Perceptual Learning Project Page

My present research focuses on the mechanisms of perceptual learning. A recent Psychological Review article (Petrov, Dosher, & Lu, 2005) exemplifies the benefits of combining quantitative experimentation and formal modeling. An experiment revealed perceptual learning effects that were partially specific to the context surrounding the target stimuli: Switch costs (interference) occurred whenever the context changed. A multi-channel selective reweighting model provides an existence proof that an incremental Hebbian mechanism can account naturally and quantitatively not only for this novel result, but for the well-documented stimulus and task specificity of perceptual learning, as well as for its detailed dynamics in non-stationary environments. The model has a fully functional perceptual subsystem that works on the images themselves and is consistent with the physiology of the primary visual cortex. Mathematical analyses and computer simulations show that the recurring switch cost pattern arises from the differential predictive value of certain context-dependent stimulus features. This conceptual understanding allowed us to make novel predictions about the role of feedback in perceptual learning, which were confirmed in a follow-up experiment (Petrov, Dosher, & Lu, 2006).

Publications

A comprehensive list is available from the publication page. Some of the main ones are:

Petrov, A., Dosher, B., & Lu, Z.-L. (2005)

The Dynamics of Perceptual Learning: An Incremental Reweighting Model. Psychological Review, 112(4), 715-743.
Abstract Reprint (pdf) Data and software

Petrov, A., Dosher, B., & Lu, Z.-L. (2006)

Perceptual Learning Without Feedback in Non-Stationary Contexts: Data and Model. Vision Research, 46(19), 3177-3197.
Abstract Preprint (pdf) Data and software

Petrov, A. & Hayes, T. R. (2010)

Asymmetric transfer of perceptual learning of luminance- and contrast-modulated motion. Journal of Vision, 10(14): 11, 1-22, http://www.journalofvision.org/10/14/11/.
Abstract Reprint (pdf) Matlab reports (html)

Petrov, A., Van Horn, N., & Ratcliff, R. (2011)

Dissociable perceptual learning mechanisms revealed by diffusion-model analysis. Psychonomic Bulletin & Review, 18(?), ???-???.
Abstract Preprint (pdf) Supplement (pdf) Matlab report (html)

Data Sets

The data from two perceptual learning experiments is bundled in the PLM implementation archives together with the software that administered the experiments. "PLExp1" tested perceptual learning in an orientation-discrimination task in non-stationary contexts with trial-by-trial feedback (Petrov, Dosher, & Lu, 2005). "PLExp2" was a no-feedback follow-up (Petrov, Dosher, & Lu, 2006). The data from several other studies are available from Alex Petrov upon request.

Software

PercLearn-full-v1.1.zip (11.2 MB) PLM implementation Readme License

The complete Matlab implementation of the Hebbian Perceptual Learning Model of Petrov, Dosher, & Lu (2005, 2006) is available from my open-source software page under the GNU General Public License. It includes a short tutorial and numerous transcripts of all sorts of Matlab sessions taken during the development and fitting the model. The psychophysical data used to estimate parameters is also available.