Scientists have long believed that the cortex, the outer layer of the brain, was responsible for figuring out the meaning in a sentence. But a new study out of UC Berkeley shows that the hippocampus, a brain structure long believed to act as a center for linking memories together, plays an active role in extracting meaning from language.
“This gives us a new insight into how memory works in humans, and how memory interacts with the rest of the brain to produce behavior,” says study co-author and Berkeley psychology professor Robert Knight.
The data suggests that the hippocampus may help “to build the links between different knowledge and various information that we have about the world, about the different words and their meanings,” says Vitória Piai lead author and former Berkeley postdoctoral fellow. When we’re listening to or crafting a sentence, “We’re pulling from this source of accumulated knowledge, things that we’ve learned through life.”
In the experiment, 12 subjects participated in a fill-in-the-blank activity, where they either listened to sentences that were “constrained,” meaning they had more limited or obvious answers (“She locked the door with the____”), or they listened to sentences that were “unconstrained” or more ambiguous (“She walked in here with the____”). After hearing each sentence, they were shown an image they had to name as the final, fill-in-the-blank word (“key”).
The researchers tracked hippocampal theta waves, which are brain currents thought to make memories by decoding and coding information, and found they were more pronounced when participants heard constrained sentences than unconstrained. In other words, when participants were struggling to pinpoint an answer, theta waves weren’t as powerful.
“As the sentence unfolds, by the second or third word, as you’re beginning to know there’s a constraint on it, the hippocampus gets active and is engaged and helps you to make a quicker decision,” Knight says.
So why has the hippocampus not been examined when it plays such a strong role in memory? Piai says it’s not necessarily that scientists haven’t considered it—but that they’ve just neglected to research it.
This experiment is notable because it could change how scientists perceive the hippocampus’s role in language and memory—but also because it was conducted using intracranial recordings, rather than the more commonly used fMRI.
“I think there’s both a methodological and a historical reason for that,” Piai says. “Historically, we’ve known a lot about the areas in the brain that are important for language due to stroke. And stroke usually affects the lateral cortex—the cortical areas closer to the scalp. But it’s very rare that a stroke will affect the hippocampus…. So for a long time, we have associated the cortical, more lateral areas, to language.”
This experiment is notable because it could change how scientists perceive the hippocampus’s role in language and memory—but also because it was conducted using intracranial recordings, rather than the more commonly used fMRI (functional magnetic resonance imaging), which measures oxygen flow in the brain.
Getting the recordings would require invasive surgery, so this research was piggy-backed on epilepsy treatments. The subjects were already undergoing surgical procedures at the medical centers at UC Irvine and Stanford for intractable epilepsy, a condition that produces seizures that can’t be reduced or controlled by medication, so the only treatment available to them is to remove a part of the brain, says Piai.
Researchers left the placement of the electrodes and the actual procedure to the neurosurgeons, and then recorded their data—making sure to examine only the areas unaffected by seizures.
“We’re trying to make an inference about normal cognition,” Knight says. “Any time the epileptic activity spreads and generalizes and goes to multiple other brain areas, that slice of time we get rid of. We don’t want any contaminated data.”
Intracranial recording can provide an edge, according to Piai, because it captures neuron firings millisecond by millisecond, as opposed to fMRI, which delivers delayed results.
Despite fMRI’s temporal lag, it’s useful because it’s not invasive, can be used on healthy people, and can record blood flow throughout the entire brain. Intracranial recordings are good because they provide data in real time, but may provide limited coverage because they don’t place electrodes into every area of the brain.
To demonstrate the delay, Knight says: “Are you ready for this? One-one thousand, two-one-thousand, three-one-thousand, four-one-thousand, five-one thousand—that’s fMRI blood flow.”
Of course, Knight notes that both fMRI and intracranial techniques have strengths and weaknesses. Despite fMRI’s temporal lag, it’s useful because it’s not invasive, can be used on healthy people, and can record blood flow throughout the entire brain. Intracranial recordings are good because they provide data in real time, but may provide limited coverage because they don’t place electrodes into every area of the brain.
“[With electrophysiology], you get a direct and meaningful measure of what the neurons are doing,” Piai says. “If you use another technique like fMRI, you’re just indirectly inferring what the neurons did from looking at oxygen in the blood that went to that particular area.”
In terms of furthering the research, Piai is interested in examining the damaged hippocampus’s role in language.
“I think it’s very important when we are trying to understand function to see what goes wrong when that particular area is damaged,” Piai says, and “To look at how it’s dysfunctional when it’s damaged.”
Knight says that this could possibly lead to new studies of how external devices can be used to help the brain perform better—as in the way deep brain stimulation has helped many people with Parkinson’s disease.
“Understanding more about memory and how we might modulate it is incredibly important,” Knight says. “Since we now know that theta activity improves your understanding of the meaning of sentences as you’re communicating—one could imagine a hippocampal stimulator, an insertable stimulator that maybe drives the hippocampus at a certain frequency range that improves its function.”
