[Text first published in Cahiers n.1 des Archives Jean Piaget (Genève 1980).]
The spirit of the present undertaking, as I see it here, is to show that genetic epistemology is alive and well. This means it is evolving – at least in my mind – and I will try to make it visible on the outside and, on a more serious note, I will try to make use of Apostel’s « Lakatosian » formulation to give you my personal version of what is genetic epistemology’s research programme today.
My prescription (in the medical sense of the term) for genetic epistemology – this being part of the research programme – is to hybridise epistemology with cybernetics, which means a shot of functionalism and of neomechanism into the old structuralist vein and, of course, we will have to see how my patients reacts to this medicine. The question is : why do I want to do that ? The first reason is an epistemological one. I believe that if we reciprocally assimilate the cybernetic framework to the constructivist framework, we may get out of it a unification, an epistemological unification of the conceptual framework of the sciences of life. The backlash would be that the objects of evolutionary epistemologies are evolving objects, including themselves. This looks very much like Alice in Wonderland’s croquet party. Who are the players ? Well, at least four evolving systems. One is cosmogenesis, which I won’t talk about because I don’t consider it to be a cybernetic system. The three other cybernetic systems that I will consider are : the genetic systems of species; I will call them phylogenetic systems. They are the first producers of knowledge on this planet, producers of biological knowledge; then of course sociogenetic systems, and psychogenetic systems in the sense of individual cognitive systems.
– 179 –
Now the idea is : these are all adaptive, problem-solving, artificial intelligence systems. One way of condensing this idea in metaphorical and shocking terms, well epistemologically shocking at least, is to say things like : the first artificial intelligence system on this planet is coextensive with the first genetic system which is in turn coextensive with the emergence of life in chemical evolution. So now as the object of epistemology we have three systems that are evolving simultaneously and that saves us as epistemologists, none of them are moving at the speed of thought. They are generating knowledge on three very different time scales : geological time for the genetic systems; sociological time – well what shall we call one processing operation ? Perhaps the succession of generations, to index it somewhere, because as you know in sociology the difference between synchronic and diachronic is somehow linked. – Synchronic is people of the same generation. It doesn’t really mean much, but people who are present in one slice of time and who are interacting, who form « la république des esprits », that is who, to use a cybernetic criterion, exchange information on a given scientific subject. Then there is this aspect of transmission from one synchronic republic to the next which is called diachronic where one elementary operation, one transition, takes say thirty years. Anyway it is less than a thousand years, less than a hundred years : less than one whole life period; and perhaps the processing time of the psychogenetic system is, alas, shorter. I do not believe nor hope it goes on processsing after death like Apostel nearly said – I do hope it goes on after twelve. So this sets up three main time scales, generated by the elementary epistemic operations of the respective machines. The last one has to do with the « speed of thought » and determines some kind of meso- or micro-genetic evolution. Something like that of the model Apostel was talking about, that we carry along during the solution of a problem. Now this is a period of time that can go right down to the speed of thought in cases where there is no problem,
– 180 –
where the solution is compiled. Usually what psychologists « observe » is whatever evolution happens in the attention span of a child, perhaps fifteen minutes. I’ll call it micro-genesis.
Now if we consider these three systems as epistemic machines, this problem shift changes Piaget’s constructivism in some way that I would like to briefly illustrate. Piaget has a tendency, when studying psychogenesis, to consider equilibration of the cognitive system as the evolving thing – the agent – and to consider sociogenesis or biogenesis as fixed, which is relatively true on this time scale, and these he calls factors. It is a good way of applying the « ceteris paribus », a good heuristic. At some point he says that the factors also evolve but he does not really study them. They do not appear inside the theory as processing or productive mechanisms. The problem shift I propose, or perhaps the assimilation by the constructivistic frame of this general evolutionary position of the problem of knowledge, changes the static factors into interacting knowledge-producing « epistemic machines ». Now the object of epistemology becomes the exchanges or the interactions, the exchanges of information : « knowledge that » and « knowledge how », between three information processing machines, and epistemology becomes radically constructivist. Epistemology must take this issue of co-evolving interactions as its object.
