One of the things I tend to do - as do many who have had some educational background in psychology - is equate some aspect of psychology with whatever it is falls into my sphere of influence. This isn't entirely unwarranted, of course - psychology is the study of human behavior, and most everything involves some component of human action.
So, then, video games. A series of features over at Gamasutra goes in-depth to look at the nature and implementation of achievements, and in the second article, the author starts to look at some issues that smack of a lot of the psychology of gaming. In fact, they all relate to the psychology of play, but I'd like to focus on the portion of the article that addresses the timing of achievements, because it falls nearest to the realm of behavioral psychology.
But first, a short digression - behavioral psychology can be a dangerous topic to broach regarding humans, because it operates on a very fundamental, inhuman level and applying it to systems as complex as human thought and interaction is fairly reductionist. It works very well for explaining simple systems, like the cognition of animals (Skinner, et al.), but humans have such a large - and conflicting - set of motivations and goals that even if behavioral psychology were applicable, it would be like trying to graph a line whose path was determined by a chaotic system. That said, I believe behavioral psychology can still be used effectively to analyze single interactions one person makes with a system - just like this:
Okay, not just like that.
Anyway, to recover from my digression. In the article linked above, the author brings up the notion of adapting the timing of the conditioning, err, achievement - "When Achievement Notification Occurs" - to effectively manipulate the type of player being fed the achievement. This prompted the thought train (one that has no doubt been traveled on many times by many before me) of the notion of achievements being used as a fairly innocuous carrot-and-stick (or Skinner box pellet, to maintain the parlance). Since their advent, they've become a relatively unobtrusive means of offering notable and immediate feedback on player performance, and the idea, as the author of the "Achievements" article alludes to, that they can be used as a learning tool is a good one.
Achievements too may be able to effectively be used as a tool to promote what would otherwise be more "emergent"-type behavior. One of the ways behaviorism is used to train dogs (and here again is why I offer the caveat I do above) is to reward gradual changes in behavior to teach a complex action in parts. For example, a dog being taught to roll over may be rewarded first for sitting down, then for laying on its side, then for turning on its back, and then for finally rolling over (this is a highly simplified version of this). Without being too reductionist, I feel that this kind of reinforcement paradigm could be applied to a player using achievements.
There's really more to this subject than a simple blog post warrants, but I think it's something interesting to think about.