The following appeared in Volume 98, Number 2 (Spring, 1999) of the APA Newsletters

John Perry is Henry Waldgrave Stuart Professor of Philosophy and the current Director, Center for the Study of Language and Information (CSLI). Perry was involved in the foundation of CSLI and was one of its first directors. I interviewed John Perry in Boston at the World Congress of Philosophy. The inteview covered a variety of connections between CSLI, philosophy and computing, including cognitive science, hyperproof and Turing machines, the Stanford Encyclopedia of Philosophy and Project Archimedes, the effort CSLI has made to help disabled people deal with computers. In these excerpts we touch on the relation of CSLI to cognitive science, and the Stanford Encyclopedia of Philosophy.

Bill Uzgalis (BU): The Center for the Study of Language and Information (CSLI) is involved with computing and philosophy in a variety of different ways. I guess one of those is the work on issues in cognitive science.

John Perry (JP): I should give you a little history of CSLI.

BU: Please do.

JP: CSLI was founded in 1983 as an interdisciplinary and interinstitutional research center involving not just Stanford but SRI and Xerox Park. That’s the Palo Alto research center of the Xerox corporation and SRI is an independent research institute. At one time it was called the Stanford Research Institute, but during the Vietnam war, since the institute did a lot of secret government research, Stanford decided it didn’t want to do secret government research. So SRI became independent and changed its name. Anyway, for decades prior to the foundation of CSLI, due to the tremendous energy of Pat Suppes Julius Moravcsik, Jakko Hintikka and people like that, there had been an interdisciplinary group at Stanford interested in very technical approaches to the philosophy of language, connections between philosophy and logic, linguistics and so forth and so on. They put on a whole series of conferences over the years. And then, into the 70s, a group built up that involved a lot of computer scientists from SRI and Xerox Park as well as from the more theoretical parts of computer science at Stanford.

BU: Terry Winograd?

JP: Well, Terry Winograd, but John McCarthy was very important as well, and there were others. Before CSLI was started the Sloan foundation gave a lot of money to cognitive science, Julius Moravcsik and people in the psychology department did a good job of harnessing some of that for Stanford. And so there was a very lively, interdisciplinary, interinstitutional group through the work of people like Suppes and Moravcsik and people in the psychology department as well as Bob Moore at SRI. Brian Smith at Xerox Park.

BU: So, that is the background?

JP: Yes. Then we managed to get the foundation to give us quite a bit of money, $20 million, in fact, divided between those three institutions: Stanford, Xerox Park and SRI, to set up this interinstitutional research center with its headquarters at Stanford. We didn’t call our center a cognitive science center, but it was clearly part of the cognitive science movement. We were just more focused on logic and less on psychology. At that time, we thought that by 1997 or ’98, the National Science Foundation would no doubt be helping cognitive science in big ways and there’d probably be big centers and we’d be one of them. Well, that’s basically not true. There is, I think, one center that it supports, but mostly the money available for cognitive science from the NSF isn’t significantly larger, maybe smaller in terms of linguistics, than it was in the 70s. So, CSLI has had to evolve and to do projects which are more practical and in the interests of industry and we get most of thediscretionary funds that we can use for doing things like creating the Stanford Encyclopedia of Philosophy from contributions from about twenty-five big companies like Sony and Microsoft, NTT and Erickson from all over the world make. And in return, we put on seminars for them and I go visit them and take some researchers along and so forth and so on. So, we’ve evolved quite a bit from a kind of pure cognitive science center. On the other hand and at the same time, cognitive science has evolved quite a bit. Cognitive science, in the early 80s, was very philosophical. It was all connected with the idea that somehow ideas of philosophers, ideas from people who studied people, psychologists and philosophers, could really be useful to AI. And vice versa: AI would be a good mechanism for studying psychology. And learning how people do things would really be valuable in helping develop computers. For example, in the case of natural language understanding, understanding how humans understood natural language would be very useful in designing programs for computers to understand natural language.

BU: So this is where situation semantics fits in?

