>> Multiple Systems of Memory

	Interview with John Gabrieli, Stanford University From Studying The Mind, VHS © 2003, W. W. Norton

Why do researchers believe that there are multiple systems of memory?

The patient HM has had a seminal role in the history of our thinking about the brain basis of human memory. Up until his injury it was thought that memory was distributed throughout the brain in a rather holistic fashion, and that if you injured a certain part of the brain you would have a corresponding injury. So if you had 5 percent of your brain injured, you'd have 5 percent worse memory, 10 percent injured, 10 percent worse memory. And it wasn't clear that any single part of the brain had a much more important role in memory then any other part. In the case of HM, there was a very restricted surgical resection of his hippicampus and nearby structures so a very small percentage of his brain was taken out, on both sides, perhaps the size of half a thumb. Nevertheless this very small and specific resection of tissue had a very large effect, which is that for all practical purposes, from that day to the present, he's never been able to remember a new event in his life or learn a new fact from the world around him. So he dramatically changed our view that different parts of the brain all contribute to memory equally and showed that there are very specific contributions of different kinds from different parts of the brain.

We currently think about many memory systems in the human brain. And what led us to this view of memory is the fact that we discovered that different injuries in different parts of the brain will prevent people from having one kind of memory working in a healthy way, but allow other kinds of memory to work perfectly normally. And so by seeing many different kinds of patients with lesions in many different parts of the brain, we discovered that many different parts of the brain have served their own kind of memory individually and thus constitute a memory system.

What are the different memory systems?

There seems to be a big distinction between learning skills and learning facts. Skill learning, which is impaired in patients with basal ganglia diseases or certain cortical lesions, seems to involve changes in the same system that does the task. So if you learn to throw something effectively with your arm, the part of your brain that controls your arm seems to show the learning itself. Other kinds of memory, like memory for an event or a situation, seem to involve many different parts of the brain at once, pulled together by the medial temporal lobe structures. And so procedural learning seems to involve plasticity or change in the system that's performing the action, be it a physical action, a perceptual action, or an intellectual action. Whereas declarative memory seems to be pulling together all kinds of different aspects of experience associated with a given event or situation.

Different kinds of skills involve different parts of the brain. We've learned, for example, that the cerebellum is terribly important for classical conditioning, whereas the basal ganglia is terribly important for certain kinds of motor skills. But the thing that's common among all these different kinds of skills is that we can see, through brain imaging, different parts of the brain initially engaged; and as the person gains the skill, those parts of the brain may become less involved and other parts become very active and appear to subserve the behavior. So we see almost a ballet of different parts of the brain coming in initially and sort of more skilled dancers emerging later on. And that shift in activation from one part of the brain to the other seems to be a sort of a hallmark of how it is that people get better at doing things that require skill.

There are different kinds of memory associated with different parts of the brain. Declarative or explicit memory is what we typically think of as memory, regular memory for everyday events and facts. We know that it depends upon, most of all, the medial temporal lobe structures and the hippocampus especially. Working or short-term memory involves keeping things in mind that we're thinking about at the moment and the information we need to have on hand in our mind in order to solve something that we're working on, like a specific goal. We know parts of the brain in the frontal cortex, in the parietal cortex, seem to be terribly important for allowing us to keep our goals in mind and the information we need to achieve those goals. Skill learning seems to depend primarily on the basal ganglia and the cerebellum.

How does brain injury and disease affect memory?

In terms of the consequence of different kinds of brain injuries on the different kinds of memory, if damage occurs to a medial temporal lobe structure such as the hippocampus, then people have tremendous difficulty in learning events and facts, remembering what they did this morning or perhaps remembering a new fact they hear in the news. This kind of damage can occur selectively for patients with injuries in the medial temporal lobes. Patients with Alzheimer's disease have damage there and many other places as well. Patients with Alzheimer's disease also have damage in the neocortex and so they have other kinds of problems in addition to memory problems, such as problems in language or thinking. Patients with Huntington's disease or Parkinson's disease have injuries in the basal ganglia. We've learned from these patients that this part of the brain is terribly important for habit or procedural learning–motor skills, perceptual skills, or certain kinds of intellectual or cognitive skills.

How do you use neuroimaging to study memory function in healthy individuals?

We use functional magnetic resonance imaging to try to understand aspects of learning in the intact human brain, and relate this to what we see as well in patients with injuries. What imaging allows us to do that we cannot see with patients is to examine, for example, temporal qualities of learning. What happens when you begin to learn something and how does that shift when you get very good at it? What happens at the moment when you learn a fact versus the moment when you retrieve memory for that fact later on? And so with imaging we can track, over time, things that are impossible to see in a patient who can no longer perform those tasks. A second issue is that with patient studies we're necessarily restricted to where diseases tend to affect the brain and how they affect the brain. We've learned tremendous information from this about normal memory organization but inevitably there are parts of the brain that are less likely to be injured, or parts of the brain that get injured together, and so by doing imaging we can get much more spatial resolution in an attempt to understand better which parts of the brain mediate or accomplish which parts of the mind.

Functional magnetic resonance imaging has shown us a lot about the amygdala, a structure in the limbic area that's related to emotion and memory. And what it's shown us are two major things. First, that to the extent somebody says that an experience that they have is very emotionally intense or arousing, we see a lot of activity in the amygdala. And second, that activity predicts how well they'll remember the experience. It's as if the amygdala were getting a signal about how you felt about something that feels important, and then using that signal to boost memory for the event. And it makes a lot of sense that we would have a part of our brain help us respond to what seems terribly important to us and make sure that we have a very good chance that we'll remember it later on.