Today’s New APPS Interview is with Ezequiel Di Paolo, Ikerbasque Research Professor at the University of the Basque Country, San Sebastian, Spain. The subject of today’s interview is the new collection Enaction: Toward a New Paradigm for Cognitive Science (MIT Press, 2010), of which Di Paolo is a co-editor, along with John Stewart and Olivier Gapenne.
Many thanks, Ezequiel, for taking part in this interview. Let’s set the stage for those readers who may be relatively new to this field. Enaction is a developing “third way” in cognitive science, beside the two dominant paradigms of classical computationalism (cognition works via the rule-bound manipulation of discrete symbols) and connectionism (cognition works via the emergence of global states in neural networks). From the enactivist viewpoint, computationalism and connectionism are united in their functionalism. Can you tell us in what ways enaction takes its distance from functionalism as a basic presupposition in cognitive science?
The dominant paradigm in cognitive science has been functionalism in its computational-representational form. Connectionism is a correction to how this had initially been understood in AI (rules implemented via symbolic manipulation) but remains placed well within the functionalist paradigm.
What is interesting to me about functionalism is not its conception of the mind as a form of computation (whether in our brains or bodies or extended in the world, it doesn’t make a big difference). What I find interesting is that simply saying that the mind is a form of information processing is to say very little. What kind of computation, specifically? Is computation as a concept co-extensive with cognition? Why not? What makes some systems cognitive and not others?
In order to answer such questions we must first realize that functionalism is precisely designed to avoid them! And this is exactly the power of the information processing metaphor and the use of representational language. It doesn’t say anything, but it allows a research discipline to function.
Asking big questions like what is cognition?, what is meaning,? what is behavior?, what is agency?, what is the social?, subverts the functionalist project because such questions point to its “foundational crime”. Like nation states and the rule of law, scientific disciplines also function by a consensual blindness about their origins. We conspire not to look at the fact that our disciplinary edifices are often built over a mud of ignorance. It’s impolite to ask a biologist about the nature of life, a physicist about what is matter or a cognitive scientist about what is the mind. I have done it and the first reaction is invariably one of consternation, as if I had mentioned something obscene.
How does enaction address these unasked fundamental questions?
The enactive approach is precisely an attempt at a new start where we can hope to ask and begin to answer some of those questions. Only recently are we beginning to realize this ourselves. Enaction began as the articulation of several dissatisfactions with the functionalist paradigm, dissatisfactions that had to do with the role of embodied experience, of subjectivity and meaning-making as fundamental aspects of cognition and therefore as fundamental explananda for a science of the mind.
We’re still facing the messy beginnings in this approach, still exploring our precursors, finding affinities with research lines in different disciplines that are enactive in all but name and clarifying open issues, getting at better foundations. Over the years the approach has gained momentum, rigour and articulation, and more importantly, it has engaged with empirical work, modeling work and practical work. The contributions to this book illustrate some of these changes and give a spectrum of the constellation of enactive ideas in different disciplines.
So, what is the enactive approach in a nutshell, if I can ask it so crudely? Your essay with Hanne De Jaegher and Marieke Rohde is something of a manifesto; can you tell us the main rubrics under which enaction proceeds?
It’s a form of non-reductive, naturalist approach to the mind. Its point of departure is the embodied experience of the situated, autonomous cognizer; inherently bound not just to its materiality but also to a network of significance that is its relation to the world, but most importantly, always active and able to alter this world and in part also itself.
The core ideas of enaction are the notions of autonomy, experience and sense-making. To these we add a general concern for embodied action and emergence as an explanatory tool. The idea is to seek continuities between the concepts that emerge when we attempt to understand life and explore when and how these concepts tell us something that we can use to understand the mind.
Enaction maintains close ties with the notion of autopoiesis developed by Humberto Maturana and Francisco Varela. However, in a widely cited article from 2005 you found it necessary to supplement the notion of autopoiesis with that of “adaptivity.” Can you tell us what that notion entails, and why it was necessary to develop that new concept?
