Georg Ivanovas From Autism to Humanism - systems theory in medicine
This is the abbrevated version of the publication Ivanovas G (2005): Trivial and non-trivial machines in the animal and in man, Kybernetes 34; 3-4: 508-520
In recent years there has been an increasing number of publications proving that behaviour has a genomic cause. The Science Magazine classified the understanding of this relation as the second-ranked breakthrough of the year 2003 with the promising title “Decoding mental illness” (Science news and editorial staffs 2003). But this issue is not new. It has been discussed in psychiatry for many years in the search for the cause of schizophrenia - today correlated with a variety of genes (Harrison & Owen, 2003). However, if we open a medical journal, especially Molecular Psychiatry, we find that a lot of behavioural patterns reduced to genetic causes. Here are some arbitrary examples:
- There is a strong connection between ADHD (attention-deficit hyperactivity disorder diagnosed according DSM-IV) and the allele 4-repeat of the IL-1 RA gene, coding Interleukin-1 antagonists, whereas the allele 2-repeat is associated with a reduced risk (Segman et al, 2002.).
- There is a significant linkage of ‘simple phobia’ to chromosome 14 markers (Gelernter et al., 2003).
- The Grp-gen is regulating ‘Pavlovian learned fear’ by influencing the lateral nucleus of the amygdala (Shumyatsky et al., 2002).
- Individuals with one or two copies of the short (s-) allele of the 5-HTT gene are much more likely to develop depression than individuals with the long (l-) allele when confronted with emotional stressors between the age of 21-26 (Caspi et al., 2003).
Of course this is the resumption of the old discussion of whether behaviour is inherited or acquired. Already in 1960s fierce quarrels erupted on this subject. Results have been forged (Di Trocchio, 1993), professors have been beaten, in short: real life roared around this issue.
The question whether traits are inherited or acquired is not only difficult to answer. It is, due to recursive processes, impossible to answer. Bateson replaced this question by: “Up to what logical level is learning effective and down to what level does genetics play a determinative or partly effective role?”
The ant Pheidole kingi instabilis is characterized by a large variety of forms. The morphology of each animal depends mainly on its function. The whole population is interconnected by exchanging stomach content, a process called trophylaxis. In this way some substances are distributed among the whole population. As a result, some information about the composition and situation of the population is provided to all ants. The group has a certain kind of ‘consciousness’ of itself. Removing the queen provokes a reaction in the other ants: The workers feed a special nutrition to some larva in order to raise a new queen (Maturana/Varela).
In an article with the remarkable title »Identification of a Major Gene Regulating Complex Social Behavior« Krieger and Ross show that a difference in the Gp-9 genotype of the fire ant (Solenopsis invicta) is responsible whether one or several queens are raised (Krieger/Ross, 2002)
These short examples of the world of ants are sufficient to show the complexity of our issue: The status of a queen is acquired, not inherited. However, it is genetically determined that a queen has to exist. This genetic determination is to be found in the workers and not in the queen herself. Whether or not a larva becomes a queen depends on environmental influences and is stochastic.
It is amusing to directly translate this into human behaviour: It is not very likely for a person to become the leader of a country. One doesn’t need any special genes, one must only be fed properly.
The more serious consequences of this short, but not superficial look at the organisation of ants shows that there is no clear distinction between genetics and environment, that genetic determinations are sometimes more easily changed than acquired characteristics, that a broad social context can be responsible for the morphologic expression of a genetic possibility, etc.
If we look for comparable situations in mankind we find it, e. g., in the acquisition of language. Up to the age of 12-14 the syntax of (native) language is easily learned. If this first acquisition does not happen (i. e. in the deaf), it is impossible to recapture it later (Sacks, 1989). It is, without doubt, a critical age. Afterwards (or even earlier) someone is French, English, German or Greek. Noteworthy is that the prognosis of ADHD beyond this critical age is considered as bad (Lösslein & Eike-Beth, 2000). It is foremost a question of definition to call it a morphological consolidation or a reduced plasticity of the brain. Adolescence does not only close a chapter. It also opens up a new one with other topics and with an increased neuronal plasticity for them (McCrone 2000). Social and sexual themes become predominant. Also schizophrenia is known not to start until adolescence. However, “clinical studies show that patients with schizophrenia manifest minor behavioural abnormalities in childhood even before the onset of schizophrenia” (Sawa & Kamiya 2003).
