The conceptual foundations of decision-making in a democracy

[p. 51]

Part Two B
Information.

2b.1) What Is Information?

Everybody ‘knows’ what information is. Yet in its appendix a nice little Dutch book on information (v.Peursen, Bertels, Nauta: Informatie) mentions thirteen definitions by various authors, all quite different, all contradictory and all dealing with only one of its many facets. The book itself plunges into its subject without declaring for any of them. In his book ‘Informatie en Energie’, p 61, H. van Praag defines his subject as ‘Information is knowledge that has been communicated’, but does not bother to specify what knowledge is, so that the only difference between the two is whether it has been communicated or not. He misses the major point, namely what the ‘thing’ is that has been communicated. Definitions in themselves may not be all that important, but if authors of this stature make such statements, if there is no agreement at all on the definition of a subject, that usually is a signal that something is wrong with the way the subject is perceived and tackled. And so it is.

All writings I have found on the nature of information take as point of departure the conscious human thought. It descends to lower levels of information processing only as far as required by the theory defended by the author. Conscious human thought is the highest and most complex manifestation of information. If we want to understand a phenomenon, common sense dictates that we should not start with the most complex example we can find, but with the simplest one. That is the approach taken in this book.

It turns out that every process of life involves an information process which, if successful, produces knowledge. The next paragraph deals with one of the simplest examples, the regulation by a bacterium of the production of an enzyme required for the ingestion of sugar. To fulfil its function of biasing physical-chemical processes to increase the viability, the neguentropy, of living beings at the expense of their environment, such processes must be regulated according to the state of this environment. To that end they are coupled to some sort of switch which determines whether to start or stop the process, and that switch is set ‘off’ or ‘on’ by some event which is used as an indication as to the propitiousness of the state of the environment. The possibility to set a switch according to some criterion enables the creature to choose whether to start or stop the process; making such a choice is a decision.

The major characteristic of any switch is that it can set an electro-mechanical or chemical process ‘on’ or ‘off’ without being itself part of the process which it regulates. This independency from the process controlled by it is the major feature of biological switches. The state of the weather may induce me to take a walk, but there is no direct connection between the two: even if weather is fine, I usually do something else. If the choice is independent of the process it controls, it cannot be directed by that process. But to be useful, it also cannot be totally random. It must therefore be determined by a criterion inherent in the entity which manipulates the switch, it implies the notion of a point of view, a purpose, an objective of a subject.

[p. 52]

The setting of the switch starts with the identification and registration of specific events which have been selected as an indicator of the state of the environment and used as criteria for operating the switch. How did these criteria materialise? One thing is certain: if they are to direct decisions, these criteria must have been selected prior to taking a specific decision. They must be contained in some form of memory, for instance the genes. They are the product of evolution and the ultimate criterion is the expected contribution of the alternatives (the possible settings of the switch) to the survival and reproduction of (the information, the ‘order’ inherent in) the living system making the choice. The logic of this evolution is the usual one: a large number of confrontations with various events will progressively eliminate beings tending to set the switch in the ‘wrong’ position, leaving the field to those who are set in the ‘right’ one. As previously explained, the functionality of the criterion emerges ex post, it need not have directed past choices but been ‘remembered’, preserved, because of its favourable results.

Notice that a switch can be useful only if the environment in which the creature lives is subject to change. For if it were not, then the relative advantageousness of the alternatives would never change. We would once and for all opt for the best alternative; from there on, the ability of choosing could only produce negative results.

To take a decision that is not arbitrary but conforms to a criterion, a living being must have more than just the ability to choose. It must ‘know’:

a)	what - given a specific state of the environment - is the correct setting of the switch. As said, the criterion setting the switch in response to a certain event is part of the creature and antedates the current decision-making process and has evolved as the result of previous decisions
b)	what is the current state of the environment in terms of the relevant factors. Knowledge about the state of the environment is obtained by registering and interpreting certain impulses from the outside world.

A living system is constantly subjected to all kinds of impulses from the world around it; the same applies to inert systems. The difference between the two is that if an impulse affects an inert system, the reaction of that system is in principle predictable and determined by the laws of physics and chemistry. Also, the impulse is usually part of the reaction it generates.

When a living system sets a switch after registering an impulse, the ensuing reactions cannot be thus determined because the impulses are not part of the reaction which they direct. We may know the physical-chemical properties of the impulse and of the living creature subjected to it. But from these properties, we cannot deduce the ensuing reaction on basis of the laws of physics and chemistry alone. Nor will the reaction to the same impulse necessarily be constant over time and over various individuals of the same basic physical-chemical composition. Whenever impulses trigger an action by a living being, they have been endowed by it with a property which these impulses do not have by themselves, namely meaning, significance, a ‘value’ in terms of the objective of the living being. That property is not part of physics and chemistry.

[p. 53]

We may use one and the same kind of object or process both for its direct effect and as an information carrier to obtain knowledge about our surroundings. We use light - rays of sun - to warm ourselves or to read a book; their value then can be deduced from their physical-chemical properties which are independent from the user. We also use it through our sense of vision. In that last case, we give these same rays of light something which they do not possess by themselves, namely an interpretation of the properties of the last point of departure of these rays: the position, shape and colour, the nature of the object from which they came, in terms important to us.

Information is that set of impulses from the outside world to which a living creature has assigned some meaning, or at least has singled out for assigning meaning to it. Any matter or energy can become information if it is assigned a meaning by a living creature, and nothing is information until then.

Try to define what food is. You will see that you can only do so in relation to some creature that can eat. And you will find that there is no physical-chemical property which can be used as a generally applicable criterion for determining if a certain material is food or not: what is food to one creature is indigestible to another. OK, you might say, but once we have specified human beings, we can agree that bread is food for all human beings. Is it? Babies cannot yet eat it. And what about people severely disliking bread? Or never having been introduced to it as food? So the decision to accept a material as food contains a subjective element. Yet food differs from information. For whether a certain material could be suitable for food for a specific being depends at least partly on its specific physical-chemical properties; we can assign a probability to the possibility that a certain material becomes food for a living being, namely if analysis shows that it has a sufficiently high content of substances which the living being needs, such as proteins, in a form which it can digest.

Not so with information. Any item that can have discriminatory power and therefore is not uniformly spread through our universe, can be a carrier of information. Whether it contains information can be determined only by establishing that there are beings using it as such. Take a gene; unless we know its function in recording sequences of amino acids, a gene is just a random assembly of molecules showing no detectable regularity or function. Any arrangement of pebbles of various shapes or colour which seems totally random to the uninitiated can in fact contain a message once a meaning has been assigned to certain arrangements.

The physical-chemical properties of an item (a ray of light, a row of knots on a rope, a column of smoke, an electrical current, a streak of graphite) can tell us only about its adequacy as a carrier of information; they can never tell us if it actually contains information. To equate order and neguentropy with information provides no escape. Take the sequence of the amino acids in a gene: at first glance it looks totally arbitrary, seems an example of perfect entropy. It is totally meaningless, except as a code to a very specific creature: it gets its significance in, and only in, the actual development, the morphogenesis, of this creature. The link between the order of these amino acids and their significance cannot in any way be deduced without seeing the creature grow and function.

[p. 54]

The medium of an information process is not intended to provide knowledge about itself, but about some other feature of our environment; that is precisely why it is a medium, and not an object of the process. It can fulfil its function of medium only if its physical and chemical properties are unrelated to the information function in which it is a medium. Only scientists examining the way in which our eyesight functions are interested in the momentum of photons. Whenever a creature uses its eyes, it just wants to know about the objects it looks at.

The detail in which we can ‘see’ objects is determined both by the variety of wave-length which we can distinguish and by the number of ‘points’ on our retina registering the impact and wavelength of rays of light. If the ‘image’ we build on receiving these impulses is to reflect the properties of the object, the state of the medium (the wave-length and intensity of the rays of light) then the construction of the representation must be determined by the (light-reflecting) properties of the OBJECT we look at and therefore must be as independent as possible of the physical-chemical reactions occurring in the observing SUBJECT when the rays strike its eyes. Otherwise the subject could never know whether he ‘sees’ the object or just himself. This is the ‘real’ foundation of the notion and the requirement of OBJECTIVITY presented further on. The full richness of the genetic code (all possible combinations of strings of twenty three-letter words) is available for its information processing function precisely because it does not rely for its significance on any direct physical-chemical connection between the order of the nucleic acids in the genes and the functions of the proteins they code for.

The theory of evolution holds that most, if not all, of the very complex species on earth today have emerged in a process where new variations arise by a chance process. That has led to erroneous conclusions.

