What do cows drink?
Your first intuition was probably to answer "milk." And then, depending on how familiar you are with bovine diets, you realized that, wait, it's the calves that drink milk—adult cows drink water.
What’s going on that makes it so hard to respond correctly? And what does it tell us about how our brains store and process language?
The current leading theories of language processing have to do with "connectionist" models of language. The idea is that when you learn a word, it gets connected to similar words and concepts, so that it’s easier to recall when you need it. These connections form every time you use the word, from what’s going on in the context at that moment. Going back to our example, most of your conversations about cows have probably included the word milk, so your brain has linked them together. When you hear one of the linked words, the other one gets activated, so you’re ready to talk about the general topics of dairy products and farm animals. It would take a lot longer to get to milk from cow if they weren’t connected, and since you’re often talking about them at the same time, it saves time. Efficiency!
But sometimes the shortcuts trip you up. As soon as you hear "cows" in "What do cows drink?" your brain goes "Oh hey, cows! We might need the word milk soon, I’ll get that ready." And then when you’ve heard the whole question, asking for something that gets drunk, your brain jumps in with "Milk! You can drink milk!" Since this one question is a very rare case where cows and milk don’t go together, the shortcut remains strong, and you’ll make the same mistake every time.
But it's not just bovine factoids that get connected to each other—our brains also create many other types of shortcuts. For example, social scientists have studied implicit stereotypes of race, sex, and other attributes based on these patterns of mental connections. And linguists have looked at how our brains organize and access words by testing which words are more associated with each other. But alas for the second-grader in all of us, it's hard to measure the kind of puzzlement that comes from a brain teaser, so researchers hide their search for word connections in other kinds of experiments.
For example, one common method involves sitting the participant down in front of a computer screen, and having them press a green button when they read a real word, and a red button for a fake word (like "wug" or "blick"). Just before the target word, participants will very quickly see a completely different word (the "prime"), which may or may not be linked to the target.
So, you might very quickly see COW just before you have to decide if "milk" is a real word. You’ll probably be faster to say "milk" is a word when COW comes first, than if you see an unlinked word like SPATULA before "milk." Depending on the particular experiment, you might see the prime for such a short time that you don't consciously notice it, but the effect still comes through whether you had the time to notice COW or not.
Now that we have our experiment designed, we can test all kinds of things for linkages. Many researchers have studied the topic and it turns out that lots of the connections we develop between words are based on sounds. So cow is connected to other words that start with a K sound, like cat, koala, college, catastrophe, and Kinnickinnic. And milk is connected to M words, as well as "mih-" words like mitten, Mississippi, and million.
But where do these connections come from? One tantalizing clue is that children link their words up differently from adults. For instance, adults don’t have connections between rhyming words, like cat, hat, bat, mat, vat, and so on. But children do: Two psycholinguists tested children from 5 to 11 years old in a picture-naming experiment and compared their responses to those of adults. Adults reacted to pairs of rhyme-related words like "cat, mat" the same as they did to unrelated words like "cat, spatula," but children reacted to "cat, mat" the way they reacted to "cat, cow"—as a linked pair.
So, by 5 years old, children have linked words based on rhymes, and then they lose these links as they grow into adults. But how early do children start linking words? And is there a pattern of which words they link first? We're still working on figuring out all the details, but here's what we know so far.
For one thing, there's some preliminary evidence that children connect sounds they learn to say earlier, like B and M (think "baba" and "mama"), before the ones they learn later, like S and R. I worked with a group of researchers conducting a paired-words experiment, but with preschoolers rather than elementary-school kids. We showed the kids pairs of words with early-learned sounds versus later-learned sounds, and found that children responded the same to the different-sounding pairs ("dog, cake") and late, similar-sounding pairs ("sock, sun"), but slower to early, similar-sounding pairs ("car, cake"). It's a bit weird—we'd expect them to actually be faster at the similar pairs, based on the results from older children—but it does show that even toddlers notice when words begin with the same sound.
In fact, we can go even younger—it turns out that babies are making connections between words before they're even saying them. Some infant-research psycholinguists did an experiment with 18-month-olds using pictures: First, they’d show a picture of a dog. Then, they’d show two pictures side by side, a door, and a boat. The baby would hear one of the words, "door" or "boat," and the researchers measured how well the baby recognized that picture. And the babies, just like the adults and older children, recognized pictures better when they’d already seen another picture with the same first sound. This is pretty amazing: after only 18 months of being alive, the baby has figured out that connecting words with the same first sound is a good strategy for using language—a strategy they’ll still be using 20 years later as an adult.
Small children and babies are doing their best with the gargantuan task of learning a language, at an age when they’re still figuring out whether they can eat their toes. Managing to find one of the most widely used language shortcuts in less than two years is an impressive accomplishment.
So next time you get tripped up by a trick question, take the opportunity to educate your interlocutor about why giving the wrong answer is a sign of a neurotypical, smoothly functioning brain. And then ask them what cows drink.