Support Provided ByLearn More
Body + Brain

How Memory Works: Expert Q&A

On September 1, 2009, Andre Fenton answered questions about how memory is thought to work, his ongoing research into "erasing" memories in rats, and more.


Recieve emails about upcoming NOVA programs and related content, as well as featured reporting about current events through a science lens.

Q: Can PKMzeta repair short-term memory in dementia patients? [Editor's note: As seen in the NOVA scienceNOW segment, PKMzeta is a molecule involved in long-term memory storage in the brain.] Mary Arronte

Support Provided ByLearn More
Andre Fenton: Mary,
Biochemical processes that transform short-term memory to long-term memory occur several minutes to hours after learning. However, PKMzeta is critical for storing long-term memories that are several hours to months old and probably older, without seeming to have a direct role in short-term memories. So the simple answer is "no."

But some short-term memory problems in dementia may be a secondary consequence of another problem, such as poor attention, or a cognitive dysfunction like impaired planning and executive functioning. It is useful to realize that mental abilities depend on both the information in memory as well as the ability to manipulate or process that information. As an example, we typically only remember, even for a short time, the information we've paid attention to. After storing lots of experience in long-term memory, some people become experts at a particular discipline, like music, a sport, gardening, babies, or neuroscience.

Experts are terrific at processing and recognizing details from their discipline that non-experts often don't even notice, so an expert might also appear to have better short-term memory, but this is a consequence of the expert ability to attend to and process certain types of information. If you notice more information, you have more opportunities to store and recall that information. Since experience and long-term memory help to create an expert brain, PKMzeta may also indirectly contribute to short-term memory through its role in long-term memory. Given there are no PKMzeta drugs, and many problems to overcome to make one, cognitive exercises and therapies may be a better, more direct way to improve failing short-term memory ability.

Q: Can you re-add the "glue" to the neurons to help memory? My son, 26, has lost his short-term memory but can remember the past. I desperately want to help him. Thank you. Tracy, Northville, Michigan

Fenton: Tracy,
People like H.M. and your son teach us that short-term and long-term memories are biologically distinct. I'm sorry about your son and wish we knew how to effectively enhance the synapses between neurons that glue their effective communication patterns so they can be more easily repeated and mediate learning. We know about many molecules that contribute to the gluing process, including PKMzeta, and so manipulating any number of them could be used therapeutically, in principle.

However, I think that to be effective we first have to make advances on the major problem that we don't yet know how the effective synapses are organized in time and space. There are approximately 100 billion neurons in the human brain, each with thousands of synapses, so there are trillions of synapses in a brain. Only a very small fraction of the synapses encode a particular memory, and a slightly different small overlapping fraction may store a second memory, and so on. When we learn more about the biology of how the activities of neurons and synapses that represent a memory are organized so the memory can be selectively activated and manipulated, we will surely get better insight into how to help people like your son.

Q: You showed that there are chemicals that can nullify the effects of PKMzeta, but is there a way for someone to get more PKMzeta working in the brain? My own memory does not seem to be as good as others in my family. My Dad had Alzeheimer's and I wonder if I am going down that road. Thanks. Anonymous

Fenton: Scientists can use a number of powerful genetic manipulations and molecular biological techniques to increase the amount of PKMzeta neurons produce. These manipulations, like the use of ZIP to inhibit PKMzeta, are being used to try to understand the basic biology of memory and information processing, but I am sorry to report that we don't yet know enough of that biology to apply the techniques to attempts at therapies. I am excited to report that we are making impressive progress, but understanding memory and the brain is a very hard problem that will be rapidly advanced if our progress can be measured in decades.

Q: How does PKMzeta actually "glue" interneuronal synapses in long-term memory? How is long-term memory accessed? Thanks. Robert Hampton, Newport Beach, California

Fenton: Robert,
The key to understanding how the "glue" works is the fact that several PKMzeta molecules can aggregate at a synapse. There's strength in numbers, because the aggregation protects PKMzeta from being degraded and attracts new PKMzeta molecules to maintain the aggregate when some molecules do eventually degrade. This aggregate is not the glue, but it is key to the glue persisting.