‘Modern theory (wishing to confine itself to physical and chemical modes of operation) is left with nothing but chance as the originator of the particular structures of all the DNA molecules. Since anything can happen by chance - even the most improbable thing - these molecules may have happened into being by pure chance. But it would seem highly improbable. The chances that merely BY CHANCE they should have become arranged in the meaningful ways in which they are arranged are beyond much doubt less than the chances that a pile of rocks rolling down a hillside will arrange themselves by chance on a railway embankment in such a way as to spell out in English the name of the town where the embankment is.’ Thus writes Michael Polanyi who then concludes to the necessity of postulating some metaphysical principle to explain meaning and life. But his interpretation of ‘modern theory’ contains a fundamental error. For his argument holds true only if we assume that evolution had to produce the creatures we see in our world today. No modern evolutionist would ever state that the way living creatures are today is the only or best way they could have been: as said, perfection and optimum are not the language of a modern biologist. The order inherent in a living being is not one which was given a priori. The limited number of creatures living today consist of the minuscule portion which happen to survive and propagate out of the immense number of others which did not. Both are only a tiny fraction of all arrangements of molecules which can be generated by chance. The number of possible arrangements resulting in a viable creature is totally unknown, but certainly large enough to give a totally different perspective to its probability of occurring by chance. The correct way to put Polanyi's metaphor is not to ask about the probability that a hill slide

[p. 55]

will produce an arrangement of rocks which spells the name of the next town, but to estimate the probability that a rock slide will produce an arrangement which we would recognise as something familiar, as having some significance, some informative value. Anyone who likes to roam the countryside with open eyes will have encountered many such instances: a cloud which is shaped like an elephant, a tree reminding us of a man, a clump of grass we mistake for a rabbit. For a discussion of Polanyi see Volume Two, chapter ‘Meaning: Michael Polanyi’, p.259.

If rays of light, arrangements of genes or smears of graphite on paper have no information, no meaning by themselves, then how does information arise, where does it come from? There is just one answer which generates no inner contradictions and is in accordance with what we know about living creatures, including ourselves: the interpretation of the medium, the meaning assigned to rays of light, genes or letters has to exist prior to the generation of specific information. The information process must be in place, including the ‘meaning’ we will accord the impulses we register, before any representation about our environment can be made. The being which uses an information process creates its meaning, which is why any information process necessarily must contain a subjective element: the concept of totally objective information is a contradictio in terminis! Consequently, there can be no objective and cardinal way to measure information, as explained in more detail in Volume Two, chapter ‘More than Physics and Chemistry’, p. 255.

2b.2 An Illustration of the Subjectivity of Information.

There is no better way to illustrate the nature of the subjectivity of any information process than to show it at work in one of the simplest living creatures, the bacterium Escherichia coli regulating the intake of milk sugar for its ‘digestion’. The process of transforming milk sugar into energy is initiated by three enzymes, large proteins whose blueprint is encoded in the genes. We will deal with one of them, galactoside-permease. First we will present in more detail the elements of the process of life which are relevant to the subject of this book and concentrate on the information-processing aspect, especially on the role which genes play in this process.

The alphabet of life consists of four ‘letters’, four ‘signs’, four nucleotides:

-	(A)denine
-	(G)uanine
-	(C)ytosine
-	(T)hymine

All ‘words’ are formed by three of these ‘letters’. Each word identifies one of the 20 amino acids which are the building blocks of all matter, all proteins, from which living beings are formed. Twenty words - as shown in the chapter ‘The language of life’ - are quite enough. (Such a group of three nucleotides is also called a codon, which is a synonym for ‘word’.)

These ‘words’ are grouped in sequences, form ‘sentences’ called genes; each such gene has a specific function, codes for specific proteins. These genes are strung along one after the other in the immense chain of sentences that forms a chromosome which is a ‘book’ of the building

[p. 56]

plan of the being contained in the nucleus of the cell. The size and number of these chromosomes vary according to the sort.

Out of four ‘letters’ we can form 64 different groups of three letters. As there are only 20 amino acids, there are 44 codons left for other uses, such as ‘punctuation’ which indicates where the ‘phrase’ ends and the next starts. Examples of punctuation are TAA, TAG and TGA. Some amino acids have more than one ‘name’.

The chromosomes consist of two strings of nucleotides adjoining each other. Each nucleotide of one string ‘holds hands’ with a partner in the other. Three of the nucleotides (C, G, T) associate only with one partner. Guanine always associates with Cytosine, Thymine with Adenine. Adenine will also associate with (U)racile acid if it cannot find Thymine, its DNA partner. The chromosome then is a long sequence of pairs of nucleotides walking side by side. Because the side on which they walk also is a distinctive feature, there are four such pairs (genetic letters):

G	-	C
C	-	G
A	-	T
T	-	A

The process of constructing an amino acid uses only one side of the chromosome as a blueprint. For the purpose of carrying information, four different signs are just as good as four pairs of them. Then, why pairs? That is necessary for another function: the replication of the cell. This replication is initiated by the chromosomes splitting lengthwise, and thus making widowers out of every nucleotide. These look out for a new partner of the same kind they had before. In this way the original chromosome is perfectly duplicated by each of these halves, the left one reconstructing the right side and the right one the left side. This process works because the bond of a nucleotide with its partner is much weaker than the bond with its neighbour: the chain thus is easier to split lengthwise than across.

Below we will sketch a segment of a chromosome in order to illustrate the ‘logic’ of the process; it is not an accurate representation of an actual biological process. For instance, the chain is twisted along its longitudinal axis in a spiral; here it is shown as a straight line for the sake of clearness. Also, the space separating the words in this sketch is there only for illustrative purposes. The cell ‘knows’ a word always has three letters and thus does not need it to know where a ‘word’ ends.

Clearly the nucleotides of the blueprint cannot themselves be used in either the production of the protein or the actual transmission of the instructions to wherever these proteins are manufactured. To safeguard their role as record, as a blueprint, they must remain a totally unaltered part of the chromosome. The only way to achieve this is to use messengers. That is exactly what happens.

[p. 57]

In a fairly complex process, the record encoded in the gene is transcribed into other nucleic acids called messenger RNA, which form a string, again double. Upon encountering a code signifying the end of a phrase, this RNA string separates from the genes to carry its message to the appropriate location in the cell. Three DNA nucleotides (C, G and A) will be replicated in that transcription; the fourth, Thymine will be replaced by Uracil acid (U) which has an affinity to Adenine, just like Thymine had. These RNA nucleotides associate with their complements just as the DNA did. In this way an RNA-‘mirror’ of DNA is produced, and just as with a mirror, that image is inverted: the image is identical to the reverse side of DNA, except that we now find Uracil acid instead of Thymine; that distinguishes the messenger from the gene and at the same time preserves the message encoded in the gene.

Just as with DNA, we cannot by looking at RNA imagine what the ‘product’ to be made will look like; RNA as such has no apparent ‘meaning’. The translation of the code into the message, the ‘interpretation’ of the code, occurs when the freed messenger RNA gets attached to a ribosome, a kind of work station for the production of proteins. In a very complicated and as yet only partially understood process, this work station interprets the code into a code for an amino acid (for instance AAC into Leucine); it then fetches that amino acid and assembles all amino acids encoded in the ‘message’ into the required protein. After having done its job, the messenger RNA is freed to resume its task as courier, and the protein sets out to fulfil its function in the cell. Below is a sketch of this process of transferring information.

1) Creation of the messenger RNA.

2) Reading the message of which protein to produce.

[p. 58]

As you see, the code used to identify amino acids is the code inscribed on the ‘reverse’ face of the gene, except for T which has been replaced by U. There is no evidence of any ‘objective’ relation between this code and the amino acid to be produced: other codes can also do. Leu(cine) for instance is coded not only by the above UUG - AAC, but also by:

The cell produces two kinds of proteins:

-	large ones which form the building blocks of a cell
-	smaller ones called enzymes which fulfil the role of a regulator.

In front of the ‘text’ encoding the production of such a protein, for instance an enzyme, there arer two codons. The first, called a promoter, indicates that a new sentence describing the production of a protein or enzyme is about to start. The second, called an operator, ‘tells’ RNA to start assembling. Neither is part of the structural element of the phrase, and therefore has no meaning in terms of amino acids.

The decision-making process of whether or not to actually assemble RNA involves two elements, one of which is the operator. The other element is a continuously produced molecule which - in its unaltered state - will cling to the operator who will become unrecognisable to RNA and thus will not start the assembly of RNA. This molecule is appropriately called an inhibitor. That inhibitor also has an affinity to a specific other molecule, milk-sugar in the example used further on, and bonds to it. If bonded to another molecule, it will become unrecognisable to the operator, it will not bond to it, and RNA will be assembled. The operator thus forms a biological switch because it can have two states: ‘free’ or ‘bonded-to-an-inhibitor’, which determine whether a specific action is taken or not and thus is the first part of decision-making.

The other part is the information process which directs the setting of the switch. In the example of the regulation of the transformation of milk sugar into energy, that regulation is initiated by the enzyme permease. This enzyme permits sugar molecules to enter the cell through its normally impermeable protective skin. It should be produced only if there is sugar to digest. The object of the information process is the presence of milk sugar. The information process is initiated by the production of inhibitors whose property is that they can bond both to the operator concerned and to sugar. If bonded to sugar, they lose their ability to bond to the operator and the enzyme is produced. If they do not encounter sugar before reaching an operator they will bond to it, RNA will not recognise the operator as such and no permease will be produced. This process is functioning because the (Brownsian) movements and the diffusion processes in the cell ensure that if no significant amount of sugar is present, many ‘free’ inhibitors will meet operators, while if there are sugar molecules, inhibitors will bound to them.

Inhibitors are not very stable; after some two minutes they decay and release the operator which can then resume its function in case milk sugar has arrived in the mean time. If no milk

[p. 59]

sugar has appeared in exploitable quantity, new inhibitors will be around to bond to the operator. The process works because of the constant and high rate at which the regulator gene produces inhibitors.¹⁾

Summing up:

-	the gene coding for a protein is preceded by an operator
-	Messenger RNA assembles a message only after recognizing that operator
-	A regulator continuously produces inhibitors bonding to operators and to sugar
-	If bonded to sugar, it will not bond to operators
-	an operator to whom an inhibitor is bonded will not be recognized by RNA

The operator serves as a switch and gives meaning to the inhibitor which serves as a medium in the information process setting the switch and whose production initiates this information process.

