I don't think GP is asserting that the multimodal encoding is "more rich" or "more accurate", I think they are saying that the felt modality is a different thin

I don't think GP is asserting that the multimodal encoding is "more rich" or "more accurate", I think they are saying that the felt modality is a different thing than the text modality entirely, and that the former isn't contained in the latter.