Move the stop-window to remove words from the lense. Removing 4 though 6, the default, corresponds to the tuned hedonometer for large corpuses.
Click on the graph and drag up to reveal additional words.