These days, people are routinely outperformed by computers in all kinds of different tasks. But one ability for which humans still hold the crown is speech recognition, a task that we carry out hundreds of times a day without a passing thought. How does this work, and what makes us so good at it?

Often psychologists learn a lot from cases where language is ambiguous. The ambiguity can slow down recognition to make it easier for us to understand what’s going on. Ambiguity also allows us to test theories about the types of information that are used routinely to ease comprehension. A classic example of ambiguity that became prominent in 2018 was the yanny/laurel ambiguity, which showed how much individuals can differ in their comprehension of the same speech. You can find out more about the science behind this example, and even adjust the prominence of the different frequency bands in the clip to find your own boundary between the two words.

Although less well known, this example of ambiguity may be even more compelling. For many people, you can decide what you are going to hear in advance. If you think you’re going to hear “green needle” you do; if you think you’re going to hear “brainstorm” you hear that. It’s a great example of “bistability” in perception, and particularly impressive because the two interpretations seem so different.

If you would like to find out more about how models of spoken word recognition work, and even download the software to try one out for yourself then check this page.

Or you can find out more about how we learn new words, and how sleep helps us to do this.

Next Section »