This complex process is vital. The production and maintenance of the proteins (enzymes) required to break down the sugar molecule and transform it into energy is just a chemical process which consumes energy. Without the information process and the biological switch, the bacteria would continue production of such enzymes even in the absence of accessible sugar. That is a very likely event, for by consuming sugar the bacteria will reduce its availability in its direct environment. In the absence of sugar the energy balance of the process would be negative. The whole process is capable of running against the second law of thermodynamics only by the above information process. Clearly, if it is to work, this information process must save more energy than it consumes which is possible because it is not part of the process itself.

The presence or absence of milk sugar is not information; it is just a fact, it is part of the immense and infinite universe about which a living being might obtain some information if it wanted to. Attempting to know everything would not contribute to life. On the contrary, it would annihilate any advantage of the information process, because obtaining information always uses some energy, for instance the production of inhibitors. The efficiency of the information process resides precisely in its selectivity, namely that it only looks for information that might at least potentially be useful; it is biased towards (the purpose of) the subject, an attribute which we properly describe by the word ‘subjective’.

1) Because a stochastic process - the Brownsian movement - is involved, a few molecules of permease will be produced - be it at very low rate - even in the absence of sugar molecules in the interval between the decay of the old inhibitor and the meeting of the operator with a fresh one. This ‘imperfection’ explains something that in the beginning puzzled me. For the sugar whose concentration should actually determine the process is the one outside the cell. Now this cell is enclosed by a membrane which keeps most things out, amongst other milk sugar, unless it contains molecules of lactose permease which ensures the transfer across this ‘border’. If the inhibitors would do their job 100% and kept all sugar out, no permease would be generated, and no sugar could enter the cell to stop the inhibitors, however dense its concentration outside may be. As it is, a few molecules of permease will always be generated, permitting the skin of the molecule to let in sugar molecules in a certain (nonlinear) relation to the concentration of sugar outside. If the concentration is high, sufficient sugar molecules will be let in to start the cumulative process of stopping the inhibitors. The bacteria will then produce permease at an increasing rate, thus letting in more sugar, which speeds up the process by neutralising still more inhibitors etc. until full production rate is achieved.

[p. 60]

The very first step in the generation of information is the need to know, the question to be answered. It is this question which determines what will be information, the choice of media and meaning we will attach to their state. In case of the above cell, asking the question is done by the operator who also determines the meaning of the inhibitor. The ability of the inhibitor protein to take on two different shapes, depending on the presence or absence of milk sugar and which can be ‘registered’ by the operator, is the property that makes it a carrier of choice for information. The two shapes it can take, free or bonded, are the potential content of the message, the information.

The carrier of information, the inhibitor, becomes part of an information process only in the presence of a recipient (the operator) which gives it meaning, significance. Only to the operator does a free inhibitor say ‘no milk sugar’, and that is the only information transmitted in the process. The operator cannot distinguish between a bonded inhibitor and any other molecule; all mean ‘no information’, have no meaning. As no living being can know everything about everything, any information process about the world outside must contain the possibility of ‘no information’.

Suppose the gene which produces inhibitors is damaged, but that RNA is still assembled along this gene and codes for the production of some protein. That protein will have a structure that is different from the structure of the original inhibitor, and would therefore not bond to the operator, even if devoid of sugar. If there is no other gene which ‘recognises’ this new type of protein, it would have no ‘meaning’ at all for the cell. There is no transmission of information, in fact there is no information. That is the general ‘default value’ in information systems of living creatures. It usually codes for a decision which minimizes the probability of a lethal consequence.

We might also imagine that this new protein would, by pure chance, fit some other operator gene, would bond to it, and consequently switch it off. Then information has been generated, but it did not give the correct message: it was a misinformation, a lie. Yet it was information. A misinformation can and very often is lethal, for instance by preventing the production of an essential molecule.

The above illustrates that information is generated only if a living being has attached some value - which you might call meaning or significance - to some impulse or matter, to some object or event. It is this activity - giving significance - which is at the root of every information process and which creates the real difference between information and other phenomena in our universe. While the notions of an a priori meaning and the motive-directed quest for information have been noted before (Kant and Habermas), only the modem theory of life has provided the exact basis for it and exposed its commonality to all living creatures.

In that most simple of information processes, registering the information and setting the switch is one and the same event: the bonding of an operator to an inhibitor. At the level of the bacteria the ‘decision’ is built into the genetic memory of the bacteria. With higher-level beings there is a whole and often very complex (decision-making) process between the registration of the message and the decision. But the principles remain the same.

[p. 61]

As stated, information systems are never perfect. Even if sugar molecules are in the cell, inhibitors may miss them, bond to the operator and produce the misinformation: ‘no sugar’. RNA might encounter an operator whose inhibitor has just disintegrated before another one has reached it. This also generates misinformation, namely that there is sugar. Even in such a simple process the information generated depends partially on the subject. Truth or falsity of the message (there is sugar or there is not) is a fact, but that truth is not known to the subject. Establishing the truth requires a separate information process, which would however be subject to the same limitations, namely that we can never rule out the possibility that the process was defective. To decide about that would require a third one etc. There is no absolutely objective way to establish the truth of information which therefore can never be known with total certainty.

The inevitable subjective elements in an information process (the selection of what to look for, of the process and of meaning) are not by themselves a liability. Only subjective elements which are not required for that process are, because they can produce misinformation. They are always pernicious in an information process aimed at establishing the facts required for a decision. (See part 4a.3: Subjective/conventional is not synonymous with arbitrary, capricious, random, irrational)

2b.3) Subject, Object and the Notion of ‘I’.

In any specific information process we can always - at least conceptually and usually also physically - distinguish two elements: the entity looking for, registering and interpreting information, and the entity about which information is sought.

1)	SUBJECT is the entity looking for, registering and interpreting information
2)	OBJECT is the entity about which information is sought.

In the context of the information process, ‘subject’ refers exclusively to the place in an information process of the entity to which it refers and therefore implies no personification, no anthropomorphism. If a cat and a mouse evaluate each other's nature as a meal or as a danger to be avoided, both are in turn subject and object.

Notice that the distinction between subject and object has any meaning, any ‘empirical content’, exclusively in an information process. If a stone falls on a mouse and crushes it, we would not call the stone a subject and a mouse an object. Rather, the misfortune occurred because an information process failed, namely the one which should have alerted the mouse to the falling stone. In that information process, the mouse would have been a subject which attaches meaning and a very negative value to the event. The stone can never become a subject. There are no information processes in the inert world.

As will be explained in a later chapter, life and information are processes with a specific property: they can be adequately described only by complex, interdependent and dynamic models of learning elements. In such circular systems, we usually cannot a priori assign any element as the primary one from which the others can be derived. The notions of subject, object, meaning and knowledge are determined by their place in the process considered and imply at best a (chrono-)logical order, never a hierarchical one.

[p. 62]

The milk sugar process and the place of DNA in it were chosen to show that these notions are general elements of any information process and are present in even the most primitive forms of life. Besides showing the extent to which the information processes of man and bacteria are similar in their logical structure and functional nature, we must also determine how they differ and what is their place in the evolutionary scheme from viras to man.

1) The subject. A subject gives significance to an event because that is functional, because it needs (or at least wants to make) a representation of some aspect of the world around it, in our example the absence (and by default the assumed presence) of sugar. Who is the subject? One of the major controversies of our time - about the duality of body and mind - finds its origin partly in the failure to define exactly who the subject is. In the case of the milk sugar system, a ready - but erroneous - answer would be: the cell. Erroneous because the subject here clearly is ‘part of the gene’ namely the pieces of chromosome regulating the production of the enzyme. It is the gene containing the operator, not the whole cell, which gives significance to a free inhibitor. Many information processes in a cell are not about the world outside the living being, but consist of one part of itself acting as a subject and requiring information about the state other parts which then have become an object in this specific process, but which often will be subjects in another information process.

Yet we tend to see the whole cell as the subject. Indeed, the milk sugar process has any function only as part of the bacterium. Living creatures form a ‘whole’ which is dependent for its survival on the function of many processes organised in an interrelated system. To achieve its purpose, to be functional, the objective of all these individual processes must be congruent with the objective of the whole creature. What is relevant for a part of a cell such as a gene will in the end also be relevant for the cell as a whole. But the relevance of the message ‘no sugar’ to the whole cell is a second-hand one: it becomes relevant to the rest of the cell because the activity of the real subject (the permease-producing gene) is relevant to the rest of the cell. The only interest of the rest-of-the-cell is the timely production of the enzymes it requires, not the details of how the gene regulates its production. Therefore the specifics of the information process, in this case the state of the inhibitor, need not have any significance for any part of the cell other than for the genes involved in the production of these enzymes. These genes then are the only real ‘subjects’. The relation between various information processes will be further explored in the paragraph ‘The concept of “I”.’

