Next

Previous

Carlos von der Becke y Oscar García- Cascades III

d.7. MORE ON ENERGETICS

1. Some subcascades, for instance the (4),(5),(6),(7), or the arousal-to- exit one inside (5),(6),(7), involve a motivated behavior, which can be described as a generalized Le Chatelier response to the energy sensor alarm showing the serotonine supply. If low, the (4),(5),(6),(7) subcascade appears, and if very high, the arousal-to-exit subcascade shows up.

2. As generalized lechatelierian response, we mean that the strong transition-inducing fluctuations may perhaps grow up, but as they mix with the old dynamic structure or order, they happen to change it, obtaining a new structure or order, which includes in itself the fluctuation, as the final result.

FRIEDRICH R, HAKEN H - A short course on synergetics, in PROTO AN, Non- linear phenomena in complex systems, North Holland (1989).

3. From a macroscopic viewpoint, the fluctuations behave as decaying and displacing the previous non-equilibrium stationary state, resembling somewhat the classic Le Chatelier-Braun explanation for the equilibrium case. An appropiate non-equilibrium phase transition is supposed to use, in a typical dissipative condition, the fluctuations suffered by a former behavior, to find a new order, i.e., a new behavior (Barral and von der Becke).

BARRAL R, BECKE CH von der: Biotermodinámica del Cerebro, Imprenta de Universidad N. de Luján, Luján, Argentina, 2a. ed. (1995)

4.There is some advance in our ideas when we say that each behavior is a structure or an order, or at least a building block or an element of a structure or an order. An element, because we already mentioned subcascades as opposed to (full) cascades, which may be the true dynamical structures and orders.

e. ADVANCED DISCUSSION

e.1. BEHAVIORAL CASCADES AND BUG'S SELF

1. Barral and von der Becke put together some von Baer-Mayr's and some DeBono-Hopfield's ideas in a single "mind selforganization of knowledge- based subnets model". Von Baer-Mayr's, because the ontogenetic embrionic development follows the phylogenetic evolutionary stages of embrionic predecessors, but the whole process also appears restricted to multiple energy-saving selforganizations during the embrionic growth dynamics. DeBono's, because he claims that the act of decision-making involves distinct mental subnets agreeing in a parliament's way of compromise. Hopfield because this autor presents the electronic tools which simplify the essentials of the living neuronal nets possibilities.

2. Back to Beer. Along the bug's main phase transition, attributed to the step (6) - see d.5.8 - and along the (5),(6),(7) subcascade, the neural correlate of an emotion is metaphorically painted. It is sheer hunger, and the bug adapts to "feel the hunger emotion". It acts accordingly. The action selected- i.e., orientation towards the detected food, and stepping into the food patch - is due to the behavioral subcascade frozen at the moment of orientation and stepping in. An external observer concludes: the bug should eat because it is hungry.

3. This teleological motivation and the assumption of a hunger emotion, are alternate ways of explaining the vagaries of neuron nets firing and resting, somewhat like a fixed evolutionary trait simulated by a human designer, Beer in this particular case.

4. Now the bug suffers the most obvious, and mathematically important, non- equilibrium phase transition towards the 11010 behavior, strongly different from all previous ones. The LC, i.e. the leg command, inactivates and the bug stops or almost so. NFCL activates. FA enters in an exponential growth: eating makes further eating easier, a cleverly programmed emotion-arousal sign. Such selfreferenced action also occurs with some higher-level brain functions. This happens when cognition bursts temporarily into more cognition; when creating novelties bursts temporarily into more creation; when being optimist/pessimist bursts temporarily into more of the same; when entering in an emotional state bursts temporarily into a cascade of emotions, as in a concert.

5. One faces now a pure brain, receiving sensor pulses related to environmental conditions, and motoring other pulses towards self-organized leg movements: Beer's artificial bug does not have any metabolism. As a teddy bear it is filled with inert matter, virtual fillings at that. Not so the artificial brain. This is not empty. It has four docens of neurons and seems to be adapting and working, with generalised lechatelierian reponses, to various environmental subsets of conditions. Beer, doing a good work as a neuroethological engineer, is responsible for a supposed bug's sense of identity, showing a small shadow, if any, of a peculiar self. It transpires in all the opportunities to find food and to avoid death.