JP: Yes, situation semantics, which Jon Barwise and I were finishing the book Situations and Attitudes about that time and Stanley Peters, who’s been at CSLI from the beginning, was involved with situation semantics and Robert Cooper and so forth and so on. It was a very hot item then and there was a lot of interest on the part of computer scientists who thought some of the ways in which we differed from other views, like partiality and attention to contex, would be very relevant for what they were trying to do. Well, I don’t want to diminish all that, I mean it’s turned out, I think, that at least all the big claims we made in situation theory have been kind of accepted and all the technical apparatus we set up have worked very well. That’s kind of my take on it. But in the larger picture, computers have gotten so that memory, both in the terms of processing power and disk space, is no longer scarce. So strategies that are very non-human can be used. So statistical methods of getting computers to deal with natural language, I don’t know if you’d want to call it understanding, have been very successful and the attempts to design computers based on theories of how human beings understand natural language, I wouldn’t want to say they’re unsuccessful, but they’re not as central to the computer industry as it might have seemed they would be from the perspective of the1980s. And interest in theoretical linguistics and theoretical philosophy of action within the computer industry and computer science departments is not what it used to be. Now, in the area of AI, we had quite a bit of impact. Michael Bratman, John McCarthy, Bob Moore, David Israel, Martha Pollack had a very active collaboration in the 80s that really developed some new architectures for what they call bound rationality. So, cognitive science has changed, the funding has changed, and CSLI is a much different places.

BU: Well, another way in which CSLI has been involved with philosophy and computing is the Stanford Encyclopedia of Philosophy. So, the Encyclopedia was your idea, yes?

BU: What do those improvements amount to?

JP: Well, the basic idea of the encyclopedia which we had at the beginning was that if it was online, it could be alive in the sense that it could be always changing, in the sense that the people who wrote the articles could update, edit, respond to criticisms simply by FTPing a new version of the article to the encyclopedia. And new articles could be added to the encyclopedia. For example, if you look at the McMillan Encyclopedia of Philosophy edited by Paul Edwards which came out in the late 60s, it is a great reference work. It was a huge amount of work. But, almost immediately, it started going out of date. If five years ago you wanted to go to the McMillan encyclopedia and find out something about David Lewis or functionalism or the identity theory or qualia or John Rawl’s theory of justice and reactions to it, it would have been pretty slim pickings.

BU: Yes, and not a lot about women philosophers, past or present.

BU: So you have a multimedia capability.

JP: Multimedia, yes. But, that is not exclusive to the web encyclopedias like the Stanford Encyclopedia. The Routledge Encyclopedia comes on a CD. I haven’t seen the CD or I don’t possess the CD, but both the Routledge and the Encarta Encyclopedias on CDs can have multimedia, can have excerpts, can have an enormous amount of material. Encyclopedias on CDs can have multimedia, can have excerpts, and can have an enormous amount of material. Again, you can have a search engine on a CD or on a website. That’s very helpful. So the ability to have multimedia and searchability are not unique to the internet. But, the endless depth and the updating are things that are unique to the ineternet. In this regard you are not any better off with a CD than you are with a printed encyclopedia. I mean, that’s not quite right because you can get a huge amount of material on a CD that would take up a lot of space on your bookshelf in a printed encyclopedia. But, still, as a practical matter the Stanford Encyclopedia of Philosophy has the potential to have endless amounts of material.

BU: If I recall right, you just received an NEH grant to improve the encyclopedia? Is that correct?

JP: Well, to improve it, is probably not quite the word. To create it. The Encyclopedia exists now as something you can visit online and it has a tentative table of contents of about three hundred entries. It has an editorial board, which is helping us find authors for some of these articles and helping us decide which articles to have. Because of this dynamic thing, all that can change and grow, but at the present there aren’t that many articles. There’s some really nice ones, but if you want to use the Encyclopedia now as a research tool, unless you’re interested in Frege or Russell or a few other topics about which there really nice articles, you will be disappointed. As a practical matter, it will probably take three to five years before we have enough articles to make it really worthwhile as something that you would go to with some expectation of getting some help. So it’s really to create the encyclopedia in that sense that we got the grant. And it’s a very helpful grant. It covers about half of our expenses and I have to cover the other half now with money I raised for CSLI from our industrial affiliates. We’ve also applied for an NSF grant which would help a lot. Our hope is that eventually the encyclopedia will be self-supporting, we’re not exactly sure how. We’ve gotten some very nice help from the Pacific Division of the APA and the Canadian Philosophical Association.

BU: I didn’t realize that you had gotten so much help from such a variety of sources. That is heartening.