That the gene with the operator is part of the subject is evident. What is the status of the gene-producing inhibitors, is it also (part of) the subject? If so, is one of them more important than the other? In the previous paragraph we have introduced the notion of a misinformation: a damaged inhibitor which is not recognised by the operator for which it is intended, but is recognised by some other operator which would then be inhibited in its function. Three other examples of misinformation will help us to determine the ‘real’ subject.

a) The damage to the inhibitor-producing gene is such that it produces an inhibitor which still has a form that bonds to the operator, but will not combine with an available sugar molecule to annulate this affinity. The operator would then continuously read ‘no sugar’, even if there sugar present. Information then is still produced, but it is misinformation.

[p. 63]

b) The inhibitor-producing gene would stop producing inhibitors or would make inhibitors which never cling to any operator even if devoid of sugar. In short, it would cease to be part of the information process. The operator on the other hand continues to give meaning, namely ‘no sugar’. Information is being generated, but it would not fulfil its function.

c) The function of operator would disappear, while the inhibitors would continue to be produced. Nothing the inhibitor-producing gene can do would then affect the process. The element of choice, the switch itself, has ceased to exist; only the chemical process of producing enzymes remains but involves no information process, functioning or not. Clearly, giving value, meaning, significance, is the primary role in the generation of information. It is performed by the subject, not the mechanics or media of the process, for there is no direct connection between the physical nature of the event and the significance which is given to it. The protein, electric current, sound wave, blot of ink or ray of light to which significance is attached fulfils only the role of postman, a role it obtained because of its proven or expected suitability for that function. By itself, it contains no information.

2) The object. Besides a subject and a medium, information supposes an object; it must be about ‘something’. The object of information is that part of it's environment about which the subject seeks information. That might be sugar, the state of an adjoining gene, the mood of a friend, the nature of a quasar, the logic of a theory or what I said two minutes ago. The difference between subject and object again is unrelated to the physical nature of the entities concerned. What makes a subject a subject and an object an object is exclusively their relative place and function in an information process and can only be defined within this specific information process. The object becomes an object only because a subject is interested in it. The function which the information process fulfils in the being concerned creates both the notion of subject and of object and determines which is what.

3) The concept of ‘I’. There are no living beings surviving on just one information process. To live, a being requires the cooperation of many functional entities which each perform a specialised task in the struggle against chaos and are regulated in that function by one or many separate but often interrelated information processes. That cooperation requires coordination.

The information processes of bacteria are aimed at very specific decisions and actions and are directly related to them. The coordination between the various processes is achieved by ensuring that all information/action units form a ‘whole’. The unifying criterion - the overall point of view for deciding what is adequate and what is not - is the survival and propagation of this aggregate, of the process which generated the whole bacteria. If we indulge in humanising bacteria and think about what would correspond to the notion of ‘I’ as used in common language, it would be whatever entity holds that overall point of view. In a bacterium there is no such entity; the ‘I’ of a bacterium can exist only in the eye of an observer and is created by him. All functions and actions of bacteria are encoded in the genes and their apparent coordination is the result of selection.

If a being is endowed with various action organs and each of them is capable of a wide range of activities, then the number of possible actions becomes extremely large and the overall

[p. 64]

coordinating criterion for choosing a specific action must then become correspondingly general. Animals have instincts which for instance might prompt them to take the least conspicuous route. It takes thousands of different decisions to determine in a specific case what is that least conspicuous route and to make all the necessary moves. Only a powerful separate information processor such as a central nervous system, a brain, can provide the required coordination. With animals, the instincts generating that coordination have also evolved through selection. The notion of a central, overall ‘I’ supervising the coordination, the ‘wholeness’ of the animal's decision making probably is totally foreign to it and again will exist only in the mind of a human observer.

With man, whose central nervous system includes the faculty of conscious, rational thinking and of imagination, the number of alternatives becomes - for all practical purposes - limitless. Any instinct or learned behaviour, all selection criteria derived from past experience, can be overruled by becoming conscious of them and by replacing them with products of our imagination. There are no physical or chemical laws which put any limit on the kind of alternatives for action which we can imagine. The overall point of view, the overall coordination function must be applicable to all conceivable circumstances, and therefore must be as ‘abstract’ as possible. Being conscious of our imagination and thinking, we must also be conscious and absolutely clear about the final entity whose objectives are to be served by the actions considered. The notion of ‘me’ which we humans have fulfils that requirement. Beings endowed with reflection and imagination must ipso facto have a coordinating function similar to the human ‘I’.

‘I’ is a function fulfilling that need. How it works is still largely unknown. But being a function, it need not be tied to a specific group of cells, only to a specific kind of cells which have the capability (the required neuronal ‘wiring’) to fulfil the function. It could be fulfilled by any part of our brain which is currently available; it does not by itself require a specific organ, need not reside in a part of our brain exclusively reserved for it. That solves a lot of paradoxes and problems which have bedevilled philosophers. In Volume two I have devoted the chapter ‘Body and Mind’, p. 361, to that subject.²⁾

2) The little I know about this subject points to the prevalence of neuronal connections, linkages, cross-wiring between neurons directing various action and information processes. To the extent that such a linkage can be switched on and off according to the ‘representation’ which one of the linked entities has made, such linkages are information processes in their own right. The whole information process then becomes an highly complex and partially hierarchic system of interrelated subsystems with various degrees of - always only partial - autonomy. It would seem to me that for any being endowed with a central nervous system, these processes can only be adequately apprehended in a close cooperation of neuro-physiology, which maps the physico-chemical entities and processes, with psychology which investigates how these processes generate representations and use them to produce decisions. My intuition is that the status of empirical science claimed by psychology should be made dependent on the extent to which the macro-concepts of psychology are explained as aggregates of neuro-physiological micro-processes ultimately grounded in the laws of physics and chemistry, which does not imply that they can be reduced to them (see Volume Two, 2a.9 B, p. 281, ‘More than physics and chemistry’).

[p. 65]

It will be clear that the notion of ‘me’ is not a synonym of individual. ‘INDIVIDUAL’ identifies a certain entity as a living creature having a clear boundary with the outside world and being self-supporting. The ‘I’ refers to an entity which gives value to the whole information process of that individual, and which - if applied to other creatures than man - exists only as an anthropomorphic metaphor or speculation. In any case, it has no connotation of ‘selfish’.

Equating ‘subjective’ to ‘random’ is an error which has arisen somewhere in antiquity and has ever since haunted philosophy and thwarts the acceptance of the evident subjectivity and functionality of any information process. We will expand on this subject in PART FOUR, chapter 4a.3, p. 133. It is this functionality to which we refer when we say that a message makes sense.³⁾

3) The functional theory of information is not an explanation of its origin. The schema of an information process, always involving a subject obtaining information about an object, is - as are all theories - a heuristic device to deal with reality, and it becomes fuzzy along the boundaries of the area of reality which it is intended to explain, in this case life. Take the emergence of the process of life and thus of information and the first recording of information in the first genes. The record formed by nucleic acids in the genes records the sequence of production of amino acids required to (re)produce a certain living being. What is the subject who needs this information? It could be the being which is to reproduce itself or the offspring which attempts to reconstruct its parent. Both seem rather far fetched. What is said in this book about information emphatically is not intended as any explanation of the origin of that phenomenon. This does not invalidate the schema presented in this book: the fact that it does not explain the origin of life and information does not disqualify it as a tool for dealing with life and information in the form in which we experience them today.

2b.4) Communication.

In the field of information, the word communication should be applied only to the transmission of information from one living being to another. To achieve communication, the message must have significance both for the sender and the receiver. Communication always requires a medium, a means for transmitting it. In all sensory information processing, also reading etc., that is some physical entity. It could be inert: sound, light or a string of pebbles. It could be another living being, but only if this living being is used as a device for transmitting information he has received, and not in its capacity of processing it. If the messenger must understand the message before transmitting it to the receiver, for instance because he must translate it, if we use him as an information processing entity, that involves two distinct communication processes, one from sender to messenger, and the other form messenger to receiver.

Communication always contains two value-giving moments, both subjective: what the sender intends to convey with the message, and how the receiver interprets it. Both are contents of the message and both are different… except to the extent that they ‘overlap’! The information content of the medium alone is zero, or rather, it cannot even be defined.

[p. 66]

The semblance of an autonomous content of a message is generated by the ‘overlap’ of the interpretations of the message by the various people involved. This overlap is neither spurious nor arbitrary, but has been carefully cultivated: the bulk of our education is devoted to achieving it. The education starts by interaction between parent and baby, long before school, and continues until we die. Our efforts to achieve such an overlap are so much at the core of our social life that it has been used as an illustration of the theory of the psyche of man as being primarily an imitator. It has given birth to a (contestable) philosophy which explains the whole process of thinking in terms of language. This overlap is the precondition for the whole cultural adaptation that makes the human phenomenon so unique. The tendency to assign an objective, autonomous content to our books etc., is the best possible testimony to our success, but nonetheless it is a mirage.

It must be clear that ‘overlap’ is not here used in the exact meaning which it has in topography; it cannot be quantified. It refers to the extent to which the participants in the communication process refer to the same phenomena of their environment and their mind. Studying how to define that overlap and how we achieve it is far more fruitful than the conclusion which has most impressed the philosophers of language, namely that this overlap never can be perfect. In the light of what has been said about information, we would neither require nor expect it to be.

