Whooshup Reorganization

To reflect what this blog has become, the format has changed to emphasize the enormous number of useful links to resources we provide. To go to the whooshup blog and conversations about these resources, just scroll to the bottom of the lists of resources!

Tuesday, January 22, 2008

Discussion: Dreyfus and Embodiment in Virtual Space

This relates to several previous posts and comments about some of the implications of this discussion for the embodiment problem - that is, how can computers be *like* humans without possessing a humanlike body, and at another level, how can people learn through computers, at a distance from their teachers, without being bodily present in a studio or classroom. Hubert Dreyfus has commented extensively on these issues, and the background to his critique can be noticed in some of these webcast lectures. The full arguments are found in his books "What Computers Still Can't Do" and "On the Internet."

Demographer and I got started on this, but I would like to invite anybody interested in this subject to join our conversation. I have collected the text, some of it repeated from other threads, some of it from emails, and put it all together as the starting comments to this post.
Posted by at

6 comments:

Karl Tyson said...

In my initial post about Second Life (1/4/08), I quoted Dreyfus from the concluding paragraphs to SectionTwo on Distance Learning in "On the Internet:"

...(C)an the bodily presence required for acquiring skills in various domains and for acquiring mastery of one's culture be delivered by means of the Internet?

The promise of telepresence holds out hope for a positive answer to this question. If telepresence could enable human beings to be present at a distance in a way that captures all that is essential about bodily presence, then the dream of distance learning at all levels could, in principle, be achieved. But if telepresence cannot deliver the classroom coaching and the lecture-hall presence through which involvement is fostered by committed teachers, as well as the presence to apprentices of masters whose style is manifest on a day-to-day basis so that it can be imitated, distance learning will produce only competence, while expertise and practical wisdom will remain completely out of reach. Hyper-learning would then turn out to be mere hype. So our question becomes: how much presence can telepresence deliver?

January 22, 2008 at 7:55 PM
Karl Tyson said...

Then Demographer came back with this (1/16/08):

Dreyfus' comment seemed to assume that there is a single bodily experience to refer to. Either telepresence gets it right, or it doesn't, but the bodily experience is taken for granted, a sort of "state of Nature".

The dichotomy between "virtual life" and "real life" is patently false. That is my main point. Virtual reality is just another iteration in the same process that brought us language, literacy, the telephone, the printing press, etc. Bodies communicating and building worlds through symbolic communication.

The bodily experience of learning is hugely socially constructed: lecture halls, disciplined corporality (Mauss and Foucault), climate control, audio-visual technology, electric lights, amplification, knowledge of protocol, let alone language, etc. I would argue that there is no natural reference point, and that second life is just one more iteration on a very complicated set of arrangements that mingle corporality and meaning and communication, and the lecture hall is merely a different iteration, as is the Agora, etc.

Culture can't escape the body, but the other direction is true as well.

And to privilege the lecture hall over second life is just as ridiculous as privileging orality over literacy, primitive tribes over civilized tribes, etc, etc.

We don't know yet the character of cyber communication, but it forms a corporally based thing with as much Heideggerian worldhood as any other context. Second life is just as "natural" as the Black Forest -- i.e. its naturalness is a non issue, and I think Dreyfus finds himself in a nostalgia for a natural state (that probably never existed, and that is a figment due to our Anxiety to pretend that we have an essential Being).

Technology has been becoming background since we shed our fur and put on skins (or the blind person picked up his cane). And we have been desperately ("anxiously") trying to ground ourselves in the natural body (among other things) since we said the words "I am ..."

January 22, 2008 at 7:57 PM
Karl Tyson said...

Then I shot back this (1/21/08):

The reason I am so focused on this issue is this: Dreyfus ultimate argument against *both* the viability of Heideggerian AI and the viability of using distance learning to achieve expert knowledge, appears to be the embodiment problem.

These two areas, AI and expert skill acquisition, sum up a large part of Dreyfus' contribution. He has basically rested his case for discounting internet learning at high levels and standard AI on the impossibility of adequately simulating bodily presence and know-how in the digital format we now have (he does not discount the possibility of *analog* computers and robots getting closer - what I think of culminating as the Ghola option - taking tissue resembling our flesh and somehow injecting neural activity).

I have looked at both What Computers Can't Do, and On the Internet, where these points (embedded of course in a lot of other stuff) come out. But perhaps the best and easiest way to get his take on these is in the Andrew Keen Interview reffed on the blog - at the end of that 30-minute audio clip he simply states that none of this will work because of the body problem, period. But it's everywhere in his writings.

You proposed that there is no body/computer issue, just a misunderstanding of what the situation really is. I think that opinion represents some of the most advanced intuitions of the generation now so embedded in digital devices that they see no essential difference between a hammer, a refrigerator, and a language translation program.

Do you operate under the assumption that AI will in fact deliver, because the constituent humans who matter - those embedded in technology, who are turning it into their background - already expect and will it to?

Or do you agree with Dreyfus that because background knowledge is absolutely riddled with body-dependent articulations and personalized, non-conceptual connectedness, and these cannot accumulate in disembodied, non-coping data structures, that to ever expect a digital device to really handle it is untenable?

January 22, 2008 at 8:00 PM
Karl Tyson said...

