Blind to Counterexamples?
The following from Newsweek is unlikely to be the whole story of alleged failures of reasoning:
Wouldn't you expect defenses against invalid arguments? The need for an ability to see through other people's propaganda is a clear counterexample to this argument.
The fact that humans are subject to all these failures of rational thought seems to make no sense. Reason is supposed to be the highest achievement of the human mind, and the route to knowledge and wise decisions. But as psychologists have been documenting since the 1960s, humans are really, really bad at reasoning. It’s not just that we follow our emotions so often, in contexts from voting to ethics. No, even when we intend to deploy the full force of our rational faculties, we are often as ineffectual as eunuchs at an orgy.
An idea sweeping through the ranks of philosophers and cognitive scientists suggests why this is so. The reason we succumb to confirmation bias, why we are blind to counterexamples, and why we fall short of Cartesian logic in so many other ways is that these lapses have a purpose: they help us “devise and evaluate arguments that are intended to persuade other people,” says psychologist Hugo Mercier of the University of Pennsylvania. Failures of logic, he and cognitive scientist Dan Sperber of the Institut Jean Nicod in Paris propose, are in fact effective ploys to win arguments.
Besides, according to other research (seen via Rust Belt Philosophy), slime molds have similar “failures of logic,” but—as far as I know—have no need to win arguments. (If I lose an argument with a slime mold over the next few days, I'll let you know.)
One possible reason for the human tendency to look for examples instead of counterexamples is that counterexamples are mostly useful for judging universal statements of the form “for every …” and I suspect such statements were less useful during most of human prehistory than existential statements of the form “there exists ….” (On the other hand, the preceding sentence was a universal statement and therefore must be tested with counterexamples.)