Critical analysis of the immunological self/non-self model and of its implicit metaphysical foundations
An examination of the concepts used in immunology prompts us to wonder about the origins and the legitimacy of the notions of self and non-self, which constitute the core of the dominant theoretical model in this science. All theoretical reflection concerning immunology must aim at determining a criterium of immunogenicity, that is, an operational definition of the conditions in which an immune reaction occurs or does not occur. By criticizing both conceptually and experimentally the self/non-self vocabulary, we can demonstrate the inaccuracy and even the inadequacy of the dichotomy of self / non-self. Accordingly, the self/non-self model must be reexamined, or even rejected. On the basis of this critique, we can suggest an alternative theoretical hypothesis for immunology, one based on the notion of continuity. The "continuity hypothesis" developed here attempts to give a criterium of immunogenicity which avoids the reproaches levelled at the self model.
self; autoimmunity; tolerance; immune system; philosophy of immunology; continuity.
Shortened version in English
Since it was suggested by Burnet in the 1940s, the self/non-self model has formed the theoretical backdrop to all immunology. This model answers the question of giving a criterium of immunogenicity by tendering two propositions : 1) the immune system does not react against the self, and 2) the immune system reacts against the non-self. But what exactly does the use of such metaphysical notions as self and non-self mean in an experimental field of biology ? Why should there be a relation between the issue of identity (the self) and the system of defence of an organism (immunity) ?
Herein lies the inherent tension in the self/non-self model. Biological identity is not merely, but is predominantly, genetic : we are all unique because we are distinguishable by our genomes, understood in their relations to the cellular and general environment, and not by the specificity of our immune cells, which is simply a phenotypic manifestation of the individuality of our genes. Thus, the question remains : why did immunologists elect to use the term selfto describe organismal defence ? We can give two reasons for this choice. The first is the extraordinary phenotypic diversity of immune components. The second is the implicit conceptual shift from identity to defence of integrity : relying on the definition of immune cells as 'defensive' cells, immunologists postulated that the issue faced by the immune 'self' was not properly the question of identity ("who am I ?") but the question of maintaining the integrity of the organism throughout its changes ("how am I protected ?", that is against which foreign entities do I have to be protected to remain myself ?). As we will show, this tension and this conceptual shift have metaphysically and inappropriately burdened the immune 'self', which is not the synonym of 'identity' that it pretends to be. Burnet introduced a dynamic vision of the self as something acquired. Following in his wake, immunologists have tried to understand the self not as a fixed and determined thing, but on the basis of the processes by which the immune system learns not to attack the self in its embryogenic or immediately post-natal periods.
Under close examination, the immunological model of self/non-self proves to be false. The starting point of our criticism is an examination of the negation in the notion non-self : what does the 'non-' mean in the phrase 'non-self'? As Aristotle's philosophy of logic shows, there are only four meanings of opposition : things can be opposed as affirmation and negation, as privation and possession, as contraries, and as relatives. To take the first meaning, the non-self cannot be the negation of the self, since affirmation and negation characterize propositions, not notions. It cannot be the privation of the self either, because for an organism the self and the non-self coexist. But it is equally impossible to say that the self is the contrary of the non-self because, as Aristotle explains, the individual substance (which is precisely what the notion of self aims to represent) has no contrary. Since all the other meanings have been ruled out, only opposition as relatives remains conceivable. One may think this is an adequate meaning : the 'non-self' is everything which does not belong to the 'self', so it is defined in relation to the 'self'. However, an opposition as relatives must be founded on a trenchant distinction between the two relative parts. The second part of our criticism uses arguments derived from experiments to show the inexistence of such a clear-cut distinction, from which we conclude that the notion of 'non-self' is meaningless.
Three elements show the inaccuracy of the first fundamental proposition of the self/non-self model("the immune system does not react against the self") :
1) Non-pathological autoreactivity of immune cells after selection in the thymus: cells are selected when they react weakly to the endogenous antigens, and not when they don't react. Lymphocytes can be understood on the model of NK cells, which do not go through the selection process by generation of diversity, but play, nonetheless, a key role in the body. Everything in the thymus is a question of activatory and inhibitory signals.
2) Peripheral autoreactivity: lymphocytes survive only if they are regularly stimulated by endogenous antigens.
3) Autoimmune diseases, if understood in linkage with normal autoreactivity.
