Georg Ivanovas From Autism to Humanism - systems theory in medicine

4. Systemic Basics

previous -- home -- content -- next

4.3 Higher orders of learning

Bateson was one of the first to investigate the principles of second and higher order phenomena. He showed that a concept of higher order is necessary to understand the course of events observed in biological and social adaptation.

Second-order or self-referential problems are those problems who need themselves to be solved. Language needs language to be discussed on. Thoughts need thoughts to be thought of (von Foerster, 1999). Science needs science to be investigated.

Such a second order sight has enormous consequences. It constitutes a different logical class (chap. 3.2), a different way to observe processes. This does not only allow different insights. It introduces simultaniously uncertainty (chap.3.3) and meaning (chap. 3.5). The second order view is not a theoetical construct. It is a model much nearer to the events in nature than the usual linear model.

This shall be demonstated with a main field of research of Bateson, the idea of a

learning of a higher order. His concepts underwent several changes during his lifetime (overview in Lutterer: 123-140). A reason might have been that one of his basic assumptions – the axiom that acquired characteristics cannot be transmitted – was wrong. Despite this shortcoming, Bateson’s categories are a most valuable tool to understand medical events. The impact of this concept shall be illustrated with the immunological reaction to germs.

Bateson distinguished five types of learning (Bateson, 1972: 279-308).

Zero learning

This is the case in which an entity shows minimal change in its response to a repeated item or sensory input” (Bateson, 1972; 283). This is seen in simple mechanical circuits or in living organisms that are overstimulated, or where the response is structurally fixed. In medicine learning 0 prevails when under similar circumstances a certain infection reoccurs. Some people develop tonsillitis having a cold, some women suffer from herpes labiales during every menses. Also short term death after an infection should be regarded as learning zero.

Zero learning is also the main assumption for most drug therapy. It is expected that the organism reacts repeatedly the same and does not adapt. Adaptation is often regarded as unfavourable.

Learning I

This is the typical learning investigated in laboratories. It measures to what extend and in which time a human is able to solve a mathematical riddle or to remember nonsense syllables. Learning a language is Learning I. This type of learning is also the learning normally investigated in medicine. It is, for example, the development of a lifelong immunity after an infection or vaccination.

Learning II

Bateson called this type of learning also deutero-learning (Bateson 1972: 159-156), a second order phenomenon. It is learning to learn. Someone who learns nonsense syllables is, after a few tests, able to learn nonsense syllables more easily, or might even remember numbers better. Or if someone learns a language he does not only learn the language, but learns, as well, how to learn a language. Then, the next language is learned more easily (Schweizer Nationalfonds zur Förderung der wissenschafdtlichen Forschung 2009).

This second order learning cannot be detected by a simple measurement. It needs a series of similar experiments and their comparison. The measurement as such does not mean anything. It is the difference between the measurements that reveals learning II.

In a remarkable experiment Bateson worked with porpoises. Normally these animals are trained in learning I. That is, whenever they fulfil a task they are rewarded. Under Bateson’s directive the animals had been only rewarded when they showed a new behaviour. It was hard work for the animals until they understood this principle. But then it was easy for them to invent new tricks (Bateson 2007). This is obviously a different kind of learning than the usual and linear learning I.

In medicine one might see learning II under the following circumstance: When after a vaccination against flu, for example, the morbidity against this kind of flu is reduced we have a simple learning I effect. When, however, all-cause mortality is reduced, as had been seen in some studies (Voordouw et al 2004), we have a learning II situation. That is, the vaccination had more profound effects than only to induce the production specific antibodies. Such effects are by no means an ‘unspecific’ reaction. They are absolutely specific, yet on a different logical level.

Examples of such a decrease (or an increase) of all-cause mortality after vaccinations have already been discussed (chap. 2.8.c). The same is true for all kinds of ‘natural’ infections like the persistence of a germ inducing a ‘cross-protective immunity’, which is a second order phenomenon (chap. 2.8.c), like the GB virus or measles leading to a survival advantage in AIDS. It can be expected that all infections have such a second order impact, exceeding the simple ‘specific’ immune response.

