The Life & Mind Seminar Network

Summary of seminar #1:

Posted in Seminars by Tom Froese on November 8, 2006

After spending a considerable amount of time writing up a summary and pressing publish, the blog decided to publish an empty post. Ahrg! Oh well…

Here’s the second (and final) attempt:

The following summary is based on some of the notes which I took during Ezequiel’s talk. It goes without saying that this is entirely my own interpretation of the significance of what was being said and discussed. If I missed, misinterpreted, etc., anything of importance (as I’m sure I have), please feel free to post some clarifying comments.

At the start of his talk Ezequiel made some general comments about the enactive approach. The AI and robotics work which is done within this framework is not just (simply put) obstacle avoidance, but there is interesting work going on which is charting new territories. One common and recurring insight of these investigations is that cognition is best understood as always already embodied-embedded.

Evolutionary robotics has emerged as the methodology of choice for generating experimental evidence for enactivism in the field of AI and robotics. It can be interpreted as a kind of subversive science; it is a tool for minimal existence proofs. It can challenge traditional assumptions about the necessity of certain elements (e.g. representations) for the possibility of particular cognitive behaviours. As evolutionary robotics allows us to push forwards into new directions, the traditional claims for the necessity of representations can be seen as a receding horizon.

We are now at the stage where enactivism faces a new challenge: by emphasizing an embodied-embedded approach to cognition, the understanding of a cognitive agent appears to be glued to the here-and-now. How can we explain intelligent behaviour in more “representation-hungry” domains? Ezequiel proposed that perhaps an analysis of the role of play for cognition can provide a way out of the current explanatory short-comings of the enactive paradigm.

However, before we can gain a better understanding of what is involved in playful activity, we need to be clear about the conceptual basis on which such an analysis needs to proceed. Central to an understanding of play is the notion of autonomy, a concept which lies at the very heart of the enactive paradigm. Ezequiel defined a cognitive agent to be autonomous when it needs to affect its own constitution in such a way that we can reasonably say that it builds itself as an entity. On this view, an autonomous agent is a precarious, self-sustaining process of identity generation, and this identity constitutes a proper, irreducible domain of existence.

Combine this form of autonomy with adaptivity and you then have the basis for what Ezequiel called sense-making, namely the active evaluation of the significance of worldly encounters. Sense-making conceived of in this manner is a process which underlies all cognition. It was pointed out that the claim that sense-making is a necessary and sufficient condition for cognition entails the consequence that most of the enactive work in AI and robotics is not actually dealing with cognition, to which Ezequiel agreed. Sense-making should be understood as the construction of a world of significance. It is one more step along the life/mind continuity towards more human-like cognition, a move towards an increasing mediacy between self and other, a necessary step towards tackling more “representation-hungry” problems in line with enactive principles.

When talking about sense-making and the construction of significance it is important to be clear about what is meant by the concept of value. More importantly, what we want is an operational definition: a value is the extent to which a situation affects the viability of an autonomous agent. On this view, value is not something which resides within an agent but emerges out of its concernful interaction with the world in which it is embodied-embedded.

With this conceptual foundation in place, Ezequiel now turned upon the focus of his talk, namely an analysis of the importance of play. The phenomenon of play can be tackled in several different ways. For example, it can be seen as an exploration of the dimensions of the I-can/I-can’t. In the sense that play is interpreted as the construction and exploration of constraints, we can understand it as the bringing forth of a new freedom through the self-imposed adherence to a set of rules. In addition, in pretend-play it is possible to see the roots of the creation of worlds of meaning. During make-belief the object of the pretend-play becomes the target of a similarity/difference dialectic which can ground the skilful coping required for value guided exploration. In this manner play emerges as a form of life.

During the discussion after the talk Ezequiel made some comments on how such considerations impact on concrete experimental work. Since play is essentially constituted by decision making it is necessary to gain a better understanding of the operations involved in such “choice-making”. However, he also pointed out that in order to have artificial agents of which we can say that they actually create value, they necessarily have to self-constitute their identities. This is still an ongoing challenge for enactive AI and robotics which has not yet been resolved in a satisfactory manner.

After the discussion a game was played which could be described as a drawing version of Chinese whispers. The “artworks” resulting from this activity were interpreted by Ezequiel as examples of value guided explorations of creative autonomy versus self-imposed constraints.

The end.

Any comments? Suggestions?



4 Responses

Subscribe to comments with RSS.

  1. edizquierdo said, on November 9, 2006 at 10:08 am

    Sorry to hear that. Although that had not happened to me using this site, because it can always happen when you are doing things online… as a rule of thumb, whenever I will write something elaborate online (e.g. blog, wiki, webmail), I first write it on some text-pad locally and only copy and paste when finished. Just in case, WordPress also has a ‘save’ button that does not immediately publish but stores it for later.

  2. xabier said, on November 13, 2006 at 10:23 am

    Nice talk!

    I hear Ezequiel talking about games first time in Compiagne during the Phiteco Seminars. I found it really fascinating. I just have a question… how and what did the final game you played exactly consisted of?



  3. Ezequiel said, on November 13, 2006 at 11:11 am

    The game was a simple example of how an activity can constrain itself, and at the same time that activity is spontaneously seeking to break constraints.

    I handed over pieces of paper to people. On these pieces you could find a single line or curve or a simple doodle. I asked everyone to

    1. look at the paper
    2. add a single contribution to what was in there
    3. pass on the result to the person sitting next to them
    4. repeat

    The results were not precisely artworks (sorry!) but they revealed different styles of working with constraints. Some people completed previous figures, others tried to break the obvious completion, other added details that transformed existing figures into a new meaning (like a dot for an eye, transforming an enclosed shape into a face).

    Nothing special, but it illustrated my points about play being a dialectics between creating new constraints and submitting to them.


  4. xabier said, on November 17, 2006 at 5:45 pm

    Nice! thanks for the explanation.

    I wonder, nevertheless, that talks in terms of constraints is sometimes to fuzzy in the sense that almost anything in biological, cognitive or social sciences can be interpreted as a constraint. Indeed what looks like breaking a constraint from a descriptive level might also be interpreted as following a higher level constraint. This does not mean that the game be useless or uninteresting I just wanted to share the worry that it is too often difficult to go beyond that point of conceptualizing certain processes as constraints and/or attractors. On the other hand, because GOFAI concepts are much closer to linguistic structures it is easier to expres and built more “complicated” explanations (where “complicated” just means longer and compositionally structured).


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: