions. The same actions would be almost impossible to synthesize or program from scratch, as experts in robotics know all too well. Another example is seen in the many reflexes with which newborn babies are endowed. These are not the only contributors to value, however. The diverse morphological characteristics (such as those of sensory organs and the motor apparatus) that link the parts and organs of the body to various brain functions are further examples. Hormonal loops can be prime contributors, but so can the way limbs are jointed in different vertebrate species. Values thus provide a basis for the development and refinement of brain-based categorization and action within a species.

It is important to stress that value is not identical to category. Value is only a precondition for arriving at a perceptual or behavioral response. Such a categorical response depends on the actual occurrence of selection. Perceptual categorization usually emerges as a result of selection during actual behavior in the real world. 

In general, although value shapes catego-rization in accord with evolution, it cannot convey or preserve the details of a real-world event. For example, value may be necessary for orienting the eyes of a baby toward a light source, but may not be sufficient for the recognition of different objects.

There are two limits to the concept of value as so far described. First, even a conglomerate of morphologically based values handed down in the phenotype (such as opposed thumbs and different kinds of joints) may not be specific enough to guide neural behavior (perceptual categorization, for example). The second limit is that evolutionarily defined, fixed value para-meters may, by themselves, be too rigid to guide sufficiently rich behavior when an animal is confronted vith unforeseen demands of the environment.

The first limit appears to have been transcended bv the evolution of the special brain centers. 

In higher vertebrates, a series of diffusely projecting neural value systems appear to have evolved that are capable of continually signaling to neurons and synapses all over the brain. 

Such signals carry information about the ongoing behavioral state of the organism (sleep, waking,exploration,grooming, and the like), as well as the sudden occurence of events that are salient for the entire organism (including, for example, novel stimuli, painful stimuli, and rewards).

These systems, whose importance vastly outweighs the proportion of brain space they occupy, include the noradrenergic, serotoninergic, cholinergic, dopaminergic, and histaminer-gic cell nuclei (see figure 4.4C). 
These are all small, compact cell groups, each of which sends diffuse projections to a substantial portion of the brain. The locus coeruleus, for example, consists of only a few thousand neurons in the brainstem. These neurons give rise to a vast meshwork of axons that blanket the cortex, hippocampus, basal ganglia, cerebellum, and spinal cord, potentially influencing transmission at billions of synapses over all levels of the central nervous system (see figure 7.3).

FIGURE 7.3 DIAGRAM OF A VALUE SYSTEM. The noradrenergic system originating in the locus coeruleus projects diffusely to the entire brain and releases the neuromodulator noradrenaline.

Neurons within some of the nuclei of value systems fire in a continnous or tonic manner when an animal is awake and stop firing when the animal falls asleep. Moreover, neurons belonging to value systems often produce a sudden burst of firing whenever something important or salient occurs to the animal. For example, neurons in the locus coeruleus fire whenever an animal enters a novel environment or something unexpected happens. 

When they fire, they release a neuromodulator—in this case, noradrena-line—over most, if not all, brain regions. Noradrenaline and neuromodula-tors that are released by different value systems can thus modify the activity of a large number of target neurons. They can also alter the probability that the strengthening or weakening of synapses will occur in response to neural activity.9 In this way, value systems are perfectly poised to signal the occur-rence of important events to the entire brain.

The importance of value systems for the functioning of a selectional brain has been demonstrated in a set of syntheticallv modeled artifacts that are capable of real-world behavior.

For example, in one such artifact, Darwin IV a value system was required to allow a system controlling eye movements to track randomlv moving targets. This value system reflected the inherited bias that "light is better than darkness" by firing whenever a spot of light hits the center of the eye, at which point a simulated modulating substance was released. This substance decayed over time, but, at suffficiently high levels, it led to the selective strengthening of synapses. W~th this value system in place, the simulated eye came to track objects after a certain number of trials. Of course, the system would also have responded as well in low light conditions if the value, "darkness is bet-ter than light" was used instead. In the appropriate nighttime environment, bats, with their sonarlike system, are as effective as eagles are in daylight or more so. In both cases, value systems are essential.

The second potential limit to the concept of value—too rigid a set of evolutionarily derived value constraints, leading to a pinched repertoire of stereotyped responses in a selectional system—can be met by evolving modi-fiable value systems. We predict, for example, that connections will be found that allow the responses of the ascending value systems themselves to be modified during learning experiences. A recent computer simulation that contrasts fixed and modifiable value systems has been used to test the effects of altering value constraints by learning. The introduction of a modifiable value system in this model led to rich behavior and allowed higher-order conditioning of that behavior that was not possible under a rigid inherited value constraint.l'

An intriguing possibility is that the various value systems of the brain work together to affect brain action by interacting combinatorially, releasing various proportions of their different neuromodulators at the same time. For example, it is well known that during active waking, the noradrenergic, sero-toninergic, and cholinergic systems are firing together. During slow-wave sleep, these three systems reduce their discharge, while during REM sleep, the noradrenergic and serotoninergic systems shut off completely, as the cholinergic system resumes its firing. Different combinations of the corresponding neuromodulators in vast areas of the brain are certainly responsi-ble for many differences among behavioral states in the responses to external stimuli, learning and memory, emotion, and cognition. The possibilities are numerous but have so far not been explored.

It is but a small step to realize that much more sophisticated interactions among value systems related to pleasure, pain, bodily states, and various emotions are possible and are likely to govern cortical responses. The effects of value-dependent learning can range from alignment of auditory and visual maps in the brain stem of the barn owl to the exquisite distinctions made by a connoisseur of wine or the emotional responses of a guilty person. Valne and emotions, pleasant and unpleasant, are obviously tightly coupled and are central to conscious experience.'

Value is a sign of nested selective systems—a result of natural selection that yields alterations of the phenotype that can then serve as constraints on the somatic selection occurring in an individual's nervous system. Unlike evolu-tion, somatic selection can deal wich the contingencies of immediate envi-ronments that are rich and unpredictable—even ones that have never occurred before—by enabling an individual animal to categorize critical fea-tures of these environments during short periods. But we again emphasize that neuronal group selection can consistently accomplish this categorization only under the constraint of inherited values determined by evolution. The nest of systems is a beautiful one, guaranteeing survival for each species in terms of what may be called its necessary prejudice—one required for sur-vival under behavioral control by a selectional brain. As we shall see, the existence of such arrangements is essential for the operation of various forms of memory found in selectional systems - forms that are essential for the evolution of consciousness. When we complete our discussion of memory, we will show how a conscious scene may be built by interactions of memory systems constrained by valne with systems carrying out perceptual categorization.