In formalizing our ideas about binding and gaze, we have been moving toward a theory of intentionality that depends on the agent's embedding in its world, rather than solely upon its internal models of that world. An agent can keep track of particular objects in terms of their functional significance - the roles that they play in the ongoing activity. And it can keep track of the tools and materials associated with different tasks by keeping them in different locations, for example different regions of a countertop. So far, however, our ideas on the subject have been limited to very simple cases, for example an agent switching its visual focus back and forth between two objects. To model the more complex patterns that are found in everyday life, we need a much better theory of the world in which we are embedded. This theory is partially a matter of biology and physics, of course, but it is also a matter of cultural practices for organizing activities in space. In this section, we would like to sketch a more general theory of these matters using the concept of ``cognitive autopoiesis.''
For Maturana and Varela [26], autopoiesis refers to the processes by which organisms act on their environments in order to provide the conditions for their own continued functioning. Cognitive autopoiesis refers to the active means by which agents structure their environments in order to provide the conditions for the own cognitive activities. These include most basically the means by which agents provide for the factorability of environments: engaging in customary activities, using the customary tools and materials for them, partitioning the activities in the customary ways, and so on. But it also includes a range of more subtle phenomena. Kirsh [19], for example, has drawn the useful distinction between actions that aim at achieving functional goals (beating eggs, sweeping floors) and actions that aim at facilitating cognition (setting out the right number of eggs at the beginning, opening the curtains so that dust will be more visible). Actions can, of course, serve both purposes, for example when one chooses to boil water in a kettle rather than a saucepan: each strategy achieves the result, but the latter will also provide a sign that it is possible to take the next action, for example preparing tea. Stabilization actions [16] also provide the cognitive conditions for other actions. One might, for example, develop the habit of leaving items by the door the moment one realizes that they need to be taken in to work.
These phenomena help in understanding what is inadequate about the concept of ``the environment.'' If one conceptualizes ``the environment'' as a monolithic whole, perhaps the way it looks when viewed from an airplane, or else the way it looks when understood through the peephole of a momentary vector of sense-perceptions, it begins to seem arbitrary, chaotic, or hostile. In a certain sense it seems static, as if it has an anatomy but no physiology. But in fact the phenomena of cognitive autopoiesis reveal that the lifeworld has a great deal of living structure, and that this structure is actively maintained by agents while also providing crucial preconditions for their own cognition. Indeed it is hard to draw a clear line around an agent's cognition; if we trace the sequence of causal events that led a given agent to pour a pitcher of milk on a particular moment, this sequence will lead back and forth between the agent and its customary surroundings. It is almost as if these surroundings were an extension of one's mind.
Cognitive autopoiesis is a complex and multifaceted phenomenon and no single theory will suffice to explain it. One useful way to think about cognitive autopoiesis is spatially, in terms of a series of buffer zones between the embodied agent and the putative dangers and complexities of ``the environment.'' For people whose lives are similar to our own, these buffer zones can be conveniently sorted under six headings:
These buffer zones do not always offer perfect protection from harm or complete support for the pursuit of goals. Shared and public spaces can be sites of conflict, for example, and these conflicts can include involuntary disruption or destruction of one's body and the other buffer zones that are customarily under one's own private control. A serious theory of activity must include an account of these phenomena as well, which are usually just as orderly in their own way as anything else.
In any event, the nested buffer zones of ordinary life participate in a large metabolism that continually interweaves cognitive and functional purposes. Among these purposes is learning. Just as the adaptation of body parts and tools to customary activities helps channel action in customary directions, so does the existing background of objects, spaces, and practices help channel the actions of children and other newcomers in customary directions on a larger scale. Caretakers regularly construct customized types of buffer zones around the young, for example, so that it is difficult or impossible for them to get into anything that could cause harm. The lifeworld of a child, for example, differs from that of an adult who can reach up to the cookie jar and into the locked cupboard where the roach spray is kept. A growing literature has investigated the processes of cognitive apprenticeship [30], situated learning [23], distributed cognition [18, 35], and shared construction of activities [15] that go on in these systematically restrictive and supportive lifeworlds.