What happens in adolescence? Are some genes switched on and suddenly produce schizophrenia? Or do they become more important as minor changes have already been noticed earlier? Does the environment prevent the development of a coherent personality by incoherent communication (Bateson, 1972; Selvini Palazzoli et al., 1989)? But why after adolescence? Do we have to live with such vague terms such as ‘multi-factoral disease’, actually confessing ignorance?
The structural connection between genes and behaviour can best be described with the concepts of trivial and non-trivial machines (chap. 4.5).
Imprinting as first described by Konrad Lorenz (Lorenz, 1949) is a trivial machine: The baby duckling will follow the first moving object it sees at the moment it slips out of the egg. This is normally the mother. Therefore, throughout its childhood it walks behind its mother. If it sees first Konrad Lorenz it will follow Konrad Lorenz, and if it first sees a moving football, it will spend a lot of his lifetime walking behind a moving football. Imprinting is a trivial machine with just one recursive loop. Not understanding this recursive loop will lead to the following fallacy: Imprinting is genetically determined. If behaviour is (falsely) defined as ‘following the mother’, there are these unexpected situations when it follows Konrad Lorenz. In this case the genes will not code for the expected behaviour. This is called penetrance. If the correlation gene-behaviour is high as in some kinds of haemophilia, the penetrance of the gene is high. If the correlation is low, then the penetrance is low. Low penetrance is normally expected when the maternal and paternal genes differ or when a behaviour is coded by many genes, thus producing a non-trivial machine.
What should be demonstrated here is that a seemingly ‘low penetrance’ might be nothing else than the misunderstanding of one single recursive loop. In the duckling the behaviour ‘following a moving object’ is 100% genetic, the behaviour ‘following the mother’ is 100% acquired. With a slight change in the definition, the inner logic is turned upside down, although the observed process remains the same.
But recursivity can also be misunderstood the other way round.
The raising of the queen in Pheidole kingi instabilis or the queens in Solenopsis invicta is a trivial machine. Nevertheless, from the viewpoint of the larva (“Who is going to be queen?”) it is a stochastic process that cannot be predicted. But the result is determined. The environment only regulates the formalities. There are no recursive loops. If there are several queens as in the species Formica fusa, the workers prefer to feed the brood (larva and queen) they are genetically more related to (Hannonen & Sundström 2003). This is not a recursive process by itself, because it does not change the behaviour of the queen, the workers and the larva. A recursive effect could only be seen through generations, if according to environmental conditions the reproductive share of one queen increases. This preference of the own brood was called nepotism. But nepotism (‘one hand washes the other’) is a highly recursive process, which continuously changes the behaviour of all people concerned. It might have a stabilizing effect in chaotic social situations (e. g., post-war) or a destabilising effect in stable conditions (mafia), it might lead the partners to the highest positions or to prison. The outcome is not foreseeable. It is as non-trivial just as the cooperation of scientists.
Heinz von Foerster always maintained that man is a non-trivial machine. His well-known example is that of a child answering the question “How much is 2 x 2?” with “green”. Such a child, he said, would be sent immediately to a ‘trivializing institution’ thereafter giving the ‘right’ answer (von Foerster, 1999).
What was initially thought of as more anecdotal - von Foerster, as he told me, was unaware of the clinical picture of synaesthetics - turns out to be an exact description of how complex behaviour is trivialized. Synaesthetics is a state of perception where different sensual qualities are inseparably connected. Shapes have colours, names have odours etc. One of these children during first grade indeed answered the question “How much is 1 plus 1?” with “dark green” (Schneider 2003).
Synaesthetics is said to be quite common if not ubiquitous in early childhood before a more social perception is adopted (Hackenbroch 2000, Motluk 2004). This corresponds to the finding that the infant’s ability to discriminate among native speech sounds improves, whereas the same ability to discriminate among foreign speech sounds decreases (Kuhl et al. 2003). Similarly 6-month old infants are equally good in recognizing facial identity in humans and non-human primates. Something they have lost at the age of nine months (Pascalis et al. 2002).
Early childhood is characterized by a great variety of non-trivial, non-linear, non-formed behaviour that brings forth a stable social behaviour only through recursive interaction. This is accompanied by a stabilization of neuronal brain organisation.
Biology does not discriminate between structure and function. Function changes the structure and structure shapes function. The genomic foundation only provides a very wide framework. Trivialization is the necessary condition for the development of social behaviour.
However, this organization is not fixed, as brain research of the last years showed (Gross 2000). It even might resume to some extend, e.g., after stroke (Kluska et al 2004). This corresponds to the observation of Keeney that synaesthetic perception can be achieved by certain rituals (Keeney 2005).