The Demographer responded (1/21/08):

He attacks *Cartesian* AI, I think, though I haven't read the interviews.

I would dissolve the "issue" of X versus Y into a multitude of empirical questions about particular bodies, particular computers, and particular networks thereof (D talked about cognition being in the world in one of his H lectures -- on a mailing list, cognition rests partly in the server...). Do I think it is possible to learn philosophy by reading and discussing it online -- absolutely, though you both lose and gain *certain* things that must be empirically explored. Do I think you can learn to ice skate online? -- maybe not so much, but you can find an ice rink to go put your body at -- a fact that shows that the question is not clear cut. Do I think you can find true love online? ... I am happily married, so that is a question for another philosopher.


[on the question: Do you operate under the assumption that AI will in fact deliver?]
I don't know how to answer that. SQL databases already mediate an IMMENSE part of the lives of everyone in the rich countries, and we skillfully negotiate them even when we don't know they are there (I talked to an interviewee recently who said that he expects to hear from an estranged relative soon because of the internet, and I am confident that he doesn't know anything about how the adoption registries work.) Do I think artificial *inference* algorithms will succeed? -- I think there is more promise in neural nets than in tables of abstract properties, and I am convinced of this because of Dreyfus.

[on the question: Do you agree with Dreyfus?]
Again, I resist the dichotomizing of what seems to me a huge field of empirical stuff to be investigated. I am far more comfortable collecting data on what happens than making sweeping statements that depend on argumentation and logic (and my distrust of such is buttressed by Dreyfus on Heidegger on Descartes, and is the reason I fancy myself an anthropologist and not a philosopher). That is why I think experimenting with Second Life and blogs is such a good idea, because who knows until you try, and who knows *in what respect* until you explore it in some detail.

I personally don't think SL approximates the body at all, and that plain text may actually be tied into the body better than silly avatars. But I am infected by Kojeve on Hegel, and the thought that there is an almost Manichean dynamic between flesh and words...

An afterthought: Computer mediated discussion allows asynchrony, where embodied discussion requires synchrony. Heidegger thought time very important...

Also -- I bet Heidegger would have absolutely hated the internet. But he was something of a reactionary and curmudgeon...

January 22, 2008 at 8:02 PM
Karl Tyson said...

Me again (1/22/08):

I need to reply to several points.

First, and most important, Dreyfus has insured his continuing relavence to the AI debate by demolishing Heidegerrian AI (neural nets plus) as well as Cartesian - and fundamentally on the issue of embodiment, secondarily on the issue of infinitude - that even if we theoretically *could* we physically could not replicate the endlessly connected background.

Here is a snip from a paper published April 2007:

Why Heideggerian AI Failed
and how Fixing it would
Require making it more Heideggerian

So, according to the view I have been presenting, even if the Heideggerian/Merleau-Pontian approach to AI suggested by Freeman is ontologically sound in a way that GOFAI and subsequent supposedly Heideggerian models proposed by Brooks, Agre, and Wheeler are not, a neurodynamic computer model would still have to be given a detailed description of a body and motivations like ours if things were to count as significant for it so that it could learn to act intelligently in our world. We have seen that Heidegger, Merleau-Ponty, and Freeman offer us hints of the elaborate and subtle body and brain structures we would have to model and how to model some of them, but this only makes the task of a Heideggerian AI seem all the more difficult and casts doubt on whether we will ever be able to accomplish it.

[Note that Brooks, Agre, and Wheeler propose nets and robots and are desperately trying to avoid D's original critique!]

Second, your alternative approach - to treat the entire dimension of human - computer interaction - in which we continually ask the computer to do more and more important jobs for us (one that comes to my mind almost continually is that we ask it to predict 50 years of climate change so we can make policy next year - an unbelievably audacious demand) - as merely a blizzard of discrete separable problems - that we can fix up, or further experiment, or extend, or put up with the defects of, whatever we find ourselves thrown into, seems to me as flawed as the Cognitive AI advocates saying that whatever small problems might stand in the way of their formal rule based project would be worked out in a few years. The single thing about Dreyfus that grabbed my attention like a lightening strike was that he predicted and explained why those researchers could not achieve their end. Yes, that's a sweeping statement, but it's also Newtonian - it threatens to apply everywhere, always, with a few relativistic exceptions.

January 22, 2008 at 8:05 PM
foundrysmith said...

Regarding presence and interaction in the Second Life setting, my own experience last weekend is that it felt different than posting to a list or chatting via Instant Messaging. I suspect that using a combination of freely available resources will enable something like a discussion section to take place online, yet it will be unlike the real thing at a university lecture hall. Even so, it should get neurons firing up in some fashion, and that is the point, isn't it?

There does seem to be a pre-occupation with the avatar skin in SL, but then again there is a preoccupation with appearance on campus, too. On the upside, I think the avatar concept may be a leveler, in that folks who have disabilities or other issues can use computer enhancements to their advantage. A blind person can hear. A deaf person can read. A mute person can type, et cetera. Or with a language translator, even cultural borders may be crossed. These issues wouldn't be important through one's avatar. One can then focus on the discussion, and the exchange and spawning of ideas.

January 23, 2008 at 6:09 AM

Post a Comment

Subscribe to: Post Comments (Atom)