there are two common explanations for this. one is that "mental disorders affect everyone" -- they're caused internally. people aren't depressed, the theory goes, because their life sucks; they're depressed because they have a fundamental chemical imbalance, and the same goes for other mental problems. just because you are a celebrity does not mean you can't fall prey to the same fundamentally intrinsic mental problems as anyone else.
the second explanation is that being famous produces a lot of pressure, and it's not surprising that a girl who was 18 when she became a super duper megastar was unable to handle this pressure, basically cracking under a fast lifestyle and a horde of paparazzi. to be so young, with so much money, and have the opportunity to go nuts and ruin your life without even making a dent in it, is a temptation that when combined with the intense scrutiny on you is hard to resist.
i'm not going to comment on my personal opinion of the validity of either of these explanations, but i think there's something much much simpler going on here. in short, it's regression to the mean. this is an extremely important phenomenon that most people don't seem to get. basically, it suggests that every observational average will tend to gravitate towards the mean in the future. for a simple example, suppose you have a coin and you flip it 10 times and get 9 heads. what's your estimate of the next flip being a head? it's going to be closer to 50% than 90%. in this case, you'd probably even ignore the possibility of the coin not being fair, and guess 50%.
but what if it were 100 times, 90 heads? now the probability of the coin not being fair is substantial, and you might guess something like 85%, or even 90%. but obviously somewhere in between here your guess will be (continuously) 60%, 70%, whatever. how much to regress to the mean is for statisticians (and depends on your prior estimate of coin fairness), but the point is that 90 percent is never a going-forward estimate. it's always going to be less than 90 percent.
okay, what does this have to do with britney? the same thing that it has to do with me. in britney's case, a few things went unbelievably right at the beginning of her career, a perfect storm which allowed her to become arguably the most famous female in america -- the mickey mouse club, the marketing of her first album, word of mouth, probably some connectors distributing her name around the country, who knows what else. but ultimately a surprisingly small number of events. so here you have someone for whom EVERYTHING HAS GONE RIGHT. she's batting 90 percent, or .900 or whatever. and it's natural when things are not a sequence of heads or tails but rather the events in your life to expect them to persist at this charmed average. no one intuitively believes that statistics apply to them.
of course, here comes regression to the mean. things start to go wrong. she chooses to marry a man who she probably liked very much, expecting this to go well just like everything else. it doesn't, and she's devastated. success breeds expectations, but because of this regression to the mean (as well as because these expectations are orthogonal to her success, and thus this success is no real portent of her romantic prognosis), maintaining her level of general success in the future is an unrealistic expectation. so she is doomed to live a life of disappointment, because she can never be as successful as she was at its beginning.
what does this have to do with me? it's exactly the same thing. it's better for me than for britney, because my successes are more numerous (if less striking) and more variegated. but let's face it, pretty much everything i've done has succeeded, with the exception of relationships, which have not all in all resulted in k-fed-style epic failures. but again, regression to the mean lingers there -- i'm disappointed much more than i'm pleasantly surprised, because i expect everything to go right for me (or at least 90 percent), because it has in the past. for both me and britney, this is very hard to come to terms with, and i think it's this, more than either of the other explanations, that is responsible for our sometimes shaky mental state.