That's a good question, but in the context of this I think the key is clearly defining what the two reward options are. Additionally, this is ideally being used with low-value rewards so the disappointment should be minimal.
What the research seems to suggest is that by properly framing an unsure reward, individuals put less emphasis on the value of the reward and focus more on the process by which they will attain the reward. The reward itself then becomes secondary to the fact that a reward was present at all.
It's very interesting stuff. I'd suggest reading the article linked and doing a bit of tangental research. I really think the key lies in keeping the rewards low-value and framing the uncertainty in a way that emphasizes the "you'll just have to play along to find out" and then delivers something of reward-worthy value (though that takes surprisingly little).