Another factor responsible for the illusion of an autonomous content of a message is that it usually deals with some object which does have a real, independent existence. It is then only human to confound the representation we make on receipt of the message with the object about which we make this representation; human but nonetheless a mistake. Suppose I call my friend and ask him: remember my dining table? If I have changed my dining table some months ago, it may well be that he remembers the old, square one, while I am thinking of my new, round table. What would the autonomous content of the message? Even if we were thinking about the same table, we are bound to make representations which differ in many aspects. One memory is better than another, we select different aspects of the object to remember, our eyes differ in accuracy and sensitivity to colour, and we probably will not have looked at it at the same time and from the same angle. However many details we tell each other, in our minds we will never generate exactly the same representation of this table.

There exists no separate, independent ‘sense’ of a table, only the actual representations of a table in the minds of my friend and myself (See Volume Two, p.267, ‘Frege, sense and referent, concepts and objects’). The one totally objective referee which can help us to decide whether a representation is correct is the table itself.

What about abstract concepts like a general notion of ‘table’ divested from all details by which tables can differ? If the abstraction is successful, we would have a concept which is free of any personal connotation, and thus totally objective. As explained in the previous paragraph, such a success is impossible. The idea of a table is created in each of us by the experience we associate with this word.

Suppose it would be possible to achieve total abstraction, total objectivity in the sense of freedom from any subjective connotations. We would have gained nothing, for such a totally

[p. 67]

abstract and objective concept would, if applicable to all possible objects which somebody might consider a table, have lost any discriminatory power against many other somewhat similar objects for which we have another name such as a tabouret, as you will see if you try it. The abstract concept of ‘table’ is a name for the class of all real and imaginary objects which we ‘feel’ should be given that name. That feeling is culturally determined but remains subjective. In its smallest details, it will be different for each of us. The decision to put an object into one class or another is often determined by the use we make of that object. As it may be put to different uses or used in different contexts, such a definition can never depend exclusively on the properties of the object, can never be objective in that sense.

The overwhelming evidence against the possibility of the objective determination of the content of a message has remarkably enough not lead to the obvious solution, namely a functional concept of communication. Instead, it has spawned the theory that whenever it is impossible to determine the content of a message in an objective way, ‘real’ communication is impossible. Such a conclusion is just as false as the postulate of an objective content. For the functional theory of information entails that even though there are two distinct subjective contents which are never totally equal, communication is possible as long as the two contents show sufficient overlap to achieve its purpose, usually a concerted action. And why expect more? Imperfection is a feature which communication shares with every human endeavour. If my wife tells me to meet her at a certain tearoom, her description of the tearoom is good enough if I will not confound it with other tearooms. This subject will be further explored in PART THREE A, p.93, paragraph 3a.2.4, ‘Establishing facts for social decsision making’.

As said, this overlap does not arise by itself. It is the fruit of considerable work and thus energy, in complete accordance with the laws of physics. It contributes to the increase of neguentropy only as long as we do not attempt to ‘improve’ it beyond what is required for the concerted action we need in our fight against chaos.

2b.5) The Holistic Nature of Life and Information.

Science came into the life of man only very recently. The phenomena which science intends to explain were there long before him and our ancestors had to find other means to explain and deal with these phenomena. Scientific theories gradually, and often only after a long struggle, are replacing these pre-scientific explanations. But science still has a long way to go. Many phenomena have as yet resisted apprehension by the scientific method. To the extent that it can get away with it, the scientific community tends to ignore such phenomena unless it can explain them by the methods currently in use. One subterfuge popular with many scientists is to assert that such a phenomenon will be proven to be totally explicable by current theories and methods as soon as sufficient time and resources have been devoted to it. So it has been with non-clinical medicine, and so it has been with the perception which people have that there is something binding all life into one, ultimately indivisible, whole.

Scientists justifiably reject spiritualistic forms of holism. That does not do away with the feeling of ‘wholeness’ of nature which any diligent observer must experience. We must find a

[p. 68]

scientific explanation for it. The dismissal of the notion of a unifying spirit in no way disqualifies the word ‘holistic’ as a description of a phenomenon exhibiting such ‘wholeness’. For by itself the word implies no reference to any supernatural agency which is responsible for that ‘wholeness’. The fact that it has been (mis)used by mystics and some philosophers cannot be imputed to the word or the phenomenon. Until a better word has gained general acceptance, we will have to use it.

Here it will be given a definition which does justice to the property to which it is intended to refer, yet which does not contain any mystical elements. On the contrary, it will be defined by a scientific and mathematical property of certain systems.

I call a phenomenon holistic if it cannot properly be grasped by analysis alone. Analysis is a method through which we attempt to understand an object or system by breaking it down - at least conceptually and often de facto - into parts. We then explain the object or system by:

-	a list of these parts and their properties
-	some general laws governing the relations between these parts
-	aggregation methods for assembling these parts into the whole object or system.

Today, most research in empirical science is based on this analytical method which in philosophy sometimes is called reductionism. I will not use that word because it is also used in the sense of ‘unjustified reduction of complex systems to simpler ones’, for instance to consider man to be just another animal. We will call a system is ‘holistic’ if we would omit essential properties if relying exclusively on this analytical method. Essential means that the representation thus made of an object is so mutilated as to be inadequate for use as a guide in decision-making because it can be expected to lead to a high number of wrong decisions, and thus to a low score of achievement of our objectives.

Holistic systems thus defined are not endowed with any supernatural property, but only with a poor response to the analytical method. Such a concept is just as scientific as the necessity to abandon Euclidian geometry and to introduce the notion of relativity when Newtonian mechanics failed to explain the observed behaviour of heavenly bodies and atoms.

The mathematical explication of the holistic property of certain systems is well known for instance in economics. It applies to all systems like living beings whose mathematical models have the following structure:

-	complex
-	interdependent
-	dynamic relations between variables
-	the system and/or some of its elements can learn in a way which involves stochastic elements.

Any living system is composed of various entities which are already quite complex, such as proteins. Each of these entities reacts in a very specific way with its environment and - most important - with other entities of the system. Not only must we take account of a great many entities, but each of them also is subject to a number of different interactions with many other

[p. 69]

entities of the system. What happens to one will sooner or later affect every other entity. An adequate description of the system therefore requires that the whole system be taken into account, even if we want to determine the response of just one of its entities to an action from outside. We cannot break up the system into a number of independent and more manageable subsystems. That is what makes it complex.

Most of these interactions occur simultaneously. We cannot then deal with the above complexity by looking at the interaction of the first entity affected by an outside stimulus with just one other entity, then observing what this second entity does to the third one, etc. The sequential method of analysis is closed to us. The reactions in the system are interdependent.

A complex and interdependent system can only be described by a set of simultaneous equations where the variable to be explained in one equation is an explanatory variable in another equation and vice versa. As every scientist conversant with interdependent systems knows, such models require many more observations, very complex calculations and the results it produces are much less reliable. Worse is that it becomes very difficult to know what we really measure. Only a controlled experiment which keeps invariant all but one or a very few external factors would help us. Such an experiment may be possible for bacteria and populations of social insects like certain ants whose life is determined by very few factors, for instance because they eat only one kind of food. In case of more complex systems with a greater variety of options, controlled experiments become extremely difficult. For a human society they are totally impossible.

Interdependency also mandates that we take account of all factors that affect the system concerned, a patent impossibility in case of a human society. In social science we have therefore to limit ourselves to a selection of what we assume to be the most important factors, and postulate that others either play no role or have remained unchanged over the period covered by our measurements: the well known ceteris paribus clause. In real life ceteris never are paribus. Conclusions from our measurements thus always contain an error; we can never be certain of the nature of this error, nor can we measure it: all assertions about such an error are - at best - (gu)estimates, usually in the form of some theoretical confidence limits. Most, and often all, non-tautological statements in a model of a living system thus contain a probability element, are ‘stochastic’. The stochastic element becomes part of all equations of the system which contain the variable whose explanatory equation contains a stochastic element.

That may be fairly innocuous in the original equation. But its effect in other equations where the variable concerned appears could be considerable. Say that variable y was defined by the equation y = Ax + d. where d is a chance deviation which we have cause to believe rarely exceeds 2% of y. Suppose that y becomes an explanatory variable in the equation defining m = B(z - y), where z is an external variable. If z and y do not differ very much in their order of magnitude, say that y usually varies between 90% and 96% of the average z, then a 2% variation of y can become a 20% to 80% variation of m. The fact that we are confident that we did not make large errors in the measurement of variables by excluding certain factors believed to be of small importance is not sufficient to give us confidence in the results of the model. Such models require sensitivity analysis, another complication.

[p. 70]

Dynamic in its structure means that the time element is part of the specification of the relations used to describe the system because we have reason to expect that the speed of these reactions is a factor in itself or because some variables are dependent not (only) on the current, but on past values, either of themselves or of other variables. We can explain many interesting phenomena, such as the pork cycle of Hanau familiar to every student of economics, only by taking account of this time factor, and thus of the history of the system. This dynamic element compounds the difficulties generated by interdependency because it multiplies the number of variables and relations of the simultaneous system. The problems of auto-correlation and inter-correlation between independent variables become even more intractable. And it explodes, in a usually uncontrollable and often exponential way, the consequences of the inevitable past errors generated by the impossibility to satisfy the ceteris paribus clause.

