TL;DR
Many psychological theories deal with concepts that are hard to objectively measure, like valuation, or ... deception.
We can use brain-decoding to measure such thought processes and directly test psychological theories.
Of course, easier said than done.
Example Research
Psychological theory: future rewards are not as valuable as immediate ones because people can't imagine them as vividly.
Difficulty: how do you measure something as subjective as 'imagination vividness'?
Answer: build a brain decoder of imagination vividness.
Test of theory: when people see future rewards, their imagination vividness is low as measured by the brain decoder.
Paper: A neural signature of the vividness of prospective thought is modulated by temporal proximity during intertemporal decision making. [PDF]
Challenges
1) Constructing a high-dimensional model to predict mental states is hard.
Usually small dataset size
Computational burden
Interpretability of model
Paper: Fast construction of interpretable whole-brain decoders. [PDF]
2) Ensuring validity of brain decoder is hard.
How do we know if our decoder is predicting the mental state that we want and not something else?
If our decoder is confounded, how do we fix it?
Paper: Distinguishing deception from its confounds by improving the validity of fMRI-based neural prediction. [PDF]