The second proposition of the self/non-self model ("the immune system reacts against the non-self") is also inadequate, as three experimental remarks, encapsulated by the notion of immune tolerance, prove:
1) Tolerance of exogenous and potentially pathogenic entities (for example, bacteria on the skin or in the gut).
2) Tolerance of grafts, especially the foeto-maternal tolerance.
From all these observations we can conclude that, contrary to what the self model asserts, the categories of ‘immunogen' and ‘exogenous' are actually not equivalent. Thus, the only remaining possibility, that is the opposition as relative, cannot define the immune non-self. The opposition in the term 'non-self' is therefore meaningless. The self model fails to define the limit of self and non-self, and so the dialectic of self and non-self initiated by Burnet seems inadequate as a whole.
However, to show the inadequacy of the immunological self/non-self model is not enough. We must suggest another hypothesis for immunology. The rival hypothesis we develop here can be called the continuity hypothesis. According to this hypothesis, the immune system does not respond to 'non-self' but to every break of spatiotemporal continuity between the immune receptors and the antigens to which they react (whether these antigens are endogenous or exogenous). The continuity hypothesis is based on two main ideas. First, immunity has a beginning, since organisms have a period of immune tolerance (whether embryonic period or immediately post-natal period); thus, immunity is not an innate characteristic of an organism. Everything that is present when the selection of lymphocytes occurs will not trigger an immune reaction later. Secondly, autoreactivity is inherent to normal immunity: the immune system demands a continuous selection, since a lymphocyte must be stimulated regularly by endogenous antigens to survive.
Why is the continuity hypothesis better than the self/non-self model ? Mainly because it is more explicative, that is it provides a single, comprehensive explanation for the triggering of immune reaction, whereas the self/non-self model is obliged to define a multiplicity of exceptions to its general rule. We think that at least five domains display the superiority of the continuity model: 1) autoimmunity; 2) cancers; 3) uninterruptedness between immunity and global regulation of the organism; 4) immune tolerance; 5) induction of tolerance (desensitization ; provisory tolerance for paternal antigens after pregnancy; tolerance for some pathogens with which we have had a long contact; greater graft tolerance after a blood transfusion from the organ-donor; etc.)
Two philosophical conceptions of identity underlie the self model and the continuity hypothesis. The self model understands identity as a substance (the preservation of a metaphysical core throughout changes, that is defence of integrity: only the changes originating from the 'inside' are tolerable), on the model of Leibniz's philosophy. The continuity hypothesis, on the other hand, leans on the definition of identity as spatiotemporal continuity, on the model of the philosophies of Locke and Hume. Since their philosophies represent a metaphysical deflation of the substantialist conception, we think that it is time for immunology to reproduce this deflation in its own domain.
Is the continuity hypothesis a simple reformulation of the self model, the only difference being that we define identity by continuity instead of substance? We think not: the continuity hypothesis explains phenomena that cannot be explained by the self model, and, moreover, if we assimilate the continuity hypothesis and the self model we risk repeating the confusion between the two meanings and 'substantializing' everything in immunology.
The continuity hypothesis has one feature in common with the network theories of immunity, initiated by N. Jerne : immune reactions are conceived of as perturbations of the system. However, the continuity hypothesis differs from these theories on a fundamental issue : while network theories describe the immune system as closed and self-defining, the continuity hypothesis aims at opening the system, mainly with the idea of induction of tolerance (a supposedly 'foreign' element can be integrated into an organism and tolerated).
Thus, immune identity is defined not by defensive reactions towards all exogenous entities, but in a ‘continuist' way, that is, as the spatio-temporal continuity of the reactions between the immune receptors and the antigenic sites (whether endogenous or exogenous). We can then conceive of the biological identity as an open identity that deals with both endogenous and exogenous elements in order to determine eventually the combined ('impure'), and always precarious, nature of a given organism.
[a] Agrégé de Philosophie, Département de Philosophie de l'Ecole Normale Supérieure Ulm-Paris, Visiting Fellow à Harvard University, History of Science Department, Science Center, 1 Oxford Street, Cambridge, MA 02138, USA. Auteur correspondant. Tél.:(001) 617.493.4157.
[b] Directeur de recherche au Commissariat à l'Énergie Atomique et Chef du Service de Recherches en Hémato-Immunologie (DSV/CEA) à l'Institut Universitaire d'Hématologie de l'Hôpital Saint-Louis, Centre HAYEM, CEA, 1 avenue Claude Vellefaux, 75475 Paris Cedex 10.