Talk:Embodied cognitive science

From Wikipedia, the free encyclopedia

Embodied cognitive science is within the scope of WikiProject Robotics, an attempt to standardise coverage of Robotics. If you would like to participate, you can edit the article attached to this notice, or visit the project page, where you can join the project or contribute to the discussion.
??? This article has not yet received a rating on the Project's quality scale. Please rate the article and then leave a short summary here to explain the ratings and/or to identify the strengths and weaknesses of the article.
??? This article has not yet received an importance rating on the importance scale.

Hi, I created a page called Embodied Embedded Cognition. Only later I found out about Embodied cognitive science. This page is pretty ok, bigger, and largely covers what I wanted to say in my 'EEC' page. There are two main differences:

  • Embodied cognitive science page starts explicitly from the practices in behavior based robotics. I personaly would not claim that Pfeiffer is *the* big man in the field, but to roboticists this might be so. (Why not put Brooks up front?) In my version of it, robotics would only be one of the practices from which ideas for the theoretical position emerged.
  • Conceptually there can be discussion about what is meant by 'embodiment'. Most of the roboticists (cf Braitenberg) stress the physical interplay between bodily properties and world properties, and how a control system can easily use these properties so as to avoid having to represent them in an internal model. Let the world be its own repres. etc... But this I would not call 'embodiment', I would call it "embeddedness" or "situatedness". See also Andy Clark who is really a big man in this field and should be mentioned on this page. Embodiment I personally associate with body-internal processes (much more biologically inspired) such as homeostatic or hormonal processes, leading to emotional states, leading to a body-based internal "bias" on the perception-action loop. Edelman stresses this point, and so does Damasio (but Damasio is much more traditional in the rest of his vies though). So embodiment for me is like: When I'm hungry I think differently. Embeddedness is: "I put my bag in front of the door so I won't forget the bag when I leave the house", and more of these things. (Also Kirsh should be mentioned in that respect).

What do others think? Is this a valuable distinction? What to do about my newly created page: integrate it?


=============----

A message from user Paroswiki I am the original author of this article. I'm really wondering if the person who left the message above this one is the user called Noticket who appears all over the edit history. Either way I would like to explain my recent edit. That paragraph needed an overhaul. There were TOO MANY NAMES here. Remez's research is too particular for this article. Carol Fowler should be linked to in the wikipedia article on psycholinguistics, not here in ECS. Ditto Alvin Liberman. Ladefoged is a linguist and doesn't belong here. Andy Clark is a philosopher/writer, not a scientist who does research.

Embodied Cog Sci basically refers to scientists (that endorse embodied philosophy) who are doing research to ellucidate the general principles/structures/models of intelligence.

Mind you, there are researchers who are proposing general models of intelligence that are quite alien to the principles of embodiment. One person I can think of, in particular is Eliezer Yudkowsky with his idea of seed AI, which basically says that strong AI can be acheived with nothing but self-modifying programming languages. This idea is, in effect, a software problem that is completely non-embodied and non-situated.

Anyways, Embodied cog sci is very much biologically-inspired but I don't see a proper way to state this in the article without delving deeply in other directions. Also, it is not exclusively concerned with humans or human use of language. Researchers like Pfeifer or Edelman wouldn't mind having a robust model of a rabbit's brain, for instance, especially if rabbit cognition exemplified a "general principle" of intelligence.


=============----

Another message from user Paroswiki

"What do others think? Is this a valuable distinction? What to do about my newly created page: integrate it?"

You defined ECC as a "philosophical position" which is not the same thing as a research area in science. Your article is much more closely related to the wikipedia article Embodiment. Notice that *this* article contains a link to Embodiment article way up at the top.

This is a message from user noticket to Paroswiki: No, I am not the person who left the message at the top of the page, though after reading those comments I think that there is a lot that I agree with, though I know little about behavior-based robotics. My familiarity with embodied cognition comes from work done by scientists in psychology and linguistics that stretches back over 20-30 years and relates to the inseparability of cognition and related biological systems. Examples include the work of Eric Lenneberg in terms of the biological bases of language and Michael Turvey in terms of action-perception linkages (later also worked on by J. A. Scott Kelso). I like the clean-up that you did on the paragraph that I worked on. As background to why some of the names were included, Fowler and Remez were added because of an early paper by Fowler, Remez, Rubin and Turvey that related action to language and cognition and was very influential in this area. Also, I have been told that Fowler has done recent work in embodiment, but a reference would be needed to verify that, so leaving it out seems appropriate at this point in time. I agree that the Remez work has been focused in other areas and that his name is not essential. Regarding the comments at the top of the page, I also believe that "behavior based robotics" is only one of the areas related to embodied cognition. There has been substantial, and earlier, work in psychology, linguistics and philosophy. The cross-pollination across these various areas makes embodied cognition the fascinating area that it presently is. Because of this, the one name that I probably disagree with you about is Andy Clark. Though he is a philosopher, his writings and ideas have had great influence in this area. In summary, I think that the excellent page that you created would be best served by including the breadth of contributions to embodied cognitive science. Thanks for creating this important entry. Noticket 04:51, 26 February 2007 (UTC)

