Copyright ©  2000 Derek Ellerman

If you have any comments or questions, please use the boxes below.  Thanks! 

Introduction.

Bertrand Russell once said that philosophy is the gray area between science and religion.  What he may have intimated is the extent to which both philosophy and religion would recede in their authority to explain the events of the world, both in the subjective and objective realms, as science provides an increasingly adequate and demonstrable analysis of the universe.  The time is rapidly approaching when the bookends of his statement will come closer and closer together with the growing implausibility of supernatural explanation.  While Christianity has traditionally relied on the supernatural to explain its core beliefs, much of Buddhism has lent itself to an practical, experiential and subjectively experimental approach, which shied away from metaphysical speculations. 

In regard to its secularism, Buddhism has proved highly compatible with the advancement of science, and may be the first of the major religious practices to be integrated into the mainstream scientific world.  This paper is based in the assumption that in time an equally adequate intellectual explanation of Buddhist insight can be achieved using a scientific framework as can be achieved using traditional textual sources.  While these two perspective will clearly differ in their language and perspective, it will be increasingly evident that they are referring to the same set of phenomena, one from an objective point of view based in evolutionary and biological analyses and one from a subjective based on the experience of the meditator.  It should be noted that both these perspectives can only provide an intellectual or conceptual understanding, and that practice is necessary to obtain experiential insight.  A conceptual understanding does not, however, degrade, reduce or explain away the insight of the meditator, but rather aids in affirming it with an unequivocal stamp of truth that transcends the purely personal understanding of the practitioner.  As sociologist Marvin Bressler once said on a related subject, the dogged preservation of age-old religious belief systems “…encourages decent men to tremble at the prospect of ‘inconvenient’ findings that may emerge in future scientific research.  This unseemly anti-intellectualism is doubly degrading because it is probably unnecessary.”[1]  In Buddhism this was never more the case.  In this spirit, the following paper is heavily composed of neurobiological theory in relation to Buddhist meditation.  It is, however, first and foremost about Buddhism and the experience of insight that results from Buddhist practices. 

Most Buddhists would not attribute the insight gained in meditation and practice to a supernatural entity, as some Christians may.  This leaves a subset of epistemological possibilities that are compatible with modern science.  The question is to what extent is our knowledge and in particular Buddhist insight is tied to the biological medium we are using to “know”?  For math, most mathematicians would assert that there is very little, in that while our brains are necessary of course, the mathematics operates as it does regardless of the particular make-up of our brain.  There are, however, different types of knowledge, and some are more dependent on the make-up of our brain-mind and the evolutionary forces that shaped it.  How we “know” love or desire, for instance, or even the parameters of our sense of Self, are examples of this type of experiential knowledge.  These forms of experience would differ greatly if our brain had been shaped by other evolutionary pressures. 

The question is whether or not particularly Buddhist insights like the perception of impermanence or that all beings possess Buddha-nature are based on 1) some form of abstract deduction (logic-based for instance) 2) a privileged understanding of the state of basic reality (corresponding to physical reality as some have hypothesized), or 3) the particular process of consciousness itself.  Most would probably agree that while Buddhists can of course arrive at the first basis for deduction, it is not primary.  The second, while growing increasingly popular with the linkages made to physics, is deceptive.  The temptation to conclude that a privileged understanding is achieved through meditation ignores the embodied state of the understanding.  It is assuming that the knowledge functions like deductive or propositional knowledge (a piece of knowledge based on an objective truth sharable by multiple subjects, not necessarily human), and that the understanding can be divorced from its medium (like some would say math can be, in that the math is valid whether used by a brain or by a computer).  This paper attempts to show that the insight that results from certain meditation practices, prior to intellectual reflection, is embodied, in that the nature of the knowledge and experience is a direct reflection of the process in which it is embedded.  In this manner, meditational insight is the process of consciousness “knowing” itself devoid of the synthetic construct of the Self.  This does not, of course, preclude the possibility that the understanding of direct experience is later used to build on or supplement language-based knowledge.  While we are just now arriving at an appreciation of this embodied understanding, Buddhists and other mystics have been intensely aware of this distinction for millennia. 

The question now arises: how dependent are the insights achieved in Buddhist meditation on our particular brain structure and in what way?  Could another organism with high-order processing and language capability, but different emotional modes for instance, still experience Buddhist insight?  This paper will argue that Buddhist insight is above all an absence of identification with the Self that evolution has given us.  The contents of that Self can and of course do vary, but the absence of the synthetic construction of the Self based on our symbolic world reveals a basic process of consciousness.  This assertion seems to resonate well with the notion that all sentient being possess the Buddha-nature.[2]   The next sections elaborate on the biological basis of the brain-mind, providing a theory for consciousness and higher-order processing based on neurological function.  From this foundation we will examine the hypotheses of the neurological correlates to Buddhist meditation, and hopefully by the end arrive at a clearer understanding of the nature and origin of Buddhist insight.

