Thursday, 15 March 2012

Olympic Countdown - Fencing

It can take less than 300 milliseconds for an elite fencer to complete an attack. If you’re going to parry it away successfully, you really need to be able to read the direction of the attack as it’s about to be launched. Afterwards will probably be too late. This, not amazing reaction times, is the key to success in fencing. So the key question is what information do the experts get from studying their opponent that enables them to do this? Several attempts have been made to find out, notably Bard et al (1981) and Hagemann et al (2010). Bard found that fencers looked mostly at the hand guard, but this was true of both experts and novices. But if you’re looking at one place, it doesn’t mean that you can’t get information from other areas as well. So it could be that the experts were fixed on the hand guard, but were also able to use their peripheral vision so that they could pick up subtle differences in angle between upper and lower arms, or even between the lower arm and the wrist. Hagemann’s group used a variety of techniques, such as occluding areas of the body and fixing eye cameras to participants as they watched videos of an expert fencer launching attacks. They found that experts fixated the upper trunk area longer than advanced or novice groups, and their performance in saying where the attack was aimed was most severely affected when this area was occluded. They then tried highlighting this area to see if it helped novices predict the direction of attack any better. However, far from improving performance, it actually made it worse. The problem was that the area of the upper trunk actually included not just the upper part of the chest, but also the arm and the sword itself. Because the arm and sword would often be in front of the chest, it was impossible for the researchers to digitally occlude them without occluding at least part of the chest as well. And because they were highlighting such a large area (with a red patch), they were probably obscuring more than they were highlighting. You probably would have guessed that it would have been some part of the upper body, sword or arm that the experts were using to judge the attacks, rather than, say, the head or feet. What you really need to know is whether it’s just one crucial area or whether it’s the angles between them that are the key. A start would be to get the expert fencer to make attacks holding something really short, but not a sword. Ideally, you’d then attach something to areas of his clothing that only showed up in certain lights. By turning these areas on and off, you should be able to tell which areas are the crucial ones. You could then use something else to for the expert to wear that highlighted those areas to see if it helped with predictions. I’m almost tempted to do some research myself. David Donner