New research from a group at Hopkins has pinpointed the area of the brain that allows us to detect subtle differences in our environment and store that information in our memory.
The study, authored by Craig Stark of the Department of Psychological and Brain Sciences and published last week in Science, helps elucidate the neural processing that goes into declarative memory, the aspect of human memory that allows us to remember everything from the capital of Bolivia to our first day of school.
Current theories of how declarative memory works suggest that two opposing but complementary processes take place in the brain. One involves the recognition or "completion" of patterns: You can equate two similar environmental cues, even though they may not be entirely identical.
Imagine looking at your mother's face from different angles or in different lighting. You're still able to identify that particular combination of eyes, nose and mouth as "mom," despite variation in shadow and perspective.
The second process is called pattern separation. This involves dissociating two similar (but not identical) events or images.
"We need to be able to separate events in our mind and not do completion. Where did I park my car today? I need to be able to have today isolated from yesterday, the day before, last week, and so on, so that I can go to the right spot," Stark said.
In that sense, pattern separation allows us to notice small changes in our environment, a process that's arguably been fundamental to our success as a species.
"Another example is that we need to learn the names of people we meet," Stark added. "Sometimes the person looks very unique and this isn't a challenge. But other times they're not. Two brothers, for example: Internally, we need to separate our representations of the two people so that we can learn things about one without having it spill over onto another."
A host of recent research has shown fairly conclusively that a brain area called the hippocampus is where all this completion and separation takes place. Researchers have known for decades that the hippocampus is critical to creating new memories, but a step-by-step timeline of memory formation has remained elusive.
Figuring out where in the hippocampus a computational process like pattern separation takes place will further our understanding of how memory works in general.
And that's exactly what Stark and his colleagues set out to do. As with most neuropsychological studies, their experiment's design was the secret behind its novel results.
Though the scientists undeniably wanted to study the neural basis of memory, they were obliged, in their choice of stimuli, to take memory out of the equation.
This is because memory capacities vary from person to person, so, in order to balance one subject's particularly good memory or another's particularly lousy one, the study's stimuli had to avoid being overtly memory-based.
In each round of testing, subjects were presented with one of three pictures, which Stark and his colleagues termed "novel," "repeated" or "lure." Novel indicated the first time the subject saw a given picture - for example, a light switch - while repeated corresponded to the presentation of an identical picture.
However, lure, as its name suggests, was a little trickier. It involved presenting subjects with a slight variation on the previous picture. In the case of the light switch, for instance, the switch was up in the novel and repeated pictures but down in the lure picture.
In all three situations, the researchers measured their subjects' brain activity using functional magnetic resonance imaging (fMRI), a technique that uses changes in how blood is supplied to the brain to pinpoint areas of increased neural activity.
Since the brain areas Stark and his colleagues were interested in were quite small, they took very detailed snapshots of activity, each of which consisted of cubes only 1.5 millimeters on each side.
If pattern completion was taking place in a certain area, the researchers hypothesized that there should be no difference in brain activity elicited by the novel picture and the lure.
In other words, if the hippocampus is completing a pattern, it should dismiss any small differences and "see" an identical picture.
On the other hand, pattern separation in any given area should cause different levels of activity between the novel picture and the lure. This would indicate the small differences between the two had been noticed.
After tabulating their data, the researchers saw different activity levels in only one region of the hippocampus, an area that included two adjacent structures called CA3 and the dentate gyrus.
This wasn't exactly a surprise. Past studies have hinted that the dentate gyrus creates what neuroscientists call a "sparse code," in which a single environmental cue elicits a reaction in a very small group of brain cells.
Take this example: You feel the same prick when you poke your finger with either a No. 2 or a No. 3 pencil. A "sparse code" in this case would be one in which a single group of touch receptors on your skin would react to a single pencil type (one group for No. 2 and one for No. 3).
A perfectly sparse code is half-jokingly referred to as a grandmother cell, because, in theory, it only responds to one particular stimulus (your grandmother, for example).
It's now thought that such a one-to-one association doesn't exist in the brain, but response to a stimulus can be limited to relatively few cells. In the hippocampus, then, one can imagine that a specific population of neurons reacts to the picture of the "on" light switch and another reacts to one that is "off."
This, in effect, is how our brains parse the distinct identities of two very similar stimuli, and more broadly, how you always know your mother when you see her.
Stark says future research will focus on how pattern separation abilities degrade with age. "We think that normal healthy aging leads to an impairment in pattern separation," he said.
"What do people have trouble with as they age? Things that require pattern separation, that is, remembering the details of an episode versus the general idea."
Stark also hypothesizes that disorders like Alzheimer's disease tax the pattern separation system more than normal aging: "We're working to detail the behavioral issues you have with separation, detail the functional MRI changes, link them and try to see how this is affected by or could predict Alzheimer's."