The Neural Basis for the Consciousness and the Self

Gerald M. Edelman is a Nobel Laureate currently developing cutting edge models of the how consciousness and mental functions arise from neural processes.  The centerpiece of his work is the Neuronal Group Selection (NGS) theory.

Selectionist Theory.

            At the heart of the NGS theory is the hypothesis that the brain operates as a selectional process, using homeostatic guides and other value systems to select upon a diversity of behaviors and internal states represented in functional neuronal groups. (Edelman, 1992) The value systems are embodied in what has been called the diffusely projecting modulatory system and other structures like the hypothalamus and amygdala that help serve homeostatic functions. (Edelman, 1989; Bear, et al., 1996; Hamann, et al., 1999) The modulatory system is composed of brain areas like the locus coeruleus and the dopiminergic systems that project diffusely to many areas of the brain, and release neurotransmitters that, among other functions, modulate the strengthening or weakening of the synaptic connection patterns associated with the adaptive value of a particular behavior or internal state.  (Bear, et al., 1996)

The strengthening and weakening of synaptic connection patterns has also been demonstrated to be dependent on neural activity and the coincidence or degree of synchronization of that activity.  The basic observation of one of these strengthening/weakening models, the Bienenstock–Cooper–Munro (BCM) theory, is that the strength of a synaptic connection, and the strength of a pattern of synaptic connections, is dependent on the activity of the synapse or synapses, with more excitation leading to stronger connections.  (Rittenhouse, et al., 1999)  Synchronized inputs will also strengthen connections, while de-correlated activity may weaken overall connectivity. There is also evidence that a small amount of activity leads to a greater weakening in synaptic connectivity than no activity at all.  (Rittenhouse, et al., 1999)  LTP and LTD tend to lead to the creation of functional neuronal groups, where a population of neurons becomes strongly connected through the coincident activation associated with a unitary aspect of behavior or experience. (Edelman and Tononi, 2000) The differential amplification of adaptive neuronal groups therefore occurs through the strengthening of synaptic connection patterns, both in activity dependent and in neuromodulator-induced strengthening. 

 The analogue to a diverse gene pool in natural selection is the tremendous diversity of synaptic connection patterns that arises during brain development and during neurogenesis in the mature brain.  The variability is a necessary result of the randomness inherent in neuron migration and cell-process wiring that cannot possibly be directed or instructed by genetics alone.  New variability is introduced into the system of the mature brain through the novel patterns of activation between neuronal groups and through neurogenesis in hippocampal and cortical areas, as well as through LTP/LDP-based competition in the case where multiple neuronal groups possess overlapping connectivity.  (Edelman and Tononi, 2000; Ericksson, et al., 1998; Gould, et al., 1999)

Connections within and between the neuronal groups associated with a behavior are selectively strengthened based on the input of the value system that evaluates the overall state of the organism as well as by LTP.  This leads to a ‘memory’ of functional activation patterns represented in the strengthened synaptic connectivity. (Edelman, 1989) There is some evidence that the brain continually activates in cycles or reverberations these connection patterns, helping to preserve their relative strength.  (Tsodyks, 1999) 

‘Reentry’, defined as the mutual statistical dependence of activity in one neural area based on the activity in another strongly connected area, binds together the various components of behavior and experience as embodied in the active neuronal groups. (Edelman, 1989) Functional and coordinated behavior naturally arises due to the activity-dependent and neuromodulator-dependent processes that helped produce the strengthened patterns of connectivity, which were originally, in most cases, coincident in time and progressively active in sequence.  NGS theory proposes the existence of a ‘dynamic core’, or a unitary but constantly changing pattern of reentrantly connected neuronal groups that are active at any one time.  (Edelman and Tononi, 2000)  The neuronal groups active in the dynamic core at any point in time are incorporated due to the historically strengthened connectivity with other contemporaneously active groups, and are therefore ‘bound together’ because of the functional association in the organism’s past.  Strengthened synaptic connection patterns will not become expressed or made aware of until they are ‘bound’ into the dynamic core, creating functional reentry with other nervous system areas.  (Edelman and Tononi, 2000)

      NGS Theory further proposes that primitive consciousness is the process whereby value-neutral sensory and cognitive inputs evoke the activation of strengthened synaptic connection patterns associated with similar inputs from the brain’s past.  These evoked connection patterns that are not value-neutral, as they are associated, also through strengthened pathways, to the value inputs linked to them in the organism’s history, and the various sensory modalities are bound together through real-time and continuous reentrant connectivity.  The current value system input then acts upon the evoked activation patterns, driving adaptive behavior from a mixture of past associations and current assessment of the state of the organism. (Edelman, 1989) Primitive consciousness is therefore the tying of a value-neutral scene from reentrantly connected sensory modalities to evoked activation patterns in higher level cortex that place the inputs in the adaptive context of the organism’s past experience, while also modifying this activation pattern based on current input from the value system. (Edelman, 1989)  The form of the activation pattern is the dynamic core, with more or less strongly connected neuronal groups continually ‘entering’ and ‘exiting’ the core of activity. 

