The patterns and locations of activations might be predictive within a sample of images, but cannot discriminate in other samples. The temporal and spatial resolution is too low.