JP: It is, but how we will finance it in the long run, I don’t know. I’m not particularly pessimistic. I think that it’s appropriate that universities pay some fees since it would still be relatively cheap. They have to pay a lot to buy a few copies of the Routledge Encyclopedia. Advertising is a possibility. Philosophical book publishers, or subsidies from philosophy departments that are relatively wealthy may be other possibilities. It is also possible that some of the software we developed could be licensed. We don’t know. So I’m not pessimistic, but still, it is a problem. We hope to have it basically free. Maybe not free to universities, but free to individuals. How we’ll do that, I don’t quite know, but the whole issue of financing things on the Internet is changing so fast that it doesn’t have me unduly worried.

BU: Well, I am glad to hear that you think that problem can be resolved and that you are getting such assistance.

JP: We are also getting tremendous help without any promise of recompense from our board of editors and from the individual contributors, many of whom really like this idea of having this free, growing, alive thing on the Internet. Another thing I should mention is that it’s conceivable to have two articles from quite different perspectives on a single subject in the Encyclopedia. It is also conceivable to have connected to the Encyclopedia various, what do you call those, not chat rooms, but...

BU: Threaded discussion tools.

JP: Threaded discussions, right. We haven’t pursued that yet, but those are all things that could be done.

BU: Yes, if you want a discussion of the articles, that would be a sensible addition.

JP: Yes, I’d like to have it so that, on any given article, you could push a button and just like at Amazon.com you can go see what the reviews of the books are, you could see what people have thought about this article. But right now that’s not a high priority for us. Eventually I’m sure that’s someone will take care of it.

BU: Right.

JP: The thing to realize is that we’re doing this at a nonpropitious time from the point of view of languages of the Internet. The HTML that is a relatively simple language that has made the websites easy to construct and use isn’t suited for everything we’d want in an encyclopedia of philosophy.

BU: So, what kind of difficulties are you encountering?

JP: Well, in the technical articles, in particular, we are having to use graph images and things like that. Much more sophisticated tools will be available that will lend themselves to sophisticated search engines and in a way it might have been better to wait until those were available. On the other hand, my feeling was that philosophy is an excellent discipline to start this kind of thing with because it’s not that big a discipline, either in terms of people or in terms of the kinds of stuff you need to cover and so it’s relatively manageable. And, so, we decided to start with just what was available with HTML and then just count on being able to upgrade to more powerful languages as they become standard on the web. But that’ll be a problem, too, because that’ll all require some doing. So the way it works now, you have an account on the Encyclopedia computer, which is a computer in Ed Zalta’s office, and you just telnet in and make your changes or write your article and there’s some review of changes by the editorial board, but pretty much you have the ability to directly change, develop, and add to your article. And the overhead is very low. Essentially, there is just one set of keystrokes from the author to the monitor of the user.

BU: I am very interested in Project Archimedes.

JP: The Archimedes Project, which we started about six years ago, has to do with the access to computers and information for people with disabilities. This is something we sort of drifted into. We were interested in the meaning of non-sentential symbols, such as graphics. And one person got interested in the semantics of American sign language; she learned ASL, learned more about the Deaf Community, and from there formed the idea that CSLI could study the use of computers in communication as a possible benefit and a possible problem for people with disabilities. That seemed to tie into the theoretical picture of CSLI: Different forms that the same information can take. It also seemed like a project that the rest of the Stanford community could understand, for CSLI had a reputation as a place that was so theoretical that no one could fathom what we were doing.

BU: I can see how computers can aid people with disabilities, but how can they be a problem?

JP: Probably the best example of that is blind people and graphical user interfaces, such as the Mac Interface or Windows. These are very handy for sighted people, but they created havoc for blind computer users. With an old-fashioned screen full of ascii characters, an automatic screen reader could convey the text to spoken words. People who used this regularly and got used to it could grasp language spoken at a about three times normal speed, so it was a very practical way to access a computer. These screen reader programs simply broke down when faced with the graphical user interfaces.

BU: Couldn’t something be done?

JP: Yes, but there is this time lag. For example, the screen reader that could more or less deal with the first version of Windows was released almost the same day as Windows 95, which made it obsolete.

