Possibly I'm an amoral wretch, but I was surprised, last month, by the shock that greeted a study about what people do when they find lost mobile phones. In what it called the "honeystick project", the digital security firm Symantec left 50 smartphones in North American cities, loaded with juicy-looking files – one called "Saved passwords", another "HR salaries" – and configured so that researchers could monitor them remotely. Stunningly, given the study's sponsors, the results indicated the importance of digital security: 96% of people tried to access data (53% snooped at that salaries file) and only 50% attempted to return the phone. The findings "confirm[ed] the worst in people", sighed one typical headline. Only Canadian media outlets managed to sniff out an optimistic angle: "Lost cellphone study deems Ottowans most honest."

I'm not sure the honeystick project reveals much about honesty (if I find a phone but don't get round to returning it, am I dishonest, or selfish?) or about the need for digital security (how many of those "snoopers" were just trying to identify the owner?). What it confirms, though, is a decades-old truth about social psychology experiments: one of the most entertaining – and media-friendly – ways to discover what people are like, as opposed to what they say they're like, is deliberately to "abandon" items in public places and watch what happens.

The godfather of this technique was the American psychologist Stanley Milgram. In the early 1960s, Milgram prepared hundreds of letters, all bearing stamps and addressed to a variety of organisations, then scattered them around New Haven as if they'd slipped accidentally from the sender's pocket. All bore the PO box number, which belonged to Milgram. The results confirmed his suspicions: three-quarters of the letters addressed to a medical research group were picked up and posted on, while only 25% of those addressed to "Friends of the Nazi party" reached their destination. (Which is still oddly large: apparently, a quarter of New Havenites didn't disapprove of Nazism sufficiently to override their public-spiritedness.) You could write books about the methodological pitfalls of this approach, but it proved durable. Milgram subsequently used it to detect which neighbourhoods of Munich were home to the most neo-Nazis, and to foretell the outcome of the 1964 US election: letters to groups favouring Lyndon Johnson got posted more than those favouring Barry Goldwater.

Since then – in studies and stunts with varying degrees of credibility – we've learned that most people return small sums of money; that your wallet's more likely to be returned if it includes a picture of a cute baby; that people won't steal a wallet that has a chalk circle drawn around it; that women are more honest than men, and that people are more likely to steal money from men than from women. Considering these studies all together, one gets the impression that people are basically honest, but too wrapped up in their own concerns to spend hours helping unknown strangers, and also (the key finding of the smartphone project, surely) incredibly nosy. And that Canadians are trustworthy. Which presumably means they don't need so much digital security – though Symantec, weirdly, didn't spin it that way.

oliver.burkeman@theguardian.com; twitter.com/oliverburkeman