Learning III

Bateson refers to learning III as an experience that totally alters the life style of a person, involving the complete being. According to him, such a change might be seen after an important religious experience. I propose that a change from a linear point of view to a systemic view is also learning III. Such a learning changes fundamentally the understanding and handling of data. It is on a higher logical level than learning II and has little to do with specific issues.

In the immune reaction learning III occurs when an acute infection is the starting point of the improvement of a chronic disease. A classical example are the worm infections improving asthma (Wilson et al 2005) or Crohn’s disease (Summers et al 2006). That is, an infection might have a beneficial effect for the function of the human in general. This type of learning (where infections improve the general health condition) is also the core of the hygiene hypothesis (chap. 2.8.b).

Learning IV

Bateson subsumes under this heading the changes in the learning structure, that is, a change of the genetical adaptation. In Bateson’s times this was only attributed to the evolutionary selection. Therefore it was expected to take many generations. Recent research on epigenetics shows, however, that this type of learning plays an important role not only from one generation to the next, but might even be important during one lifetime. Genes have memory (Bird 2007). Accordingly, learning IV is closely interwoven with the other levels of learning and cannot be separated decisively.


Excursus 1 (the reality of learning of higher order): This reflection leads necessarily to the question, in how far these categories represent a reality. Tomaras believes that they are only a metaphor (Ivanovas et al. 2007). In this view the concept of a higher order would be a kind of mnemotechnical device, a reminder not to forget certain aspects in an otherwise complex situation, as the interactions of the helmintho-bactero-viral flora. Moreover, in the reductionist approach, under the microscope or in the test tube no higher order exists. There are only reactions, often unrelated, at times inconsistent. In fact, the immune response cannot be understood on a cellular level. “As one dissects the immune system at finer and finer levels of resolution, there is actually a decreasing predictability in the behavior of any particular unit or function (a gene, a cell)” (Germain 2001).

That is, in order to describe what happens in the immune response, its organisation has to be understood. This is not possible without teleological (chap. 4.7), and polycontextural (chap. 3.5) concepts introducing the notion of ‘meaning’. In such a view it becomes necessary to distinguish between different logical levels, between the learning of a first, second or third order. Although it might be a strange procedure for an analytically trained scholar to use such seemingly vague concepts. Nevertheless, it is a difference whether a certain trait is trained (as in behaviour therapy) or whether the ability to find own and new solutions is supported (as in systemic psychotherapy; chap. 5.2). Likewise different approaches exist also in general medicine and in drug therapy (chap. 6.8; 6.9). For the general practice it is of minor importance, at first, whether these different categories represent a reality, whether they are of the observer or of the observed. Important is the impact they have for the general practice.


Excursus 2 (epigenetics): The knowledge on epigenetics is still quite rudimentary although related observations mushroom. The following sample shall just give an impression of what is attributed to epigenetics today.

The mechanism of how acquired characteristics are inherited is attributed to the small RNA (sRNA). It can activate, deactivate or even cut off parts of the genome. It is the regulator of the master regulators (Couzin 2008). This discovery was classified as “The Breakthrough of the Year 2002” by SCIENCE magazine (Couzin 2002).

Until now, we have but hints what is transmitted and how. It is known that a given genotype can give rise to different phenotypes depending on environmental conditions. The current evidence is, however, sufficient to state that the epigenetic mechanisms are of major importance for all kind of health issues (Blech 2008b). Single nutrients, toxins, behaviour or environmental exposures of any sort can silence or activate a gene without altering the code (Duke University 2005). Such responses to the environment may be expressed in the offspring rather than in the parent and might persist across a number of generations, even if the environmental factor itself has altered (Bateson et al 2004).

But the process is even more complicated. Arabidopisis, a plant, violates mendelian genetics by restoring a DNA-sequence of the HTH-gene that is found in the grandparents but not in the parents (Lolle et al 2005). That is, the inner organisation is able to change and restore the genome according to certain plans, best characterized with the notion of pattern (chap. 4.9). It seems that the whole process of genetic transmission is extremely recursive. It even became quite doubtful what ‘inheritance’ really means (Jabolka/Lamb 2005), as genes are “only marionettes in the hands of enzymes” (Kaati 2002, my translation).