Now lets consider the glue. There are many other molecules at a synapse, and PKMzeta interacts with some of them. One key type of molecule on the "receiver" post-synaptic side of a synapse is a receptor protein. Receptors bind the neurotransmitter chemicals that the presynaptic "sender" neuron releases into the synapse. This binding causes electrical activity changes in the receiver neuron, as part of the communication of information between the neurons. In the case of PKMzeta-mediated synaptic strengthening, the strength of the communication depends on the number of receptors in the post-synaptic side of the synapse. Remember there's strength in numbers, so the more receptors the stronger the communication after a neurotransmitter is released.

At least one of the things that PKMzeta does is to approximately double the rate that a "trafficking" protein guides the receptors into the synapse. Thus when PKMzeta aggregates in the post-synaptic region of a synapse, the synapse ends up with about twice as many receptors for as long as PKMzeta stays aggregated there. This increase of receptors means that the next time the sender neuron releases a neurotransmitter, the receiver neuron will have a stronger, larger electrochemical response. We call this strengthening long-term potentiation (LTP) of the synaptic response. LTP can be measured as an increased electrical response in the post-synaptic neuron when the presynaptic neuron is activated.

We don't know exactly how the information is read out, but in my opinion, memory expression is essentially a replay of the electrical activity that occurred when the experience first occurred in the brain. From this viewpoint, the glue/storage mechanism helps explain the readout, because LTP occurs at weak synapses that are strongly activated by some experience, and the experience converts the weak synapse into a strong one by the mechanism I just described. Experience changes synapses so that the pattern of strong and weak synaptic communications can be more easily reproduced. This is one way the information can be replayed. How one set of information is selectively accessed without interference from others is a key big question we are working on.

Q: Does the repetition of audio/visual data—rote learning—help the transfer of this data from short-term memory to long-term memory? What could stop this transfer? Ed Robinson, Edmonton, Alberta, Canada

Fenton: Ed,
The short answer to the first part of your question is "yes," but the biology is also more complicated than a simple "yes" suggests. Repetition aids the formation of long-term memories, but after a point repetition also causes habits to develop, and habit learning seems to be different from just a strong long-term memory. In particular, different parts of the brain are crucial for long-term memories of experiences (e.g., the hippocampus and related structures) and the storage of habitual behaviors (e.g., the basal ganglia and related structures). To what extent these different "memory systems" operate in parallel and how they interact are questions of active research.

Q: What is the scientific advantage of being able to "erase" memory, i.e., what is the next step? Dick Pohl, Omaha, Nebraska

Fenton: Dick,
I'd like to start by stating clearly that until very recently memory was a concept we invented to account for behaviors we can observe in others and ourselves. We used to say that memory is what happened when experience changes behavior, because we did not have a clear physical explanation for what memory might be. Today we can define long-term memory as the information that PKMzeta stores by maintaining LTP, a strengthening of synaptic communication. This is surely not a comprehensive definition, and many people are working to do better; nonetheless it is something concrete, physical, and measurable.

My colleagues and I did not set out to find a way to erase memory; we were far more ambitious, and we aimed to define memory. The journey starts with the idea of activity-dependent synaptic plasticity made popular by Donald Hebb in 1949: Neurons that are active together change how they communicate so that they will be more likely to be active together in the future. Then in 1973 Tim Bliss and Terje Lømo discovered long-term potentiation (LTP), the activity-dependent strengthening of electrochemical communication between neurons as a potential mechanism for Hebb's idea. Many neuroscientists investigated the physical basis of LTP in hopes that it might be how information is stored in the brain (see David Glanzman's profile on the website). After more than a decade of studying LTP in slices of brain, Todd Sacktor, my collaborator, determined that PKMzeta was both necessary and sufficient for maintaining LTP.