We felt that whatever else we did at CSLI, we could help raise the consciousness of these problems. To give you a good example of why that is necessary, go back to 1983. In 1983 the Xerox Corporation gave CSLI one hundred workstations that used their innovative graphical user interface—icons, windows, the whole thing. This was the roughly the same Xerox work that Steve Jobs saw and took to Apple. Imagine if we could have gotten engineers and computer scientists and cognitive psychologists to start thinking about how blind people would be able to deal with this new technology? But no one thought of it. And I suppose if anybody had asked about blind people and the GUI, somebody would say oh, well, they just won’t use them. But what choice does one have now? For example, Larry Scadden who is a blind psychophysicist at the National Science Foundation just came into his office one day and found he had been "given" a windows style computer, so his software would be compatible with everyone else in his division.

The point is, a lot of time was lost because no one thought of the impact of the new technology on people with disabilities. With the pace of change now, and the internet, these problems come up every day. So there is plenty of room for a project like the Archimedes Project to help alert students and faculty who are doing the research that may lead to new products, that the work they do may have an impact, negative or positive, or computer users who are disabled.

BU: Are there theoretical research problems involved?

JP: Yes, because you need to think about how the icons and windows actually function. To take a simple example, suppose you simply reproduce the windows and shapes on the screen so that they are available to a blind person through the sense of touch. Does that convey the same information? Think of the trash can. Would a blind person realize that an icon that felt like that represents a trash can?

BU: Right, it might be a beer can.

JP: It might be a beer can, or a zillion other things. A person who is blind might have a lot of experience with trash cans, but it doesn’t follow that a tactile version of a tiny trash-can representation will mean anything. So, that’s a simple example of how the issue of sameness of meaning and content across different systems of representation comes alive in the study of disabilities.

BU: That’s all fine, but does this really have implications for philosophy?

JP: It has made me sympathetic to a view we might call materialistic Cartesianism. You separate the ways people have of gaining information, and the ways they have of performing actions, from the core of the person themselves. The real person is something that’s inside the skull. That person has the problem of interfacing with the world. That means getting information from the world, and having an effect on the world. The standard way is—or used to be—to use the five senses to get the information and move your body in various ways to have the effects. But if a person has sensory or motor disabilities, they can’t use these standard ways. But the need is the same.

BU: What do you mean "used to be"?

JP: Well none of us really live in a world where our sense organs and our limbs suffice as an interface to the world. It used to be real simple. Your sense organs and told you what was going on around you and you have muscles and those allowed you to move your limbs and affect what was going on around you. This is what you need to fix breakfast or start a fire. So in those days disabled people were really, truly disabled in the sense that they were in a basically much different position than non-disabled people.

But our world isn’t really like that at all. Most of the time, we get information about the things we need to find out about through a complicated process that depends on a huge social infrastructure that’s built up with taxpayers’ money. And you usually rely on a similar set of structures to have an effect on the world. We spend our lives getting information and doing things, often with our fingers on a keyboard, that have an effect thousands of miles away. So I may get some information from England and fax it to Hong Kong. So the success conditions of my actions have to do with changes that occur in Hong Kong. The information that tells me what to do to get the right effect in Hong Kong comes to me from England and my senses and muscles just have to do with end of the informational process and the inauguration of the process of having an effect. In this world, our world, the disabled person really isn’t in a basically different situation than the non-disabled person. We are all dependent on a huge technological infrastructure. And what makes a disabled person handicapped, unable to do things that other people can do, is just the details about how the last stages of getting information, or the first stages of inaugurating change. And for a wide variety of tasks, if the person can interact with a computer, just one computer, they can interact with all the computers on the net, and learn and do enormous numbers of things. So computers are a tremendous equalizing opportunity for disabled people and at the same time, computers, if they’re not properly designed, can really hurt disabled people—as we saw with the GUI problem for blind computer users.

So the study of disabilities does give one a lot to think about, in the sphere of epistemology and philosophy of action, as well as, of course, ethics and public policy. You can find some stuff on this on my webpage, <www-csli.stanford.edu/users/john/>. Also I’d like to mention Ron Amundson and Anita Silvers, as two philosophers who have done interesting things on the philosophy of disability. Ron has done important work on this sort of environmental conception of disability and handicap I was describing above. I’m not an expert on the field at all, but I do know that there are a lot of interesting problems for philosophers there.

John Perry

CSLI