Now about my epistemological unification of the life sciences. Let me briefly back off to my own « psychogenetic » experience. I have long had the intuition that cybernetics has one object and the way I tried to make it explicit was pretend to be a bit pragmatic about it and say, look, what will cyberneticists accept in practice as objects of their discipline, as cybernetic machines. What I get out of this pseudo-experiment is a pattern. If you look at the phenomena cyberneticists try to approach, to assimilate, you will not find cosmogenesis or meteorology. They seem to have a very consistent reticence about a whole range of systems; exactly
– l8l –
those where cybernetic treatment would entail a return to finalism and artificialism. If we talk about the « hardware of the cosmos » then we are talking about the cosmos as a computing machine which means that, in some sense, there is a great artificer who has built it and presumably uses it as some kind of gadget. So to characterise the objects that cyberneticists will accept, let us go back to Wiener, to cybernetics as « communication and control in the animal and machine ». Obviously animal is zoo- or anthropo-centric : plants, bacteria and viruses all have internal control and communication processes, so we have to say organisms. Of course machine is where it all started. But there is one more thing which makes this whole affair recursive and that is organisations : organisations of organisms, of organisms and machines, of machines and machines. This leaves the progression open at the top, whereas you can consider an organism as an organisation and go right down to some basic hardware level where if you go deeper you are out of the cybernetic category. This hierarchic and recursive approach is the important point. But the constructivist theme that appears here and that relates with another point in Piaget’s theory, which gets another « éclairage » from this point of view, is Piaget’s « cercle des sciences », his hypothesis about the circular organisation of sciences. The scheme connects physics to biology, biology to psychology (and sociology), psychology to mathematics. Then the diagram folds back onto itself and mathematics connects to physics. The connections between physics, biology and psychology are expressed by Piaget in terms of « continuity » : there is a functional continuity between life as an adaptation and psychology – intelligence – as an adaptive system. There is a (constructive ?) continuity between natural, logical or mathematical systems and formal structures (inter-subjective structures I would call them now after Apostel’s talk), and there is some kind of epistemological continuity between physics and mathematics in the sense that mathematics plays the role of a sort of mobile or portable a priori for the organisation of
– 182 –
physical experience. Finally there is again, and this you do not get in Piaget’s explicit formulation, a continuity between physics and I am going to say the hardware of biological systems. What he does say explicitly is in his very fascinating science fiction speculation on why mathematics is adapted to reality. His conjecture is that it is so « because the organism at some time was coextensive with physics ». That is, the general coordinations of actions were,at some point in evolution, the same as general laws of interaction between objects. Now this, I believe, is wrong. I am just going to state the thesis but I shan’t pretend to explain it. To put it in a nutshell, mathematics is part of the software of biological systems. It is a software construct. In Piaget’s original terms : an « instrument immatériel des échanges entre organisme et milieu ». Physics is part of the hardware but in fact only some selected physical parameters make up the hardware of organisms. Thesis : software is hardware independent. And conversely. Full stop. So, in specific particular cases, the nature of the hardware may limit or otherwise pattern the computational power and in fact does limit the ultimate power of the software that can be executed on a given machine in a given time. This de facto relation exists in specific cases for specific machines but there is no general de jure relation between any hardware and any software in this sense : given a specific computational scheme or procedure it can de jure be executed (with varying degrees of practical functionality) on completely different hardwares that entail the use of completely different laws of physics. You can build an adding machine with gears and ratchets and some mechanical linkage; you can do the same thing with hydraulic valves, you can do this with electromagnetic relays or with electronic ones. In all cases the physical laws governing the parameters selected as a physical substratum for information are different. If we followed Piaget’s idea, something like the laws of carbon chemistry at the level of chemical evolution should be strong enough to reflect such general laws of
– 183 –
the universe that we could build up mathematics by using some basic perhaps combinatorial laws of organic chemistry and then project them in a totally alien domain to discover that they would be preadapted to cosmology and to subatomic phenomena, or to anti-matter systems. I do not have a better idea, but I am not happy with this one and the main reason is this hardware-software independence. Its converse is true, of course. The identically same set of physical phenomena can be used to code or to compute completely different functions. Just take another aspect, or scale, or other parameters of the same phenomena. In general, the same goal can be reached by very different means, actions, and the same action can be used to realise very different goals. We know all too well this intuitive equivalent of the hardware-software mutual independence. That was an aside, I am afraid.
