Excerpt from Memory: The Key to Consciousness, by Richard F. Thompson and Stephen A. Madigan (Princeton University Press, 2007). Reprinted with permission from the authors.
Chapter 9: Mechanisms of Memory
In an espionage movie an American secret agent discovers a horrendous terrorist plot to destroy a U.S. city. With this discovery he knows how to stop the terrorists. Unfortunately, before he can tell anyone else about it, he is killed. Government scientists extract protein memory molecules from his brain and inject them into the movie’s hero. He acquires the memories from the dead agent’s brain molecules and stops the terrorists.
Sound far-fetched? Actually experiments have been done that gave some credence to this sort of scenario. The initial studies involved little flatworms called planaria that have a very primitive nervous system. These little creatures are able to regenerate after being cut up. Planaria were “trained” using electric shocks to make a certain movement. When “trained” planaria were cut up, the regenerated worms remembered the task. As it happens, planaria are cannibals. So the experimenters then ground up trained worms and fed them to untrained worms. It was reported that the untrained worms then performed the learned response.
Studies like this gave rise to the notion of “memory molecules,” the possibility that particular memories could be stored in protein molecules. This rationale led to further studies with rats, which were trained to approach a food cup. The brains of the trained rats were ground up and injected into untrained rats, who were then reported to have acquired the learned response.
Unfortunately, these dramatic experiments were not well controlled. Attempts by other laboratories to repeat these findings on planaria and rats failed completely. Indeed, a Nobel Prize-winning biochemist, hoping to identify memory molecules, devoted some 20 man-years of his laboratory’s efforts to training planaria, with complete lack of success.
Donald Stein and his students at Clark University did a critical experiment that shed much light on this puzzling situation. They trained rats in a simple task where they had to remember not to step forward in a little box (if they did, they got shocked). Brains and livers from these trained animals were ground up and injected into untrained rats. Both the brain and liver recipients showed some degree of “memory,” but the liver group did better than the brain group!
How could this happen? No one believed that memory molecules were stored in the liver. The answer is that injecting foreign protein tissue into an animal causes an immune response and other problems and can be very stressful for the animal. Stress can markedly influence activity and performance in simple learning situations. The untrained recipient animals did not “remember” the task at all; they were simply less active in the situation and didn’t step forward in the box.
Other studies involved extracting RNA from the brains of trained animals and injecting it into untrained animals. There was a bit more logic here. RNA is the messenger molecule that takes information from DNA and uses it to make proteins. So if memory involves changes in the genome, the DNA, the extracted RNA should have this new information. Unfortunately, none of these studies could be replicated. As with the planaria, proper control conditions had not been used.
We now know that such experiments could not work. Proteins and RNA are large complex molecules and cannot pass the blood-brain barrier, a very special structure that prevents many harmful chemicals and large molecules for passing from the blood into the brain tissue. It is a kind of connective tissue lining all the blood vessels in the brain. We also know that when large foreign molecules are injected into the blood, they are generally broken down, so even if they could cross the barrier, the “memories” in the molecules would have been destroyed. Attempts to transfer memories from one brain to another by means of “memory molecules” did not work. But these studies encouraged Time magazine to suggest a solution for what to do with old college professors.
The possibility that memories could be coded in DNA and RNA raised yet again the old notion of inheriting acquired characteristics, the idea that memories could be coded by changes in the DNA of the genome as a result of experience. As far as we know, this does not happen. But the genome is very much involved in memory formation, particularly long-term memories. Studies on animals ranging from invertebrates to goldfish, rats, rabbits, and other mammals all agree that synthesis of proteins is necessary for the formation of long-term memories. The specificity of memories is in the connections among networks of neurons.
