Error Feedback: Theory
Charles P. Nelson,
Explorations in Learning,
Mar 14, 2007
The whole post is a but fuzzy but manages to make its way to what I would consider to be the correct conclusion. "Although not innate, expertise takes time to develop." The author cites Anderson and Schunn: "For competences to be displayed over a lifetime, time on task is by far and away the most significant factor." And in particular, "for learning a language, 1. the crucial element is practice rather than some language module [and] 2. the process cannot be accelerated." The article does note that "all practice is not equal" (they mean, "not all practice is equal") and that "'effective time on task' is promoted through... feedback." But they don't mean 'feedback' per se. They mean correction. Which, computationally, is represented by (for example, something like) back propagation. Which is the sort of thinking that would be expected of the author of Human Associative Memory.
Today: 4 Total: 91 [Direct link] [Share]
