Everyone is interested in the how people make classifications. What do people pay attention to? How do the represent prior experience? How do they weigh evidence in favour of certain decisions?
This is our general approach.
During a standard categorization task, subjects are given a stimuli, asked to categorize it, and then provided feedback on their answer. Most categorization research has focussed on the categorization decision, but we decided to turn our attention to the feedback portion of the trial. Specifically, we are interested in the way subjects in a categorization task use their eyes to process feedback, and the relationship between gaze patterns during feedback and category learning.
We presented our first paper on the topic at the Cognitive Science Society Annual Meeting in 2008 **bold that and make it a link to paper**. Recently, we have begun analyzing eye movements during feedback using a much larger data set, which we intend to submit for publication in the fall of 2010. Thus far, we have found the following nifty results:
* People processing feedback spend much more time looking at the features of a re-presented stimulus than they do at the category labels.
* Eye movement behaviour during the first few trials of the experiment is predictive of both learning success (whether subjects eventually master the task) and learning speed (how quickly subjects master the task). For example
* During the first five trials alone, learners have more fixations than non-learners
* Also during these first five trials, learners’ fixation durations are longer than non-learners’.
* Learners who make more fixations during the first five trials learn faster.
* Eye movement patterns during the feedback phase are highly correlated with the eye movements made during the categorization decision. People seem to be looking at the stimuli in much the same way during both phases.
* The time spent fixating to the correct category label on incorrect trials is predictive of learning success as well.
This is interesting, exploratory research, and we still have a great deal to find out about it. Feel free to drop us a line to ask any questions you may have.