Francisco Varela
Ethical Know-How
Action,Wisdom, and Cognition
Stanford University Press 1999


pg 45-52
Keywords: the notion of a cognitive subject - Contrary to what seems to be the case from a cursory introspection, cognition does not flow seamlessly from one "state" to another, but rather consists in a punctuated suc-cession of behavioral patterns that arise and subside in measurable time. This insight of recent neuroscience - and of cognitive science in general - is fundamental, for it relieves us from the tyranny of searching for a centralized, ho-muncular quality to account for a cognitive agent's normal behavior - oscillations are the symptoms of very rapid reciprocal cooperation and competition between distinct agents activated by the current situation, vying with each other for differing modes of interpretation for a coherent cognitive framework and readiness-for-action. This dynamic engages all the subnetworks that give rise to the entire readiness-for-action in the next moment. It involves not just sensory interpretation and motor action but the entire gamut of cognitive expectations and emotional tonality central to the shaping of a microworld. 


MORE ON NON-UNITARY SELVES  AND COGNITIVE AGENTS
I want to pursue further the new approach to the notion of a cognitive subject by focusing on the cognitive activity that occurs in that very special space I call the hinges of the immediate present. For it is in the immediate present that the cognitive subject actually lives. But before we proceed we need to revise, as we did for the question of skills and their importance, some entrenched assumptions concerning cognitive mechanisms inherited from the computationalist orthodoxy.

The main point is this: the sensorimotor coherencies of micro-worlds and micro-identities we discussed in the First Lecture disguise the arising of a large set, or ensemble, of transiently correlated neurons within the brain. These ensembles are both the source and the result of the activity of the sensory and effector surfaces.

Consider, for example, Aplysia, a water mollusk with a "small" nervous system (a few thousand neurons). When Aplysia brings its siphon in touch with a surface (or when the siphon is independently touched), it contracts its gill. This is the so-called gill-withdrawal reaction, one among many of the behavioral patterns normally present in these animals. Traditionally, this sort of behavior is described as being mediated by a "reflex-arc." A recent study has deter-mined that in Aplysia the gill-withdrawal reaction activates a significant proportion of the entire nervous system. This ensemble of activated neurons arises in a coordinated and mutually influential manner, and their coactivation abates after a few seconds. 

Thus the neurons of even this invertebrate ganglion must be conceived as a network of overlapping ensembles which arise in various coherent configurations depending on the animal's context.

The lessons learned from such humble mollusks can be applied to animals with larger and more complex nervous systems, in short, to animals with brains. Indeed, all that changes as we move "up" the chain of being is the number of mediating interneurons and the architecture of the nervous system being studied. In humans, for example, some 1011 interneurons interconnect some 106 motor neurons which are linked to 107 sensory neurons distributed in re-ceptor surfaces throughout the body. The rise and decay of neuronal self-organization has an even more pronounced effect in larger brains. 

In the cat's brain, for instance, 5 -100 million neurons are active during a simple visuo-motor task of pressing a lever. Such neural assemblies arise in a patch-work of regional areas, evincing the enormous distributed parallelism proper to vertebrate brains.

In fact, it is fair to say that a recently established fact of the brain's constitution is what I like to call the Law of Reciprocity: if a region (say a cortical area, or a specific nu-cleus) A is connected to another region B, then B is also connected to A, but by a d fferent as~atomical route. Consider, for instance, the mammalian visual system. Consider, further, the well-known flow of impulses from the retina to the so-called first "relay" station in the visual system, the dorsal thalamus (call this the region A), and then on from the thalamus to the primary visual cortex (call this B), and then on to other cortical regions. There are, in accordance with the Law of Reciprocity, connections from B back to A, from the cortex back to the thalamus, and they are even more numerous than those from the thalamus to the cortex. 

This bidirectional thalamo-cortical neuronal traffic is not a mere anatomical nicety: the visual performance of an animal depends on the integrity of this feedback loop.Thus, the dynamic underlying a perceptuo-motor task is that of a network, a highly cooperative, two-way system, and not that of a linear process in which information is abstracted from sense data in a unidirectional sequence of stages. The dense interconnections among the sub-networks of the brain ensure that every active neuron will operate as part of a large and distributed ensemble. For ex-ample, although neurons in the visual cortex do have dis-tinct responses to specific "features" of the visual stimuli (position, direction, contrast, and so on), these responses occur only in an anesthetized animal with a highly simplified (internal and external) environment. 

When more normal sensory conditions are allowed, and the animal is studied awake and behaving, it has become increasingly clear that the stereotyped neuronal responses to "features" are highly labile and context sensitive to effects such as bodily tilt or auditory stimulation. Even a change in posture that is not accompanied by any change in sensorial stimulation will alter the neuronal responses, demonstrating that even the supposedly downstream motorium is in resonance with the sensorium.

