Perceptual learning models are constrained by the location and feature (e.g., orientation) specificities. These specificities are regarded as fundamental properties of perceptual learning, and are assumed to suggest a retinotopic early visual cortical locus of learning. However, such specificities would also suggest a highly inefficient brain that has to learn every stimulus feature at every retinal location. We developed a novel double training program to demonstrate that perceptual learning is nearly completely location and orientation transferable. Regarding location specificity, we had subjects practice conventional feature training (e.g., contrast) at one location, and additional training with an irrelevant feature/task (e.g. orientation) at a second location. This additional training enabled a complete transfer of feature learning to the second location. Regarding orientation specificity, we had subjects practice conventional feature training (e.g., foveal orientation) at one orientation, as well as additional training with an irrelevant feature/task (e.g., contrast) at a second orientation. This additional training again enabled a complete transfer of orientation learning to the second orientation. Our findings challenge location and orientation specificities as the fundamental properties of perceptual learning and question the inferred critical involvement of the retinotopic early visual cortex. Rather they are more consistent with the Mollon-Danilova hypothesis that learning occurs at a central site. We propose that perceptual learning involves training-induced improvement of task-specific, but feature and location non-specific, decision-making in high brain areas.