Perception can hardly be called ‘learned’ or acquired. But its organization is much more flexible than we had expected. It is based on recursive social interaction.
Such recursive processes involving structure and function have a major impact on the outcome of genomic research on behaviour. The reliability of this kind of research has been proven to be low. The results of one group can rarely be confirmed by another group. Normally technical problems (Colhoun et al. 2003) or study design (Hattersley/McCarthy 2005) are held responsible. But the misunderstanding of recursive processes is another main factor. One hint in that direction is that the correlation genome-behaviour is higher in smaller studies than in larger studies (Ioannidis et al 2003). In general, smaller groups are more homogenous in their behaviour than large populations. What does this mean in terms of recursivity?
If we take a small traditional village in the Cretan mountains, the behaviour of the inhabitants, sexual self-image, expression of aggression, etc. is very homogenous, at least to the foreign observer. This is an acquired pattern as such a person changes behaviour going to the city, especially if this is done in early life. The situation is similar as in language. According to Chomsky (chap. 4.2) we could postulate a ‘faculty of behaviour in a broader sense’, that is, the possibility to interact, to acquire a language, to express emotions, to mate etc., and ‘a faculty of behaviour in a narrower sense’, that is, the human possibility to produce a large variety of social contexts. In contrary to the duckling who has a restricted number of recursive loops, humans have many recursive loops that sometimes are restricted after adolescence. Investigating a small population creates the delusion that a certain kind of behaviour is stable out of biological reasons. In such a stable situation the correlation gene/behaviour might be quite high. However, it is only an effect of the frame of observation.
This is comparable to laboratory research, where variable factors are held stable in order to investigate some traits of a system. The relation, however, will be true under these circumstances only. Unexpected behaviour will arise over and over and it will be labelled with expressions as ‘low penetrance of genes’.
The other similarity to laboratory research is that all studies investigate the relation of one gene for a defined behaviour. For example, different studies show a relationship between the genome and schizophrenia, but always for a different gene (Harrison/Owen 2003). As the studies are designed to find a trivial relation of cause and effect, they can only provide tendencies. But, since recursive processes are involved, they cannot show any kind of structural interconnection between genome and behaviour.
The problems, which are arising as a consequence shall be demonstrated with a study on depression (Caspi et al., 2003): Individuals with the s-allele of the 5-HTT gene only become depressed when confronted with emotional stressors. They develop twice as often depression after having suffered from four and more stressing events. The study investigated stressors like employment, financial, housing, health and relationship. It was found that the tendency to develop depression increases with every stressing event. This is not true for individuals with l-allele. They hardly change, if at all, even when repeatedly making painful experiences. The l-type shows (according to the criteria of the study) a trivial reaction to environmental influences (as the reaction is always the same), the s-type a non-trivial behaviour (as it changes with every new experience). This is not surprising. We all know that some individuals are sensitive and vulnerable and others are indifferent and indolent. It only would have been a surprise, if these different constitutions had no genomic differences. Now, the only conclusion allowed is that the 5-HTT gene is connected with emotional processing.
The study investigates one trait out of a complex pattern: the sensitivity for depression. It comes to the conclusion that (in a society with certain undefined values) a sensitive person is more inclined to become depressed when some defined values (job, money, spouse, health) are in question. This finding can be formulated differently: A society with certain values brings forth depression in its sensitive members. It could well be imagined that sensitive persons in a different society or with a different study design, instead of getting depressed, will get religious. May be then 5-HTT would turn out to be the compassion gene. This is indicated by the fact that mildly depressed people are more compassionate than normal people (Queen’s University 2005).
All these are rough hypotheses. But the expectation that a gene through its impact on metabolism brings forth a certain behaviour is a rough hypothesis as well.