If I may continue to use vision as an example, I can take the provious discussion up one level of generalization, to note that in recent years research has turned to the study not of centralized "reconstruction" of a visual scene for the benefit of an ulterior homunculus, but of a patchwork of vi-sual modalities, inclading at least form (shape, size, rigidity), surface properties (color, texture, specolar reflectance, trans-parency), three-dimensional spatial relationships (relative positions, three-dimensional orientation in space, distance), and three-dimensional movement (trajectory, rotation).

It has become evident that these different aspects of vision are emergent properties of concurrent subnetworks, which have a degree of independence and even anatomical separability, bot cross-correlate and work together so that a visual percept is this coherency. This kind of architecture is strongly reminiscent of a "society" of agents, to use Minsky's metaphor. This multi-directional multiplicity is counterintnitive but typical of complex systems. I say counterintuitive because we are used to the traditional causal mode of input-processing-output. Nothing in the foregoing description suggests that the brain "processes" information in such a way; such pop-ular, computer-like descriptions of the workings of the brain are simply incorrect. 

Instead, the architecture of the brain supports a different kind of operation: signals move "back and forth," gradually becoming more coherent until a microworld has been constituted. The entire exercise takes a certain amount of time, which accounts for why every animal exhibits a natural temporal parsing. In the human brain this flurry of cooperation typically takes about 200-500 msec, the "nowness" of a perceptuo-motor unity. 

Contrary to what seems to be the case from a cursory introspection, cognition does not flow seamlessly from one "state" to another, but rather consists in a punctuated succession of behavioral patterns that arise and subside in measurable time. This insight of recent neuroscience - and of cognitive science in general - is fundamental, for it relieves us from the tyranny of searching for a centralized, homuncular quality to account for a cognitive agent's normal behavior.

Our present concern at this point is with one of the many consequences of this view of the disunity of the subject, understood as a cognitive agent. The question I have in mind can be formulated thus: Given that there is a myriad of contending subprocesses in every cognitive act, how are we to understand the moment of negotiation and emergence when one of the many potential microworlds takes the lead and constitutes a definite behavior? 

In more evocative terms: How are we to understand the very moment of being-there when something concrete and specific shows up?

The answer I wish to propose here is that within the gap during a breakdown there is a rich dynamic involving concurrent subidentities and agents. This rapid dialogue, invisible to introspection, has recently been revealed in brain studies. Some key aspects of this idea were first introduced by Walter Freeman, who over many years of research, managed to insert an array of electrodes into the olfactory bulb of a rabbit so that a small portion of the global activity could be measured while the animal behaved freely. He found that there is no clear pattern of global activity in the bulb unless the animal is exposod to one specific odor sev-eral times. Furthermore, he found for the first time that such emergent patterns of activity are created out of a background of incoherent or chaotic activity by fast oscil-lations (i.e., with periods of about 5-10msec) until the cor-tex settles into a global electrical pattern, which lasts until the end of the sniffing behavior and then dissolves back into the chaotic background. The oscillations then provide a means of selectively binding a set of neurons in a transient aggregate that constitutes the substrate for smell perception at that precise instant.  Smell appears in this light, not as some kind of mapping of external features, but as a creative form of enacting significance on the basis of the animal's embodied history. 

What is most pertinent here is that this enaction happens at the hinge between one behavioral moment and the next, via fast oscillations between neuronal populations that can give rise to coherent patterns. There is growing evidence for this kind of fast resonance to transiently bind neuronal ensembles during a percept. For example, it has been reported in the visual cortex of cats and monkeys; it has also been found in such radically different neural structures as the avian brain and the ganglia of an invertebrate, Hermissenda. 

This universality is important, for it points to the fundamental nature of reso-nance binding as a mechanism for the enaction of sensorimotor couplings. Had resonance binding been restricted to, say, mammals, it would have been far less interesting as a working hypothesis. It is important to note here that this fast resonance is not linked to sensorial triggers in any simple way: the oscillations appear and disappear quickly and quite spontaneously in various places of the brain.

It seems that between breakdowns these oscillations are the symptoms of very rapid reciprocal cooperation and competition between distinct agents activated by the current situation, vying with each other for differing modes of interpretation for a coherent cognitive framework and readiness-for-action. This dynamic engages all the subnetworks that give rise to the entire readiness-for-action in the next moment. It involves not just sensory interpretation and motor action but the entire gamut of cognitive expectations and emotional tonality central to the shaping of a microworld. 

On the basis of this dynamic one neuronal ensemble (one cognitive subnetwork) finally prevails to become the behavioral mode for the next cognitive moment, a micro-world. When I say "prevails," I do not mean to say that this is a process of optimization; it resembles more a bifurcation in a chaotic dynamic. It follows that such a cradle of auton-omous action is forever lost to lived experience, since by definition, we can only inhabit a microidentity when it is already present, and not when it is in gestation. 

In other words, in the breakdown before the next microworld shows up, there are a myriad of possibilities available until, out of the constraints of the situation and the recurrence of history, a single one is selected. This fast dynamic is the neural correlate of the autonomous constitution of a cognitive agent at a given present moment of its life.

Francisco Varela: Ethical Knowhow - page 52 

HOME