They used the N-back task, a standard task used for inducing mental effort (e.g. [1]). So in this case they have N=1 for the easy group and N=3 for the hard group. Then they compared glutamate concentration across groups.

I'm don't disagree that "thinking hard" is poorly defined. But they did control for staring at a screen.

[1] https://journals.plos.org/plosone/article?id=10.1371/journal...