The neuron isn’t picking out any coherent linguistic or semantic category – it simply fires on whatever words happen to have a strong value in that same embedding dimension (e.g. “I,” “struggled,” “click,” “strengthening,” dates, certain numbers, etc.).