Let us take one more step. I was talking about Piaget’s continuities; now I am going to try and show their relation to my radical constructivism. What Piaget describes as continuities, I will reformulate in the cybernetic framework as « construction relations ». The relation between constructor and construct, between artisan and artifact, between the engineer and the engine, between the man who makes the machine and the machine itself, is functional and I call it delegation of function. It is prosthetic. Now, there are muscle machines and brain machines – prosthesis for muscle and prosthesis for mind, in the sense that bulldozers are a prosthesis for muscular power – extensions of our limbs, of our hands- and there are of course extensions for intellectual power like computing machines, but they are both information processing machines. This is one of the epistemological criticisms I would adress to Papert’s very stimulating article in « Logique et connaissance scientifique » where he tries to found the distinction between cybernetic machines and physical machines on the relative proportion of information processing : on information flow versus energy flow. Now my prostheses for muscle are his energy flow machines : they are things like steam
– 184 –
engines, and his information flow machines are my prostheses for mind – Babbage engines, not Diesel ones. But of course all machines are information processing machines. Even engines have a cybernetic expression in terms of computing machines because mechanical signals have to be sent by cams and pushrods from one part of the machine to coordinate its action with the other parts. And these submachines must also be integrated : the ignition system must be coordinated with the valve system. In fact we now have printed circuits to do many parts of this kind of computation which used to be done mechanically. So, all machines, even steam engines, are computational engines at least in part… There is a very nice anecdote about 12 year old Stephenson who, as the story goes, was given the task of opening and closing the admission valve of a proto steam-engine with a string when the piston was at the ends of its course, and he had the idea of tying the string to a well chosen moving part of the machine, closing the feedback loop on itself. From then on he was free : he was no more the printed circuit. This is the prosthetic delegation of function relation illustrated in concrete terms. Now I am going to generalise this prosthetic relation into a general construction relation. The relation between constructor and construct is purposive or teleonomic, and the basic fundamental, epistemological relation for cybernetics is the means-end one. From this, my conclusion is going to be that the main referential shift, an epistemological shift, in the life sciences in the last twenty years has been the constitution of a conceptual framework, of an intersubjective conceptual framework, for goal-directed finalistic, teleological, intentional, functional, etc. phenomena and this is a scientific revolution in its own right. « La vieille téléologie », Jacob’s mistress – in fact she who has been the biologists’ mistress for well nigh 2000 years – functional reasoning or adaptation, has made an end of it and has become respectable. The only trouble is that she has not married a biologist, she has married a cyberneticist. However, the conceptual
– 185 –
framework has become practically intersubjective during the last twenty years. We have witnessed in vivo an event that only happens once in many life times : the Piagetian emergence of a Kantian a priori category of human reason. Philosophically, if not historically speaking, it all started, to my mind, with the Wiener, Rosenbluth and Bigelow article on « Behaviour, purpose and teleology » – which was mistaken on epistemological grounds. I think Taylor’s refutation was perfect, it was brilliant. It was better than Wiener’s article though Wiener clearly put it on the epistemological terrain. But of course he backed down, he was beaten by Taylor and had to fall back on « this framework was simply proven very useful to engineers ». Now that is not a thing to say to a normativist opponent because it is backing out of the epistemological warfare. My belief is that Wiener was right and he should have stuck to his guns on the normative plane and defended his epistemological revolution – and that as such it now has to be digested by genetic epistemology as part of its research programme.
