Scientists Decipher How our Brains Decide Which Words to Use


How the brain decides which is the one word you’re looking for between the smorgasbords of related concepts stored is a complicated cognitive task that is not really understood. Seeing an epilepsy patient with a grid of electrodes directly atop their brain had researchers looking into this question. They found that wide, overlapping bands of the brain work in parallel to retrieve the precise word from memory.

When you look at a picture of a coffee cup, the neurons that store your memory of what a coffee cup is begin firing. It is however not an isolated process as many other neurons that code for related items and ideas (tea, bowl, mug, fork, plate, supper) are all activated at the same time. How your brain decides between this myriad of related concepts to find the one word you’re looking for is a complicated and poorly understood cognitive task. Neuroscientist Stephanie Ries, of the School of Speech, Language, and Hearing Sciences at the San Diego State University led a new study that investigated this question by measuring the brain’s cortical activity. The team found that wide, overlapping bands of the brain work in parallel to find that one correct word in your memory.

Most adults can effortlessly and quickly recall as many as 100,000 words they regularly use when prompted, but scientists have long been baffled by how the brain accomplishes this. Previous work showed that the brain organizes words and ideas into semantically related clusters. When we try to recall a specific word, the brain activates this cluster, significantly reducing the size of the haystack in which it has to find the needle.

To determine what happens next in the process, Ries and her colleagues sought help from a population of people in a unique position to lend their brainpower to the problem: patients that have to undergo brain surgery to reduce their epileptic seizures. Neurosurgeons monitor their brain activity before surgery to figure out which region of the brain triggers the patients’ seizures. This is done by the patients wearing a grid of dozens of electrodes placed directly on top of their cortex, the outermost folded area of the brain.

While the patients were waiting for a seizure to occur in a hospital and were therefore hooked up to this grid, Ries asked if they would be willing to take part in her research. Recording brain signals directly from the cortical surface gives neuroscientists like Ries an unparalleled opportunity to look at exactly where and when neurons are communicating with each other during a task.

Ries noted that during the period the patients were waiting anyway, they had time to do cognitive research that would be impossible to do otherwise. She described this as an extraordinary window of opportunity.

Nine patients agreed to participate in the recent study. Ries and her team would show the patients items on a computer screen in 15 minute sessions. These included ordinary things like vehicles, musical instruments and houses. The participants were then asked to name these as quickly as possible. Their brain activity was tracked throughout the process.

The team measured the separate neuronal processes involved when the item’s conceptual cluster was first activated, and then selecting the proper word. They were surprised to discover that the two processes actually happen simultaneously and that a much wider network of brain regions than what they were expecting was activated. They did expect the two regions known to be involved in language processing to light up, i.e. the posterior temporal cortex and the left inferior frontal gyrus. Several other regions not traditionally linked to language were however also activated, including the middle and medial frontal gyri.

Ries noted that the research shows that the word retrieval process in the brain is not as localized as what was previously believed. There is no clear division of labor between brain regions, but the process is much more complex.

Finding out how exactly the brain accomplishes this task could one day help speech language pathologists develop a strategy for treating disorders that prevent people from easily accessing their vocabulary.

Rise concluded that although word retrieval is usually effortless, it is routinely compromised in patients who suffer from anomia, or word retrieval difficulty. This is the most common complaint in patients with aphasia induced by a stroke, but is also common in both normal aging and neurodegenerative diseases. It is therefore critical to understand how this process works to understand how to treat it.

The research was published in the Proceedings of the National Academy of Sciences.