EDIT
$100 isn't necessarily a few days wages for those who are working in a well paying full time position but for a college student that could mean a few days of going out.
But I haven't read the study, so I could be wrong.
Money might be nice in such a situation, but if you're not going to starve to death or freeze to death and you don't have 300k$ in medical debt suffocating you, it might not be as valuable as a couple days extra pay is to a poverty-line worker in the US.
Imagine yourself playing the game but with $1,000 (or more) on offer. Would you refuse $300 to punish your partner for keeping $700? Now change the amount to $10. Would you refuse $3 to punish your partner for keeping $7? Now switch it around: how much of the money, $1,000 and $10 offers, would you give to your partner?
When the stakes go up, your partner is screwing you harder. He's not just screwing you out of $2 or $20, he's screwing you out of $200 or $2000.
Fuck him.
But I would claim that in "WEIRD" societies, we're constantly engaged in a game of iterative prisoner's dilemma with a string of strangers. So you're not forgoing your gain for nothing: you're forgoing your gain for one round to ensure equitable treatment in future rounds.
It's like a two-layered prisoner's dilemma.
First, imagine the simpler dilemma: you have to players, the Chooser (who chooses the split) and the Decider (who decides to take it). You just play the same game over and over again, choose and decide, choose and decide, with no role switching. If you're the Decider, you could settle for a 30-70 split, but over the dozens or hundreds of iterations that's a lot of money lost. It's worth it to forgo some short term gain to keep the Chooser in line. Eventually you end up with a 50-50 split, and your profits are maximized.
In the second layer, imagine you have two pools of Choosers and Deciders. In each game, a Chooser and a Decider are randomly matched up. In this game, you might imagine there's no incentive to forgo short-term gain for long-term, but there is. The trick is that you -- the Decider -- are not just in a Dilemma with your Chooser on each round, you're also in a Dilemma with your other Deciders.
Imagine you have two Deciders, A & B, who have each been offered the same 30-70 split by their Chooser.
If Decider A and Decider B both reject it, maybe on the next round they get each other's Choosers, who then offer them each a 50-50 split. In the long run, their profits are maximized.
If Decider A and Decider B both accept it, on the next round their Choosers might offer them an even worse split. They profit, but far less than before.
If Decider A accepts and Decider B rejects, Decider A profits maximally (next round he gets a 50-50 split from Chooser B, without forgoing any profit) and Decider B profits minimally (he forgoes profit this round and gets a 30-70 split or worse next round from Chooser A).
This is now reduced to the classic Prisoner's Dilemma. If the Deciders cooperate (by rejecting bad offers from Choosers they may never see again) they all profit, if they all defect they're all worse off in the long run, and if some defect (and accept the bad offers) while others cooperate they profit even more to the detriment of their good compatriots.
Throwing away $200+ dollars to spite a stranger cheating you out of $200+ more dollars isn't just about spite; it's about forcing them to cooperate with other strangers so you have a better ecosystem of strangers to cooperate with.