However, there is a great danger when such trivializing definitions of behaviour become the starting point of therapeutic interventions. This is obvious in the case of psychotic experience, where the vicinity to creativity is a well-known fact (Jamison, 1993) and the list of people who have been diagnosed as manic-depressive reads like a who’s who of Western art (cited after Simon, 1995: 192):
Hans Christian Andersen, Antonin Artaud, Honoré de Balzac, Charles Baudelaire, Irving Berlin, Hector Berlioz, Anton Bruckner, Lord Byron, Paul Celan, Joseph Conrad, Noel Coward, Charles Dickens, Emily Dickinson, Isak Dinesen, Ralph Waldo Emerson, T. S. Eliot, Edward Elgar, Georg Friedrich Händel, William Faulkner, F. Scott Fitzgerald, Paul Gauguin, Vincent von Gogh, Nikolai Gogol, Maxim Gorky, Ernest Hemingway, Hermann Hesse, Hölderlin, Victor Hugo, Henrik Ibsen, Henry James, William James, John Keats, Ernst Ludwig Kirchner, Heinrich von Kleist, Otto Klemperer, Wilhelm Lehmbruck, Malcolm Lowry, Gustav Mahler, Wladimir Mayakowski, Herman Melville, Michelangelo, Charly Mingus, Modest Mussorgsky, Edward Munch, Eugene O'Neill, Charly Parker, Boris Pasternak, Cesare Pavese, Sylvia Plath, Edgar Allan Poe, Jackson Pollock, Cole Porter, Ezra Pound, Alexander Puschkin, Bud Powell, Sergej Rachmaninoff, Gioacchino Rossini, Robert Schumann, Alexander Scriabin, Mary Shelley, Robert Louis Stevenson, August Strindberg, Torquato Tasso, Lord Tennyson, Dylan Thomas, Leo Tolstoi, Georg Trakl, Peter Tschaikowsky, Ivan Turgenjew, Walt Whitman, Tennessee Williams, Virginia Wolff, Emile Zola
Intervening in the basic organisation of the serotonin metabolism (what is current practice) or gene regulation (what most likely will be done) might lead to far reaching consequences on creativity, compassion and other central human values. These consequences cannot be proved or disproved as they do not show up under the conditions of study designs.
Quite similar is the case with autism. Those famous people missing on the list of the psychotics are found in the list of persons said to have suffered from autism or Asperger’s syndrome like Einstein, Freud, Yeats, Russell, Wittgenstein, Ramanujan or Turing (Fitzgerald 2004).
In an survey on the therapy of depression under the title “Future brightening for depression treatments”, the author comes to the conclusion: “While scientists continue to argue about the architecture of depression, however, enough has been discovered in recent years to give them a number of new levers to manipulate” (Holden 2003). Others see it more cautious: "It's almost unfortunate that the illness has the label 'depression',... It focuses our attention on the mood disorder and makes us forget the rest of the body, she says. Yet depression has as many consequences below the neck as above“ (Farley 2004).
As depression is an emergent state, every intervention on the basis of a linear understanding of the process cannot be something different than a manipulation.
This habitude to manipulate has caused severe critics with strong statements such as “The Nuremberg code doesn't apply here”, comparing the treatment of psychiatric diseases with the medical experiments of the Nazis (Whitaker 2003). It is a matter of taste to go so far, but reading some examples of animal experiments in neuroscience, it is, in fact, difficult not to feel uneasy about the moral values, combined with dreams of unrestricted feasibility:
- In promiscuous male voles the introduction of a gene directly into the brain leads to high levels of vasopressin receptors in the ventral forebrain. This is associated with a lost of interest in other females. The male stayed with a previous partner just as a monogamous relative that has the same high density of vasopressin receptors (Linn et al 2004).
- Procrastinating primates can be turned into workaholics. It is sufficient to injected a short strand of DNA into monkey's brains, in order to temporarily switching off a key gene in a region of the brain called the rhinal cortex. With the gene turned off, the monkeys were unable to anticipate how many trials were left before a reward was given. As a result they worked hard throughout the task, making consistently fewer errors at every stage (Liu et al 2004a).
As a conclusion it can be said that the relation between the genome and behaviour is complex. Sometimes it is like a rigid trivial machine (one or several queens have to exist), sometimes like a trivial machine with one to multiple recursions (imprinting, acquisition of language), but very often like a non-trivial machine (nepotism in humans).
But, if one recursion takes place, the relation depends largely on our definitions (following the mother versus following a moving object). However, the idea of relation is not inherent in the observed data. It is a consequence of a theoretical construction applied to the observed (as imprinting). There might be a strong relation but a lose correlation and vice versa.
The more recursive processes are genetically foreseen or the less trivial these processes are, the lower is the correlation genome behaviour, although the relationship itself does not change. The process might be 100% determined, but the outcome cannot be predicted.
In a stable frame, recursive functions might seem trivial and a high correlation genome behaviour might be thought of as a trivial relation.
It is always possible to analyse an organism from behaviour down to the genes. But it does not work the other way round if recursive processes are involved. This has been demonstrated for polycontextural structures where the classic principle ‘the way up is like the way own’ is no longer true (chap. 3.5). That is, genes are in principle undeterminable.