Augmented Hebbian Reweighting Model (AHRM, Petrov, Dosher & Lu, 2005, 2006) passes stimulus images through a representational system of orientation and spatial-frequency tuned units, with non-linearities and spatial pooling. These activations, along with inputs to a bias and feedback unit are weighted by the task-specific weighting system to yield a decision. The AHRM has predicted the dynamics of learning in non-stationary training, the various roles of feedback in learning, and performance in external noise paradigms. (After Petrov et al, 2005).