Friday, January 23, 2009
question about multiple, overlaying feedback conditions
I am starting to become focused on the question of how to study individual parameters of cane travel using our apparatus, and how (or whether) this can all be brought together into a wholistic picture of the trainee's behavior and adherence to classic 2-point touch technique. It seems to me, and we will know more about this as Zach continues his inquiry, that we will always be studying only one parameter at a time. so, we would not be able to set up the apparatus to look at, say, both arc width and wrist isolation simultaneously, because, 1, it is not possible to simultaneously be carrying out multiple real-time calculations necessary, and, 2, because there is no practical way for a user to process feedback about two different features of cane use at the same time. It's hard to imagine most people understanding, for example that two different beeps mean they are swinging too wide left or right, while a vibration would mean that they were not immobilizing their wrist. this is especially hard to imagine being at all effective in young children. On the other hand, I can picture a system working that uses verbal coaching to do this: "veering left", "widen arc left", "try to keep the cane from rising so high off the ground", whatever statements are the most effective in actual training.
Subscribe to:
Post Comments (Atom)
8 comments:
I reviewed the lit on motor skills learning at the very beginning of the project last year. I think your proposal is supported by the theory realted to motor-skills acquisition. Students in easrly stages need sufficient practice without having to perform other tasks, such as attending to two aspects of cane use parameters or noticing sensory information.
“One leading notion is that some of these performance changes [improvements] are related to transitions between two types of processing modes: an initial, controlled, more effortful mode and eventually a more automatic mode . . . ” (2008, p. 1544).
The goal is for the student to establish the muscle memory until it “becomes increasingly resistant to interference from competing or disrupting factors with the continued passage of time” (Walker, Brakefield, Hobson, & Stickgold, 2003, p. 616).
Adi-Japha, E., Karni, A., Parnes, A., Loewenschuss, I., & Vakil, E. (2008). A shift in task routines during the learning of a motor skill: Group-averaged data may mask critical phases in the individuals' acquisition of skilled performance. Journal of Experimental Psychology: Learning, Memory, and Cognition, 34(6), 1544-1551.
Walker, M. P., Brakefield, T., Hobson, J. A., & Stickgold, R. (2003). Dissociable stages of human memory consolidation and reconsolidation. Nature, 425(6958), 616.
So, are you saying that the literature supports the idea of starting with a single feedback conditions about one aspect of their performance, and once that has been learned, we can then introduce another kind of feedback? Please elaborate.
Yes, I think the goal of the device is to help achieve automaticity for the motor skill as well as accurate performance. I was referring to the phenomenon called "interference" (to learning). Skills often consolidate over time or after a night's sleep, but to make this happen you should focus on a single motor task at a time. For example, Walker, Brakefield, Hobson, and Stickgold wrote, “. . . when subjects . . . were trained on a second motor sequence immediately after the first, interference was seen, so that overnight improvement in accuracy only occurred for the second sequence, and not the first” (p. 617). I suspect the negative effect would be similar if you attempted to teach the sequences immediately after one another or simultaneously.
My practical expereince agrees with this. I think the O&M specialists here would agree (or perhaps not!) that when a student has problems acquiring a specific skill, an effective approach is to isloate the skill and/or break it down into even smaller isolated components, and focus on this singleton until some mastery can be achieved.
I'd like to hear what others think about this.
Walker, M. P., Brakefield, T., Hobson, J. A., & Stickgold, R. (2003). Dissociable stages of human memory consolidation and reconsolidation. Nature, 425(6958), 616.
I'm all for academic rigor, but could we keep the citations to a minimum? I think they impede readability and hence understanding of the actual ideas being discussed. No one doubts that we can ultimately support our decisions with the proper literature when the time comes, but in my experience, citations don't really help discussion unless everyone involved is familiar with the work being cited.
Back to the actual topic at hand, though...
The system is going to have to be doing some relatively sophisticated modeling of the Wii/participant system no matter which parameter is being monitored. Unless there is really a reason to focus on only one parameter at a time, I don't accept a technical limitation argument for limiting the feedback to a single parameter.
I would support Steve's second idea about the verbal feedback, but I don't know how you would make this work for a deaf participant (maybe a bluetooth Braille display?? -- but that wouldn't work for really young kids...).
I'm reminded of learning to ride a bike or ski: you don't only focus on one thing at a time, or you keep falling over. Rather, you need to (and can) attend to several skills at once. Certainly, one of them may be dominant at any given time (e.g., steering or peddling; keeping your skis parallel or placing your poles), but total skill isolation is rarely possible or desirable.
What if we don't provide realtime feedback at all? Since the actual course will be of limited length, I wonder how much feedback we could realistically provide en route anyway. What if we measured all the parameters for each trackable skill over the whole length of a single track (call it one trial) and gave a report card (verbal or ASL) at the end of each completed run? This would solve the deaf/hearing feedback problem, as well as the distraction problem. I admit, it isn't as sexy as realtime feedback, but it might be more effective and more widely applicable.
Sure Josh. I incldued them because I assume others on the bolg have not read the articles. We have several university faculty (I count at least four) on the blog who I thought would want to go to the actual actilces, but if you find them bothersome I will refer to the authors but leave off the longer citations. I hope this makes for easier reading.
Josh, the only feedback that has been documented to work with veering (and so far as I know we have nothing on the other variables) is the realtime feedback. Feedback after the fact is exactly what Guth, who has studies veering for a few decades, said does not work. “As an O&M instructor, I routinely reported to my blind students that they had veered, basing my reports on the consequences of their veering (e.g., “You
veered into the street”). As I was to understand later (e.g., Gesink, Guth, & Long, 2003) and as Kallie et al. (in press) point out, this emphasis on the consequences of veering is not very useful for instruction or for research because an infinite number of paths lead from Point A to Point B, and the same path can be created by different walking movements.” it is very clear that at least in the case of veering, only the real-time feedback is likely to be effective.
josh, as usual I am horrified at your suggestion, which probably means that in two weeks I will be claiming it as my own. but the idea of abandoning real time feedback in favor of a "report card" is hard for me to even consider. As always, I am focused on developing the technology in such a way that we could expand the audience by coming out with other products based on the same principles and intellectual property. Regarding your general argument regarding, I need to digest that over night, and will have additional comments tomorrow. This is a crucial question, obviously, and I think it deserves careful consideration and debate.
Post a Comment