The attention has been driven to learners' conceptions when one tried to make sense of errors and not only focused on getting rid of them. Taking an ecological perspective, which considers meaning as a construction resulting from a process of adaptation (modelled by the constructivist theories), the background of an error couldn't be only contingent. Indeed there are errors which are due to a default of attention or a kind of accident. But important errors from an epistemic and learning perspective are those which repeat themselves in a significant way within a population and/or within a specified type of situation. These errors are good candidates to be symptoms of a certain understanding or a certain meaning; in short: a conception. It is this conception that we have to characterise. Indeed, we cannot tell much about what happens in human beings heads at a knowledge level (in Newell terms). What we can start from is what we observe: the behaviours as they are perceivable in the course of an activity taking place in a situation we can document. These behaviours are always a component of an interaction, a dynamic sequence of actions and feedback exchanged by the human being and an environment (which could be "natural", "social", "virtual", etc). Eventually, what we observe is not just a human being, but a human being interacting with an environment in order to achieve a certain goal. The human being and his or her environment forms a system dedicated to the achievement of this goal. Then a conception can be characterised by the following four elements.

- The cluster of the tasks or problems which are at stake (we will label it "P")
- The set of the operations which can be performed (we will label it "R")
- The means of representation which are available (we will label it "L")
- The control structure which describes the means to take decisions (we label it "Σ")

The quadruplet (P, R, L, Σ) constitutes the ck¢ characterisation of a conception.

.