From noticket to Paroswiki. I did a little further checking (by using the "history" button on this page), and it looks like the contribution that you were asking about at the top of this page was from someone called "J.vandijk" . Looking around a little further, it looks like there is a lot of overlap between embodied cognitive science, embodied embedded cognition (I never heard of that one before until the comments of J.vandijk), embodied cognition, situated cognition, and embodied philosophy. If you can find J.vandijk, you might want to consider working with her/him on some kind of consolidation across some of these topics. Unfortunately, my experience is narrower, relating mostly to embodied cognition coming from the perspective of language and cognitive psychology, so I don't think that I can be of much help except in those areas. Speaking of that, don't forget to include Varela should you decide to add more and also keep in mind the work of Justine Cassell. She produced the excellent book "Embodied Conversational Agents". She used to be at the MIT MediaLab, but has since moved to Northwestern University where she directs a new center. From my point of view, when I think of embodied cognition, her work and that of her colleagues immediately comes to mind. It is apparent, however, that we bring different perspectives and viewpoints on what comprises "embodied cognitive science." I hope that this information was of some use. Noticket 18:29, 26 February 2007 (UTC)


I agree that behavior-based robotics is only a particular area of ECS. Perhaps there needs to be an overhaul of the top of the article as well. "1. The building of robotic agents capable of engaging in tasks that require realtime adaptive behavior (as opposed to engineered solutions seen in industrial machines.)" Perhaps that needs removed completely! And then we could remove the second item as well, since it is, in some way redundant to the defn at the top. The bullet list could be substituted by something like: "In contrast to cognitive science ECS models psychological systems in a holistic manner that includes both the brain and body as a single entity."

It seems to me there is a danger in allowing this article to be overtaken with subject matter related to Psychology proper, which is concerned only with human beings, or with linguistics, which is concerned only with human language. ECS seems mostly focused on "general intelligence", as a property of any animal with a central nervous system, rather than an exclusive attribute of humans. Perhaps lip-service needs payed to this.

There is also a danger of this article appearing to be a minor area of Artificial Intelligence. This may be the current problem with the article itself. I have linked to it from AI articles on wikipedia. Pfeifer's book is really a text on AI. Perhaps the methodology that Pfeifer is endorsing is that if an autonomous agent is successful using some model, then it follows that the model is what? Correct? Biologically feasible? I don't know because he never explicitly addresses this. Pfeifer not only mentions Edelman by name, but several chapters are an explanation of the connectionist "brain" used in NOMAD. Edelman's use of agents was to test models of the brain that were already known to be grounded in biology.

Regarding Fowler and Remez, you may want to consider this article: Situated_cognition

Andy Clark would be a fine person to mention in either Embodied_philosophy or Embodiment. But I don't think he needs mentioned here because he simply doesn't do scientific research. I will welcome your thoughts before making any deletions. paros 08:26, 27 February 2007 (UTC)

Contents

[edit] ==

From noticket to paros. I really like your suggestion on the re-write of the first sentence and think that the one you proposed (above) is excellent and is much better than what is presently there. I also think that your suggestions on the movement of Fowler and Remez seem like good ideas. The Andy Clark suggestion seems a little more problematic to me. I know of Andy's ideas because of other cognitive scientists and think that his writings have influenced scientific directions in embodied cognitive science. An analogy might be made to someone like Jerry Fodor, a philosopher, who, though not a scientist, has profoundly influenced scientific aspects of linguistics and psycholinguistics and is often mentioned and referenced in those contexts (for example, for his thoughts about modularity). Just because someone is a philosopher doesn't mean that they can not have a strong influence on particular scientific directions. I know absolutely nothing about AI, so I can't comment on that portion except to say that I also hope that this page does not become a subset of AI. Instead, my personal feeling is that the article should differentiate ECS from CS. In addition, the article should acknowledge the contributions from a variety of areas including cognitive science, psychology, linguistics, AI, robotics, and philosophy. What seems most important to me is those scientific approaches to studying cognition as holistic systems in which there is a crucial inseparability of individual and environment. Coming from a background in psychology and linguistics, I think that there are excellent long-standing examples from those areas, but agree that they should not in any way dominate the content of the article. Feel free to make any changes that you think are appropriate. Noticket 04:03, 28 February 2007 (UTC)


[edit] Where is Lakoff and Johnson?

