Contributed by: Marek McGann, University of Limerick, Ireland.
The enactive approach is a way of thinking about what cognition is generally. It relates cognition to processes of living and biological autonomy,
and in particular to the interaction between autonomous agents and their environments. In this view, cognition is not something that occurs “inside” of an agent,
but is a product of that interaction, much like a handshake is the product of an interaction. There are five key "pillars" to the enactive approach:
A dynamical systems perspective and emergence, embodiment, biological autonomy, "sense-making" (the creation of meaning), and experience.
Some key implications result from these principles for how we think about and how we investigate cognition.
A very brief history of the enactive approach
If you've only recently come across the enactive approach, you might be a tad confused about just what it entails.
The term “enactive” has become something of a buzzword in the Cognitive Science literature of late, and as such is actually used in a few different ways by different researchers.
This briefing is about the enactive approach as inaugurated by Varela, Thompson & Rosch (1991)? in their book The Embodied Mind. It is related to but separable from the dynamic
sensorimotor approach to perception argued for by the likes of O'Regan & Noë (O'Regan & Noë, 2001; see also Noë, 2005), Hurley (1998), and it developed largely independently of
Bruner's account of enactive learning (Bruner, 1966, 1968).
The enactive approach broached in The Embodied Mind stressed the fact that cognitive activity doesn't take place in a vacuum, but in a world where a cognitive agent is trying to get something done – whether that's get to work, make a cup of coffee or figure out what this “enactive approach” thing is (in the jargon, this is an embedded, embodied approach to understanding cognition). The approach also makes central the essential relationship with the experiences of the cognitive agent and the meaningfulness of that experience. Thinking isn't abstract for the thinker – it really means something and feels like something. The concept of biological autonomy is also important in the enactive approach, the fact that we are not just pushed around by our environment, but have some control over our own behaviour as living agents.
Though the concepts of biological autonomy (in particular, autopoiesis) that Varela had worked on earlier in collaboration with Humberto Maturana (Maturana & Varela, 1980, see also the more readable: 1987)? seemed eminently consistent with the enactive approach, they were not explicitly discussed in The Embodied Mind, but gradually melded into the enactive way of thinking over the following decade (Varela, 1997; Weber & Varela, 2002; Thompson, 2007) Though there is inevitably some differences in the way enactive theorists think, at present the enactive approach can be understood as placing a strong focus on the interaction between an autonomous biological agent and its environment. It addresses such difficult questions as meaning and experience, and sees cognition as dynamical and emergent from that interaction.
Five key elements of the enactive approach
Thompson (2007) outlines five essential components to an enactive approach to cognition. Adapted from Thompson (2007, p.13) these five principles are:
Once you're taking all of these elements seriously, you're taking an enactive approach to understanding the mind.
- Living beings are autonomous. They actively generate and maintain themselves, and in doing so “bring forth” their own cognitive domains.
This is to say that all cognitive systems must be agentive systems, having some capacity to direct and control themselves, in order to maintain their own living existence.
Put simply, the agent makes itself, it doesn't depend on someone else to tell it how to exist or what is essential to it – all that arises directly from how the agent itself works.
- The nervous system is an autonomous dynamical system. A nervous system is not like a symbolic computational system, which accepts information from the "outside world"
and processes that information in some way. Rather, it is governed by its own internal dynamics (the complex tangle of feedback loops between its neurons).
The central nervous system is a mess of interconnection. Any neuron in the CNS is more affected by other CNS neurons than any “input” coming in from the peripheral nerves.
Thus, the state of a nervous system cannot be determined by external events, it can simply be perturbed by those events, and its activity at any given time is a complex result
of its own activity and those perturbations.
- Cognition is the exercise of skillful know-how in situated and embodied action. Cognition never occurs in abstraction, but rather in some context which
includes not only the agent's environment by their own embodiment. Cognition is the adaptive coordination and control of actions, but thus cannot be discussed
or considered without reference to both the physical instantiation of the agent, and that agent's own environment.
- The world of the agent is not a prespecified external realm, represented internally by the agent's brain, but is a relational domain "brought forth" by the agent's
autonomy and coupling with the environment. This “bringing forth” is also referred to as “sense making”, the agent's ability to give meaning to its environment.
There is simply no way to specify external reality without making at least implicit reference to the perspective, capabilities and intentions of the agent.
Any description of the world must always acknowledge the perspectival nature of any such description. Trying to understand this can get people and knots,
but it's simply stating this: There's no such things as red in the world – there's wavelength and reflectances, but it only becomes red when we see it.
There is no such thing as flat in the world – just those parts of the world we can stand on or put things on and expect them to stay there; get too small or too big,
then it stops being flat. End result: the world you live in depends as much on you as it does on the physical environment.
- Finally, experience is not a scientific afterbirth, curious but ultimately possible to ignore while the science goes on. Experience must, rather,
play a central rôle in our understanding of cognition and mind, it must be part of the fundamental data of an adequate mind science.
Cognition is the adaptive interaction between an autonomous agent and its environment. This means that our thinking is not something that occurs in our heads and
is then expressed in our behaviour, but is constituted in the adaptive interaction itself.
