Epistemology is the branch of philosophy that studies knowledge. It attempts to answer the basic question: what distinguishes true (adequate) knowledge from false (inadequate) knowledge? Practically, this questions translates into issues of scientific methodology: how can one develop theories or models that are better than competing theories? It also forms one of the pillars of the new sciences of cognition, which developed from the information processing approach to psychology, and from artificial intelligence, as an attempt to develop computer programs that mimic a human's capacity to use knowledge in an intelligent way.

When we look at the history of epistemology, we can discern a clear trend, in spite of the confusion of many seemingly contradictory positions. The first theories of knowledge stressed its absolute, permanent character, whereas the later theories put the emphasis on its relativity or situation-dependence, its continuous development or evolution, and its active interference with the world and its subjects and objects. The whole trend moves from a static, passive view of knowledge towards a more and more adaptive and active one.

Let us start with the Greek philosophers. In Plato's view knowledge is merely an awareness of absolute, universal Ideas or Forms, existing independent of any subject trying to apprehend to them. Though Aristotle puts more emphasis on logical and empirical methods for gathering knowledge, he still accepts the view that such knowledge is an apprehension of necessary and universal principles. Following the Renaissance, two main epistemological positions dominated philosophy: empiricism, which sees knowledge as the product of sensory perception, and rationalism which sees it as the product of rational reflection.

The implementation of empiricism in the newly developed experimental sciences led to a view of knowledge which is still explicitly or implicitly held by many people nowadays: the reflection-correspondence theory. According to this view knowledge results from a kind of mapping or reflection of external objects, through our sensory organs, possibly aided by different observation instruments, to our brain or mind. Though knowledge has no a priori existence, like in Plato's conception, but has to be developed by observation, it is still absolute, in the sense that any piece of proposed knowledge is supposed to either truly correspond to a part of external reality, or not. In that view, we may in practice never reach complete or absolute knowledge, but such knowledge is somehow conceivable as a limit of ever more precise reflections of reality.

The following important theory developed in that period is the Kantian synthesis of rationalism and empiricism. According to Kant, knowledge results from the organization of perceptual data on the basis of inborn cognitive structures, which he calls "categories". Categories include space, time, objects and causality. This epistemology does accept the subjectivity of basic concepts, like space and time, and the impossibility to reach purely objective representations of things-in-themselves. Yet the a priori categories are still static or given.

The next stage of development of epistemology may be called pragmatic. Parts of it can be found in early twentieth century approaches, such as logical positivism, conventionalism, and the "Copenhagen interpretation" of quantum mechanics. This philosophy still dominates most present work in cognitive science and artificial intelligence. According to pragmatic epistemology, knowledge consists of models that attempt to represent the environment in such a way as to maximally simplify problem-solving. It is assumed that no model can ever hope to capture all relevant information, and even if such a complete model would exist, it would be too complicated to use in any practical way. Therefore we must accept the parallel existence of different models, even though they may seem contradictory. The model which is to be chosen depends on the problems that are to be solved. The basic criterion is that the model should produce correct (or approximate) predictions (which may be tested) or problem-solutions, and be as simple as possible. Further questions about the "Ding an Sich" or ultimate reality behind the model are meaningless.

The pragmatic epistemology does not give a clear answer to the question where knowledge or models come from. There is an implicit assumption that models are built from parts of other models and empirical data on the basis of trial-and-error complemented with some heuristics or intuition. A more radical point of departure is offered by constructivism. It assumes that all knowledge is built up from scratch by the subject of knowledge. There are no 'givens', neither objective empirical data or facts, nor inborn categories or cognitive structures. The idea of a correspondence or reflection of external reality is rejected. Because of this lacking connection between models and the things they represent, the danger with constructivism is that it may lead to relativism, to the idea that any model constructed by a subject is as good as any other and that there is no way to distinguish adequate or 'true' knowledge from inadequate or 'false' knowledge.

We can distinguish two approaches trying to avoid such an 'absolute relativism'. The first may be called individual constructivism. It assumes that an individual attempts to reach coherence among the different pieces of knowledge. Constructions that are inconsistent with the bulk of other knowledge that the individual has will tend to be rejected. Constructions that succeed in integrating previously incoherent pieces of knowledge will be maintained. The second, to be called social constructivism, sees consensus between different subjects as the ultimate criterion to judge knowledge. 'Truth' or 'reality' will be accorded only to those constructions on which most people of a social group agree.