Yes, there are indeed close ties between enaction and the theory of autopoiesis (a theory that proposes that the defining characteristic of living organisms is their organization as self-producing systems).
In part this is because Francisco Varela was one of the originators of that theory and of the enactive approach. Many of Varela’s motivations in criticizing traditional cognitive science and philosophy of mind are the exact same motivations that made him think (together with Maturana) of alternative, systemic ways of conceptualizing life.
So we have a link through the intellectual history of one thinker and several colleagues. However, a certain maturity has been reached in enactive thinking and reaching this maturity has demanded that we think and in some cases carefully re-think the role of its core concepts. This can sometimes mean putting the autopoietic conceptual toolkit to do a job that it may not have been originally designed to do, which entails, inevitably, amending, extending or criticizing this set of concepts. This is a fascinating exercise.
What was happening to autopoiesis before this mature self-reflection?
For years, on some interpretations (certainly not Varela’s), the theory of autopoiesis looked like an impenetrable fortress. You were either outside it and it seemed to you like a sort of cult or dogma (its textual apparatus is uninviting to say the least), or you were within it (maybe because you’ve been teleported there) and it looked to you that every question in the world could be answered by deriving the appropriate statement from the theory’s axioms.
This, in my opinion, was not a good situation because a good set of conceptual tools needs to generate enough friction to be useful. Autopoiesis as a theory was running the risk of spinning frictionless in a void, with the result that most scientists (especially biologists) simply have ignored the idea to this day (though this is changing in some areas of biology, like origins of life studies).
How did it extricate itself from this stalemate?
It turns out that Varela intuitively saw a connection between ideas like self-production and self-maintenance of identities and some aspects of the mind, like being teleological in nature, being intrinsically concerned with the world (as opposed to being indifferent to it as in the case of functionalist cognitive architectures that must always be artificially kicked into motion by the action of “boxes” with labels such as “motivation”).
In his later work Varela tried to argue for this connection. He got inspiration from the bio-philosophy of Hans Jonas, who had been saying very similar things; linking the logic of metabolism (a stable form made out of a material flux; dependent on matter but in some way also under-determined by it) with basic aspects of existence: concern for ourselves, for our world, etc (see for instance, Renaud Barbaras’ chapter in the book).
In attempting to establish this connection, Varela effectively opened up the autopoietic fortress, because he tried to make it do a job that (I argued later in my 2005 paper) it wasn’t quite fit to do. Varela’s move was a change of perspective and the resulting parallax effect revealed that the theory was not really accounting for some interesting biological phenomena (like illness, fatigue, ageing, etc) or important aspects of cognition (like norms, the nature of agency and so on).
What was the result of this change of perspective?
Maybe investigating these phenomena was not the main objectives of the theory but the enactive approach, if it was to use some of the autopoietic concepts as its basis for a naturalization program, needed to make some changes. One of them was to introduce the concept of adaptivity. Initially, it seemed it was an idea that would merely patch up Varela’s argument that life already implies mind (more technically: that autopoiesis implies sense-making or relating to the world in terms of values and norms), but it turned out that adaptivity could be used to develop other concepts such as a grounded notion of behaviour, temporality, agency, habits, and has had a strong role to play in enactive ideas about social cognition.
In short, the notion of adaptivity intends to capture those cases where the living system is not only self-producing, but is also capable of acting either in the world or within the system itself and in this way recognize situations that may result in some risk for the system, avert dangerous situations or, if they occur, try to take remedial action. The notion seems teleological, but it can be formulated in strictly operational terms, thus it serves as a complementary concept for the theory of autopoiesis. I argued that without some version of adaptivity, you couldn’t have a sense-making organism, no matter if it was autopoieitic.
In a formula, sense-making implies autopoiesis plus adaptivity, which is to say that mind implies a precarious self-generated identity able, at least partially, to act adaptively and avoid situations that may lead into some breakdown.
