Cognición Sin Neuronas

Cognition without Neurones:
Adaptation, Learning and Memory
in the Immune System
JOHN STEWART
Institut Pasteur
...f.
.......,

Abstract
This paper proposes a definition of cognition as a system capable of both action and perception, in which the coupling of action and petreption ŭ such that the emergent bthaviour of the system in its environment satisfies a meaningful constraint. Theimmune system of vertebrate animals reacts to nonself antigens by producing a specific immune response which triggers destruction of the antigen; it reacts to sel f antigens by incorporating them into the regulatory
dynamics of a self-organizing network. lt is argued that this báaviour meets the requirements of the definition, and
hence that the immune system is a cognitive entity capable ofadaptation, learning and memoty. Consequences of this
perspective are that the molecula and cells of a multicellular organism are not, as such, "cognitive"; that the full articulation of neurophysiology into cognitive science requires a study of emergent behaviour at the level of meaningful
interactions between a system and its environment; and that the objects of cognition, with or without neurones,are
brought into existence by the coupled perceptions and actions of the cognitive system itself.
Key Words: Adaptation, Cognition, Learning, Memoty; Computer Simulations; Network; Immune System

Cognición sin neuronas: adaptación,
aprendizaje y memoria en el sistema
inmunitario
Resumen

Este artículo propone una definición de cognición como un sistema capaz de acción y percepción, demodo que esta
confluencia acción-percepción determina que el comportamiento emergente del sistema en su medio satisfaga una constricción significativa. El sistenza inmunitario de los animales vertebrados reacciona ante los anagenos ajenos produciendo una respuesta inmune especifica que desencadena la destrucción del antígeno; reacciona ante los propios ant ígenos incorporándolos en el mecanismoregulador de una red auto-organizada. Se argumenta que esta condurta se ajusta
a los requisitos de la definición y, por tanto, que el s ŭtema inmunitario es una entidad cognitiva capaz de adaptación,
apnmdizaje y memoria. Las consecuencias de erta perspeaiva son: las moléculas y células de un organismo multicelular
no son, por s mismas, cognitivas; una plena articulacién de la neurofisiología comociencia c ognitiva requiere un estudio de la conducta emergente en el nivel de las interacciones significativas entre un sistema y su medio,- y los objetos de
la cognición, ya sea con o sin neuronas, se materializan al confluir las percepciones y acciones de/propio sistema cognitivo.
Palabras clave: Adaptaci én, c ognición, aprendizaje, menwria, simulación en ordenador, red, s ŭtema inmunitario.Acknowledgements: The work on modelling the immune system described in this paper was carried out in
collaboration with Fraricisco Varela and Antonio Coutinho, and would not have been possible without
their intellectual stimulation, friendship and support.
Author's Addreu: Unité d'Immunobiologie, CNRS URA 359, Inscitut Pasteur, 25 rue du Dr Roux, F75724 Paris, France.
1993 by Aprendizaje,ISSN: 0214-3550

Cognitiva 1993. 5 (2),187-202

188
I. INTRODUCTION

It is widely accepted as self-evident that neurones (or neuromimetic automata
in the case of computer simulations) are a necessary and sufficient basis for cognitive
phenomena, and hence that neurophysiology is both automatically a constitutive
element of the cognitive sciences, and virtually the only biologicaldiscipline attaining to that status. In this paper, I shall question these assumptions by providing a
possible counter-example. The science of immunology is characterized by the fact
that immunologists spontaneously attribute "cognitive" capacities to the immune
system: recognition, learning, memory and self/non-selfdiscrimination (Klein,
1982). In order to examine the validity of the suggestion...