You're at the grocery store, and around the corner walks a familiar face. She looks just like someone you've met, and you try desperately to place her. Did you meet her at that cocktail party two years ago, or is she a long-lost college classmate? You walk up to her and wave, but in return you're greeted with a blank stare.
James Knierim, Ph.D.
Professor of Neuroscience, The Zanvyl Krieger Mind/Brain Institute, Johns Hopkins University
A professor of Neuroscience who focuses on the hippocampus and memory, Dr. Knierim has studied the effects of zero-gravity environments on spatial perception and how landmarks are used to create "cognitive maps" in the brain. He is currently researching the processing of memory in the hippocampus.
How do our brains make decisions about who we know and how we know them? It's more complicated than you might think. New Research published today by Dr. James Knierim, a professor of neuroscience at Johns Hopkins' Zanvyl Krieger Mind/Brain Institute, provides some insight. His team used confused rats to dig deep into the mysteries of the hippocampus, the part of our brains that stores crucial memories of our lives. Their findings suggest that the way we remember our experiences is surprisingly similar to how rats remember locations in a maze.
We spoke with Knierim about his research, cognitive maps, Jennifer Aniston and how our brains are like a democracy.
HOPES & FEARS: How do we tell the difference between different faces? What happens when this process fails?
JAMES KNIERIM: [Telling the difference between similar things] is one of the important jobs that our memory system has to do. You see somebody in a store and their face looks kind of familiar. [You think], "that kind of looks like somebody I met five years ago". You have to make the decision, is it really that person? You might first think "boy, that person looks familiar". Then as this input from this person's face reactivates these memory networks in the hippocampus, it starts to fill in the details.
This is the process that theoreticians call pattern completion. You get a pattern of input from the visual stimulus of this face, and then the hippocampus completes the pattern as more and more of the details get filled. "Oh yeah, I think that was the person I met five years ago at a cocktail party, and we had a conversation about this." You'd say, "I know that person", and you go up to them, and you say "hey, how are you doing?", and they look at you like you're crazy. It's the wrong person. It's not the person that you met, it's just somebody who just kind of looks like them.
Now, what happens? Now your brain says "all right, I don't want to make that mistake again". Now you're going to encode this new face as a completely different memory. You want to use completely different brain cells now to encode this experience because as embarrassing as it is now that you said "hi" to that person, what if tomorrow you see that same person again and you do the same thing? You'd be even more embarrassed.
This is what the theoreticians call pattern separation. You want to create a new, completely separate brain activity pattern that allows you to distinguish those two memories [of the different people].
This is what your brain is doing all the time when you're seeing things. Is this the same thing that you saw before? Is this just a bit of a modified version of what you saw before? Or is this really something different?
That's the kind of questions we were addressing in our study.
H&F: Tell me about the study.
JK: We study the brain activity of cells in the rat hippocampus. When you record the activity of hippocampal cells, each cell fires when the rat is occupying a specific location in the environment. One cell will fire when the rat's in the corner, another cell will fire when the rat's in the middle of the room, and other rat cell fires when they're at the edge.
The whole environment is mapped out by the firing of these cells. It's thought to be a cognitive map or a mental map of the environment. The people who discovered this "rat GPS" system won the Nobel Prize last year in physiology medicine.
We recorded these brain cells, and we trained rats to run around in circles on a track that had local cues on, little different textures and patterns and so forth. After the rat was very familiar with this terrain, we did something that was similar to if one day you opened the door to your apartment or house, and all of the furniture had been rearranged.
That's what we did to the rat. We changed this circular track that it ran on, and then we rotated the cues on the wall and put the rat back in the environment. We looked at their hippocampal cells to see what they did. Is it going to retrieve the right memory? Is the rat going to recognize that this is the same environment as before, but it's been rearranged, or is it going to say wait a minute, this is so different, this must be a different environment, and I'm going to create a completely new memory and new representation, a new map of this environment.
H&F: What did you think you'd find?
The theoretical work, years ago, said that there's a part of the brain, part of the hippocampus that's called the dentate gyrus, whose job is to always create a new representation for a memory whenever there's a change in the environment.
Then, according to the theory, it sent this new representation to the next part of the hippocampus called the CA3 region. According to the theory, the job of the CA3 region was to say "okay, you're telling me this is a new environment, but I'm going to also weigh the evidence that it's the same environment, just somewhat altered. I'm going to do this pattern completion thing. I'm going to try to retrieve the memory of who this person was." The job of CA3 is to be the final judge, to weigh the evidence.
The theoretical work says that the CA3 region weighs the evidence and the whole region comes to a decision and then sends the decision out to the rest of the brain. Work from our lab and some other labs over the past few years has actually supported that model.
H&F: What did your new findings say?