But what is the relationship between LTP and memory? I had been studying memory in awake-behaving rats for over a decade, and Todd and I decided to investigate whether PKMzeta also stored memories. There are three logical approaches to answer the question. If PKMzeta stores memories, then inhibiting it should erase memory; forming a memory should generate PKMzeta, and putting PKMzeta into neurons should create memory. Erasure was the easiest of the three tests, so we investigated whether the PKMzeta inhibitor ZIP could erase a long-term memory. We are working on the other tests and investigating how PKMzeta is organized in space.

Q: How does your memory-erasing research fit within the larger conceptual framework of the brain? What do we still not know about memory that poses challenges for you? Joel Marsh, Stockholm, Sweden

Fenton: Joel,
As I explained in my reply to Dick Pohl above, our research confirms that activity-dependent synaptic strengthening (and weakening) can store information in the brain. The PKMzeta mechanism for maintaining LTP also suggests, at least to me, that memory expression may indeed be a form of replaying the electrical activity that occurred when the memory was formed by direct experience. The work by Itzhak Fried's group at UCLA provides strong evidence of this. Many challenges remain. I think it is essential to take on the problem of understanding how the patterns of synaptic changes and electrical activity in the vast networks of neurons and synapses are organized and coordinated in space and time so that distinct memories do not interfere with each other and yet can be associated and recombined to form new knowledge.

Q: Now that we know that PKMzeta is crucial for long-term memory formation, is there any kind of pharmaceutical intervention in the works for people who have had brain injury and have difficulty retaining memories?

Also, is PKMzeta only involved in long-term memory, or is it also involved in short-term memory? Anonymous

Fenton: Re your first question, I'm not a clinician so I consulted Todd Sacktor, who is a neurologist. He tells me that cognitive rehabilitation is the standard intervention currently, but others are working on pharmacological treatments for injury-induced memory loss. Indeed, I have been working on rat models of traumatic brain injury with Drs. Peter Bergold and Samah Abdel Baki at SUNY, Downstate Medical Center to identify synergistic combinations of standard drugs that might be effective at reducing the cognitive impairments.

Re your second question, PKMzeta seems not to be directly involved in short-term memory. See my reply to Mary Arronte above.

Q: Dear Dr. Fenton,
I wrote my doctoral dissertation (Univ. of Penna., 2007) based on the oral histories of two triracial families—my own and one other—so I was very interested in your experiments concerning memory, featured on tonight's episode of NOVA scienceNOW.

My question may actually be two. In many cultures and ethnic groups, some individuals become their family's or community's historians, capable of remembering large, detailed, and interrelated amounts of information, which often remain remarkably accurate over time.

Is there a genetic or hereditary aspect involved in this type of long-term memory? And, if genes and chromosomes can transmit phenotypic and physiological "data" through heredity, is there a component to memory that can also be hereditary—not just in terms of ability, but actual memories or parts of memories as well?

Thank you for your consideration and for a fascinating educational program. Sam Lemon, Ed.D., Aston, Pennsylvania

Fenton: Sam,
I do not expect there is a genetic or biological basis for the information storage you describe. That would imply that a person is born with specific cultural, episodic knowledge for facts and events. I imagine the ability to learn and remember such information may be heritable, but I don't know any mechanism that can explain heritable storage of that type of information.

Q: If it is/will be possible to erase a specific memory, then would it be reasonable to suppose that it might be possible to implant a specific memory? For example, the memory of having gone to college with what would have been learned had the memory been real. Mike

Fenton: Mike,
If we knew the memory mechanism (and I think we do), then in principle it should be possible to implant information in memory if we also knew how the information is organized and we had the necessary technology to do the implantation. People are working to understand the basic biology, which has to be a lot better before we can imagine how to implant detailed knowledge like going to college, without destroying most of the information that was previously stored. See my reply to Dick Pohl above.

