bzs@BU-CS.BU.EDU (Barry Shein) (11/11/87)
Date: Tue, 10 Nov 87 22:00:17 EST From: Michael Travers <mt@MEDIA-LAB.MEDIA.MIT.EDU> To: viv-core@MEDIA-LAB.MEDIA.MIT.EDU, prog-d@MEDIA-LAB.MEDIA.MIT.EDU Subject: interesting but scary interface technology I thought of this technique a few years ago, but I was too scared by the possible repressive uses of it to pursue it, or even mention it much. Now I see other people are. ----------------------------------------------------------------- This message is aimed at you people out there on the UsetNet community who are doing (or know of someone doing) research in the area of Electromyography and Covert Oral Behavior. An abstract of what we are doing follows --- if you know of anyone who is working on something similiar, we would most certainly like to get in touch. Please contact: Dr. Howard I. Thorsheim Department of Psychology St. Olaf College Northfield, MN 55057 USA (507)-663-3144 ---------------------------- ABSTRACT FOLLOWS ------------------------------- The Possibility of a Thought-Recognition Interface Walter D. Poxon Craig D. Rice Academic Computing Center St. Olaf College Howard I. Thorsheim Department of Psychology St. Olaf College Northfield, MN 55057 An ideal for human-computer interaction is approaching the here-and-now. To date, any interaction between humans and computers has taken place through the use of keyboards, or more recently, voice recognition hardware, and other specially-designed input devices. Keyboards provide a reasonably efficient means of interacting with a machine and allow for the entry of very complex commands and data sets. Voice recognition systems offer a more natural means of communication, but have limitations in their ability to recognize large or specialized vocabularies. The drawback common to these and all current human-computer interaction systems is that they require that human thoughts be transformed into overt actions in order to be recognized by the computer; that is, users must type their thoughts, or say them, or draw them. A more natural human-computer interface would have none of these drawbacks. It would be able to recognize and respond to covert human behavior rather than requiring overt signals. Determining the possibility and practicality of such an interface is one of our current goals. Through the analysis of electromyographic (EMG) signals of the covert oral behavior phenomenon [McGuigan and Winstead, 1974], [Thorsheim, McGuigan, and Davis, 1975], produced while a subject speaks or thinks preassigned syllables, it is anticipated that unique, reproduceable digital patterns may be identified. By linking these patterns to the syllables or words that they represent, a new system of communication may be built whereby the machine interprets these signals as its source of input in place of the codes generated by the traditional keyboard interface. The current stage of our research is to characterize the EMG signals generated during covert oral behavior and assess the feasibility of using currently-available microcomputer resources to process and interpret these signals using common analog to digital conversion equipment and Fourier analysis techniques. Long-term goals include the building of subject-independent libraries of "command patterns" and the referencing of these libraries to allow for the control of a text editor without the use of a keyboard. ----------------------------------------------------------------------------- Craig D. Rice Math Deptartment Computer Systems Manager Academic Computer Center Systems Programmer USMAIL: St. Olaf College Northfield, MN 55057 UUCP: ..{ihnp4,umn-cs}!stolaf!ricec AT&T: Work: (507)-663-3096 Home: (507)-663-2191 Data: (507)-663-2191 ("cinta" midnight-6am)
OWENSJ@VTVM1.BITNET (John Owens) (11/12/87)
>From: Michael Travers <mt@MEDIA-LAB.MEDIA.MIT.EDU> >To: viv-core@MEDIA-LAB.MEDIA.MIT.EDU, prog-d@MEDIA-LAB.MEDIA.MIT.EDU >Subject: interesting but scary interface technology >Through the analysis of electromyographic (EMG) signals of the covert >oral behavior phenomenon [McGuigan and Winstead, 1974], [Thorsheim, >McGuigan, and Davis, 1975], produced while a subject speaks or thinks >preassigned syllables, it is anticipated that unique, reproduceable >digital patterns may be identified. I remember reading a science fiction story several years ago (in Year's Best S.F. 1955 or something like that) in which "Big Brother" used such a recognition device to read the thoughts of the main character without her knowing it. The story concerned a woman who navigated a ship through hyperspace (or somesuch) by psychic means. In the course of doing this, she fell in love with another psychic navigator, and they created their own private world (which was part of the navigation process, but took on special meaning for them). When she found out that they (I don't remember who "they" were) had been monitoring her thoughts and invading the privacy of their communications, she was crushed. Anyway, at the time I thought that reading thoughts by detecting "subvocalizations" was purely an invention of the author, but perhaps not. Does anyone remember the story and the year so we can compare it to the papers referenced above? Maybe the researchers got their idea from this storybe
riddle@woton.UUCP (11/12/87)
I agree that wiring people up in such a way as to detect "covert oral behavior" would constitute "interesting but scary interface technology." But how much of thought do you really believe is expressed in syllables? It's my opinion that only some thought is verbally (let alone orally) based. The Sapir-Whorf hypothesis notwithstanding, I think that on introspection most of us would realize that there are plenty of ideas, experiences, fantasies and other thoughts which we can think but can't verbalize. Visual and tactile images would be simple examples. Even linguistic thinking doesn't always map successfully into words -- how many times have you had a thought "on the tip of your tongue" but been unable to verbalize it? Furthermore, some of the "scarier" potential applications of this technology might be easy to thwart, given a concerted effort to do so. If communities of speakers have managed to come up with cant and jargon to avoid being understood by outsiders in the past, it should be even easier for an individual to purposely obfuscate the language used for his or her internal monologue. Reading the abstract you posted, one thing jumps out at me: I doubt that the system described would be more "natural" than keyboard or speech interfaces because it would avoid the drawback "that human thoughts be transformed into overt actions in order to be recognized by the computer." In fact I suspect that at best it would contain all of the problems of current speech recognition systems -- limited vocabulary and an ability to handle only an artificial subset of speech behavior. If this were the case, in order to use the interface people would have to *overtly* control their "covert oral behavior." --- Prentiss Riddle ("Aprendiz de todo, maestro de nada.") --- Opinions expressed are not necessarily those of Shriners Burns Institute. --- riddle@woton.UUCP {ihnp4,harvard}!ut-sally!im4u!woton!riddle
riddle@woton.UUCP (Prentiss Riddle ) (11/13/87)
OWENSJ@VTVM1.BITNET (John Owens) writes: > I remember reading a science fiction story several years ago (in > Year's Best S.F. 1955 or something like that) in which... I read a similar story, but the twist in this one was that the hyperspace pilot carried on a stormy dialogue with another crew member whom he never saw, who turned out to be the personfication of his own subconscious "subvocalizations". This was his employers' solution to keeping pilots sane in the solitude of deep space. I suspect that this is well-trodden ground in sf. --- Prentiss Riddle ("Aprendiz de todo, maestro de nada.") --- Opinions expressed are not necessarily those of Shriners Burns Institute. --- riddle@woton.UUCP {ihnp4,harvard}!ut-sally!im4u!woton!riddle