Staying with Varela’s contributions for a moment, one of the things that most attracted me to enaction (or as it was called in an earlier incarnation, “the embodied mind”) is its incorporation of phenomenology into its presuppositions. Briefly, the “neurophenomenological” approach of Varela put the first-person perspective of phenomenology into communication with the third-person perspective of laboratory science. What is at stake here?
This turn towards experience is one of the central aspects of enaction. You will see several chapters in the book exploring this connection, such as those by Renaud Barbaras, John Stewart, Véronique Havelange, Michel Le van Quyen, to mention a few.
I think there are several things that can be said about this. I find at least two aspects of this connection particularly interesting.
The first one is the concern with taking embodied experience seriously, not merely as “data” that deserves a functional explanation, but as part of a method of reflection and guidance of theoretical and empirical work. I think enaction was born with this concern with experience and to me it is still one of its strongest (and most challenging) points.
However, I would like to immediately add that the enactive approach is definitely not the same as phenomenology. But it does attempt to enter into a serious dialogue with phenomenology. This is the second interesting aspect for me. The methodological questions that are raised by even trying to conceive of this dialogue and how in actual practice and in concrete cases such questions are often resolved without major trouble while in the broader perspective they seem almost unsolvable in principle.
The neurophenomenological approach is one example of what we may call a “combined practice”. It puts together third-person scientific rigour and a solid place for first-personal experience (see Le van Quyen’s chapter for concrete results of this approach). Vasu Reddy’s attempt at developing second-person approaches to understand the experience of social engagement in infants would be another such combined practice. And that’s the thing. The dialogue between scientific and phenomenological approaches can definitely work in practice even if it doesn’t quite yet work in theory!
In your opinion what is it that each side gains from this encounter?
I definitely think that all sides stand to gain something and not simply that phenomenology informs enaction. However, a change of mindset might still be necessary both on the part of scientists and traditional phenomenologists for this to work.
The issue may be less of a problem with other practices of experience such as mindfulness traditions or some forms of psychoanalysis, but both phenomenology and third-person science have a strong barrier to dialogue and this is that they pretend to be totalizing practices, i.e., practices that give us access to knowledge from first principles and prior to any other human practice. Thus traditional science would like to explain experience using third-personal methods, and phenomenology would like to explain how we could do science at all as part of the structures of our lifeworld. These are both very dogmatic attitudes and you won’t shake some people out of them.
But some dialogue is occurring, no?
Yes, these dialogues are indeed taking place. Not between hard-core phenomenologists and hard-core scientists, but between people who care more about the practice of dialogue (including its enjoyments and frustrations) than about first principles.
In particular, I would like very much to see phenomenology changed by this dialogue, not just science. I think this suggestion may be anathema to many phenomenologists but that’s because they may have indeed forgotten the lessons of taking their own experience seriously at a personal level. The structures of our experience are plastic and this means that knowledge and practices (for instance, scientific knowledge and scientific practices) can be re-inscribed into our personal and social bodies so that these structures of experience can change.
We need to make some distinctions here. The attempt to reason at the limits of phenomenology exclusively from within phenomenology is not the same as thinking the relation of phenomenology and science.
Take for example Michel Henry’s intra-phenomenological work, where his radically immanent notion of self-affection forced him to defend an extreme ontological stance and admit that indeed at the core of our experience there is a radical gap separating our innermost self-experience from the appearance of things in the world.
Contrast that with the attempt to benefit from the parallax effect of jumping between phenomenology and science, where we might realize that the radical self-enclosure Henry finds in self-affection also happens in material systems seen as autopoietic systems. This self-enclosure in no way signifies a disconnection from their world but a dialectical relation of self-distinction and co-dependence with the world. When we think closely about autopoietic systems and their worlds, we realize that we can rethink self-affection. In fact, the ambiguity of its radical self-enclosure had already been noticed, for instance by Derrida.
In short, by drawing parallels with explanations and descriptions in science, we complement our phenomenological insights not with new information but with new tools for interpretation and thinking.