Q: Can our memory be affected by what we eat? If so, which foods or micronutrients can help improve our memory? Roxanne, Victoria, British Columbia, Canada

Fenton: Roxanne,
Memory is a biological function of the brain, and so foods that affect brain function have the potential to affect the ability to form, store, and recall memories. I don't know of any specific foods that directly improve memory itself, but there are many foods thought to be good for brain function, such as foods that are rich in some omega-3 fatty acids.

Q: I have two questions that may be related to each other. I always had a mind like a steel trap, but as I am getting older it seems that I can't hold onto memories from yesterday. At least not the way I did a few years back. Why is this?

My other question is: How do drugs used for pain (opiates) interfere with the formation of long-term memories? I have neuropathy in my feet, and I have to take them for the pain, but it interferes with my memory. Thank you. Anonymous

Fenton: Unfortunately, memory abilities decline with age, along with other brain functions. The causes are probably diverse and include damage and the fact that the aged brain expresses different proteins, which changes brain chemistry. In addition, aging has also been associated with abnormal information processing in the brain.

Opiates may stimulate the release of the neurochemical dopamine, which is rewarding and can encourage learning when it is occurs naturally. However, opiates are narcotics that can reduce awareness and make it hard to be attentive. We tend only to remember what we pay attention to, so opiate-induced inattention may be a major factor in your memory difficulties.

Q: As an aging man I seem to have acquired the common experience of not being able to recall proper names sometimes, or having to wait until a name pops into consciousness. Do you have any insights into name recall? Are words, or their electrochemical equivalent, stored in neurons or neuron aggregates? Why, if I quit trying to recall a name, does the name often become conscious? Why does the aging brain have these lapses? Why do I have the feeling, when I forget a name and try to remember it, that I am searching through a file cabinet or database, running through the alphabet or making related associations? I would sure be interested in what you might know about this phenomenon. Thanks! Phil Carman, Hollister, Missouri

Fenton: Phil,
I don't know any information on aging-impaired name recall specifically, but I don't expect that words are fundamentally different than other forms of acquired information stored in the brain. I thus expect that words, like other forms of information, are stored by the changes in synaptic strength that I have discussed above.

I can only speculate as to why recalled words may come to mind apparently without effort. My guess is that although you may not be aware of it, you are unconsciously working to recall the name. A similar thing happens to me when I am writing difficult computer programs. I often encounter a problem that I work on for hours, unsuccessfully. If I stop working on the problem and do something else, especially sleep, the correct solution often spontaneously comes to mind as if I had been working on it unconsciously.

My explanation is that brains solve problems and retrieve information by activating patterns of neural activity in many different neural pathways, akin to your searching through a file cabinet. When you search for information or a solution to a problem consciously, your attention limits the patterns that can be activated and thus limits where you look for the answer to your current best guess at where to look. I sometimes do the same thing while searching for keys in my office: I only look where I think I can find the keys, and that limits my ability to find the keys if they are somewhere I do not anticipate. If the search is unsuccessful, and I do something else, attention changes the limits of what neural activity patterns can be activated, and I can unconsciously search elsewhere to find the solution.

Q: Why did my 93-year-old grandmother with Alzheimer's remember how to play the piano and sing but not know who I was? And, since music helps cognitive learning, does it also help memory?

P.S. She was an accomplished musician who played the piano for the silent movies. Anonymous

Fenton: Your observation highlights the fact that there are different kinds of information stored in memory, and different types of memories are stored in different parts of the brain. To your grandmother, music was probably like a language or a learned skill, closer to walking or riding a bicycle in how the brain represented that information than recalling who or what she recently saw. She was an expert, and probably could play music effortlessly, without thinking about it because of her many years of practice. Alzheimer's affects the hippocampus and related areas of the neocortex, which are more critical for storing episodic, what, when, where memories from experience and less involved in storing habitual, very familiar information that support habits and skills.