Synthesis of Proteins
Genes, DNA, are simply long chains or sequences of four different forms of a type of molecule called nucleic acids: adenine (A), guanine (G), cytosine (C), and thymine (T). A given gene may be hundreds of molecules long and has a specific sequence of these four nucleic acids. These sequences determine the protein made from the genes. Basically, all that genes do is generate proteins. RNA molecules transfer these genetic instructions from the DNA to make proteins in cells. Proteins, of course, can do many things.
Some form actual biological structures like neuronal synapses, and others serve as enzymes, controlling chemical reactions in the cell. Consequently, blocking this process of making proteins, which prevents long-term memory formation, means blocking the activity of the genes.
Many studies in which long-term memories were prevented by blocking protein synthesis involved injecting these blocking drugs systemically (that is, into the body and ultimately the bloodstream). These are powerful drugs, and they make animals sick, which might account for their effect on memory. However, infusion of these drugs into places in the brain where long-term memories are being formed can also prevent the memories from being established. One of the present authors (RFT) infused such a drug, actinomycin, into the cerebellar nucleus in rabbits, where long-term memory for the learned eye-blink response is formed. This procedure completely prevented learning of the eye-blink response. However, if the drug was infused in this location in trained animals, it had no effect. It only prevented learning of the conditioned response. Interestingly, a new protein was formed in this region as the rabbits learned the task. This protein is an enzyme that is normally involved in cell division. Since neurons do not divide after birth, perhaps such enzymes do other things in neurons.
One possibility of why gene expression—making proteins—is necessary for the formation of long-term memories is that new structures must be formed in the neurons to store the memories. Synapses, the connections between neurons, are an obvious candidate for new structures in neurons.
Synapses and Memory
The number of possible synaptic connections among the neurons in a single human brain may be larger than the total number of atomic particles in the known universe! This probability calculation assumes that all of the connections are random, and indeed many do seem to begin at random in the developing brain. But the actual number of synaptic connections in a typical adult human brain, is of course, very much less, roughly a quadrillion (1 followed by 15 zeros), still an impressive number.
The human genome contains a great deal of information, perhaps the equivalent of an encyclopedia set. However, the information capacity of the genome is orders of magnitude less than the number of synaptic connections between the neurons in the brain. So these connections cannot all be determined genetically. Instead, experience must shape the patterns of nerve cell connections in the developing brain. To be sure, the organization of major structures and areas of the brain and their patterns of interconnections are determined genetically. It is at the finer grain, the details of synaptic connections on the dendrites of neurons, where experience comes into play. In Chapter 3 it was noted that brain synapses increase dramatically in number over the first few years of life, the same period when much of our learning occurs. This is particularly true for the brain systems critically involved in memory formation, the cerebral cortex, the hippocampus, and the cerebellum. Synapses form, alter, and disappear throughout life.
We have a good understanding of how the patterns of synaptic connections form in the primary visual cortex. The axons from the visual thalamus conveying information to the cortex terminate on primary receiving neurons. Initially, each neuron has equal synaptic connections of input from each eye. Then as a result of visual experience after birth, input synapses from one eye or the other dies away, so each of these primary neurons is activated by only one eye or the other.
The key is visual experience. Covering one eye results in all the input from that eye dieing away from all the primary neurons, so the eye, actually the visual cortex, becomes blind. Seeing is necessary for normal development of the visual cortex. But how? It turns out that nerve cell activity, spike discharges, or action potentials are necessary. When seeing the world, the neurons projecting information from the eye to the visual cortex are very active, conveying many spike discharges per second to the primary receiving neurons in the visual cortex. Covering the eye prevents these action potentials from occurring. Suppose both eyes are functioning normally and the neurons projecting information from one eye to the cortex are inactivated, perhaps by infusing a drug that reversibly shuts down neurons, preventing them from generating spike discharges. The result is the same as if we cover one eye. The development and maintenance of the normal visual system is dependent on normal activity of the neurons in the visual brain.
Recall the rich rat-poor rat studies. A normal stimulating environment is necessary for normal development of the cerebral cortex. A poor environment results in many fewer synaptic connections. A similar type of process is thought to occur in the brain when we learn.
