Next month sees the publication of Are You Smart Enough To Work At Google? by William Poundstone, a book about job interviews that seems well-suited to the current climate. With so many applicants chasing so few posts, it's reportedly getting more common to encounter the kind of "extreme interview questions", as Poundstone calls them, previously widespread only in Silicon Valley. Such as: "You're shrunk to the size of a 10p piece and thrown in a blender… The blades start moving in 60 seconds. What do you do?" Or: "How would you market ping pong balls if ping pong itself became obsolete?"

We've entered an interview arms race. The new breed of questions arose because traditional ones were so bad at identifying the best candidates: it was easy to "get good at interviews" without being best for the job, or vice versa. Now this book enables you to get good at the new questions, too, thus rendering them useless. Google's next move should be to stun applicants with something unexpectedly retro: "Where do you see yourself in five years' time?" Hint: the answer doesn't involve being trapped in a blender.

But the interview conundrum will probably never be solved, because it's an archetypal case of the mental quirk that psychologist Daniel Kahneman calls "answering an easier question". Faced with a cognitively demanding question, involving uncertainty – "Will this person do the job well if hired?" – interviewers unconsciously substitute an easier question, and answer that one instead: "Did this person impress me in the interview?" We all do it, all the time. Watch any news discussion programme – questions such as "Who'll win the next election?" are often answered as if "Which party's most popular now?" had been asked.

In one classic experiment, students were asked, "How happy are you these days?" then "How many dates did you have last month?" There wasn't much correlation, suggesting that dating wasn't the major factor in their lives. But when the questions were switched, the correlation was strong. Why? Because, Kahneman argues, when people hear, "How happy are you these days?", the question they answer is, "What's your current mood?" The dating question made students feel happy or sad, and they answered the next one accordingly.

As the philosophy blog Less Wrong noted recently, all sorts of "cognitive biases" follow this pattern. We ask ourselves, "How scared should I be of a terrorist attack?" But we answer, "How scary did it look last time I saw a terrorist attack on television?" I'd go further: this "substitution principle" governs whole swaths of our lives. You ask how you can minimise your impact on the environment, but the question you answer is, "How can I feel like I'm doing my bit?" – which helps explain the phenomenon known as "moral licensing", whereby people who use canvas bags at the supermarket feel entitled to take long-haul flights. Or you ask if you had a productive day, but what you answer is, "Did today feel unpleasant and/or effortful?" which isn't the same at all.

So how can we get rid of this eccentricity? Sadly, it's probably too deep-rooted, so let's answer an easier question instead. What can we do about it? Stay aware. Next time you're trying to solve a problem, remember to check that the problem you're solving is the problem you've actually got.

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