Q: Dear Dr. Fenton,
My son Alex, now 22, had open-heart surgery at the age of 2. In high school Alex began to complain that it took him considerably longer than other students to memorize material. In college Alex was diagnosed with rote verbal memory syndrome. Twenty years ago a patient during the heart surgery remained on a heart-lung machine. These machines together with blood pumped tiny bubbles of air through the patient's brain, causing short-term memory problems. Could this be a cause of Alex's problems with recall? How can he be helped?

I give him memory supplements: Phospatidylserine, Vinpocetine, DMAE, Huperzine A, and Neuropower by Synergy. Do you think these supplements may help? Alex has just started dental school, where the ability to memorize a large amount of material is key. Alex has to read it several times to be able to record it in short-term memory. I will greatly appreciate any suggestion you may have. Eva Orlik, Indianapolis, Indiana

Fenton: Eva,
I'm sorry to hear about Alex's difficulties. As I'm not a physician, I consulted my collaborator Todd Sacktor; he is a neurologist and very knowledgeable. Todd said that it's possible Alex had structural brain damage from the early operation, but an MRI or CT scan would have shown it. In Todd's opinion, cognitive mnemonic training is probably better than any over-the-counter drug. I agree with him that there are multiple proven ways to improve mental functioning. One is to improve the chances of storing information in memory by paying attention, and another is to improve the ability to manipulate the information that is stored by learning to organize information and think effectively. It seems like Alex is able to compensate for poor memory by old-fashioned and effective repetition and attentive focus. It's hard work, but it is the surest way I know to succeed.

Q: Wonderful show. What are the implications of a photographic memory? Has your study looked at the variables that different people can learn at a different rate? Ray Curry, Reno, Nevada

Fenton: Ray,
We've only studied memory in animals, as we are interested in understanding the biological mechanisms and can manipulate the biology in animals in ways we cannot in human subjects. Photographic memory is a human phenomenon, and I don't understand it. I don't have facts to offer, but I can tell you how I think about it. I am disinclined to think that people with photographic memories have a better basic biology for storing memory. Rather, I suspect they have an enhanced ability to be attentive and an enhanced ability to organize the information they store in their "snapshots" of experience.

Q: I have the disease CMT and am having some memory problems. Could the faulty synapses caused by the disease be the problem? Paula Davidsen, Berkeley, California

Fenton: Paula,
I am sorry you are having difficulties. As I am not a physician I consulted my collaborator Todd Sacktor, who is an excellent neurologist. He said that Charcot-Marie-Tooth disease is a defect in peripheral myelin, the fatty, insulation-like material that surrounds nerve cell axons to increase the efficiency of electrical transmission within nerve fibers. Todd has not heard of a central nervous system (brain or spinal cord) defect in CMT that would cause a memory problem.

Q: Hello Dr. Fenton,
I teach developmental mathematics at a community college. Currently, I am taking a cognitive psychology course. The first topic of discussion is short-term and long-term memories. As an educator, my lecture reinforces skills using repetition and relationships. I observed from my adult learners that applying meaningful relationships to mathematical concepts, they perform better on their assessments and have higher retention of the material in future courses. Does the chemical molecule PKMzeta play a significant role in retaining memories involving math operations over the long-term? Does the repetition prevent the decay of the memory? Anonymous

Fenton: While we have not tested the role of PKMzeta in storing mathematical information, or in people for that matter, we have determined that PKMzeta stores the precise kinds of information needed for accurate recollection of several forms of long-term memory, in several parts of the brain. I expect that mathematical concepts fall into the category of information stored by PKMzeta. Repetition activates similar if not identical neural pathways over and over, which is precisely the kind of neural activity that encourages synapses to store information by changing so that the activation pattern is more likely to occur again in the future. See my replies to Dick Pohl and Joel Marsh and Ed Robinson above.