Cognition is a bit like a handshake. You can't make a handshake by yourself, and you can't really do it by accident. Handshakes involve reaching, and they involve a
rather subtle form of negotiation: how firmly do you grip? How many pumps? How long do you hold on before you let go? Once you know how to shake hands though,
you can do lots of interesting things, such as decide whether the person you are shaking hands with is confident or timorous, friendly or suspicious, interested or disdainful.
So it is between you and the world in the case of cognition.
This means that it is better to think of cognition as something like a skilled activity. Skills strongly emphasise the processes involved, but they are distinguished by the kinds of
structures they have. The skill of playing chess is in the playing of the chess, but the fact that it's chess-playing rather than, say, draughts-playing,
is because of the structures involved (the board, the rules, the pieces). The structures characterise the activity while the process is the activity itself.
What makes it skilled is that we can get done what we want to get done. When we are learning, just managing to play chess may what we're after; once we've learned,
winning will be the thing.
Understanding cognition from an enactive approach means that we need to understand both the things that structure, guide or scaffold the activities in question
(like environmental obstacles, social norms and biological predispositions), but also what drives those activities. We need to understand the motivations that power the "reaching"
in the handshake of thinking, and that give us our understanding of what differentiates a skill from, for example, a computational process. Put another way, it isn't cognition
until somebody cares (and the caring has to be done by the agent themselves, we can't do it for them). From an enactive perspective, the emphases on autonomy and sense-making focus our thinking about thinking in this regard.
Implications for our investigations of cognition
Taking an enactive view of things means that some of the things that cognitive scientists normally take for granted will have to be thought through a little more.
For instance, the typical assumption in experiments on cognitive psychology is that everyone essentially has the same standard issue set of computational machinery
and we use it in more or less the same way in all situations. An enactive view sees the person as a tangle of skills and motivations which will be in a more or less
unique combination in a given individual We'll have a lot in common because our biology and our culture standardises things significantly, but to understand the mind in general,
we will have to be prepared to take account of those individual differences and understand them not as being the same machines in slightly varying packaging,
but as basically varying machines. This does not mean that we throw out fifty years of cognitive research, but it may mean that we reinterpret that research somewhat.
It also means that we should be more careful and explicit about the kinds of motivations and skills that we expect participants in our experiments to have, how they are
likely to interact with the experimental situation and how that interaction may differ in other kinds of situation.
In the domain of artificial cogntive systems, it means taking the autonomy of the system seriously, and trying our best to develop a science with such autonomy at its centre.
The behaviour-based robotics of the likes of Brooks (1999, 1991) make the task to be performed fundamental to the operation of the agent themselves,
and as such share some common ground with the enactive way of thinking. Evolutionary robotics approaches (see Harvey, Di Paolo, Wood, Quinn, & Tuci, 2005),
particularly those which emphasise the evolution of the agent's own value system (e.g. Di Paolo, 2003)? will allow us to examine cognition as emerging in a
dynamic and skillful interaction between the agent and the world, rather than something which must be pre-built into the agent in some way. It is the existence of some kind value that
"belongs" in some sense to the agent, instead of our values as researchers trying to understand cognition, that means we are studying enaction rather than computation.
Brooks, R. A. (1991). Intelligence without representation. Artificial Intelligence, 47(1-3), 139-159.
Brooks, R. A. (1999). Cambrian Intelligence: The Early History of the New AI. . Cambridge, Mass: Bradford Books.
Bruner, J. (1966). Toward a theory of instruction. . Cambridge, Mass: Belknap Press.
Bruner, J. (1968). Processes of cognitive growth: Infancy. . Worcester, Mass.: Clark University Press.
DiPaolo, E. A. (2003). Organismically inspired robotics. In K. Murase & T. Asakura (Eds.), Dynamical systems approach to embodiment and sociality (pp. 19-42). Adelaide: Advanced Knowledge International.
Harvey, I., Di Paolo, E., Wood, R., Quinn, M., & Tuci, E. (2005). Evolutionary robotics: A new scientific tool for studying cognition. ARTIFICIAL LIFE, 11(1-2), 79-98.
Hurley, S. L. (1998). Consciousness in Action. . Cambridge, Mass: Harvard University Press.
Maturana, H. R., & Varela, F. J. (1980). Autopoiesis and Cognition. . Boston: D.Reidel.
Maturana, H. R., & Varela, F. J. (1987). The Tree of Knowledge: The Biological Roots of Human Understanding. . Boston: New Science Library.
Noë, A. (2005). Action in Perception. , 289. MIT Press.
O'Regan, J. K., & Noë, A. (2001). A Sensorimotor Account of Visual Consciousness. Behavioral and Brain Sciences, 11(5), 939-973.
Thompson, E. (2007). Mind in Life Biology, Phenomenology and the Sciences of Mind. (1), 568. Harvard University Press.
Varela, F. (1997). Patterns of life: Intertwining identity and cognition. BRAIN AND COGNITION, 34(1), 72-87.
Varela, F. J., Thompson, E., & Rosch, E. (1991). The Embodied Mind. . Cambridge, Mass: MIT Press.
Weber, A., & Varela, F. J. (2002). Life after Kant: Natural purposes and the autopoietic foundations of biological individuality. Phenomenology and the Cognitive Sciences, 1, 97-125.