So let us turn back to this indigestion process : reciprocal assimilation. One way of starting the cycle is to ask : where is cybernetics in the « cercle des sciences » ? Indeed where is it ? We have physics, we have biology, we have psychology, we have mathematics and all the rest, but where is cybernetics ? Putting it like this is inducing your hearer into a category mistake. You are giving him rope to hang himself with. I am going to try not to hang myself too publicly but I fear I am certainly going to be tied up. Cybernetics is already there, but it is the same kind of thing as causality, or time, or space – it is a category of explanation or understanding, it is an assimilatory category – so you are not going to find it under the name of a science. If you ask the same question about causality – where is causality in the « cercle des sciences » ? – it is also there, but not within the diagram : it is outside since it is not a concept of science, but of scientific explanation. Now one way of making the assimilation more concrete is to translate
– l86 –
Piaget’s continuities into what I have called the relation between constructor and construct. And to go on to say things like : the constructor of life on this planet is one small side issue or side branch of cosmological construction or cosmogenesis : geogenesis. I mean, this is the way people who theorise on the chemical origins of life would put it : in Oparin’s formula « life is the inevitable consequence of the evolution of matter on this planet, of the laws of chemistry operating on this planet ». In fact what he is saying is : life is the inevitable consequence of geological evolution which has an inevitable consequence which is the formulation of the « Oparin-Haldane primeval organic soup ». Which in its turn has an inevitable consequence in starting a general combinatorial organic or rather carbon chemistry « computer » in which computational schemas emerge, metabolisms, which we can only through a category mistake consider as computing schemas as the very first computing schema is coexistent with the emergence of life. « Life » is the appearance, the assembly, through the action of the carbon combinatorial system, of one schema which is autopoïetic -that is in Maturana’s terms. Now autopoïesis is a concept invented by Maturana to capture the essence of living systems without being too anthropomorphic, biomorphic rather or even too geocentric, to give a more general definition than that of life on this planet. His solution to this problem is to define an autopoïetic system as a system – it’s a kind of regulator, metaregulator – whose prime function is to preserve, make invariant, its own prime function. A kind of primeval biological brain twister. Now genetic replication, the genetic computation schema, the schema of reproduction, is autopoïetic in this sense, in this very sense. An essential property of genes is that they make more genes in their own image, and we call this reproduction. There is no point for a gene, or for a gene system, to make a gene system that is not itself self-reproducing. If it makes a copy of itself and this copy does not have this property of being self-reproducing it has failed. There-
– 187 –
fore it is not autopoïetic. Therefore it is not yet the origin of life. There is a beautiful theorem by von Neumann, an epistemological by-product of which is that to have this autopoïetic property a machine must have a representation of the actions it does to copy itself. So it must have a programme, it must have a difference between representation and execution, it must have two different physical systems and processes within it, one of which is the representation of the action of the other; and this is the origin of cybernetics, the origin of information, the origin of the difference between hardware and software, between programme and execution and all the rest. Before that, and here is my epistemological point, if you talk about information, if you talk about programmes, if you talk about logical schemata, or anything of the sort, before this point in chemical evolution you are making a category mistake, and are henceforth epistemologically damned because you are using information processing terms to talk about a system in which there was no information – yet. This is an echo to von Foerster’s dictum : « There is no information in the environment. » I translate it as « information is not a concept of physics ». Von Foerster adds : « Saying there is information in the environment is the same thing as saying there is transportation in my garage when my car is in it. » Of course, when my car is not in it makes it even worse, but this is exactly what you are doing when you talk about information before the origins of life – the car is not in the garage yet.
So to go on restating Piaget’ s « continuities » in cybernetic or computational terms : cosmogenesis assembles the first self-replicating automaton, then I have tried to show – and I think it is very easy to do so – that if you take such a computational schema (consider it as a formal machine, as an abstract machine) and you realise it in physical, in fact, organic chemistry hardware (in fact we should say carbon chemistry hardware because organic chemistry did not yet exist), then you automatically, inevitably, have a Darwinian evolution or a Darwinian iso-
– 188 –
morphic evolution, from the very nature of physical things. Because of random physical interactions hard-ware is inevitably subject to « entropic degradation ». On the software side, this means messages are noise injected, so replication is perturbed, copies go wrong and become variants. We now have Darwinian variations. Now for selection : from the computational nature of the self-replication schema – from one you get two – you have this exponential multiplication. Then, from the necessarily finite nature of real physical systems, you get competition for a finite quantity of matter and energy between these self-replicating machines variants; from this competition you get selection and you have a hill-climbing Darwinian meta-schema on the self-replication schema. It is a second level, autopoïetic schema composing the replication schema with itself. And this is Darwin’s great discovery, the sort of circular meta-schema at a higher logical level, defined on the basic self-replication schema. You now have a genetic system; immerge it in the flow of time and you get a phylogenesis. Now genetic systems were the first artificial intelligences on this planet. They solved the problem that genes have, which is to make genes in their own image. An important sub-goal emerges based on the fact that to reproduce you have to exist for more than the reproductory time; or at least that much time – even if you are very swift about it – the necessary time for full reproduction. A partial reproduction, even minimal, means total failure. You may be as much of a Don Giovanni as you want or can be, but you have to survive and it is a sub-goal of reproduction. We might say that organisms were built to satisfy a main sub-goal which is to preserve genes invariant over a long enough time so that they can reproduce, which is the top level goal. So that the instinct of conservation that we « observe » is a sub-instinct of reproduction – in fact it is a sub-routine of the cybernetic machine. We can assign all the homeostatic functions that Cannon described as « the wisdom of the body » and all the homeostatic behaviours that have to do with these « essential
– 189 –
(biological) variables » to the satisfaction of sub-goals of the reproductive function.
