Understanding how the tasks that we perform while perceiving a stimulus modulate brain activity is of wide interest to neuroscience. Reviewers found the experimental setup interesting, with the clear hypotheses about how tasks can impact neural activity. The small effect sizes observed were identified as a key limitation in drawing conclusions from this experiment. Reviewers found this worrisome particularly when coupled with marginal accuracies and few subjects. Reviewers identified that BERT results were not very conclusive, and significantly more could be done. This would make the paper much more interesting to an ML audience. Reviewers did not find the methods themselves to be particularly novel, but did find that the application and clear test was valuable. In general, the work is suggestive of how one might more deeply connect multitask models to neural data, something that has not occurred to date and could provide fertile ground for both creating new models and furthering our understanding of the brain.