THE SELF-ORGANIZING BRAIN
David Kahn, Stanley Krippner, and Allan Combs
Abstract: We argue the REM dream experiences owe their structure and meaning to inherent self-organizing properties of the brain itself. Thus, we offer a common meeting ground for brain based studies of dreaming and traditional psychological dream theory. Our view is that the dreaming brain is a self-organizing system highly sensitive to internally generated influences. Several lines of evidence support a process view of the brain as a system near the edge of chaos, one that is highly sensitive to internal influences. Such sensitivity is due to several factors. First, the dreaming brain normally gates our external input and thus operates without the stabilizing influences of external feedback. Second, the pre-frontal cortex is only minimally activated during REM sleep, and hence the brain operates with weakened volition, reduced logic, and diminished self-reflection. Third, because the neuromodulatory inhibition mechanism is turned off during REM, the brain responds spontaneously to the least provocation. In addition, the dreaming brain is also subject to powerful intermittent cholinergic stimulation which may stimulate creative patterns of dream activity.
Over the past three decades numerous empirical and theoretical investigations have made it apparent that self-organizing dynamics are fundamental to processes at many levels of the organic as well as the physical world (e.g., Kaufmann, 1993; Laszlo, 1987; Maturana, Varela, & Uribe, 1974, Prigogine & Stengers, 1984). Recent work shows this to be no less true for the brain (e.g., Freeman, 1991) Kahn & Hobson, 1993; Pribram, 1995; Varela, Thompson, & Rosch, 1991), and indeed for the process structure of human experience itself (e.g., Combs, 1996; Combs & Krippner, 1998). The present paper examines such self-organizing dynamics in the brain with the aim of understanding the REM dream experience and how it differs from waking consciousness. We begin with the brain.
The Self-Organizing Brain
Many lines of evidence argue for the idea that the brain is a self-organizing system comprised of self-organizing subsystems. To begin with, how could it be otherwise? Though the brain is commonly conceptualized in terms of neural networks and circuitry, there seems little doubt that this circuitry is not rigid, but is significantly influenced by neurological development, day to day learning experiences, and many types of neuromodulation. Thus, the apparent neuroanatomical stability of the brain hides beneath itself many dynamic processes of change. Moreover, the widespread and continuous presence of both single unit firing and mass activity suggests that process itself is an essential feature of the brain, as important as anatomy. While machines and passive electrical circuits can spend indefinite periods of time in inactivity, self-organization and self-creating (autopoietic) systems such as ecologies and living organisms are constantly in motion, as indeed is the living brain.
Many of the activity patterns exhibited by the brain are indicative of complex underlying self-organizing processes. The EEG rhythm, for example, tends to be roughly cyclic, but is not precisely so. It's global form is easily recognized, but the exact shape of its waves differs from cycle to cycle, defying precise prediction. Moreover, it is unlikely that it ever exactly repeats itself. This situation of global familiarity combined with non-predictability, in a pattern that never precisely repeats itself, is exactly what defines a chaotic process, one whose action describes a strange, or "chaotic" attractor (Kellert, 1993).
An attractor is a pattern of behavior toward which all nearby patterns (or trajectories) converge. If they converge to a perfectly cyclic pattern we have a cyclic attractor and in a physical system we are dealing with something like a clock, that always settles into a regular rhythm. When mathematicians discovered equations for attractors that never settle down in this fashion they humorously called them "strange," and these have continued to be known as strange or chaotic attractors.
Such attractors appear to be a common if not universal feature of complex self-organizing systems such as living cells, ecologies, and evidently brains as well (e.g., Abraham & Gilgen, 1994; Basar, 1990; Freeman, 1995; Pribram, 1995; Robertson & Combs, 1995; but also see Mandell & Selz, 1997).
Additionally, the human EEG exhibits significant fractal structure (e.g., Basar, 1990; Screenivason, Pradhan, and Rapp, 1999), further suggesting that it is the result of complex self-organizing processes (Anderson & Mandell, 1996). With regard to REM sleep, at least one investigation (Babloyantz, 1990) found rapid eye movement (REM) sleep EEG to exhibit higher dimensionality than slow wave sleep, suggesting the play of a larger number of underlying influences, as one might expect if EEG activity in any way reflects the complexity of accompanying dream experiences. Anderson and Mandell (1996), who have made detailed studies of the temporal structure of REM state electrical activity in fetal rats, believe that such activity reflects self-organizing hierarchical integrative processes in the developing nervous system. Interestingly, preliminary evidence indicates that this integrative process may follow an abnormal developmental course in the case of autistic individuals (Tanguay, Ornitz, Forsythe, & Ritvo, 1976).
The fractal constituency of the EEG also suggest the possibility that the brain resides in a state of self-organized criticality (Bak, 1996). A system is said to be in a critical state if a small stimulation can set it into fluctuation on all length or temporal scales--in other words, if the response distribution is fractal. The classic example of a critically poised system is a sand pile ready to cascade into an avalanche when a single grain of sand is dropped onto it. Bak points out that the brain must also be critically poised. Otherwise it would not, for instance, respond globally to the appearance of a single visual image which carries but a minute amount of actual physical energy. Unlike the sand pile, however, the brain is not a randomly organized static structure, but an enormously complex ongoing dynamical process system, a product of its own self-organizing tendencies, and thus can rightly be said to exhibit self-organized criticality. With regard to the importance of self-organized criticality in biological systems, Stewart Kauffman (1993) observed that "selection achieves and maintains complex systems poised on the boundary or edge of between order and chaos. These systems are best able to coordinate complex tasks and evolve in a complex environment" (p.xv).