User:Wikivangelist is a reputable editor of articles. He thought the mentioning of Lakoff and Mark Johnson at the bottom of the next paragraph was a dangling modifier. I agree with his/her edit, but since Lakoff and Johnson are not strictly ECS researchers, I cannot see the appropriateness of mentioning their names BEFORE the other people on this page. Due to this, I have completely removed mentioning them at all. I know this is going to bother some of you so here is some further explanations: I have to be honest, I both own and have read books by George Lakoff. His works are almost entirely concerned with the basis of semantics in language. He is very much a LINGUIST and does not deserve a mentioning that high up in this article. Lakoff's work is not concerned with ellucidating general principles of intelligent behavior. If anything, Lakoff shows peculiarities in human language.

Now what about Mark_Johnson_(professor)?? This is a toss-up as far as I am concerned. His work seems too focused on language in humans. His most recent book is written to PHILOSOPHERS as an audience. See the review for it at amazon for instance. Mark Johnson is first and foremost a writer whose concern is with the big questions of epistemology. He may be better mentioned in Embodied philosophy or maybe even in the Philosophy_of_science?? ECS, on the other hand, is concerned with scientists performing experiments in research laboratories. These scientists would endorse embodied philosophy, but they are not in the philosophical trenches battling it out with philosophers. Edelman himself has repeated (on camera) that "You don't need a philosopher to tell you what consciousness is. It is what you lose when you go to sleep and what you gain when you wake up." Academic philosophers would find this kind of roughshodiness apalling in their own circles. Edelman can afford such latitude since he is a scientist, not a philosopher.

I refer you to the amazon.com page for "Philosophy in the Flesh:The Embodied Mind and Its Challenge to Western Thought". You can see the reviewers battling it out in their own reviews, making argument sabout consciousness in the usual epistemological rants. There is too much mention of Kant, Plato, and Descartes here to qualify this material as "hard science".

I invite your complaints, comments, suggestions. - paros 10:45, 5 March 2007 (UTC)

[edit] Response to paros

Hi--my view is that if this was an article an "Embodied Robotics" rather than "Embodied Cognitive Science", it might be a little more appropriate to omit LJ (though even there I know Rod Brooks regularly assigned readings from them for his grad students in the AI lab). In any case, I think there is no problem with giving LJ a brief mention along with other wide-ranging theorists like Varela (let's not forget the ubiquitous forays into Buddhism in his book with Rosch and Thomson) before moving into some of the more scientist in a lab stuff. Cognitive Science is a term that intentionally includes both high and low level theorization, philosophy and linguistics as well as neuroscience and computational theory.

Embodied CS starts (in my view) with that reflective quote from Turing about how he might be wrong, and it might be maintained that a body is necessary for cognition after all. So I don't agree at all with your statements that "ECS, on the other hand, is concerned with scientists performing experiments in research laboratories," or that it is [exclusively] a "hard science," or that in an emergent field so interdisciplinary anybody at all are in fact "strictly ECS researchers."

While I'm not impressed by Lakoff and Johnson's recent joint book either, if you read his anthology and journal articles Lakoff is very involved in embodied cognitive science in neural networks and machine-language learning theory with his Neural Theory of Language (joint articles with Feldman, Narayanan, Ellen Dodge, etc.). As to Johnson, some of his recent articles (such as the joint article with Rohrer) also discuss the scientific evidence in much more depth than his sole-authored books aimed at philosophers. I don't think Amazon reviews are an accurate source of information for what is really going on in the field, as most of the real action in any science happens in journal articles, conference proceedings and symposia.