Now we come to the « continuité vie-pensée » : at some point the phylogenetic system generates a cognitive system. I am going to define a cognitive system as a system that acquires new information during the individual’s lifetime and at least not over more than one generation. I will call the latter biogenesis of knowledge, and the former of course psychogenesis of knowledge – obviously it has to do with the appearance of conditioning or learning. I will ascribe instinct to the biogenetic problem solving system; but I will call psychogenetic anything that can do a little more knowledge acquisition during one generation, during the life of one individual. Now obviously this learning system must have a biological basis. You may say learning is species specific. My formula is : « no neurons without genes to specify them ». This is a short cut for saying you must have a biological basis for the hardware necessary for learning. In fact you must have more than that, you must have a set of innate programmes to write in and access information, and that is what we globally call « conditioning ». But this kind of software must have some kind of innate basis. It is not enough to give a tabula rasa; you must give a minimum of programming power – write in, access read, erase or write over, etc. Something like a minimal direct and inverse Turing machine. In this sense I would ask for much more innate computational power and programme that Piaget does. I believe there is a formidable amount of biological computational complexity in the CNS – more than what Piaget would put in his initial nucleus of sensori-motor reflex schemata. In fact I would tend to do a double-take and to say, look : the amount of knowledge that Piaget and all of us study is infinitesimally small, the amount of declarative structural psychogenetic knowledge that we have is immensely tiny compared to the amount of biogenetic knowledge we have. If I were a radical reductionist and I could imagine a measure of the quantity of information we contain, then if the quantity of information
– 190 –
in our metabolism for instance (in our computing schemata right down to the molecular biology level) and that in our declarative knowledge, were represented in the same units,the latter would simply be invisible at this scale of representation.
Summing up : at some point in evolution, some universal hardware is generated by my phylogenetic machine and this is where psychology starts. When this sort of machine appears, when learning appears, a beginning of psychogenesis appears, perhaps a beginning of socio-genesis inevitably appears, simultaneously I don’t know, but at least we now have a job, we psychologists and epistemologists…
I will now try to recapture what I wanted to say in a nutshell. I would like to mention Dawkins’ book on the « Selfish gene ». It is a remarkable book except for its last chapter which is epistemological reductionism of the worst sort. What he does in the first part of the book is to give a remarkable exposition of what Darwinian theory, synthetic theory, has to say about biogenesis of cognition. It is a fascinating book, how instincts, as software morphologies, evolve. Instincts become evolutionary stable strategies, which he introduces in a beautiful metaphor – he says : « What are we, as biological organisms ? We are robots, huge lumbering remote-controlled armoured robots. And who pulls our strings ? Genes do. What is our function ? To protect our creators. » Genes are no more the free living entities they were in the primeval organic soup. Now they are fragile microcomputers that have produced huge macroscopic automatic machines to protect them, and part of the homeostasis of these robots is what we call instincts. « Instinct de conservation » is in Dawkins’ terms an evolutionary stable (hereditary) strategy. All the things we do to capture our preys, to escape the attention of predators, to thrust our attention on future mates, all this behavioural, cognitive perhaps, automatic computation is the result of the competition of sets of E.S.S.’s. The
– 191 –
radical geneticists’ formula (geneticists in the biological sense) for the relation between organisms and genes puts us in our proper place. « Les organismes sont des artefacts contingents que les genes ont inventes afin de se conserver pour se reproduire » : organisms are contingent, accidental, serendipitous artefacts… that genes have invented to preserve themselves so as to replicate, which makes of course of self-preservation a sub-goal of self-reproduction. This explicitly introduces the sub-goal link. Conservation is a sub-goal of reproduction as far as genes are concerned. Of course this is a completely anthropomorphic way of putting it. But it makes the proper point. Now self-preservation of the organisms and all its network of evolutionary stable strategies is the necessary biological basis of cognitive systems, but it is also a sort of nucleus of hereditary schemata Piaget considers as the basis of the origins of intelligence. One of the things he says – I’ll put it my way – is that psychology (as opposed to biology) begins when the child – the infant – reproduces in terms of habits things he could do in terms of instincts. He reconstructs, says Piaget. He