Looking through the mind's eye via multimodal encoder-decoder networks