Although the subject of the coupling between agents and their environment has been tackled earlier, it was not explored until the early middle part of last century when the rise of Cybernetics established the foundations for a mechanistic, situated and embodied understanding of behavioural systems.

Cybernetics died out due to a lack of proper development in the field of nonlinear mathematics and in our understanding of biological systems, in addition to a lack of computational simulation power.

 In the last decade, various research groups, notably in Evolutionary Robotics have extended the dynamical perspective with models which take time and physical properties seriously, shedding light into many consequences of the dynamical systems approach to life and cognition (DSaLC).

DSaLC can be tackled/explained in various manners, and from different perspectives. In what follows, we briefly portray one of these perspectives. For other approaches please refer to the links bellow.

Suppose that you come across a new type of agent! Beer proposes that as scientists and engineers interested in the adaptive behaviour of this agent we have discovered we ask questions such as: What patterns of behaviours does the agent exhibit? How are these patterns of behaviour generated? Why do these agents generate these particular patterns of behaviours within their particular environments? In other words, what is the fundamental nature of adaptive behaviour? And how do agents generate the appropriate behaviour at the appropriate time? For time is essential, and makes a difference between an adaptive behaviour and a suicidal one. DSaLC's main concern is precisely the synthesis and analysis of living systems (Beer, 1997).

 The framework of DSaLC is deliberately minimal. Thus, it will limit the amount of assumptions in our models. Agents and environments are coupled dynamical systems and an agent is designed to maintain its viability in the environment in which it finds itself. It is difficult to imagine a theoretical framework for adaptive behaviour that makes fewer commitments than this.

Nonetheless, a number of nontrivial consequences follow immediately:

• Adaptive behaviour can not be broken down into (a) agent's internal dynamics and (b) environmental influences.

• Agents must be active and modify their environments to be cognitive and alive!.

• Agents have internal state, and so will be able to respond differently to identical sensory stimuli depending on the state it is in when the stimuli is received.

• All the relations with the environment (e.g. perception) are often viewed as a means by which an agent becomes informed about that environment; in a dynamical perspective, those relations are processes whereby agent dynamics are continuously shaped by the  environment and shapes the environment in turn.

• Importance of the differences in time-scales between agent's learning mechanisms (slower dynamics) and behavioural dynamics (faster dynamics).