Lastly I've had the opportunity to be a fly on the wall a few times at conferences where these guys have interacted in small groups. They listen to each other. I've seen Edelman and Johnson discuss the finer points of how neuronal group selection and image schemata might produce consciousness, or Lakoff and Pfeifer debate whether a neural network/neural network mathematical analogue or a robotic/computational implementation is more relevant to model the cognition involved in a particular motoric embodiment effect such as walking. I never got a chance to meet Varela before he died, but I understand Lakoff and he were well acquainted through Gilles Fauconnier. My point is not that I know all these guys personally (I don't, but that doesn't really matter at all) but instead that they know each other, respect each other, talk and argue with each other on the conference circuit.

So my suggestion to Paros is this: Put LJ back in and let's also add back a sentence on the philosopher Andy Clark too, who also really belongs in this article as someone who also helped bring the body back into cognitive science. I'd also like to see a sentence or two on what Lenneberg's contribs are too. I also agree we need more text and more sections and subsections, and agree with you that we can't simply have the article be a catalog of names as once read.

And two asides to your comments above: I'd argue Edelman (as a wet science guy, or realistically the supervisor of lots of wet guys' theses) thinks that having a robust model of a rabbit brain is necessary for theories of cognition and consciousness (see his insights from the frog's neural maps in Neural Darwinism and Topobiology). Oh, and Kirsh is a philosopher by training as well...

(Oh and thanks for calling me a reputable contributor! grin) --- Wikivangelist 11:35, 8 March 2007 (UTC)

Why are the articles Embodied philosophy and Embodiment somehow innappropriate to Johnson and Andy Clark? I think they could even be mentioned in Philosophy of science. Johnson is literally 1 or 2 clicks away from this article. Why is it insufficient to say that ECS is rooted in Embodiment and then leave it at that? Why do we need to bring up all these peoples' names this high up in the article? Why can't you mention Lakoff in Psycholinguistics? I've never seen a single word written by Lakoff that doesnt have something to do with semantics in human language. You say Lakoff can be considered the "roots" of ECS? I honestly disagree. Lakoff is a linguist taking the embodied approach, but he is no way the "roots" of the entire edifice. You keep demanding to mention "research" done by Mark Johnson. What research has he done? He only seems intent on showing how findings by scientists effect old questions in epistemology. Well that's nice, but that is not research.
I do not beleive say that Lakoff and Pfeifer where present at the SAME CONFERENCE. I am even more skeptical that if they were, that they were talking to each other. Can you tell me what conference this was, and where and when it happened? paros 02:09, 13 March 2007 (UTC)

[edit] External links

We need more external links. Post any that are relevant to anything mentioned in the article itself. (DO NOT include links to home pages of particular scientists. If people want those, they can look up the person's name and follow the links there.) paros 02:18, 13 March 2007 (UTC)

[edit] Message to Wikivangelist

It would help if you created a USER PAGE on wikipedia so that I can respond to you personally. I'll have to put this here instead. I am handing the page over to you now. Please do not rename it. I repeat DO NOT RENAME IT to "Embodied Robotics" or whatever. This particular article has risen to slot 2 on a google search. This means your work and my work is now accessible to people all over the world. If you renamed it, it would have to start its climb all over again, and may never make it back there. Again for emphasis, do NOT rename this article and destroy its public exposure! You are in a position now to write a history of ECS, something which I can't do. Start from Turing and include the principal ideas up through Verela and into Mark Johnson, or whoever I am leaving out. Go ahead and place it right at the top of the article, or make a section titled "history" or do whatever you feel is appropriate. I agree that this entire article can flow from the Turing quote. paros 02:52, 21 March 2007 (UTC)

Ok, ok, we won't rename it. I'll write a bit on the history of ECS and submit a draft soon so we can collaborate on it. Btw, I created a user page as per your request. I like most of your changes, I'll make some minor edits before I get to drafting a short history of ECS--that will take me some time. --Wikivangelist 17:48, 24 March 2007 (UTC)

[edit] Edits March 20, 2007

I added a reference link to the turing quote, and moved some things around and fixed some grammatical errors. The top of the article use to read "The building of robotic agents capable of engaging in realtime adaptive behavior". On deeper reflection, this is a wildly innaccurate description of what researchers are actually doing! The old version made it seem as if these scientists literally BUILD robots like this in factories who are then sold to the military. Obviously this is not going on at all anywhere in the world. If you really look at what they do with robotic agents, is that they study how they develop in very controlled lab environments. Robots are used by researchers to see how the mind and brain interact with an environment. (Has anyone ever really built a robot capable of engaging in realtime adaptive behavior? Even the DARPA vehicles are not adaptive!) Anyway. Just some thoughts. paros 02:52, 21 March 2007 (UTC)