        Higher consciousness, as exhibited in humans, requires linguistic ability and the integration of these symbolic capabilities into the processes of primitive consciousness.  Reentrant activity between language centers and the rest of the associational cortex have enabled humans to attach symbolic representations to the brain’s own activity patterns, allowing for consolidated reference to inner states, events and objects in the new context of grammatical relations. (Edelman and Tononi, 2000)  On the simplest level the association of a symbol with an activation pattern in the brain’s conceptual centers is not a process any more complex than that which we have already outlined through reentry and LTP.  Grammatical relations may utilize other processes which, neurologically, have yet to be described in any detail.

Neuronal Group Selection (NGS) theory lies in opposition to the model of the brain as a computer.  Computer analogies hypothesize that the brain processes bits of encoded information, which can be likened to software, and the basis of mental functioning is the execution of this “software.”  While there are superficial similarities between computers and brains, upon closer scrutiny the analogy breaks down very quickly.  A few examples of the problems are the lack of a programmer to assign categories on an unlabeled environment or assign error feedback for behavior; the lack of an observer or homunculi who will read the translation of the “code” of neurons; lack of intentionality, which is the hallmark of consciousness; and the simple fact that information contained in genes is not enough to specify the precise wiring of the brain as a computer-like model would necessitate.[3] 

NGS theory provides a convincing alternative to the computer model by arguing that the brain selects a subset of neural functioning from the huge diversity of potential brain states (synaptic connection patterns) while adaptively matching it to certain value criteria in the absence of a superordinate coordinator.  Because NGS theory provides a theory for the operation of consciousness that lacks a superordinate coordinator or homunculus to read “encoded data,” it lends itself immediately to the Buddhist perspective.  Buddhist insight has found that there is no essential self; it is an illusion.  NGS theory arrives at the same conclusion, one that is necessary to any adequate explanation of the mind.  The following processes are foundational to NGS theory, and will lead us directly to the neural basis for meditation and insight.  

COMPONENTS OF NGS THEORY:

The following section individually reviews some of the key ideas in NGS theory in order of development or increasing complexity.  Note that some of this material will overlap with what has been presented above.

Developmental Selection.

Formation of neuroanatomy is constrained and guided broadly by genetic expression and chemoaffinity (migration and placement of neurons based on chemical signals from nearby cells).  But at the finer levels of synaptic connection there is tremendous diversity that is not programmed or directed by genetics.  On the contrary, the state of diversity is the foundation for activity-dependent neuronal selection, where neurons that fire together tend to wire together, thereby leading to the formation of specialized, functional neuronal groupings as observed in sensory and motor maps.[5]  Selection occurs in that the formation of certain functional neuronal groups over others in neural space is dependent on the activity of the neurons, which in turn is dependent on the behavior of the organism.  An example of this process is the topobiological organization of sensory maps, where nearby locations of sensory input on the finger, for instance, are also nearby in the neural pathways that carry their input and in the structures that process the input.[6]  This process has also been described as Hebbian modifications.[7]

Experiential Selection.

Experiential selection overlaps with early development and extends throughout life.  It occurs largely within the neuroanatomy created during developmental selection and is based on the activity-dependent strengthening and weakening of connections between neuronal groups.  The selectional process is constrained by diffuse projections from the value systems, which favor the strengthening and weakening of synaptic connections under particular pre-set conditions delineated by evolutionary determinants and expressed through  the value system[8], corresponding to the diffuse modulatory systems in the brain.[9]  Here the strengthening of synaptic connections is indirectly dependent on concurrent firing, in that the value systems concurrently effect those groups firing together.  But it is the expression of the value systems that determines when and to what degree patterns of synaptic connection between neuronal groups are strengthened or weakened.  A crude example is: if your blood sugar levels drops too low, the value system will strengthen the synaptic connection patterns associated with behavior that raises the blood sugar level, thus ensuring the repetition of this behavior. 

Reentry.

Sensorimotor maps and certain other areas of the brain are “massively” connected in parallel and reciprocal circuits.  Reentry is the process whereby selection on neuronal groups in one area effects selection on neuronal groups in others and vice versa.[10]  This contemporaneous and mutual interaction is the foundation of many of the processes that lead to consciousness.

Perceptual Categorization.

Perceptual categorization is the ability to categorize or isolate salient objects in an unlabeled environment.  It is one of the most basic functions of the mind and has been one of the most difficult to adequately explain.  NGS theory posits that perceptual categorization occurs when particular sensorimotor patterns (embodied in maps and their reentry) are selected for by value criteria.[11]  The strengthening of the synaptic connection patterns underlying these sensorimotor patterns creates behaviors that isolate functional salient characteristics or inputs from the environment at the expense of non-salient inputs.  Saliency is  determined by the value criteria, which in turn has been evolutionarily selected for.  Edelman has been able to construct robots that perceptually categorize based on neuronal group selection and in the absence of programmed error feedback, lending strong evidence to his theories.[12] 

Non-Representational Memory.

The nature of memory is key to understanding the effects of meditation.  NGS theory argues that memory is not representational, in that data or programs are not encoded like in a computer, to be “read” and “executed” later for repetition of performance.  Rather, memory is a reflection of how an input similar to one associated with past patterns of synaptic connection can trigger a repeat performance, even in a different context and time period.[13]  Memory as such is entirely embodied.  This model of memory will have powerful implications for the meditation process.

Global Mapping.

Global mapping describes the interaction between the hippocampus, basal ganglia and cerebellum and the cortex during a sensorimotor process, stressing the integrated and continuous nature of associating motor behaviors and sensory input in the course of behavior.  The cortex does not execute a command to the motor system based on sensory data.  They both continually and reciprocally effect each other in the execution of a behavior.[14]  Global mapping relates to memory in that long-term changes in the strength of synaptic patterns due to past behavior tends to favor mutual reentrant activity between those groups (represented in patterns).[15]  This plays a role in the automation of behavior, an element that Buddhist practice seeks to reduce, and the stability and maintenance of the Self process.

Concept.

A concept is the brain’s mapping of its own activities or global maps.  It allows for an abstraction of global mapping from the immediate context in which it originally arose.[16]  If, for instance, a global mapping of the sensorimotor process involved in a cat’s surprise is mapped, the brain will be able to relate this mapping pattern with a value input¾perhaps in this case a negative one.  Value-category memory results when output from the diffuse modulatory value systems adaptively strengthens or weakens the connectivity of a concept like in the example above.[17]   The process helps to explain the connection between meaning and memory, and suggests why traumatic memories (flooded by value systems) are so difficult to lose and prone to repetition.  The minds propensity toward the classification of the world and behavior in this manner has important implications for Buddhist practice.

Primary Consciousness.

Primary consciousness manifests when the maps of the sensory modalities, reentrantly connected, interact with value-category memory, tying present sensory input to associations and value-laden concepts from past experience.[18]  It creates a remembered present.  This confers the evolutionary advantage of constantly relating current input to past associations, allowing for avoidance of past dangers and gravitation towards rewarding situations.   A certain level of neural complexity is required, which is the balance of differentiated functions and their integration (statistical dependence on each other’s behavior for their own).  Differentiation without integration leads to lots of information but not tied into a scene.  Integration without differentiation leads to a state where functional specialization is at a low level, thus not providing the variety components necessary to construct a scene.[19] 

Dynamic Core.

The dynamic core is the current but constantly changing pattern of reentrant neuronal groups that are active at any one time.[20]  The process of their reentry (between structures described broadly above) is the basis of consciousness.  Many properties that we would expect from this system are found in our consciousness.  We will examine these properties when we look at their connection to the meditational process. 

Conscious and Unconscious.

The relationship between the conscious and the unconscious (automated behavior) is a theme running throughout Buddhist practice.  NGS theory provides a simple but elegantly convincing explanation for the question of why neural activity that is effecting behavior is not included in the dynamic core.  The basic observation is that there are circuits that begin with output from cortical areas but are not heavily reentrantly connected with those areas.  Instead, the pathways run in long loops as in the basal ganglia, and are functionally segregated and independent from other areas.[21]  These structures then connect back to the cortical areas in select locations, not massively like reentry would necessitate.  What this all means is that when an output is made from the dynamic core to these areas, the areas then carry out the behavior in an isolated fashion from the core, therefore not in the conscious mind.  There is feedback, not reentry.  Another type of automation may occur in the cortex with the formation of splinter cores.  It is well known that individuals can have functions that indicate conscious activity and yet not have this information reach their awareness (as in the case of blind sight, where though people are blind they are able to avoid objects in their path).  It is possible that these splinter core remain functionally isolated from the dynamic core, contributing to behavior but not to awareness.[22]  The process of deautomation of behavior, both mental and physical, is central to Buddhist practice.