Let’s now move from the encounter of the first- and third-person perspectives that neurophenomenology stages to the second-person perspective. Another very important move in enaction is developing ways to analyze the cognitive dynamics of social encounters, but precisely in a way that does not fall into either the first-person perspective of “simulation theory” (I understand someone else by consulting in myself an internally generated model of their action) or the third-person perspective of “theory theory” (I understand someone else by making inferences based on observations of their external behavior). Can you tell us more about the work that you and your colleague Hanne De Jaegher have done in developing a “social interactionist” paradigm that takes seriously the second-person perspective in what it calls “participatory sense-making”?
Yes, this is described in our chapter in the book, but other chapters are also relevant (e.g., Bottineau’s and Stewart’s chapters and to a large extent also Hutchins’ as it deals with mediated cognition). Although there has been a recognition in enactive circles that you cannot have a science of the human mind if you don’t take into account the intersubjective and cultural life of human beings, this recognition, as in most areas of cognitive science, fell short of providing a theoretical account of social cognition and intersubjectivity.
Our motivation with Hanne has been to see whether we could use the concepts that were very much still in development at that point in the enactive approach to take an alternative perspective not only on social cognition but on our social life in general.
Hanne discovered during her DPhil thesis that there was something peculiar to how social cognition is studied in psychology and neuroscience: there was nothing especially social about it!
Exactly as you point out above in your notion of the “foundational crime”: the notion of the social was not critically examined beforehand!
Indeed, what was being investigated under the name of “social cognition” was in fact the same as individual cognition, only applied to social situations. In fact, what counts as “social” is never even properly defined (beyond trivialities such as the presence of other people). In particular, the paradigmatic cases of social cognition involved a detached observation of a social scene, rarely a participation in it. The underlying assumption, typical of cognitivism, is that we must first understand in order to act. This assumption is criticized (and often exactly overturned) in the enactive approach.
If you don’t critically examine the cognitivist assumption of the primacy of understanding, you keep on thinking that social interaction can be explained secondarily as a skill that demands individual cognitive mechanisms of mindreading. Not only is this a very linear and classical model of the social mind, but also the first thing that, as enactivists, we noticed is how this clashes with our everyday experience of the social world. We’re thrown into it and often must skillfully participate in multiple interactions without really being able to take the perspective to think, make inferences or simulate mental states.
What was your response?
So Hanne and I set out to try to answer two questions using some of the concepts that have been recently developed in the enactive literature: in particular the concepts of autonomy and the concept of sense-making.
The first question is what makes a social encounter social? We noticed that it is not merely the presence of another but that some notion of engagement is needed to speak of an interaction. We are coupled to our world via sensorimotor loops as well as other physical and biological forms of coupling. What makes us autonomous agents is our capacity to regulate our encounters with the world, our coupling, following some norms that are given by our autonomy (our self-sustained identities). An interesting possibility is opened when we encounter other autonomous agents: we can establish a mutually regulated link that can itself acquire a transient form of autonomy. We call this a social interaction.
What are some examples of autonomous social interactions?
It’s interesting to notice that in our experience we are often thrown into interactions that we’re not particularly keen in pursuing, and for some reason or other we are “stuck” in them. Examples are narrow corridor situations when we cannot coordinate how to get past someone who’s walking in the opposite direction, or similarly trying to avoid entering into a fight when there has been a historical pattern of antagonism, or the trailing off phone conversation when we keep repeating “good-bye” to each other but take a long time to hang up. The other person in these interactions is very likely in the same situation that we are, they want to end it.
But it is a sign of the autonomy of such interactive patterns that, in spite of our individual intentions, they take a life of their own and refuse to simply end. If we were mindreading, as the tradition sustains, this would never happen.
So for the enactivist, a social interaction can be defined very precisely as the encounter of two or more autonomous agents that regulate their coupling and in so doing the dynamics of this coupling acquires an autonomy of its own, while the agents don’t lose their autonomy in the process.
How does this autonomy of social interaction relate to social cognition?