reconstructs in terms of learned routines what was given in terms of pre-programmed routines. And the programmer, then, was phylogenesis. Now the progrannner is you and me – I mean – as we were. The second point I want to make has to do with the relation between the homeostatic autopoïesis of organisms and the morphogenetic autopoïesis of genes : when a subfunction « goes autopoïetic » it is a declaration of independence, it becomes autonomous. The sub-goal relation closes upon itself. Now reproduction, the main biological phylogenetic function, has self-preservation as a subfunction. A sub-function which organisms act out. When this network of programmes for self-preservation goes autopoïetic,it folds back upon itself and it becomes autonomous – autonomous not in the sense of complete liberty but autonomous in the sense of vis-a-vis who made it, its creator. So my theological formula is « autopoeisis is the essence of sin » because now we can have
– 192 –
conflicts between the creator and the created. We can have conflicts between self-preservation and reproduction. I’ll put this in a nutshell, gentlemen. If you were given the choice here and now between self-preservation and reproduction, between instant decapitation and instant emasculation, what would you choose ? Would you choose the fate worse than death ? Well, of course I would. So this is a conflict between a sub-function, self-preservation, which has become autopoïetic, and its hierarchical superior and creator : reproduction. Now, for all this biological hardware that we have in our heads – we call it the brain – to have evolved, for the genes to have made these neurons, all this hardware must have given the self-preserving machines we were some kind of selective advantage, some kind of differential reproductory advantage. So I would make cognition a sub-function – direct or mediate. Perhaps not directly of self-preservation but certainly my answer to my own question seems to show that I prefer to go on thinking instead of reproducing. The two E.E.S. systems start interacting in some way and the result is the cognitive system. Initially it must have been non-autopoïetic but heteropoïetic, dependent, not autonomous. But at some point during evolution some kind of critical proportion must have been reached between the paleo-reptilian brain and the neo-simian cortex : between the set of hereditary data structures and programmes for self-preservation, reproduction, etc. and the new set of acquired, learned or aped data structures and procedures. Some kind of critical proportion must have been reached before the psychogenetic mass went autopoïetic and the prime function of a cognitive system became the maintaining of its prime function. I would hypothesise that this is linked with something like the beginning of consciousness. If we are radical constructivists then we have to put down the deep problems of psychology, what is consciousness, what is thinking, what is will, and free will, and we have to reframe them and to ask things like when did consciousness appear and through what shapes did it evolve in phylogenesis. It is a
– 193 –
phenomenon that is assembled somewhere during evolution and we are not going to postulate that the first self-reproducing schema was conscious, so it must have appeared somewhere in phylogenesis between then and now. Where, when, how ? It has to be assembled at some point, and it must be something like a software-hardware mutual selection. From this point we can have things like the indelicate paradox I was talking about and something like what I call the Archimedian paradox. They tell us Archimedes was beheaded during the fall of Syracuse – Alexandria would have been better because I could have had sociogenesis – libraries – but history does not always do what it should. He was beheaded during the fall of Syracuse, and what was he doing ? Well he was discovering the « principe d’Archimede ». Mixing my metaphors, he was pursuing his idea instead of fleeing his pursuer – this is a biological paradox, isn’t it ? He was pursuing his idea, he was making cognition the prime function instead of making self-preservation the prime function of the cognitive system. There is a moral to this history, he was a military engineer…
Now back to Piaget, shall we ? In this context I can now reframe what I think is a very deep question of Piaget’s in his equilibration theory. I can do it in what to me are more simple terms, but only by going back to the biological model. In Darwinian, in the synthetic theory – the modern one – we believe we know explicitly what is the prime function of the autopoïetic system. Now the prime function is differential reproduction and it has what cyberneticists call an evaluation function which tells it what is progress, what is better – let us not be anthropomorphic : in purely biological terms, biological progress is just reproducing more. Genetic systems work by a sort of signal detection on gene frequency modulation and they are set to extremalise relative frequency. When relative frequency of a gene goes higher the gene is « better », whatever that means – so Don Giovanni was not quite wrong though he is often criticised for putting the means before the ends. In the
– 194 –