A complex, interdependent and dynamic model consists of equations, variables and parameters. The equations and variables tell us what entities we can expect to change in response to which stimuli. They define the structure of the system. To be of any use, such a system must also tell something about the direction and size of these changes. That is provided by the value we give to the parameters, a job performed by the ‘measurers’, in social science usually statisticians. (Parameters are the values preceding the variables in an equation; in the above example they would be A and B, they are what fleshes out the system and enables us to draw quantifiable conclusions from it.) The accuracy with which we can estimate those parameters evidently depends on the accuracy of the measurements we use to calculate these parameters. What is most interesting in living creatures is their behaviour. Measuring various aspects of the behavior of living beings is infinitely more intricate than with inert materials. Living beings have the unfortunate tendency to refuse to hold still. If we force them to do so, we have ipso facto changed their behaviour.

Some variables we want to measure do not have any clear one-to-one relationship with any kind of yardstick: think of human attitudes. Any measurement of such a variable must be roundabout, thus increasing the chances of error and the risk that we do not measure what we intend to measure. The more variables we take into account, the greater the problems of getting adequate measures of them. Reducing the number of variables by unjustified application of the ceteris paribus clause is not a solution: it will reduce the reliability of the results even more.

Because of complexity, interdependency and dynamism, and the measuring problems involved, the analytical method when applied to living systems will give incomparably poorer predictive results than if applied to the inert world. Yet at least in theory these difficulties could be solved, be it at high cost.

Many of the greatest minds of our century, like my teachers Jan Tinbergen and Tjalling Koop-mans, have refined the techniques for coping with the problems of complexity, interdependency and dynamism. The practical difficulties of computations and the handling of a large number of data have become manageable by the advent of computers. But the problem of knowing exactly what we measure, of being sure of measuring that variable which on theoretical grounds should figure in our model, remains basically unaddressed or if it is, unsolved. Mostly it is swept under the carpet by relegating discrepancies between what we want to measure and what we do measure

[p. 71]

to the ceteris paribus world of phantoms, or simply by resorting to black-box models. It should be understood clearly that black-box models operate purely and totally on the principle of induction. In the paragraph on induction we will revert to this subject and show that black-box models are a second-rate substitute for more explicit and reasoned knowledge.

Whatever the problems, there is no theoretical cause for holding that complex, interdependent and dynamic systems are never, and in no way, amenable to the analytical method, and that they justify resorting to a new qualification such as holistic. But there is another, decisive, reason for introducing the concept of holism, namely this fourth characteristic of living systems: they can learn and do so on the basis of a unifying principle: a built-in self-oriented selection criterion which is applied to happenings like a mutation or imagination which contain a substantial stochastic (chance) component whose unpredictability is geometrically multiplied in interdependent dynamic systems. It is this ability to learn which invalidates the current analytical method as a sufficient basis for prediction.

As stated, we must estimate the parameters of any model, also of living systems, by measuring the values which the variables have taken in various known situations. Obviously we can measure only the past, in time series as well as in cross-section analysis. The value which we assign to the parameters therefore always describes the relations as they were in the past, before application of the model to today's problems. That application ipso facto changes the information directing the reactions of a system which can learn and thus the value of parameters of variables representing behaviour of living entities; these parameters then will change over time. We cannot even exclude any change in the structure of the system, certainly not over a period long enough to allow evolutionary change. While our familiar analytical method then can give us a picture of the past, it provides no rational basis for inducing expectations of future behavior from past experience, at least not until complemented by other methods for dealing with living and thus learning systems, methods which have yet to be developed at the time this is written.

The unifying effect of the self-centred direction of this learning suggests the name of ‘holistic’ to distinguish such systems from other systems, such as a totally random system, which are also not amenable to prediction or control by the analytic method but lack life's unifying, directional element. Let me repeat that the specific result of this unifying element only emerges ex post, after the event.

Living beings who lack a specialised information processing organ can ‘leam’ only through a mutation in their genes. At the level of the individual bacteria, there is practically no learning. But because they multiply so fast, and because there are so many billions of them, a population of normal size learns quite fast, as we have seen with the emergence of penicillin resistance. The mutation leading to resistance to penicillin may have occurred before the invention and application of penicillin. Having no survival value, that feature would usually disappear as redundant. Note that the disappearance of penicillin would restore the pre-penicillin situation only after all genetic and written records influenced by the appearance of penicillin have been erased, which - being a degressive, gradual and stochastic process - can take very long and has no predictable end.

[p. 72]

In an information system of living beings almost all aspects, functions etc. are related to each other. The perception of higher-level creatures, i.e. those endowed with a central nervous system, clearly is of a holistic nature. Whatever they focus on, they perceive it as a whole unless in a second information process they shift their focus on details of the first representation. If they focus on a detail, they will perceive this detail as a whole.

We are quite aware of the holistic feature of our information processes: we perceive personality as a unity, and in psychology have developed the concept of ‘Gestalt’. At the level of a society, we have no problem perceiving the difference in culture between two nations, yet can seldom adequately describe it by a limited number of specific aspects or causes. The difference in culture between a Frenchman and a man from Holland goes beyond the kind of artists they have spawned, their eating habits and their approach to the opposite sex (the French supposedly being more verbal, ‘le baratin’).

While it may seem trivial, I can tell from experience that the difference in humour is also an important factor; it can run so deep that it hampers the social life of a cultural transplant. But how do we define that difference? The one ready explanation is the kind of humour to which we were exposed in our youth. But that difference is not really identifiable, let alone predictable, for it does not depend on some intrinsic features of humour; it just was different. If two cars differ only because they have a different hood or fenders, they can be easily compared, and are - for all practical purposes - identical. If two women grew up in exactly the same circumstances and are entirely identical except for the fact that the nose of one of them is twice as big as the nose of the other, differences in the way they imagine themselves to be perceived by their peers may cause a substantial divergence in the development of their personality even though their olfactory capabilities are identical.

2b.6) Parallel Versus Sequential Information Processing; Reason and Rationality.

We have trouble imagining a holistic - instead of analytic - method to deal with the holistic character of living systems. But it must be possible: we often do so in practice, namely whenever we trust our intuition. Our brain - if left to itself - will always treat reality as a whole. Even if it has to gather information about reality in bits, it will automatically integrate these bits into the whole which reality is supposed to be. We are as yet largely ignorant about how our brain does it, but then we have only just developed some of the physical and computational tools required for dealing with the gigantic number of brain elements and their interconnections. Instead of ignoring it, it is imperative that we develop the theoretical tools for dealing with holistic systems.

The representations of reality constructed in our brain are always generated on basis of selection criteria which - for beings endowed with a cortex - depend (amongst others) on the problem of the moment. Our immense network of connections between neurons enables us to do so simultaneously; we ‘run’ many distinct information processes in ‘parallel’ to each other. We can thus present a very large number of information bits to the synthesising function which must make a whole out of them.

[p. 73]

These representations must be correct, for only then will they lead to decisions which bring the subject closer to his objectives (in nature usually survival and reproduction). Success in this venture is a criterion which automatically and almost tautologically passes judgement, but only after the action has been taken.

Both the capability and the need to justify the validity of information can only arise with beings capable of explicitly doubting the validity of that information, which means that they must have imagination and be capable of reflective thinking. Justification presumes analysis of information and does not occur in holistic information processing. (The practice of justifying information probably originated in the necessity of having it accepted as valid by fellow men when engaging in social ventures, for instance when deciding about the plan of a hunt.) The faculty to submit information to such critical appraisal is called reason, and decisions engaging reason are called rational. Reason and the command of a symbolic language are complementary and the question ‘which comes first’ is a chicken and egg problem. In any case, they are the major tools of man in his struggle of life. In their predominance lies his main distinction from other living beings.

Holistic processing of information is the ‘natural’, pre-human way. Conscious thinking has been superimposed on the holistic process. It puts a brake on it by concentrating on a predetermined and fixed subject, splits it into distinct parts and steps and then processes all intermediate steps ‘sequentially’. It forces us to put constraints on what our brain would do ‘naturally’, we must repress ‘natural’ impulses. My hunch is that here lies the real cause of much of our stress and ‘alienation’ which so often is imputed to scapegoats like capitalism or rationalism. The apple of knowledge which is based on reflective and analytical thinking gave us enormous power over the rest of the world, but also exiled us from the paradise of purely intuitive thinking. The indubitable effect of meditation techniques as well as the consumption of alcohol and various psychedelic drugs could be due to their ability to temporarily (and at will) break these shackles and free our brain for unrestrained intuitive, holistic activity.

In that sense Horkenheimer and Adorno (Dialektik der Aufklaerung) may be right when stating that rational, analytic thinking is a form of exercising power, of creating a distance, a schism, between the thinker and the object of his thoughts, of generating (the perception of) alienation. It makes plausible their theory that such objectivation implies the propensity to lose the sense of our being part of the world to which that object belongs. It supports the dialectic nature of conscious thinking. But the above view of reason does not lead to the same pessimistic conclusion which they drew.

The synthesis - and the solution to the paradox generated by our reason - is to acknowledge the functional nature of that reason and to become aware of its exceptionality within life's information processes. Once we have acknowledged it, we can then proceed with reintegrating it in the holistic process of life. Awareness and ‘interiorisation’ of the holistic nature of living systems, and finally of the whole living world, provide the antidote to Marxist pessimism and its unwarranted emphasis on just one of the three elements of dialectic, the negation. Reason not only is a cause of alienation, it also is the only realistic hope for a cure.

