:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/52090347/16701522743_f007498f67_o.0.png)
Neuroscientist Nathan Rose and his colleagues at the University of Wisconsin recently lost track of a memory.
I mean that literally. One minute, they saw a trace of a memory “light up” in an fMRI scan. In the next minute it was gone. And in this one simple observation, Rose and his colleagues are challenging a long-held scientific belief about how the brain works. And can open to door to new understanding of why we remember, and why we forget.
The curious case of the “missing” memory
Rose lost the memory during a fairly simple experiment. Participants were brought into the lab and given an image of a face and a separate name to memorize. They were told they’d later be tested on their ability to recognize that name and that face.
All the while, a fMRI scanner was peering into their brains. An artificial intelligence program then used the fMRI images to distinguish between when the participants were thinking about the face, and when they were thinking about the name. When participants were asked to recall either the face or the name, the computer program could “see” them thinking about each.
But then the participants underwent a different test. And this is where things get weird.
The participants were told that they were only going to be tested on their memory of the face and didn’t need to remember the name. When that happened, it was like a switch turned off in the participants’ brains. The fMRI could no longer “see” the memory of the name. It was “as if the item has been forgotten,” Rose says.
Except here’s the really surprising part: The name wasn’t actually forgotten. The participants could still remember it later on when prompted again. The memory was still there — there was just no observable trace of it in the brain.
This may sound like a minor observation, but it could fundamentally change the way neuroscientists think about how the brain works. The study, published today in Science, shows there’s a whole “dark” brain activity that neuroscientists can’t see with current neuroimaging technology. And if scientists can better understand this dark network of memories, perhaps they can find ways to make us retain more information, or retrieve lost memories.
The conventional wisdom on working memory may be wrong
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/6840407/shutterstock_444143494.jpg)
Working memory is the function of our brain that holds information that we need immediately — if someone tells you the names of the hosts before you walk into a party, that’s where you keep them. Think of it like the desktop on your computer. It’s where you bring long-term memories out of deep storage to work on them. It’s also where new information is first stored in your brain before it can be filed away. Working memory is the space where we work on our thoughts and put them to use.
For a long time, neuroscientists thought they knew how working memory works. Based on neuroimaging studies, it seemed that working memory was always associated with a flurry of neural activity.
“The inference, then, was that a short-term memory existed only if it produced some sustained brain activity that could be identified with our various forms of neuroscience measurement methods — including functional neuroimaging such as fMRI,” Jarrod Lewis-Peacock, a neuroscientist at the University of Texas Austin, explains. If there is no trace of a thought in a brain scan, it implies the thought is no longer there. Or that fMRIs — which are key tool of neuroscience research — don’t tell us all that much about the working of the brain.
What Rose’s new study suggests is that the memory is still there in the absence of any observable neural activity. “[It’s] a big deal because people assumed elevated activity is how [working] memories are stored,” Bradley Postle, a study co-author, says. (And it’s not the case that the memories were simply transferred to long-term storage.)
“This line of research is a big deal,” Nelson Cowan, a psychologist who studies memory at the University of Missouri Columbia who was not involved in this study, tells me. “This kind of work could be helpful in diagnosing what’s going on when a person cannot remember something you just presented to them.” Because there’s a huge difference in treatment approaches for a patient who does not have the memory stored somewhere in her brain, and a patient who does have that memory stored somewhere, perhaps silently, but just cannot call it to attention.
It’s also a big deal because of what Rose and his colleagues did next. They brought that silent memory back to life with a zap of magnetic radiation.
The memories can be temporarily restored
Transcranial magnetic stimulation (TMS) is a neuroscience tool that sends pulses of magnetic radiation to temporarily “turn on” or “turn off” a person’s neural circuits. “For example, if find your thumb area in your left motor cortex, and if I [turn on TMS] I can cause your thumb to twitch,” Rose says.
Guided by machine learning, Rose’s team was able to train the TMS to locate areas of the brain where participants were storing the memory of the face or the name. Simply put, Rose and his colleagues devised a way to turn back on the memory that had grown silent.
When they turned on the TMS on the silent memory, an EEG brain scan showed activation in the area of the brain associated with that memory.
So they ran the memory trial again, zapping people with TMS when a memory became silent. When the TMS was on, it seemed like the participants became confused about which item — the face or the name — they were supposed to remember, and they performed worse on the test. Some of Rose’s colleagues jokingly called it the “Frankenstein effect,” suggesting they brought this memory back to life.
What’s more, Rose and his colleagues found evidence that these “silent” memories also get silently “deleted” when person is told they no longer need the information.
So, to use the computer analogy again, it’s like these working memories, as with a computer, can either be brought out of recycling or they can be deleted when no longer needed. And this whole process happens beyond the detections of typical neuroscience research tools. It’s a part of the brain’s “operating system” that we’ve never seen before.
As for the ability of TMS to reactivate forgotten memories in patients, Rose says that’s still a far-fetched idea. “It’s going to be hard to avoid having the [study] spinned as ‘TMS can bring your memories back,’ which is going far beyond what we’re showing,” he says. It’s better to think of the TMS more as sonar, pinging the brain to see if the memory is even in there.
Where do the silent memories go?
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/6544785/shutterstock_131390873.jpg)
One theory is that the brain is constantly and silently reorganizing its neural networks to hold the impression of these memories, but keeps them inactive. That’s roughly how neuroscience thinks long-term memory storage works. But these results are making Rose wonder if that’s how many systems of the brain work: that they are constantly moving thoughts from “active” states to “silent” states, possibly by reorganizing the connections in the brain.
Christos Constantinidis, a neuroscience researcher at Wake Forrest University, says Rose’s results don’t negate the importance of neural firing in working memory, but that they suggest the neural firing and this “silent” process work together to make the brain more efficient. “There are some instances where [sustained activity] may not be enough or may not be efficient,” he says. “This silent mechanism bridges the gap.”
There’s also an alternative explanation for what happens: The brain still is firing neurons to sustain the memory, but they are too few in number for brain scans to detect. (There can be hundreds of thousands of neurons in a single voxel, the smallest unit of brain activity a fMRI can detect).
Overall, this new study is a reminder neuroscience is just starting to understand the brain, and it’s a recognition that the typical tools of neuroscience — fMRI, EEG, all of the tools that depend on electrical signals or blood flow — just can’t see many of the brain’s processes. And that means we’re just scratching the surface in terms of our observations about the brain. It’s still mostly a black box.
“Just showing one part of the brain ‘light up’ and inferring that’s where the item [memory] is — that’s a rudimentary idea,” Rose says.
“The brain is the most complicated object that we know of in the universe,” Postle says. We’re going to have to be cleverer in our methods if we ever want to fully understand it.