genetic system we have an explicit evaluation function differential reproduction clearly linked to the prime function. In the homeostasis of organisms we also have an implicit evaluation function for self-preservation. Now, for sociogenetic systems, what is progress ? What is the evaluation function ? And for psychogenetic systems ? Precisely, Piaget’s central problem in equilibration is that equilibration is « majorante ». Very well. This simultaneously raises two questions – what is the prime function of a cognitive system and what evaluation function does it have as a means to compute whether this prime function is being satisfied. This is the main reason why I would say that Piaget tried to do what Darwin did : Piaget’s equilibration theory is to psychogenesis what Darwin’s equilibration theory is to phylogenesis. An isomorphism between what Piaget tries to do and what Darwin tries to do. It is also the reason why I believe Piaget’s tertium quid between Lamarckism and Darwinism is an answer to a question his theory has already answered : his equilibration theory is already isomorphic to the synthetic theory of evolution.
Since I have very little time left I would like to comment, as a conclusion, on some ideas of Apostel’s that I found very exciting for the research programme of genetic epistemology. The first one is (I don’t remember how it was formulated, but I’ll reframe it) : Is sociogenetic evolution rational in just the same way as psychogenetic evolution is rational ? I think this has to do with rational in the homo economicus sense, in Pareto’s sense. Rational is in Pareto’s sense the proper means towards an end. And when the means are not proper they are irrational. Rational here does have something to do with what logicians mean but, by a formidable detour which Apostel has made in one of his recent presentations at the Centre : What is inference, what are theories, what are deductive systems about ? They are functional in some sense. Clearly inference is a means toward some cognitive end. Now a point here is this one : we should not expect synchronic sociogenetic theory formation,
– 195 –
individual theory formation and historical diachronic theory formation to fulfill the same functions, and even to have the same mechanisms or produce the same knowledge structures. I would distinguish the synchronic and diachronic aspect of sociogenesis. I think Apostel was actually talking about the synchronic aspect, « la republique des esprits » : the set of professionals who are interacting and exchanging information in this generation. There is also this diachronic aspect of transmitting organised knowledge to the next generation : one is cooperative, the other is perhaps more imperative, more dogmatic… we have to programme our sons so that they can start standing on the giant’s head. They may be dwarfs but they will see more because they will be standing on this sociogenetic giant which is the set of knowledge that has accumulated and been integrated until now, and that is deeply functional. You clearly find this distinction in Piaget : preservation and transmission of acquired knowledge as opposed to creation of new knowledge. I would say of the prime functions of this kind of theory formation and of the two kinds of transmission – synchronic and diachronic – that we should expect them to be different from the kind of function an individual cognitive system has to fulfill for the individual, and one would expect them to produce different means for the same end. Even if the functions were the same, we should expect them to produce different structures for the same means. By distinguishing different epistemic machines with non identical functions we get a new range of concepts that we can start playing with, which illuminate epistemic genesis from a different angle. For instance, one of Papert’s observations had to do with this interplay of three problem solving systems working on different time scales. Papert once said that if equilibration goes to this final beautiful formal stage it is in some sense not such a miracle because the social environment is programmed to give the cognitive system proper hints, coups de pouce, to channel it into creodes that lead into this final stage. We can now generalise this
– 196 –
idea; it is only if you take equilibration in isolation and you consider the other two epistemic machines to be factors and you forget that when you look at one of them history does not stop for the others that it becomes such a miracle that all children converge so elegantly into an adaptive singularity. It is not such a miracle. It is even less of a miracle if you introduce the biogenetic, the phylogenetic factor into the interaction net. A genetically programmed sequence of differential maturation would be an excellent heuristic for factoring the problem of acquiring complex knowledge by opening limited problem spaces at the right time and in the right sequence. So if we want to teach, we have to find out what the hints are in our social programmes – institutionalised hints. And we also have to try to find out what biological hints our development gives us, what maturation means in terms of helping an autopoïetic or equilibrative system to solve the problems of learning to learn.
