[kolbe]

Is this baseline you speak of a personal baseline for each participant in the study? Or is it like a canonical brain state that all other brains are compared to?

And pardon my leap of logic here (I'm not a neuroscientist), but why would a personal baseline affect the results? A personal baseline, intuitively, would not add to the goal of the study and could negate results that they're looking for. That is, a logical person will probably be more logical in his baseline, too. So, logical_observation - logical_baseline = 0. But what they're looking for is simply whether logic is being used in response to the experiment. And I wouldn't think the goal of the study is to find out whether people are more logical than their baseline, just whether they're more logical than other people who make different assessments.

Whereas, if we understand, based on previous work, that X things happen in an MRI when thinking logically and Y things happen when acting emotionally, why can't we simply correlate observations to these known states?

[amirmc]

The sibling comment is correct, in that you're always measuring differences. The 'baseline' isn't a canonical one but specific to your experimental design (and research question).

For example, if I were interested in which regions respond to pain and pleasure I have several ways I could design a study. I could have a 'pain' condition (drink something bitter), and a 'pleasure' condition (drink something sweet) and then just do comparisons between them (pain - pleasure, and the inverse). However, I could also include a 'neutral' condition (drink plain water) and now I can compare the emotional conditions against the 'neutral' one and extract the regions that are generally involved in emotion.

Also, your comment highlights another problem when research work is badly disseminated. The concept of logic is not being examined whatsoever, yet you've taken 'cognitively driven' (already a false statement) to mean 'logical'.

[fluidcruft]

What you observe in fMRI are always changes, not absolute measurements. So you end up comparing differences between changes.

Suppose you measure the height of two mountains their bases, but you want to make inferences about which mountain has the greatest elevation. To do that you have to compare the relative heights of the bases where the measurements were made.