The other question that we addressed was, given this notion of social interaction, can we say anything about social cognition? If we relate to our world as agents, in terms of norms, we are sense-makers. We act and regulate our actions in terms of the value of a given worldly encounter for our continuation as the forms of life that we are. What happens in interaction (and this is backed up by decades of empirical research across ages and across cultures) is that our patterns of embodied movement, our utterances and our postures get coordinated. Sometimes coordination is lost, then recovered. This is a dynamical phenomenon.
But the consequences of this have not been properly fleshed out until recently. Because as embodied sense-makers, the result of our coordination with others (often unwittingly or even sometimes explicitly unintended) implies that the other is in some way participating in my sense-making process. It is thanks to the active participation of the other that my sense-making is changed, not merely as a result of reflection or inference, but directly as an embodied encounter where I’m often literally re-oriented in my sense-making and in turn I do the same to the other.
This is what we call participatory sense-making. It’s a broad term and it covers a variety of possibilities of participation, from the orientation of someone’s sense-making (e.g., I move a shiny new toy in front of the baby’s eyes) to joint-sense-making such as situations of play, working together (we often cannot trace where an idea came from when we work in a fruitful and strong collaboration) and acts that are inherently social, like the act of giving.
We can’t go through each article in this rich collection, but perhaps we should talk about Diego Cosmelli and Evan Thompson’s tour de force “Embodiment or Envatment?” in which they show that the “vat” in the “brain-in-a-vat” thought experiment would actually have to be “a body in the world”! I’d like to invite you then to discuss some of the outlines of this marvelous argument.
This is a very nice contribution to the book. In a way it’s an alternative introduction to enactive ideas, by working through the contradictions of functionalism as exemplified in the well-known brain-in-a-vat thought experiment. In effect, this story is meant to reveal to us with more clarity the meaning of accepting that our mental states are first and foremost constituted by brain states. If that were the case, we could suitably wire a brain into experiencing a Matrix-like world when, in reality, this brain would be safely stored in a vat and kept alive.
Diego and Evan’s argument takes a fascinating route. They propose: OK, let’s try to do exactly that! What would you need to keep this brain alive? What would you need to accurately simulate the right electrical inputs to the neurons to simulate a world? Well, the argument goes, in the end you require nothing short of a living autonomous body situated within a world (I recommend following the argument in detail).
The whole exercise doesn’t prove or disprove the question of whether mental states are only brain states, but it breaks down the “naturalness” that such an assumption seems to have if we accept uncritically the brain-in-a-vat situation. It turns out that this assumption cannot be supported by this intuition pump because when you think about it, what the thought experiment suggests is exactly the opposite assumption, i.e., that you cannot do without a body and a world in order to have mental states, and that their roles might well be constitutive of those states.
What do you think this article demonstrates about the role of thought experiments in cognitive science?
Good thought experiments are essential for science. Especially when we are dealing with such complex, multi-level and non-linear systems such as organic bodies in a complex social world. We simply don’t have the right intuitions and we end up most of the time thinking in terms of billiard-ball causality and simple linear explanations.
But thought experiments can obscure as much as they can reveal. So we must be careful. We should always recognize the difference between a thought experiment and an explanation. The thought experiment can support or give us intuitions about the direction where an argument should go, but it cannot replace it.
It turns out that the brain-in-a-vat thought experiment was indeed a much better one than we previously imagined! It ended up pointing our intuitions in exactly the opposite direction than its originators intended. Poor examples of an idea often under-represent this idea, but rich ones often contain an excess that makes us eventually realize that our initial idea may not have been quite right. That’s when thought experiments lead to real progress. In my reading, Diego and Evan have shown that the brain-in-a-vat is a case of a rich thought experiment.
Many thanks, Ezequiel, for helping us to see some of the contours of enaction (the emerging paradigm) and Enaction (the book). I’m very much looking forward to the reviews of the book and the continued progress of the paradigm.
Thank you, John.
[Editor's note: this will be our final New APPS interview for the time being. We'll resume in mid-September with interviews with Graham Harman, Mark Wrathall, and others.]