Thank you. It's quite interesting, actually, how the easy-to-compute discriminative models can be fooled so easily, while they hypothesize that generative models (which are much more general, and closer to our best models of human cognition), despite being much slower, ultimately provide better performance at the boundaries of the classification region.
I'd go on to venture that our minds probably use discriminative models for "intuitive" judgements, while reasoning that can explain the variance in percepts uses generative models, thus obtaining better performance where necessary despite using more energy. Or possibly, the generative models can be used to train the intuitive discriminative ones, slowly allowing the less intensive part of the mind's processing to adjust its class boundaries to suit what's really known.
I'm not exactly a deep-learning fan, but could you post the links?