BEER RD: Intelligence as Adaptive Behavior. Academic Press (1990)

6. The structure of this self-sense can be studied in all detail because nothing is inaccesible to understanding in the neuronal map. The map shows us, in Dennett's terms (referring to mammals, p. 444)

DENNETT D: Conciousness Explained, Little-Brown, (1991).

7. Now replace the word monkey by the words artificial bug. Ask, correspondingly, and if you have enough courage, "How bunches of connected Hopfield's operational amplifiers, see the world?". And, again, that's that. Exagerating, you may begin to see a virtual, perhaps also real, shadow of a sense of self, designed in this bug when the neuroethological engineer finished his work. The bug behaves as having a self while the programm is connected. It is neatly different from the environment. It behaves as if it would try to live as much as possible. Its brain does all that. As a quite peculiar computer creature, this particular one shows, each run more and more, the shadow of a feeling of self. "How can I feel that I have a self, until I show that I behave and behave and behave?"

8. As Forster put it, "How can I know what I think until I hear what I say?". Similarly this mute and shadowy insect could ask: "How can I know what I planned until I show what I did?". The whole net showing a cascade of behaviors, always unplanned but often repeating a cycle, is the unplanned neuroethological correlate of this virtual or real new bug's sense of self, which now, reflecting, one can deduct mostly from intuition, but also from reason. From reason, because we can oppose the bug's self with the bug's environment. The last is like the teddy bear and does not have any self whatsoever, only filling or virtual filling. But the virtual bug, related to this sense of self, is filled with a pure brain. At that, it survives better if the distinction between the bug's self as such and the environment as such, is preserved.

FORSTER EM, as in DENNETT D: Conciousness Explained, Little-Brown, p.245.

9. The artificial neural net functions, together, preserve the survival trend given by the human designer. No selforganization or selfcreation of goals is assumed. Its selfcreated goals are, at most, oportunistic and unplanned goals. There is a goal to follow the food odor. There is another to avoid overfeeding. There is another goal to bypass the obstacle. These multiple goals, as impasses have to be oportunistically resolved, in an unplanned way.

NEWELL A. Unified Theories of Cognition, Harvard University Press, Cambridge, Mass., (1990).

e.2. COMPARISON AND EXTRAPOLATION WITH RESPECT TO PREVIOUS MODELS.

1. Barral and von der Becke model refers to the human superior intellectual abilities or human theorem proving mechanisms. It can be enriched with Beer's actual neurothological engineering viewpoints, when both are brought to a joint consideration. It happens to be a microscopic and subsymbolic view. Initially it was only a macroscopic and symbolic description,open to new investigations.

2. When the bug's neuroethologic model is compared with the superior human one, some far-reaching conclusions appear. Both show mechanistic similarities and differences. For instance, one includes subsymbolic motivated responses but excludes memory; and the other, on the contrary, is built around a new idea's maturing process with memory at the symbolic level. Extrapolating, one is tempted to say that as happens with the bug's model, the normal human way of acting may be based, weakly, on following a master plan of action previously provided by genetics and culture, and strongly, on solving the immediate problem, bypassing the general planning responses.

3. The visual analysis of the bug's behavior when both subjected to the food odor field and to an obstacle put on the way to a direct access to food, is highly illustrative of an unplanned-navigation-through-life viewpoint.

4. The general plan of orienting the walk towards food "so as to avoid starving" (or the neuroethological correlate of this eventual emotion) is clearly secondary when memory is absent. The only strategy provided by the neuroethological map is to lose the intense odor signals and follow the obstacle's edge until a discontinuity is found. The short term strategy of being subjected to two alternative signals, the food odor detected by the chemical antennae sensors and the hard obstacle detected by the tactile antennae ones, needs an energy saving experimental solution, i.e. one with less memory.

