Intuition, observation, focus testing, play metrics, experience...all of these improve our ability as designers to create entertaining games.  However, I often wish I had a lab where I could run scientific experiments that gather data on the physical and chemical responses in players' brains.

For example, on a recent plane trip, I found myself pondering the differences between video games and board games.  I noted that while some board games and video games are similar in nature, video games are better able to put the player into a fantasy role (soldier, wizard, captain, etc.).  Without a computer to create a detailed simulation around that role, board games have fewer tools to fully realize those roles.

I then thought about what kind of challenges are most satisfying in a video game that does put the player in a fantasy role.  Perhaps they are the challenges that increase the player's immersion into that role? 

At that point, I began to think about challenges more generally.  How do our brains react (physiologically)  when we complete a challenge? And how does that change if the game does/does not recognize that completion? And how does it change if the game provides a reward? And what happens when the player fails a challenge? Several challenges? Consecutive and periodic? What about different types of challenges: estimation, planning, observation, memory, coordination, motor learning?

And then of course I launched into the fantasies of brain wave scanners coming standard with the latest gaming system, allowing us to alter the game based on the player's mental state instead of relying on crude gauges such as gameplay success and failure.  And then compiling that brain data into a large database that allowed us to steer players towards games that players with similar response profiles enjoyed.

And yes, I am interested in what psychology has to say, too.  In fact, I often look at MMOs as big psychological experiments, where World of Warcraft is the one where the guy never wants to leave the comfy chair and Everquest was the Stanford Prison Experiment.  It is certainly interesting to watch how human nature manifests itself in worlds with different parameters than our own.  

However, I'd rather get right into hacking the brain directly instead of going through the psychology API.  Unfortunately, not many of the experiments I've read about in this field pertain to the art of game design - if you know of some or are doing some, I would certainly love to know more about them!