Dear List: I have been wishing to enter this discussion for some time now, but my current thesis work (an investigation of human interlimb coordination using optical motion capture) has left me little time to address what I think are the extremely salient issues now being discussed on the list. I hope that I will find more time in the future to contribute to this discussion (but apologize in advance if I fall into the black hole of research)! To begin, I would like to cite a couple of studies that (i) will illuminate the perspective from which I personally approach the issues of gesture, expression and dynamics (both artistically and scientifically), and (ii) may provide some interesting material for reference by others on the list who are exploring these topics: 1) J. Streepey and M. Gross (1998) conducted a study entitled "Influence of emotional intent on dance kinematics". Dancers were asked to perform what Streepey and Gross call a "whole body gesture" (in this case basically a demi plie in parallel position). Subjects performed the movement under 3 expressive conditions: neutral, fear and anger. Position data from reflective markers placed at 12 body landmarks were captured, and kinematics for the ankle, lower spine and shoulder were analyzed. To cut to the chase, Streepey and Gross found that "range of motion increased with anger and fear. Peak velocities differed between emotions, with the lowest velocities in neutral movements and the highest velocities in anger". One interesting question that arises here is what is the difference between expressive intent (on the part of the performer) and the perception of expression (on the part of the audience). Would a naive observer have interpreted these same movements as "fear" or "anger"? 2) One study that addressed this question was undertaken by Dittrich, Troscianko, Lea and Morgan (1996), entitled "Perception of emotion from dynamic point-light displays represented in dance". One of their fundamental questions concerned whether it is possible to judge the emotional state of a human body from motion information alone; no faces, no fingers, no information about muscle tension, just a bunch of dots moving on a screen. In this study, naive viewers were shown video clips of expressive movements performed by 1 male and 1 female dancer. For the sake of comparison, the clips were presented in 4 conditions: full light (i.e. regular video footage), full light inverted (regular video footage upside down), point-light upright (just the dots m'am) and point-light inverted. The 6 movements presented in the clips were intended to express surprise, fear, anger, disgust, grief and joy. To summarize in the simplest possible terms, when subjects were asked to identify the emotions being expressed in these different clips (via a rating questionnaire listing the 6 emotional categories), accuracy of recognition was above chance in all conditions (88% correct for full-light conditions, 63% correct for point-light conditions). While the use of the questionnaire may have aided subjects in identifying the intended emotions (by narrowing viewer's choices to only 6 out of the nearly infinite number of emotional possibilities), this study nonetheless provides some good evidence for perception of emotion on the basis of dynamic information alone (or at least that is my hopeful interpretation . . . I am open to counter-arguments). There is one more issue I would like to address here, which has come up in several postings--namely, the context in which gesture (I'm staying clear of trying to define this term for now) is presented. This point has already been summarized admirably by Armando: >expressiveness is definitely linked to a relationship with the environment. . . . > All of these domains (divided into further detailed layers) are "readable" or interpretable only INSIDE a particular >cultural group and moment (space and Time). Doubtless this issue troubles the sleep of many of you who are engaged in trying to create computational models that can replicate expressive human movement! To continue in this vein, the events that precede a particular movement must influence the way in which that movement is perceived. In a dance context, have we (the audience) already seen the movement earlier in the piece, such that --when repeated--it takes on a "significance" simply by creating a "history" for itself in the chronology of previous events? Does it achieve "meaning" purely by merit of its contrast with the immediately preceding movement (e.g. an obsessively repeated movement suddenly followed by stillness)? Has this movement, even though it may be completely arbitrary in and of itself, become a "motif" or "signature" movement that identifies one particular dancer? The most obvious starting place for a discussion of the contextually derived meaning of a gesture concerns the cultural associations attached to that gesture. However--even if no cultural association exists for that particular gesture--might there still be associations attached to the body part that is performing the gesture? Yes, it keeps me awake at night too. Before I wade into this quagmire any further, and on a more practical level (for the programming inclined), let's return to the parallels between dance and language (from the earlier days of this thread). Science's greatest fetish seems to revolve around the identification of "fundamental units". Is gesture the fundamental unit of dance? As a working hypothesis, for the moment let's say it is. Now, in the realm of speech production, a lot of interesting research concerns the non-serial nature of speech production. By way of introduction to this line of thought, Rosenbaum in Human Motor Control (1991) gives the example of pronouncing the word "tulip": if you look in the mirror while saying this word, you will see that your lips round before you say "t" (by contrast, see what happens when you say "teach".) This sort of anticipatory control argues for parallel processing in the production of speech (which, it should be remembered, is in essence a physical activity). How are computer programers able to provide anticipatory (feedforward) control in 2-D or 3-D models that are producing a series of micro-dances/gestures so that these discrete movements flow together? While I am not neither a programmer nor an animator, I would guess that this is one of the fundamental problems in generating naturalistic movement for computer generated figures. I would appreciate the illuminated insights of anyone who cares to offer them. In closing, I hope that this long-winded (and long-overdue) diatribe provides some further food for thought. Also looking forward to hearing more about the idea of putting together a roundtable (per Antonio's suggestion) on this topic. Regards, Robynne M. Gravenhorst Executive Artistic Director, The Anatomical Theatre http://www.anatomical.org
This archive was generated by hypermail 2b30 : 04/16/02