Another consequence of this distinction has to do with 1ogico-mathematical structures and the epistemic subject. As you know, Piaget explicitly takes as a « mobile referential » to characterise and trace the trajectory of psychogenesis, part of the very epistemic space in which it unfolds : the conceptual framework of inter-individual science synchronic to psychogenesis. Seen through this framework, the epistemic subject becomes in some sense a structural invariant over stage-equivalent psychological subjects, and last level intra-individual logico-mathematical « natural » structures converge onto first level inter-individual mathematical formal ones. Our distinction suggests we take a closer look at this articulation. Natural intersubjective structures, common-sense knowledge (common in the communitas derivation), may then appear as an organising framework, chanelling psychogenetic reconstruction through functional categories and pathways, with specialised scientific knowledge appearing initially in some individuals as a « metaframe », a reorganiser and renovator reinterpreting already acquired psychogenic structures. This might answer
– 197 –
the strong subjective discomfort I have always had with the « sujet epistémique » : the sheer infernal competence of this theoretic twelve year old has always been hard to stomach -at my current levels of performance.
Common-sense knowledge is practised more than it is preached, and this both points at a possible difference in nature between psychogenetic structures and those of science, and at an important common character of knowledge representation in different epistemic systems. Piaget has just invented a new joke about the vexed problem of whether structures exist « in the child’s head » or only in the psychologist’s. We should, he says, put the same question to physiologists who observe children’s « physiological functions » and ask them : are the corresponding organs you postulate really in the children’s bodies or « only in your heads » ? Or perhaps both ? Now, we survive, think and eventually reproduce on a formidable fund of metabolic and neurovegetative know-how of carbon chemistry, while our neurons seem to have some pretty good notions about computability and physics as far as balance kinematics, acoustics and optics, etc. are concerned. The organs in the physiologists’ heads that Piaget’s joke points at, are remarkable embodiments of procedural knowledge. Indeed Piaget’s difficulties with the status of sensori-motor space : is it a grouping (structural knowledge ?) or a scheme (procedural knowledge ?) – which he finally decides on – show he has been well aware for many years (procedurally ?) of the distinction. A good deal of psychogenetic structure may well be in the child’s head but in the implicit, non-articulate form of procedural knowledge as opposed to articulate declarative scientific theory. How to navigate versus vector composition.
Declarative, propositional, symbolic, « knowledge that » (in Ryle’s sense) and its systems of notation, formal systems, writing, even language itself may be avery late emerging post-simian instrument of symbolisation. And Piaget’s old problem in psychology, as well as that
– 198 –
of the new computational psychology, and in fact of all science when it theoretically constructs technological practice, is to translate a formidable amount of phylo-, socio- and psychogenic procedural knowledge that is coded in largerly undecipherable codes – the neuro-physiological code is uncracked, and procedures are, to say the least, not immediately legible in the genetic one – into this tiny portion of declarative symbology which is all we have for scientific communication. What is worse, it may not even be adequate, in the sense Henriques mentioned yesterday, for representing procedural knowledge. There may be hope in higher order logics, but then we get general procedural undecidability in exchange, embedded in the logic itself. We can dodge this perhaps by using procedural languages directly, but then we just preserve and transfer declarative obscurity. This is David Leiser’s paradox, the one his thesis pushes annoyingly back at me : suppose you have a complete procedural simulation of yourself, then you have to call in Piaget on artificial psychologies to explain it, because the procedural artificial entity will be obscure to you, as obscure as your own intrinsic procedural Doppelganger is to your declarative one.