[p. 74]

That appeal to reason will be ridiculed by postmodernists and hotly contested by theocrats. Their main argument can be summed up as: ‘look where it has got us: war, degradation of our environment, reduction of man to a means of production and consumption plus other assorted disasters of today's world’. They have a point, but their conclusion that all this is due to reason and rationality is based on an error of deduction. An error so evident that it must be induced by an own agenda: power to their personal fashion or religion. For they did not even investigate and thus missed the most likely and also real cause: that these disasters follow from a misuse of reason, namely the one which was exposed in my book of 1977 whose translated title is ‘Progress without Blueprint’. That misuse is the attempt to apprehend and deal with the living world exclusively with methods (which I called ‘technical thinking’) developed to deal with the inert one, a misuse which illustrates the need for the chapters on life and information. The error of postmodernists is evident because we cannot even conceive a civilised human being without the faculty of reason and the consequent concept of rationality.

Unfortunately, gearing our way of thinking to the demands of living and thus ‘holistic’ systems is unrewarding toil compared with other means for gaining status and money. Philosophers and scientists, and their schools, compete between themselves for acknowledgement of the predictive or explanatory power of their theories. When dealing with the inert world, the largely observer-independent nature of the objects of inquiry makes these objects an impartial umpire, which imposes a certain discipline which today is lacking in philosophy and social science, a problem dealt with in the part about truth. In keeping with this unchecked power struggle, the holistic nature of living systems presented above has spawned a philosophical holism which claims that reductionism (the analytical method presented at the beginning of the previous paragraph) should be rejected as a means to deal with human society. From the above, it will be clear that the only conclusion which can legitimately be drawn from the holistic nature is that the analytical method - as used today - is insufficient and that we can never expect it to provide us with the control over its objects which we have achieved over inert ones, a conclusion which itself follows from the application of these methods. We can only conclude that we must add to them, not that we can dispense with them.

2b.7) Holarchy.

That the living world is organised in hierarchic systems of more or less independent and self-centred individuals is common knowledge, so common that it has been ignored as obvious. Yet it has profound implications for our view of society. The first one - to my knowledge - who highlighted this aspect of life was Arthur Koestler in his book ‘Janus’. He coined two words (holons and holarchy) which seem to have some merit. It is difficult to better Koestler's language, so I will quote from his book to explain them.

(p.34) ‘We have seen that biological holons, from organisms down to organelles, are self-regulating entities which manifest both the independent properties of wholes and the dependent properties of parts. This is the first characteristic of all types of holarchies to be retained; we may call it the Janus principle. In social hierarchies it is self-evident: every social holon - individual, family, clan, tribe, nation etc. - is a coherent whole relative to its

[p. 75]

constituent parts, yet at the same time a part of a larger social entity. A society without holarchic structuring would be as chaotic as the random motions of gas molecules colliding and rebounding in all directions’.
(p.57) ‘The holons which constitute a living organism or a social body are, as we have seen, Janus-like entities: the face turned towards the higher levels in the Holarchy is that of a subordinate part in a larger system; the face turned towards the lower level shows a quasi autonomous whole in its own right. This implies that every holon is possessed of two opposite tendencies or potentials: an integrative tendency to function as part of a whole, and a self-assertive tendency to preserve its individual autonomy.’

This holarchic structure of our living world is well documented by Koestler in his book. It is a real, existential feature which is amenable to the methods of empirical scientific investigation. The two tendencies generate a polarity within the holon. That polarity, Koestler says (p.58):

‘.... is not a product of metaphysical speculation, but is in fact implied in the model of the multi-levelled holarchy, because the stability of the model depends on the equilibration of the dual aspects of its holons, as wholes and as parts. This polarity or coincidencia oppositorum is present in varying degrees in all manifestations of life.’

So far Arthur Koestler. How does his concept fit into the subject of this book? First note that with all such systems the higher level both chronologically and functionally follows from the lower level: the system is ‘madeup’ by its individuals. We can conceive individuals without the system, and usually individuals can survive for some time after the system has disappeared, like hair cells after the death of an animal. But we cannot conceive a system without its individuals. Any system will instantly disappear if all its subunits disappear; often it will disintegrate after the death of just one or a few groups of indispensable cells or individuals.

Secondly, living systems can be split into various hierarchic levels. Information processes take place at all levels. What makes such a lower level entity a holon instead of for instance an organ? That is the possession of an own, self-directed, value-setting and information processing apparatus such as the cells in our body have. Our limbs also are subunits of our body, but we will not call them holons, for they have no own value-setting apparatus.

Eco-systems engage no value-setting and information processing at all. An ecosystem never is a subject in an information process and the value-setting apparatus of the individuals making up the ecosystem shows only self-directed motivations: a predator hunts the weaker - probably sick or degenerated - prey because that saves him energy, not out of any concern for the health of the prey population. That he hunts in a way which serves the well-being of the prey population is the result of the selection process of evolution which favoured those predators whose manner of hunting ensures a continuous supply of prey. In ecosystems we cannot detect any value setting, any functional behaviour, directed at the system, so they are irrelevant from the information point of view, they are not holarchies. Subunits can be holons only if they:

-	have an own value-setting, information-processing apparatus
-	include the maintenance of the whole of which they are part in their ‘objectives’.

[p. 76]

If the subunits have a certain amount of autonomy and an own ‘will’ to survive and to propagate, as is the case with cells in a body, the system has to solve problems of competence and coordination, as each subunit fights its own -if not lonely - battle to survive and reproduce. Any enduring solution of these problems is based on the perceived (by reason or evolutionary selection) advantage which the holons derive from being part of the system, sometimes for their own well-being and always for successful propagation of their species. In stable holarchies the individual can realise its full potential only by participating in the system, while the higher-level system depends on activities of subunits directed towards the well-being of the whole system. Individuals in hierarchic systems must be ‘motivated’ not only to maintain themselves, but also and above all the system of which they are part. Their value-setting must be directed by two kinds of objectives:

-	maintaining (and reproducing) themselves = self-assertive tendencies
-	maintaining and expanding the system = integrative tendencies.

Often the motivation oriented towards the maintenance of the system will conflict with the direct interests of the individual, sometimes to the point of requiring him to sacrifice his life for the well-being of the system.

Until man, nature knew of only two types of information-processing apparatus: the genes and specialised (nerve) cells. Take animals: the biases directing the information process (the objectives of the individual animal) and sometimes even its specific actions, are encoded in the genes or formed in the process of developing from a fertilised egg into a grown animal. This process is largely the same for all members of a species, generating compatible integrative behaviour without leading to internal conflicts within the individual holon. Any conflict between the self-assertive and the integrative tendencies has been settled before its birth by the selection process of evolution which favours the propagation of those individuals whose tendencies are well balanced.

The human race is in a radically different situation. As (descendent of) a social animal, man inevitably has integrative ‘instincts’ built into him. The new level of information processing which appeared with homo sapiens (conscious reflective thinking, imagination and culture) has greatly reduced the scope and effectiveness of these instincts because:

-	it has provided him with an inexhaustible supply of new behaviour alternatives for which he did not develop any instincts
-	it has enabled him to overrule those instincts which he does have. Being conscious of them, they can be exploited by other individuals to further their own ends under the guise of serving society.

Man cannot rely on physical evolution to solve that problem. He must use his culture to do so by creating some form of government which performs for society certain society-directed functions which for social animals are performed by their central nervous system.

With government man has created a new information-processing entity. As long as the government only executes actions which have been decided upon by all citizens, as in a direct democracy, the basic structure is not that much different form animal societies: the coherence of society is determined by the integrative tendencies of the constituent individuals. But if - between

[p. 77]

the individual holons and society as a whole - we insert another level of information processing, like a parliament-controlled government taking decisions on its own, a new source of conflict is generated, pitching the interests of those holons which govern against those which are governed. The problem is exacerbated if the governed holons split into groups who delegate decision-making to representatives. The integrative tendencies of these delegates are then directed at ‘their’ group, not towards society as a whole. That kind of problem is at the root of Max Weber's bureaucratization with its conflict between substantial and functional rationality and explains why the latter tends to dominate.

The ability of our reason to overrule our instincts has generated another problem unknown in other holarchies: the ‘free rider’. Both instinct and reason may tell an individual that a certain sacrifice to society would also be to his own ultimate advantage. But reason will also tell the individual that he can gain even more if the others make the sacrifice while he does not. Investments in social behaviour which all ‘holons’ consider advantageous to society will therefore materialise only if society can ensure that all contribute their due to that investment: the free rider problem forces man to create and accept institutions which have the power to enforce certain types of behaviour. The necessity of authority is not particular to any specific type of organisation nor need it be the result of a pure power struggle, as anarchists assert; it simply follows from the nature of man.

The capacity of imagination plus conscious reflective thinking, coupled to the possession of a symbolic language which enables men to pool their experience and imagination, astronomically increases the advantage of living in a society and puts a premium on our integrative tendencies. Yet that same capacity of reason makes it possible, and thus creates the risk, that the integrative tendency will be perverted by groups of individuals able to manipulate large masses with low critical power to serve their own objectives, interests or ideals.

It is regrettable that instead of integrating his concepts further with prevailing theories, Koestler in the second part of his book got lost in the quest for a simple way to prevent catastrophes of the kind we have experienced in the last hundred years and which all entailed the perversion and exploitation of man's integrative tendencies by a group of fanatics to further their own objectives, their own ‘ideals’.