5. The general plan tends to have a minor priority. In some circunstances, a direct inversion of the angle of advance can be detected. With another set of numerical values for the neuron parameters involved, the relative proportion between the first immediate problem of following the food odor and the second immediate problem of following the edge of the obstacle which interferes on the way to food could be resolved otherwise, with starving as the consequence. So one can say that the immediate contingency tends to have a major importance.

6. Human behavior may be, too, a cascade of behaviors. The same is typical of emotions (Ortony et al.), one of which triggers a cascade of others. Emotional and rational behaviors have many common aspects, since both are lechatelierian ways to solve mixed problems. Both cascades show energetic contributions and both have an antientropic first branch followed by an entropic second branch, which finally relaxes when the problem is finally solved. Both show the possibility to abort instead of maturing, especially during the stepping up energetic branch.

7. So behavioral cascades have energetic aspects involved. To use memory, and to use symbolic ways of reasoning and problem-solving, one needs higher energetic requirements than when avoiding those needs. Since sometimes there seems no need and no neuroethological elements to characterize neither the prior nor the posterior behavioral trajectories, the strategies based on solving the immediate problem are not only necessary, but also sufficient. Nevertheless, Table I shows that, in the bug's behavioral main sequence, each behavior has a standard predecessor and a standard succesor. So the unplanned-navigation-through-life argument has to make an exception with motivated behaviors in the middle of cyclic cascades, like eating and feeling satiated.

8. The theme has some very peculiar implications. The bug perhaps could pendulate around the maximum odor center, coming and going and again coming, because the numerical values for the neuroethology do not provide a memory-like trigger. Changing those values the memory-like trigger appears. So the come and go pendulus remains active. Because of the numerical values incorporated, it does not show memory. It advances according to static, non historical ways of behaving. It does not remember that the Fick's law, a moment ago, was better as a food neighborhood signal. Instead it continues walking without turning back. This is good on the long run: in complex systems, as known long ago, a good result may be the consequence of a mistake. It is good because starving may be avoided.

9. Another similar situation happens when turning towards the wrong direction, towards the external border and so entering in a "corner limit cycle" which ends in starving. The morals are now different. The wrong initial action leads to a standard wrong final response. This should be an alarm. It should be corrected by evolution's self organization trend (Stuart A. Kauffman) towards either a provision of memory-like neuroethological engineered behavior - negated in our bug - or a new limit- cycle escaping behavior.

e.3. DRIVING FORCES FOR SUPERIOR INTELECTUAL BEHAVIORS

1. The engineered neuroethological maps due to Beer for his artificial bug, are shown in action through a cascade of behaviors. Under favorable conditions, the cascade may cycle back into a new cascade, running as a biological clock. This general situation, may be compared to the previous higher intelectual behavior's model map of Barral and von der Becke. Here also one sees a cascade of behaviors solving a problem, and cycling again, either to better the previous solution, or to try a new problem as an input which urges for a new solution, using the same brain marks left by the previous solution.

2. Based on the ideas of Edward DeBono, and the ones of Barral and von der Becke (1991), a human high intelectual behavior is seen as a parliament of the mind interaction between - say - five neural subnets. From an initial condition of soft thinking activities called distracted mind (Cerebro, Sci. Am. ), the five subnets are alerted by an interest to solve a higher intelectual problem such as a concrete theorem proving activity.

3. Each of the five, the cold rational Newell's "knowledge level" first maturing subnet, the positive looking aspects of the maturing idea second subnet, the limbic emotional feelings and passions related to the maturing idea third subnet (Freeman), the previous wrong experience based outcomes with their particular negative contributions fourth subnet, and the creative maturing fifth subnet, are involved, all, in voting in favor or against the different forms of the solving idea, in the process of either to mature, or to starve.

4. The ballet between the five subnets in their asynchronic parallel and joint operation to move the brain from the initial distracted state into an intermediate excited state, getting eventually to a final relaxed state, is full of neuroethological details to be discoverd and recognized as essential steps in the "ballet" cyclic dynamics.

