The neuron lights up on mid‐sequence content words—especially less frequent, information‐rich tokens like nouns, verbs, numerals, or list markers (e.g. “half,” “lawsuit,” “1:,” “students”).