The reported study follows up a phenomenon regarding the behavior of gamblers that is so well-established that the entire slot-machine industry builds and programs their machines to take advantage of it: near misses (two lemon and a strawberry) and machine that allow the gambler to control the process partially (e.g., to make one of the spinning columns stop at a certain point) motivate certain gamblers to keep playing almost as much as consistent wins. They measured fMRI bloodflow and found that, indeed, the brains of gamblers respond similarly to wins and near misses. However, nothing in their research explains why the brains of gamblers respond in this parallel fashion, or even if perhaps the similarity in the blood flow pattern is present because the gamblers perceive winning and near misses as similar, that is, reverse causality.
Yet, a purely behavioral study of the phenomenon, one that perhaps measured the effects of things like task complexity, intelligence, income level, and so on, would be much less likely to excite readers, and, even though the research would be much less expensive to run (fMRI machines are expensive!), it would probably be much less likely to be funded.
Why? My explanation is what I will call "voodoo neuroscience". The experiment I linked to above is a perfect illustration. You just take expensive technology and apply it to find some new correlation or enhanced precision of measurement, even though it adds nothing or very little to our understanding of the underlying phenomenon, and it is viewed as exciting, revolutionary, and important. A linguist I knew used the expression "physics envy" for such voodoo scientific endeavors.
I'm not saying that we should not do the cognitive fMRI experiments, but I am saying that in terms of costs and benefits, the magnitude of the benefits should not be inflated as they currently are, and as they were in the fMRI slot machine study.