Skinner is not entirely rejected by everyone (Ariely uses variable schedules as to describe email addiction in Predictably Irrational), and it seems to make sense when you're discussing reduced problems like slot machines or email. However, it feels too reductionist for all problems.
I'm largely working more with Daniel Pink's Autonomy/Mastery/Purpose framework, which seems to do a lot better at describing why people are encouraged to perform more complex tasks over and over, particularly games. Games tweak AMP very directly, and they fail quickly when one isn't in alignment. Autonomy is the freedom to make decisions, mastery is the space to get better at something and continue learning, purpose is discovering that what you're doing matters to you.
What people are seeking is usually feedback. Feedback about themselves, feedback from others. Feedback is usually the only reward you get, but feedback is what is required for AMP to work; you need to know what choices you are actually making (autonomy), you need to know how well you're doing (mastery), whether what you're doing is affecting anything that matters (purpose). Feedback is what drives people that care about karma on Hacker News or Reddit.
HN is a decent example of this if you're karma-oriented: you can write what you want (autonomy), but you can't do much to encourage people to agree or upvote (mastery), and if no-one upvotes you, you lose your purpose. Instead, if you focus on a goal like "writing makes me better at understanding my own ideas", you really can get a real AMP thing going. How many blogs have died because the author thought no-one was reading?
I think designing for AMP is the real desire engine; a self-perpetuating intrinsic motivation built on feedback.