Decoding moral judgments from neural representations of intentions