In these philosophies, knowledge is seen as largely independent of a hypothetical 'external reality' or environment. As the 'radical' constructivists Maturana and Varela argue, the nervous system of an organism cannot in any absolute way distinguish between a perception (caused by an external phenomenon) and a hallucination (a purely internal event). The only basic criterion is that different mental entities or processes within or between individuals should reach some kind of equilibrium.

Though these constructivistic approaches put much more emphasis on the changing and relative character of knowledge, they are still absolutist in the primacy they give to either social consensus or internal coherence, and their description of construction processes is quite vague and incomplete. A more broad or synthetic outlook is offered by different forms or evolutionary epistemology. Here it is assumed that knowledge is constructed by the subject or group of subjects in order to adapt to their environment in the broad sense. That construction is an on-going process at different levels, biological as well as psychological or social. Construction happens through blind variation of existing pieces of knowledge, and the selective retention of those new combinations that somehow contribute most to the survival and reproduction of the subject(s) within their given environment. Hence we see that the 'external world' again enters the picture, although no objective reflection or correspondence is assumed, only an equilibrium between the products of internal variation and different (internal or external) selection criteria. Any form of absolutism or permanence has disappeared in this approach, but knowledge is basically still a passive instrument developed by organisms in order to help them in their quest for survival.

A most recent, and perhaps most radical approach, extends this evolutionary view in order to make knowledge actively pursue goals of its own. This approach, which as yet has not had the time to develop a proper epistemology, may be called memetics. It notes that knowledge can be transmitted from one subject to another, and thereby loses its dependence on any single individual. A piece of knowledge that can be transmitted or replicated in such a way is called a 'meme'. The death of an individual carrying a certain meme now no longer implies the elimination of that piece of knowledge, as evolutionary epistemology would assume. As long as a meme spreads more quickly to new carriers, than that its carriers die, the meme will proliferate, even though the knowledge it induces in any individual carrier may be wholly inadequate and even dangerous to survival. In this view a piece of knowledge may be successful (in the sense that it is common or has many carriers) even though its predictions may be totally wrong, as long as it is sufficiently 'convincing' to new carriers. Here we see a picture where even the subject of knowledge has lost his primacy, and knowledge becomes a force of its own with proper goals and ways of developing itself. That this is realistic can be illustrated by the many superstitions, fads, and irrational beliefs that have spread over the globe, sometimes with a frightening speed.

Like social constructivism, memetics attracts the attention to communication and social processes in the development of knowledge, but instead of seeing knowledge as constructed by the social system, it rather sees social systems as constructed by knowledge processes. Indeed, a social group can be defined by the fact that all its members share the same meme (Heylighen, 1992). Even the concept of 'self', that which distinguishes a person as a individual, can be considered as a piece of knowledge, constructed through social processes (HarrŽ, 19), and hence a result of memetic evolution. From a constructivist approach, where knowledge is constructed by individuals or society, we have moved to a memetic approach, which sees society and even individuality as byproducts constructed by an ongoing evolution of independent fragments of knowledge competing for domination.

We have come very far indeed from Plato's immutable and absolute Ideas, residing in an abstract realm far from concrete objects or subjects, or from the naive realism of the reflection-correspondence theory, where knowledge is merely an image of external objects and their relations. At this stage, the temptation would be strong to lapse into a purely anarchistic or relativistic attitude, stating that 'anything goes', and that it would be impossible to formulate any reliable and general criteria to distinguish 'good' or adequate pieces of knowledge from bad or inadequate ones. Yet in most practical situations, our intuition does help us to distinguish perceptions from dreams or hallucinations, and unreliable predictions ('I am going to win the lottery') from reliable ones ('The sun will come up tomorrow morning'). And an evolutionary theory still assumes a natural selection which can be understood to a certain degree. Hence we may assume that it is possible to identify selection criteria, but one of the lessons of this historical overview will be that we should avoid to quickly formulate one absolute criterion. Neither correspondence, nor coherence or consensus, and not even survivability, are sufficient to ground a theory of knowledge. At this stage we can only hope to find multiple, independent, and sometimes contradictory criteria, whose judgment may quickly become obsolete. Yet if we would succeed to formulate these criteria clearly, within a simple and general conceptual framework, we would have an epistemology that synthesizes and extends all of the traditional and less traditional philosophies above.