[edit] embodiment family of articles

There are proposals to reorganize these. Please discuss at Talk:Embodied philosophy ---- CharlesGillingham (talk) 10:52, 5 December 2007 (UTC)

[edit] paros responds

To CharlesGillingham, yes. We really need to talk about this issue. I will respond both here and in your larger discussion. This article here ECS, has to be subsumed under robotics. This type of research has exploded in Japan and Europe in the last 2 years. Seemingly nothing is happening in America in this vein. The closest thing they have is perhaps DARPA Grand Challenge? Rodney Brooks is apparently the last guy to do anything in this research area in the USA. There was a strong attempt by some people to come in here and try to subsume this topic under psychology. But honestly, it just is not a topic of psychology, nor even one of philosophy. ECS is a robotics/strong AI research area. I do not want wikipedia to produce the illusion that Rodney Brooks is the only inventor and spokesperson of ECS. The big names here are Gerald Edelman and Rolf Pfeifer. Edelman is a neuroscientist, but he rarely gives a lecture without talking about a "conscious artifact" which his audiences all understand without question to be a robot. Okay well this discussion is going to digress now, so I will continue up on your page instead.

To everyone else, through reading more I have come to a much deeper understanding of the Principle of Cheap Design and Redundancy. I am amending this section in a big way and I am adding citations to publicaitons as well, since wikipedia is now complaining about this. There seems to be an attempt to draw this wiki article under the umbrella of psychology. Do we need to get into that again? paros (talk) 02:50, 17 February 2008 (UTC)

Let's talk about this at Embodied philosophy. ---- CharlesGillingham (talk) 15:47, 17 February 2008 (UTC)
Back here again. I've done a bit of research. Here's what I think.
We agree that there are two subjects here. One is an approach to robotics. One is a position in cognitive science. We agree that there should be two articles, each with only the appropriate material. (We agree, for example, that the paragraphs here about Lakoff and Johnson and this article's references belong in a cognitive science article, and that the Turing quote and the stuff from Pfeifer belong in a robotics article.)
We disagree, however, about the title of the robotics article. It's confusing if you have an article titled "embodied cognitive science" that isn't about cognitive science (i.e. it's about robotics), especially now that "embodied cognition" has become a major force in cognitive science.
(We also disagree about the title of the cognitive science article, but I gather that's not your main concern. I lean towards embodied mind thesis. "Embodied psychology" makes it sound like it was developed by psychologists, rather than cognitive scientists like Lakoff.)
It seems to me that even Pfeiffer would agree that the term "embodied cognitive science" describes a form of cognitive science (rather than just robotics). Pfeifer writes:

"By 'cognitive science' we mean the interdisciplinary investigation of intelligence, or more generally, the mind. We are mostly interested in that part of cognitive science that applies a synthetic methodology, that is, the methodology of 'understanding by building" Pfeifer 2001, Understanding Intelligence, pg. 5

In other words, he wants to use robotics (and AI) to study the mind, which makes his work a kind of cognitive science. He's not coining a new term here (contrary to your comment at talk:Embodied philosophy). He knows what cognitive science is. He's also aware of how the term "embodied" has been used for the past 25 years in cognitive science. He's not claiming to have invented "embodied cognitive science", he's just trying to help clarify it and argue in its favor. He's making an argument within cognitive science, and his primary tool to winning that argument is examples from AI and robotics. He's not saying this is the only tool we need to win the argument. Similarly, an article about "embodied" approaches to "cognitive science" should mention several tools.
Which brings me to the third way in which we disagree. "Embodied" is an important term to AI, cognitive science and philosophy and Pfeifer's work is just one example among many. I think Pfeifer's work should be covered primarily in the article on Pfeifer. He should be mentioned in both articles (the one about "embodied" approaches to robotics and the one about "embodied" approaches to cognitive science), but he should be introduced as one researcher among many. ---- CharlesGillingham (talk) 11:57, 19 February 2008 (UTC)

[edit] Some changes

I have begun to rebuild embodied philosophy so that it accurately describes the interdisciplinary movement that argues that "the kind of mind we have depends on the kind of body we have". I have largely left this article alone, since paros strongly believes it should remain focussed on the robotics side of things. I've removed a few things from this article that are more appropriate for the article I'm working on, and made a clearer link to it at the top. ---- CharlesGillingham (talk) 09:40, 18 May 2008 (UTC)