undergoing changes, large or small, living organisms must maintain themselves in a comparatively steady state. Such steady state can be measured along numerous parameters, which can be grouped into a few major categories of homeostasis. Some refer to the organism’s interior state, some to its relation with the outside world.
The category of homeostasis first described by Claude Bernard, and for which Walter Cannon (1932)1 coined the term, concerns a set of physiochemical measures interior to the organism. This category can be termed physiological homeostasis. Another set of measures also interior to the organism that are usually held even steadier, by healing processes, are those relating to body structure: this category can be termed morphological homeostasis. A third set of measures also held steady is well recognised by field biologists, namely the tendency for animals of any one species to remain within the limited range of environmental conditions to which the species is adapted – its ecological niche. Maintenance of an animal within its ecological niche is effected by a variety of behaviour patterns, the activation of which are sensitive to features of the environment. This category can be termed ecological homeostasis.
These three categories apply to all species of animal, and the first two apply also to plants. The biological control systems that maintain these categories of homeostasis are, for morphological homeostasis, physiological systems; for ecological homeostasis, behavioural systems; and for physiological homeostasis, both physiological and behavioural systems.
These three categories of homeostasis are intimately linked: by maintaining an animal within its ecological niche, the behavioural systems maintaining ecological homeostasis are acting in ways that greatly facilitate maintenance of morphological and physiological homeostasis.
I believe that at least two other categories of homeostasis can be recognised; they become increasingly evident in the higher animal phyla and play a great part in the lives of higher vertebrates and man. One of these categories refers to a special aspect of the organism’s interior state, the other to a special aspect of its relation to the environment. We start with the latter.
Individuals of a species do not roam at random throughout the whole area of ecologically suitable terrain. On the contrary, they usually spend the whole of their lives within an extremely restricted segment of it. For example, a vole lives within a few square yards of thicket, a troop of baboons within a few square miles of prairie, human hunters and gatherers within a few hundred square miles of forest or plain. Even migrating birds, which may travel thousands of miles between nesting and wintering grounds, use only special parts of each; many nest each year at or very near the place they were born.
Nor do animals of higher species mix indiscriminately with others of their kind. Individual recognition is the rule. With certain individuals close bonds may be maintained for long stretches of the life‐cycle. With a number of others there may be a less close but sustained relationship. Other animals may either be of little interest or else be carefully avoided.
Maintenance of an animal within that particular part of the ecological environment which it happens to frequent and in proximity to those particular individuals of its species with which it happens to associate I propose to term personal‐environmental homeostasis.2 Ethological evidence shows that preferences of personal companions and of environment usually develop early in life and are mediated by learning processes of an imprinting kind. In general, whatever is familiar is preferred to whatever is strange.
In order for an animal to maintain personal‐environmental homeostasis, and in order, too, for it to find its way to particular parts of that environment and to treat appropriately different individuals in it, the animal must have available two working models, one of the environment and the other of the self as agent. There is good reason to believe that working models of both these types are built in the brain. In addition, there is evidence suggesting that, once built, these working models remain relatively stable, namely, relatively impermeable to dissonant information.
Maintenance of working models in a stable and relatively unchanging state I propose to term representational homeostasis. Representational stability appears to be maintained by cognitive processes that accept information compatible with an existing model and that reject, or scrutinise with great caution, information that is, or at least seems, incompatible.
The survival value of morphological, physiological and ecological homeostasis is, of course, not in doubt. The survival value of the latter two categories of homeostasis – personal environmental and representational – is by contrast, open to debate. Nevertheless, there is good reason to believe that at least the first of the two aids survival. By maintaining a familiar physical environment and familiar companions an animal is more likely to be able to find food and drink, and more especially to achieve protection from natural hazards – from predators, from eating poisonous foodstuffs, from falling and drowning, from cold and rain.
The survival value of representational homeostasis may seem more problematic. Nevertheless a good case can be advanced. It seems not unlikely, for example, that perceptual constancy, the advantages of which are not disrupted, is itself an aspect of representational homeostasis. Indeed, another way of describing representational homeostasis would be ‘conceptual homeostasis’.
When the uses to which a working model is put are considered, the advantages of ‘conceptual constancy’ become apparent. Once a working model has been built it becomes a tool with which information is processed, classified and filed, plans are framed, and their execution is monitored. Whenever a working model is undergoing revision of more than minor degree it is to that extent unserviceable. Perception and inference are less certain or even confused; planning is less prompt, execution less practiced. Furthermore, since shared plans can only be conceived and executed in collaboration with others provided that working models are also shared, an individual holding an idiosyncratic model of the world or of himself is likely to find himself facing the world alone.
Whilst it seems likely that the revision of working models tends always to be resisted, their cautious extension in familiar directions may be accepted fairly readily. Science is a social process whereby extensions of working models can come to be agreed; whilst in a scientific community even agreed revisions of working models are, in the long term, not impossible.
Two sciences have been concerned with the phenomena of personal‐environmental homeostasis: they are ethology, notably the work on imprinting, and the object‐relations approach within psychoanalysis, notably the views advanced by Fairbairn. In neither case, however, have workers been greatly concerned with the homeostatic properties of the phenomena studied. Insofar as psychoanalysts have invoked the concept of homeostasis, they have formulated it either in terms of maintaining a hypothetical psychic energy, or tension, within certain limits (e.g., Freud, Menninger) or else in terms of self‐esteem (Engel; Sandler & Joffe). Within the scheme advanced here, maintenance of self‐esteem is viewed as constituting a special case of representational homeostasis. Formulations regarding quantities of psychic energy or levels of tension are not found useful.3
Other disciplines interested in the phenomena are the school of cognitive psychologists studying cognitive dissonance (Festinger, 1957)4 and certain traditions in the philosophy of science (e.g. Kuhn, 1963).5 It is noteworthy, however, that in both cases there is a tendency to be more concerned with the disadvantages than the advantages of ‘conceptual constancy’.
Within any of the five categories of homeostasis described states are never maintained more than relatively stable nor, except rarely, do set‐points and limits persist unchanged during the life‐cycle. In describing
