In yesterday's post on afterimages and aftereffects, I mentioned that demonstrations of neural adaptation for a particular feature (in the post, I used the examples of color and motion) is generally taken as evidence of the existence of specific neurons or groups of neurons that detect/process that feature. With motion or color, which are very basic features of the visual environment, this isn't very surprising, but in this post, I'm going to talk about some recent research demonstrating neural adaptation for a much more complex and surprising feature. But first, a little background on another cool visual illusion.
In the early 1970s, Gunnar Johansson strapped lights to people's joints, and filmed them walking against a dark background1. The result was what is now called point light animation. To see what occurs in point animation, take a look at these four images2. Can you tell which two are from points on people's joints, and which are scrambled?
Images II and IV are from a point light walker (the other two are from scrambled walkers -- see the link at the beginning of the next paragraph). Most people can't tell stationary point light walkers from those that are scrambled. However, once you set them in motion, it's a different story. Take a look at this (press play).
For other examples, go here3. The first two Quicktime displays at that link (from which the above stationary images were taken) indicate what sort of information we are using to perceive biological motion from these point light animations. The scrambled and normal animations use the same local motion (each point does the same thing), but the global structure of the motion is very different. So the brain is using this global structure to detect biological motion, even when the cues are limited to a few points of light. But our ability to get information out of this global structure is even more impressive than merely being able to determine whether the motion is human (and what the human is doing, as you can see in the QuickTime displays at the link above). We can also detect gender4, emotional states5, and even individual identities6, just by looking at point light animation. If you want to see some illustrations of gender and affect in point light animation, go play with the demo at the Bio Motion Lab.
Now on to the recent research. In an experiment published in the June issue of Nature Neuroscience, Jordan et al.7 had participants watch one of three types of point light walkers, a male, a female, or an "ambiguous" walker that was a combination of male and female walkers (50% male and 50% female). After watching the animation for 11.67 seconds, participants watched another point light walker (the test walker), composed from varying combinations of male and female walkers, for one second. They were then asked whether the second walker was male or female.
To measure accuracy, Jordan et al. computed the "ambiguity point," or the percentage of "maleness" in the test walker at which participants were equally likely to answer that the test walker was male or female. For the participants who first watched the ambiguous walker, the ambiguity point was 49% (so participants were really accurate). For those who had first watched the male walker, the ambiguity point was 55%, and for those who had first watched the female walker, it was 43%. Both were significantly different from 49%. So, watching the male walker caused participants to perceive more female motion, meaning that it took a higher percentage of maleness for them to reach the ambiguity point, and watching the female walker caused participants to perceive more male motion, resulting in an ambiguity point with less maleness. In other words, the participants displayed a gender aftereffect!
In order to show that this aftereffect was a result of the global structure, and not simply information from the local movement of individual lights, they repeated the experiment with "coherent" male, female, or ambiguous walkers and "dephased" (scrambled) male, female, or ambiguous walkers. The dephased walkers were created by "by randomizing the gait cycle phases of individual lights in the original coherent adapter," which served to disrupt "the global coherence but left the local motion of each light unchanged" (p. 739). Participants watched the coherent or dephased walkers for 11.67 seconds, and then watched a composite walker for 1 second. Again, Jordan et al. computed the ambiguity point (49% for ambiguous walkers), and as in the first experiment, they observed an aftereffect for those who had watched the coherent male walker (57%) and female (42%). The dephased walkers also resulted in an aftereffect, but it was significantly smaller than that for the coherent walkers. Thus, the global structure of the walkers were responsible for much of the aftereffect.
These results led Jordan et al. to conclude:
[O]ur results reflect adaptation occurring after the level of local motion processing. These findings are consistent with the existence of neurons selective for gender, as derived from biological motion. (p. 739)
So there you have it, neural adaptation for gender in biological motion. It will be interesting to see whether other features easily detected in point light animation (e.g., affect) also show adaptation effects. It would be really interesting to see whether the motion of highly familiar individuals shows adaptation effects as well, as it would indicate that such neurons or networks can be created through learning. Another interesting possibility to explore would be whether autistic individuals display adaptation to social cues from biological motion. Blake et al. have shown that autistic individuals' ability to detect biological motion in point light animation is significantly impaired8. Fewer neurons (or networks) that detect biological motion, and social features like gender and affect in biological motion, in the visual systems of autistic individuals might help to explain some of the theory of mind deficits found in that disorder. That's speculation, of course, but it would be an interesting line of research with potentially important therapeutic applications.
1Johansson, G. (1973). Visual perception of biological motion and a model for its analysis. Perception & Psychophysics, 14, 201-211.
2From this website.
3Other examples can bee seen here and here.
4Kozlowski, L. T., & Cutting, J. E. (1977). Recognizing the sex of a walker from a dynamic point-light display. Perception & Psychophysics, 21, 575-580.
5 Pollick, F.E., Lestou, V., Ryu, J., & Cho, S.B. (2002). Estimating the efficiency of recognizing gender and affect from biological motion. Vision Research, 42, 2345-2355.
6 Hill, H., & Pollick, F. E. (2000). Exaggerating temporal differences enhances recognition of individuals from point light displays. Psychological Science, 11, 223-228.
7 Jordan, H., Fallah, M., & Stoner, G.R. (2006). Adaptation of gender derived from biological motion. Nature Neuroscience, 9(6), 738-739.
8Black, R., Turner, L.M., Smoski, M.J., Pozdol, S.L., & Stone, W.L. (2003). Visual recognition of biological motion is impaired in children with autism. Psychological Science, 14(2), 151-157.