Federmeier Examines Predictability in Language Processing

Kara Federmeier, professor of psychology and full-time faculty member in Beckman’s Cognitive Neuroscience Group, studies how predictability figures into language processing in the human brain.

Kara Federmeier says she’s “utterly predictable.”

“I had wanted to be a scientist from a fairly young age,” said Federmeier. “My dad was a chemistry teacher, and I liked science.”

So perhaps it’s not surprising that Federmeier, professor of psychology and full-time faculty member in Beckman’s Cognitive Neuroscience Group, studies how predictability figures into language processing in the human brain.

Federmeier began her career by showing that the brain gets ready for—predicts—future words when people are reading or listening to sentences. Her work now is focused on understanding how this prediction takes place and how it matters for what people understand and remember about what they hear or read. 

One of her current studies is examining how substituting a word for an expected word affects someone’s memory.

“We use sentences like ‘When the two of them met, one of them held out his … .’ Most people think you’ll say ‘hand,’ but then you give them ‘badge.’ ”

The word “badge,” said Federmeier, is still understandable in the context of the sentence. “You can still figure out what it means, but it creates a slightly different scenario. We’ve shown that there is a brain response that occurs in situations like this, where people predict one thing and get something else that is surprising but makes sense.” 

“Prediction is great if you get what you’re predicting. If you predict ‘hand,’ and you get ‘hand,’ your processing for ‘hand’ is really easy. You were already ready to perceive it, you were all ready for the meaning of it—but the trade-off comes when you don’t get ‘hand’ and now you have to go back and kind of undo that work.”  -Kara Federmeier

The question is, what does that brain response do?

“We don’t know if that response is the brain trying to ‘turn off’ its expectation for ‘hand’ or trying to make sense of the unexpected word, or both. We can test that by seeing what people remember about these words later—are you better or worse at remembering that you saw ‘badge’ when the brain’s response is greater? And when do you misremember seeing the word ‘hand’ instead?”

These kinds of experiments, Federmeier explained, not only help answer questions about how the brain comprehends language, but have other applications as well.

“When designing a textbook, do you want to surprise people, to make things more memorable? Or will that hurt students’ ability to remember what you want them to, because they were predicting something else?”
Federmeier’s lab, the Cognition and Brain Lab, incorporates electrophysiology, measuring the flow of ions in brain tissue by placing electrodes on a subject’s head, into the studies. This allows them to very precisely follow brain processing over time.

“We can follow the brain as it goes through the whole process of understanding a word … first figuring out what letters are there and in what order and then determining what the word means, while also realizing that the meaning of the word isn’t what had been predicted and deciding what to do about that,” said Federmeier. “For language, time is everything. Words are coming in very quickly and the processes used to understand them have to be even faster. So we need methods that have really good timing.”

Interestingly, Federmeier and colleagues have found that older adults are less likely to make predictions when they are processing language.

“We find that older adults as a group are not doing as much prediction,” said Federmeier. “They’re processing a little more passively. So they’re constructing the meaning of the sentence, but they’re not forming these really strong expectations about what’s coming up next. This makes comprehension less efficient, maybe, but it could also help them if they have to deal with something unexpected.”

Federmeier’s lab is trying to figure out why this happens.

“There is a subset of older adults that look just like young adults, and they tend to be adults that score high on tests that are known as verbal fluency,” said Federmeier.

In a those tests, someone is given a category and has to list as many items in that category as they can in a short amount of time.

“We think that fluency gives us a measure of the strength of the connections between frontal lobe areas that do language production and the temporal lobe areas that are processing the meaning of words and comprehending,” Federmeier explained. “So it seems like these adults have really healthy, active connections, and they tend to predict.”

However, most older adults aren’t making predictions because for some reason it’s harder for them to do so. Likewise, young adults can be made to temporarily stop predicting if their predictions turn out not to be useful.

“Prediction is strategic,” said Federmeier. “We can change the kind of language people are getting, so that the brain seems to recognize that its predictions aren’t working, and it will stop making predictions in those cases.”

Federmeier said it makes sense for this to happen because there are pros and cons to prediction.

“This is one of those interesting trade-offs,” said Federmeier. “Prediction is great if you get what you’re predicting. If you predict ‘hand,’ and you get ‘hand,’ your processing for ‘hand’ is really easy. You were already ready to perceive it, you were all ready for the meaning of it—but the trade-off comes when you don’t get ‘hand’ and now you have to go back and kind of undo that work.”

Federmeier explained that the brain deals with the pros and cons of prediction by using the two hemispheres of the brain differently.

“It’s useful to look at the hemispheres because you essentially have two different processing mechanisms taking place within one single person,” said Federmeier. “That’s easier than having to compare across groups of people who might be doing different things, with all the concomitant differences between people that are hard to control for.

“It’s the left hemisphere that seems to be very active in its processing of language, doing prediction, particularly in young adults. It’s the hemisphere that controls speech, controls language production, so it makes sense that it is good at taking the meaning from the sentence and using that to figure out what words should be coming up next and getting ready for those words. But the right hemisphere is also comprehending language, just more passively—kind of like the older adults.”

Federmeier’s research combines her love of language (her mother is a retired English teacher) and her love of science.

“There’s just something particularly fascinating about language, which is really the pinnacle of human cognitive abilities,” she said. “It’s the thing that other species don’t seem to be able to do, at least to the same degree, and it brings together all the other cognitive abilities—perception, attention, memory, and concepts.”