"There's a well-known study by Shiv and Fedorikhin showing that if you tax people's working memory by making them remember a 7-digit string of numbers, people are more likely to make a less-optimal decision (they chose to eat chocolate cake versus the more healthy fruit salad option)."

I'm not sure how the Shiv experiment proves anything. I personally would always choose the chocolate cake, no matter how much time I had to spend on the decision - I just don't like fruit salads, at least not as much as cakes. And one doesn't eat desserts all the time, so it's not much of a problem anyway. 'Optimal' != 'healthy', and I bet (pun intended) that when choosing dessert many people optimize for feeling, not for health.

(Maybe this effect was documented using some other experiments, but sometimes when I read psychology stuff, I start do doubt both my and researchers' sanity, because I can't figure out how little shifts of behavior over e.g. a chocolate cake, tested on a small (< 1k) sample are really giving us any valuable insight, and are not just random noise)