A typical neuron in the cerebral cortex is shown in Figure 9-1, the same one shown in Chapter 3. The dendrites receive connections from hundreds or thousands of other neurons via axons forming synaptic connections on these dendrites. Each little bump or spine on each dendrite is a synaptic connection from another nerve cell axon terminal. As you can see, the dendrites are covered with these little spine synapses. Rich rats have many more dendritic spine synapses than poor rats. Indeed, people with college degrees have more spine synapses on dendrites in certain areas of the cortex than do people with only high school degrees, who in turn have more synapses than people who did not finish high school. The memory trace, the physical basis of memory storage in the brain, may simply be more synapses. But this begs many questions—for example, how and where are they formed? We return to these questions later.
Does the proliferation of cortical synapses in the rich rat and well-educated human reflect specific memories or simply more general factors like intelligence and well-being? It has been difficult to answer this question, in part because for most kinds of memories it is not known exactly where they are stored in the brain. Indeed, some neuroscientists think that complex memories like those of our own experiences (declarative memory) may be stored in widely distributed networks of neurons, probably in the cerebral cortex. The jury is still out on this question. But there are a few examples of where particular memories are thought to be stored in particular places in the brain.
Mark Rozenzweig of the University of California, Berkeley, and William Greenough of the University of Illinois pioneered the rich rat-poor rat studies. The first studies simply showed that the cerebral cortex was thicker in the rich rats. Greenough showed that this was due primarily to the growth of more synapses and more nonneural supporting cells called glia. (But remember that the rich rats were really normal rats, and the poor rats had fewer than normal numbers of synapses and glia.)
Greenough and his associates have provided a clear example of where a type of memory may be localized. When rats are trained to reach with a forepaw through a small hole in a piece of clear plastic to retrieve a bit of food, the region of the cerebral cortex that represents movements of the forepaw becomes critically engaged. After the animals have been trained, damage to this small region markedly impairs their performance in this task. As animals learn the task (no damage) there is a dramatic increased growth of synaptic connections among the neurons in this region.
Pavlovian conditioning of the eye-blink response provides an example of associative memory (see Chapter 7). As noted one of the present authors (RFT) and his many associates have been able to localize the basic memory trace for this form of associative learning to a particular place in the brain, a group of neurons in a cerebellar nucleus. Lesions of this small region completely and permanently abolished the learned eye-blink performance.
Studies by Jeffrey Kleim at the University of Lethbridge and John Freeman at the University of Iowa demonstrated that there is a dramatic increase in the number of excitatory synapses in this localized region of the cerebellum as a result of learning the conditioned eye-blink response. This particular memory appears to be stored by the increased number of synapses in this nucleus.
Excerpted from Memory by Richard F. Thompson and Stephen A. Madigan. Copyright © Richard F. Thompson and Stephen A. Madigan, 2005. All rights reserved.
Richard F. Thompson, a behavioral neuroscientist who has spent nearly a half-century researching the physical basis of memory, is Keck Professor of Psychology and Biological Sciences at the University of Southern California. In 2002, he was the first to identify and map neural circuits involved in classical conditioning, made famous by Russian psychologist Ivan Pavlov. Thompson has written six books, edited several others and published 450 research papers to date. He won the American Psychological Foundation’s 2010 Gold Medal Award for Life Achievement in the Science of Psychology.
Stephen A. Madigan is associate professor of psychology at the University of Southern California. In 1992, he opened the Vancouver School for Narrative Therapy through Yaletown Family Therapy (YFT) in Vancouver, Canada, as the first Narrative Therapy training site in the Northern Hemisphere. He has worked as a script consultant to over 60 TV and movie productions. In addition, Madigan has produced six International Narrative Therapy Conferences and one International Therapeutic Conversation Conference on Brief Therapy.
Be sure to ‘like’ us on Facebook