5. Minsky presents the general rule for an appreciation of the events involved. In computation, the more interconnections between elements operating when solving more than one task in synchronic parallelism, the slower the resulting synchronic activity. This may explain the difficulty of the strict parallel behavior of five very large internally connected and, moreover, externally interconnected subnets, with the intention of solving a superior intelectual lechatelierian problem with a selforganized parliamentary subnets structure, as suggested by Barral, von der Becke and Motto with their multidimensional crossword-like model.

6. Suppose the brain begins the processing of such a lachetelierian input. This is not general, because - in DeBono's words - thought tries to abolish thought. So a defensive and energy consuming mechanism appears. According to it, a cascade of contributions from different mental subnets help agreeing and slowly finishing the task of coping with the input. Slowly because of Minsky's rule. Energy consuming despite of the selforganization promoted, which almost always helps to diminish the energy requirements. A cascade of parliamentary actions and fluctuations result in a couple of possibilities. Either the action aborts trivialy, or constructively matures into a higher intelectual yes/no proposal as a consequence. On the contrary, the lechatelierian answer to the "hunger alarm" bug's artificial brain inputs happens among much less complicated and less interconnected changes involving the FO and NWC neurons firing activity. The difference, according to Minsky's rule, explains the difference between common low intelectual and uncommon high intellectual activities. The difference is a concentration of interconnections difference. Any other bug activity helps to complete the argument.

e.4. INTRODUCTION TO HIGHER INTELECTUAL NEUROETHOLOGICAL ENGINEERING MAPS.

1. Beer shows six bug's leg pacemakers (Fig. 2). They inhibit in a rational way the possibility of a sudden collapse during a walking activity and produce a standard selforganized gait for different velocities of advance. With a vague resemblance, one could, providing the capability, engineer five pacemakers that selforganize the ballet of five intelectual subnets involved in a superior problem-solving task. This is the number of subnets of the studied models. The six leg pacemakers which produce diferent selforganized gaits, suggest the idea of asking for five higher subnet pacemakers which produce different mature problem-solutions, as methaphoric "gaits" of mind activity.

2. Beer shows that the consummatory biting appears, if and only if, the bug shows the neural correlate of hunger and, moreover, shows the neural correlate of having food directly under its mouth. This is analogous to sum the neural correlate of the last positive vote of the most delayed subnet to the neural correlate of the other four subnets positive vote, in order to generate the consummatory action. Whatever this action may be, it requires as a prerequisite the sum of the five frequencies in order to overcome the neural correlate of a threshold value.

3. Beer engineered a neuroethological solution to the increasing driving force for food search, detected through the decay of signals which are correlates to slower serotonine-producing catabolism. Now the bug is eating. The signals increase. But the neural correlate of its hunger emotion, instead of decaying, suffers an enhancement. This agrees with the popular wisdom regarding eating, in which the question is to start with it, following comes alone. This same mechanism should help during the antienthropic problem-solving phase. Beginning to know should help with following with the knowledge accumulation, conditio sine qua non for any antienthropic mental tasks.

4. Beer simulates reflexes, taxes, and fixed action patterns. Their combinations cover most of the possible alarms. They should be recognizad as such, and relaxed in a lechatelierian way, with the production of a particular selforganized subcascade or cascade of behaviors. This makes no distinctions between the bug's necessities and the superior brain's necessities. Suppose a conflict of balloting among the five main intellectual subnets. This could, in principle, be relaxed through an appropiate neuroethological engineering design. At last, "our higher cognitive functions are our own particular human elaborations of this most basic competence for effectively coping with the world." (Beer, p. 11)

5. One may recognize that the bug's behavioral cascades do not optimize in time: Beer did not introduce memory. As the food diminishes as a consequence of consumption, the inexistence of memory avoids learning a new way to behave. It restricts - for instance - the wandering territory to the neighborhood of the food patch. This becomes smaller and smaller with age. When the food patch is large, no memory is needed. When it happens to decrease, a well used memory should be a good adaptive lifesaver. Memory, learning, symbol manipulation facilities and changing behaviors during the lifetime (adaptation) are a necessity when entering higher intellectual behaving cascades and leaving aside invertebrate-like artificial neuroethologies as incomplete.

Next

Previous