The first paragraph of that second part comes right out of my heart:

(p.77) ‘No historian would deny that the part played by crimes committed for personal motives is very small when compared to the vast populations slaughtered in unselfish loyalty to a jealous god, king, country, or political system.’

He comes to the conclusion that it was an evolutionary mistake to provide man with a powerful brain without at the same time endowing him with a mechanism to subjugate his emotions, especially his integrative tendency, to his reason. To remedy to this situation, he proposes to add a chemical to drinking water which would inhibit emotion from taking control of our faculty of reason for any prolonged period of time.

Everything about this proposal is wrong. Our emotionality is not an evolutionary mistake, for without a strong motivation, why would anyone submit to the discipline of rational thinking at

[p. 78]

all? Given the problems of free riders and the ones exposed by Weber, how can man be expected to subjugate his personal interest to that of society if his integrative tendencies are too weak? Even if we can find a chemical which can counteract prolonged emotional activity, it would have unwelcome side effects which would make its acceptance unlikely. For it would repress all strong motivations, also the one to make the sacrifices necessary for maintaining our society and for caring for our children. The obvious remedy against any kind of totalitarian appeal is to give priority to democracy over any other consideration when designing our social system, and to shore it up with a clear, positive and operational definition. Only thus can we generate the loyalty to democracy required to provide the motivation for making the necessary sacrifices in its defence. That is not Utopian. For loyalty to democracy is dictated both by reason and by a very strong subjective motivation: our desire for freedom. Today's sorry state of democracy is mainly due to the confusion about its nature. As defined in PART ONE, democracy is an organizational principle which acknowledges both tendencies of man and provides a framework for reconciling them.

Koestler's diagnosis as well as the proposed cure must have seemed sufficiently preposterous to have disqualified his whole book for serious consideration by the scientific and philosophical community. No reference to it can be found in any of the scientific and philosophical books I have read. Thus the valuable concepts of holarchy and holons have sunk into oblivion. Yet these concepts deserve to be revived and accepted as legitimate, for they provide the conceptual basis for an adequate representation of any living hierarchic system, and particularly of a human society.

How a holarchic system is to cope with holons endowed with the capacity of reason and imagination is, in a nutshell, the fundamental problem which human, democratic society must solve. Administrative, procedural solutions will not do. It must emulate the internal consistency of ‘natural’ holarchies, must to cultivate the integrative tendencies necessary to achieve the high level of coordination required by a world without natural frontiers and with instant communication, while at the same time preventing the perversion of these integrative tendencies towards the ends of specific groups or individuals. The problem primarily is to coordinate objectives and values, and only secondarily to coordinate the efforts to achieve them.

2b.8) Knowledge Versus Information.

To choose amongst the action alternatives available to him that one which - under the prevailing circumstances - maximises his chances of achieving his objectives, a subject has to ‘know’ the relevant facts about these circumstances. The information process is intended to create a representation of these facts. The information which a subject can gather is limited by:

-	the ‘impulses’ which its registration apparatus is able to register
-	the set of meanings available to it.

In the case of the milk-sugar process, the set of all possible representations is limited to two: ‘no sugar’ or ‘no information’. Visual information is limited to:

-	what can be transmitted by rays of light
-	those rays of light which the cells of our retina can register

[p. 79]

-	the detail which can be accommodated by these cells
-	the available ‘meanings’, inherited or generated by experience.

Meanings are determined by the arrangement of brain cells connected to those of the retina, their ‘wiring’ and the current state of the neurons in our brains which create images upon receiving impulses from these receptor cells. That process involves not only the neurons directly connected to the cells of our retina, but also a whole process of interpolation and extrapolation by other brain cells based on previous experience. The arrangements (for instances in rows of various angles) and the wiring of neurons may be inherited from our progenitors. The meaning generated by the activation of the various arrays of cells connected to those of our eyes is developed in interaction of these activations and the experience obtained by other senses. It is by now common knowledge that a baby has to learn to see, that he needs other experience, for example by touching, to give ‘meaning’ to what his eyes register.

We cannot look everywhere at the same time and therefore can never see all there is to see. We have to select what to register: where to look and what to look for. That selection takes place at the initiation of any information process. A predator looks for clumps of grass because past experience has shown that it is a hiding place for rabbits. Inhibitors ‘look’ for sugar within the cell. Except to the escherria coli, a free inhibitor is just another molecule. Even to the bacteria itself, its meaning will emerge only after the inhibitor has encountered an operator. But to an observer who is aware of the milk-sugar information system, a free inhibitor which has not yet encountered an operator is more than just another molecule. An appropriate definitions would be: information is the set of all possible representations which can be generated by a specific information process, any sign or symbol which we expect to have a meaning for the subject in the information process we are considering. If the rest of the system is functioning, the inhibitor has a meaning even before it has encountered an operator. But that meaning can figure as an object only in a statement about the information process. Information as potential knowledge exists only as a speculation in the mind of an observer. It depends for its existence on a subject making that speculation. As potential knowledge, as the set of all representations possible in a specific information process, information is a human abstraction created by the observing subject to deal with the phenomenon of information processes.

To have any impact on our decision-making, information must be stored. The only information which can be stored in the milk-sugar process is ‘no sugar’ and results from the bonding of an inhibitor to the operator. In that simple process, registration and storage are one and the same process. With complex information processes, such as vision, there can be no such direct storage, given the immense quantity of visual information which we can register from our environment. Our memory would become rapidly filled to its capacity if we stored it all. We have to select what to store. The selection process is especially important for beings endowed with imagination and will be the subject of PART THREE.

Only information (representations) which we have selected and stored in our memory will become knowledge. Knowledge is a concrete concept: where there is knowledge, a representation really exists, whether we are aware of it or not. That knowledge does not need (and usually will not have) any physico-chemical connection with the object. Knowledge simply is a state

[p. 80]

of the information-processing apparatus of the subject. That state is ‘real’, has an own existence independently from any observer. Whenever an inhibitor encounters an operator, that operator either will bond to the inhibitor and generate the knowledge ‘no sugar’, or it will not bond; the operator will then assume the state which corresponds to ‘there is no information’. The state of neurons in our memory is a ‘fact’, but we can become aware of that state only to the extent that they emerge on the surface of our consciousness.

In common language there is substantial overlap in the meaning we give the substantives ‘information’ and ‘knowledge’; in fact, they are often used indiscriminately as synonym. The definitions proposed here are conventional and will prove their usefulness in more complex information systems, especially in rational thinking.

The simplicity of the milk-sugar system may make it practically foolproof, but it also imposes severe limitations on the bacteria: if its surroundings change, if milk-sugar is displaced by some other kind of sugar, the bacteria cannot use that alternative, at least not until the lengthy and energy-consuming process of mutation and selection has generated a bacterium capable of digesting this new type of sugar. From beings using only very specific yes-no type of information connected directly to a specific decision and action, life has evolved to beings manipulating their environment through a very complex information system generating indirect knowledge about general - sometimes practically universal - aspects of their world, like our laws of physics. Note that such generalised knowledge will be useful in decision-making only if coupled to very specific information about the concrete situation.

Contrast the milk-sugar system with the knowledge we acquire at school, like history. Quite probably, this knowledge will have some influence on our actions later on. But in class and while making our homework we are largely ignorant of the way in which we will benefit from it. Whenever this knowledge does influence our actions, we are usually unconscious of it and we would be unable to pinpoint this influence even if we tried. It will show up in the way we select other bits of information, in our approach to other people and in our reactions to statements they make. It will colour the impression we make on them and thus influence the way they react to us. It will determine the many assumptions we inevitably have to make when confronting the reality about us.

Or think of our senses and of the enormous variety of signals which they can collect, interpret and translate into images, sounds, tactile impressions etc.: it seems that just about anything we can imagine is potential information. Which indeed is correct. Except in a totally abstract way, it is impossible for us to conceive that there may be a limit to our imagination; we thus tend to think that the limitations of our ability to imagine and develop concepts about the universe coincides with the limitations of that universe.

In that we are mistaken. One of the great concepts developed by Kant is precisely that we can make representations about our world only in terms of categories such as space and time which are part of our ‘being’, and in terms of categories derived from them. There may thus exist in the universe creatures who use other concepts to view the world, which would then look very much different to them. All men have roughly the same idea as to what is potential information.

[p. 81]

But that is not a property of that information; it is a tribute to the effectiveness of our genetic and cultural processes.

The more universal - and thus abstract - a bit of knowledge, the wider it is accepted and applied, the more harm it can do if it is wrong, the more severe the process should be which selects which information will be accepted for storage as knowledge. That selection is never perfect and depends on circumstances; its acceptance therefore must always be conditional and subject to revision.

The capability of reflection, imagination and symbolic information processing gave man scientific knowledge, the most powerful tool ever put at the disposition of living creatures for dealing with the outside world. In accordance with the above such powerful knowledge requires commensurate selection procedures for transforming information (potential knowledge) into actual scientific knowledge that will be stored and applied by society. The next part deals with the normative aspects of the selection of knowledge as a basis for the de factii in social decision-making.

Note that besides decision-making, much information processing (art, play etc) is involved in the maintenance of our information processing capabilities, our nervous system (especially our brain) and our culture. As it does not affect the subject of this book, it will be ignored.