All this is simply another way of understanding the notion that even small influences can exert sizable or even dramatic effects on ongoing patterns of brain activity. The best known example of this is the butterfly effect, which refers to the idea that no matter how small an external influence (such as sensory stimulation) might be, its influence, when compounded through many recurrent cycles of system activity, can grow to virtually unlimited proportions (Kellert, 1993; Peak, 1994). The effect was originally discovered by meteorologist Edward Lorenz (1963) in models of fluid convection. It came to be known technically as sensitive dependence on initial conditions and is a distinguishing feature of chaotic behavior. In the popular literature, as most present readers will know, the "butterfly effect" refers to the notion that the stroke of the butterfly's wing, say, in Brazil, might cascade a few days later into a hurricane in the Bahamas--or alternatively quell a potential hurricane there.
More important than the butterfly effect, however, is the seemingly paradoxical effect known as stochastic resonance, that has been demonstrated in electronic circuits as well as in nerve cells (Moss and Wiesenfeld, 1995). It refers to the fact that the presence of vibration or noise keeps the system in motion and tracking an overall course of least resistance, rather than getting stuck in small groves or "minima." For instance, an object on a vibrating tabletop will sometimes "walk" about, especially if the table is not level, following the overall line of least resistance down the slope of the surface. Stochastic resonance can actually improve the effective signal to noise ratio in a communication situation. In the brain it may allow ongoing processes to "relax" into inherently natural patterns of activity, an important point to which we will return shortly.
First, let us consider the possibility that the brain's activity, like that of other extremely complex systems such as the weather, can be understood as an exquisitely intricate strange attractor, one exhibiting an intricate array of "wings" or "compartments" (Goertzel, 1994). During wakefulness the shape of this attractor, especially in the sensory cortices, is powerfully constrained by sensory input, which itself is often highly patterned (e.g., Gibson, 1966, 1979). Freeman and his colleagues (Freeman, 1991, 1995; Freeman & Barrie, 1994) have mapped such attractors in a variety of different sensory cortices. They found that the sensory regions of the brain are critically poised to respond robustly and in an ordered fashion to even the smallest stimulation. In the REM state, however, such attractors are not constrained by sensory input. In this state the self-organizing dynamics of the brain are set into motion not by external stimulation but by its own internal situation. Interestingly, it is possible to find such self-organizational dynamics at work in the waking state as well. Freeman, for instance, discovered that new learning experiences actually modify previously established cortical activity patterns. For example, a rabbit's original cortical response to an odor is altered when the odor is experienced in a new context, such as a classical conditioning situation. Freeman interprets such changes to signify that the meaning of the stimulus is as important in the production of the brain's response as the physical structure of the stimulus itself. Speaking informally, Freeman (1997) once observed that if one sees Hamlet, then sees Rosencrantz and Guildenstern are Dead, returning to Hamlet finds it to be a different play.
The Dreaming Brain
During REM sleep the brain is as active as it is during the waking state. (This paper does not pursue the knotty debate over the meaning or even existence of non-REM dreaming, but for an excellent critical review of this question see the recent paper by Hobson, Pace-Schott, and Stickgold (2000). However, information processing is inner-oriented as distinct from the outer sensory orientation of waking. In this state a number of factors combine to make the brain acutely reactive to internally generated influences. To begin with, the stabilizing effects of external sensory input are actively inhibited. Also, there is a shift away from widespread aminergic neuromodulatory inhibition which dominates the waking brain, toward cholinergic modulation that predisposes the sleeping brain to easy activation (Hobson, 1994, 1988).
In terms of activation patterns in the REM sleeping brain, recent investigations using PET scans (Braun, et al, 1997, 1998, Maquet, et al, 1996, 1997) show notable arousal of the extrastriate visual cortex, especially in the ventral processing stream. Notable activation is also seen in limbic and para-limbic structures, most significantly in the anterior cingulate and the amygdaloid complexes. Meanwhile, activity in the dorsolateral prefrontal cortex is markedly reduced. Taken together, these finding point toward emotional arousal during dreaming, while at the same time suggest a reduction o memory as well as diminished capacity for logic and self-reflection. These conclusions are entirely consistent with many studies of the subjective qualities of REM dreaming (e.g., Hall & Van de Castle, 1966; Tonay, 1991).
Interestingly, Braun et al (1998) also reports decreased activation of the primary visual cortex during REM. This observation may seem surprising, since a deactivated primary visual cortex due, say, to a stroke, results in the absence of visual awareness. It is, however, consistent with the suggestion that the conscious experience of vision is more directly associated with the extrastriate association areas, and their connections with the frontal cortex, than with the primary visual cortex itself (Crick & Koch, 1992; Koch, 1998; Revonsuo, 1998). In line with this, lesion studies show that damage to the extrastriate cortex, as well as damage to the parietal operculum and to the mediobasal frontal cortex, result in decreased dreaming (Solma, 1997; Hobson, et al 1998a). Patients who reported a global cessation of dreaming had damage in the parietal convexity or suffered disconnection of the mediobasal frontal cortex from the brainstem and diencephalic limbic regions (Solms, 1997; Hobson et al 1998b).
PGO Stimulation, the Dream, and the Self-Organizing Brain
Sleep affords the opportunity, within certain limits, for the brain to act of itself, and dreams are the result.
Edward Clarke, A Study of False Sight, 1878