With these kind of fMRI studies, there's usually a written assessment before or after the scan to verify your understanding of the concepts. They're making claim 1, that you can detect how learning happens. There's also usually a baseline scan before the class or whatever education method takes place.

Source: I work in a very similar fMRI lab.

I have a related, but different, question about what the classifier is discriminating on. The article quotes Kraemer as saying “in the study, we found that when engineering students looked at images of real-world structures, the students would automatically apply their engineering knowledge, and would see the differences between structures such as whether it was a cantilever, truss or vertical load.” This makes me wonder whether the algorithm is discriminating on knowledge of engineering terminology, rather than what the subjects understand of the physics involved. I am curious as to how the non-engineer subjects who got the answer right compare to the engineers.

Update: on reading the paper, it seems that the goal of the experiment was to distinguish between engineers and novices, and it conflates 'knowledge' with 'understanding', so the claims that the discrimination is on the basis the subjects' understanding of the issues appears to be editorial overreach.

Sure, I believe you can rigorously find brain patterns that are correlated with understanding. The obvious, "correlation is not causation" applies, however.

The problem is such pattern aren't any guarantee of understanding. Just "feeling confident" seems like such an easily correlated thing. More people who learn a subject sufficiently to do well on a test might "feel confident" but certain kinds of people might just feel confident without that actual learning.

But they're testing the participants to see if they actually understand. Re-read gp

They're testing understanding for the learning set, but applying a correlation to estimate understanding on the test set.

What is scientific causation but correlation that works to an acceptable quantity of Sigma?

Causation also implies well sequential causality in some way and reproducibility. A subtle but crucial difference.

If most people on the battlefield are covered in their own blood will soon die that doesn't mean gathering a gallon of a person's own blood through several donations and then throwing it at them at a Gettysburg musuem will kill them.

As Feynman discussed in "The Character of Physical Law", there is more to it than that, but I think one can say that the empirical evidence underpinning all science can be characterized as correlation that passes various statistical tests.

Causal studies need to control for confounding variables, for one. There are all sorts of spurious non causal correlations with low p values.

Also, it is a matter of state of mind and mood - you might not believe you understand the concept because you do have bigger picture, but you might also understand it as intended by the test, but not believe that you yourself understands. We have an image of ourselves

fMRI wrecked by impostor syndrome and Dunning-Kruger.