Learning IV in medicine would be given, at least according to Bateson, when the following generations benefit from an infection of their ancestors. This subject has already been touched (chap. 2.8.c). Measles, varicella and other viral diseases had been rather lethal when they first came to the new world. The same was true with syphilis which – according to the prevailing theory – was brought back by Columbus in return. After some time all these diseases became less aggressive. Even people who never had any contact with the germ before and probably did not possess specific antibodies had milder forms. Former theories could explain this only by selection. It was supposed that the more susceptible people just died away, such that the surviving population was genetically more resistant. However, this theory had never been really convincing. It is not probable either that all germs change when they remain in a new population. It is likely that this learning has a strong epigenetic component. But also the whole ecological context, as the helmintho-bacterio-viral flora and other factors, might play a role. As this has been defined as learning III before, there seems to be no clear distinction between learning III and IV. After all, there might be no need for such a logical category as learning IV. It had had a sense when genetics was still a trivial science, when causal chains like “DNA makes RNA makes protein” were prevalent. But as today the words gene and genome have lost their decisive meaning and are used rather inconsistently (Angier 2008) the notion of pattern (chap. 4.9) might be more appropriate to describe the course of events in transmission.

Regardless of all the recursive imponderability around heredity, there exists a concise case of learning IV. A considerable amount of the genes is of viral origin, as viruses are able to enter the genome (Zimmer 2008). This is best known for the endogenous retrovirus. These genes are possibly beneficial for the humans. There is, for example, some evidence that they are essential for a healthy pregnancy (Society for General Microbiology 2008). As the understanding of the virus – genome interaction is new, more results are to be expected in the near future.

The question, whether it is useful to maintain the logical class of a learning IV is no trivial question. It goes right to the core of the relation between structure and function. One might discard this distinction as outdated. Then the viral genes entering the genome would be only an extension of the organism’s steering mechanism. But such a progressive view is rarely found.


It would be a mistake to believe that learning through germ contact only induces positive effects. An infection often has severe side effects like glomerulonephritis after a streptococcal tonsillitis (learning I). An infection might promote or worsen other infections (learning II). Infections might lead to a total change of the human function (learning III). This is seen in many chronic diseases, and lately a whole range of such chronic diseases are attributed to infections. This so called germ theory (Ewald 2002) is actually the opposite of the hygiene hypothesis. It believes that chronic diseases are caused by bugs (Ewald 2002). Helicobacter and peptic ulcer is the classical paradigm. But infectious agents are increasingly attributed to all kind of chronic states with a disturbed immune balance (Morens et al 2004), mainly cancer (Clifford/Francheschi 2007) e.g., childhood cancer (McNally 2005) or brain cancer of the adult (McNally 2006). But also Alzheimer’s disease (Wozniak et al. 2008), diabetes (Sobngwi et al 2008), multiple sclerosis, systemic lupus erythematosus, rheumatoid arthritis (Niller et al 2008), chronic fatigue (Hickie et al 2006) and obesity (Wigham et al 2006) are attributed to germs. Even psychiatric diseases are linked to infections like toxoplasmosis (Brown et al 2005), Lyme disease or a simple flu (Ginsburg 2004). And all this might have negative influences onto following generations (learning IV), might contribute to chronic diseases in the offspring (chap. 5.5.d). Moreover, the ‘viral genome’ is able to do a lot of harm. Under certain conditions, e.g., through an infection with a wild virus or through severe mental stress these viral genes start to produce molecules that are considered as foreign by the immune system, what might eventually lead to an autoimmune disease (Furlow 2000).


This analysis shows that the reaction of the immune system is absolutely nontrivial. Germs might be helpful or harmful (hygiene hypothesis vs. germ theory). They might be harmful when perceived on the level of learning I and helpful on another level or vice-versa. They might be harmful for the individual and helpful for the community or vice-versa. That is, one has to have sound categories in dealing with issues of the immune reaction. Observing only one factor in a reductionist frame is necessarily defective.

previous -- home -- content -- next