JK: We've discovered that the different parts of CA3 seem to come to different decisions. One part of it seems to decide "yeah, this is the same, this familiar face is the same person you met before".
Whereas the other part of CA3 makes the opposite decision. It says, "no, this face looks familiar, but it's really somebody different". Yeah, it looks familiar, there are some certain familiar aspects, but it's so different that you should treat it as a totally new face and create a new map of it.
The interesting thing is the two different parts of CA3 actually connect themselves to different parts of the next processing stage. It kind of hedges its bets. One part of the CA3 says, "that's the same person" while another part says "well no, it's really a different person". Then it sends all of that information to the next processing stage which presumably has to make further comparisons and computations before you either decide it's the same person or it's not.
H&F: That's really interesting. It's almost like your memory is a democracy where different parts are voting on what the right answer is.
JK: Exactly. The thought before was that it was like our democracy, with an electoral college. In each state people vote, but then it's all or none. One person, whoever gets the greatest number of votes, wins.
Now we think it's not quite that simple. Maybe some of the votes go one way, some go the other, and there's still some ambiguity in the answer. At least at this stage of processing.
H&F: It's more like a parliamentary system then.
JK: [Laughing] Yeah, correct, something like that.
H&F: I've heard of what neuroscientists call the Jennifer Aniston neuron [in studies, one neuron lit up only when stimuli involved the actress]. Is there one neuron that's remembering someone's nose and one that's remembering someone's mouth? How detailed does the encoding get?
JK: Good question, and we really don't know the answer to that very well. Certainly the Jennifer Aniston story was very interesting. It did show that some of the cells in the human brain are extremely specific for one kind of memory. In that case, it wasn't just a face, it was more of a concept.
That study actually does go along very well with a lot of the theoretical models about memory. If you want to store lots of memories in the brain, you have to have very high specificity. Otherwise, you would start having lots of interference with other memories. What if you saw Jennifer Aniston's face and your memory thought it was someone else, maybe one of the other actresses on that show?
H&F: On 'Friends'.
JK: Yeah, on Friends, right. A cell that encodes many memories might retrieve the wrong memory. That's what the pattern separation part of the dentate gyrus is supposed to be doing.
As you start getting into the individual features, "is that a nose, or is that a mouth?" it appears that at that level the individual cells are not as selective. As the brain is encoding and recognizing "that's the nose, that's the mouth" and all, it goes further on into the memory regions, where we're talking not so much about perception. Storing this as a memory, that's when the representation might get more specific.
[The memory is] what then allows you to not just say "this is a mouth or a nose," but "this is a specific combination of a mouth and a nose and a forehead and a hairstyle" that allows you to recognize a specific person, and then also remember a specific experience you had with that person, or a specific experience watching that television show.
H&F: So it's all about the recognizing the connections? Why is that?
JK: The memory is encoded over a large number of cells and connections. So it's not like in a computer where you've got a memory stored on a certain disk. If that space gets damaged, the memory is gone.
The way the brain stores memory is this interconnected network of neurons. Any one neuron or any one connection isn't critical. You can have damage to that and it can survive, the memory can survive the loss of any one of these neurons and still be able to retrieve the memory.
H&F: It's like a distributed network.
H&F: Why is it so important for us to be able to recognize so many different faces? Why did our brains develop this skill?
JK: In general, it's just a survival mechanism. For us, faces are very important, recognizing social cues, remembering who you interacted with before.
For any animal, including the rats I was talking about, it's important to recognize where you are in an environment. "This is where my food stores are located, this is where my escape holes are." It's critical to have these memories and represent them so you don't get lost.
H&F: I'm just wondering, say you're a rat, and you're walking around the maze and encoding the different parts in different neurons, those neurons are not actually in the shape of a maze, correct?
JK: No, no, no, they're not. They're distributed, as far as we can tell, more or less randomly in the hippocampus. It'd be a lot nicer for us if they were distributed that way, we could see them much better.
H&F: What are the next steps after completing this research?
JK: We're taking this in a couple of directions. As I mentioned earlier, different parts of the CA3 region seem to make these different decisions, and they send projections to different parts of the next processing stream, which area is called the CA1 region. We're now analyzing data from the experiment to understand what the next processing stage does with this information it gets. The ultimate goal is to try to come to a brain systems level understanding of what goes on in this type of memory when the brain has to make a decision.
We're also just starting a new project, which is in collaboration with a neuroscientist here at Johns Hopkins who's an expert in memory and how it's affected with aging. We are going to be doing the same experiment in rats that are aged. Some rats get very poor memory as they get old, and other rates seem to be just fine. We're hoping to use this experiment to try to understand what it is about the rats that had memory impairments as they get old. This might be a way of taking the basic science work that we're doing now and applying it to ways that might help people with cognitive dysfunctions that can occur as one gets older.