Many people who enroll in high-level memory competitions use the method of loci. For instance, every Pi Day—March 14, or 3.14—Princeton University hosts gatherings of people to see who can recite the greatest number of decimal places in this mathematical constant. In 2009, researchers from several universities collaborated on a functional MRI study to identify the brain areas that support the ability to memorize pi to a staggering number of decimal places. During such an experiment, participants lie on their backs inside an MRI scanner and perform a behavioral task. As specific brain areas are activated, their oxygen consumption increases. The brain detects this increased use of oxygen, and orders more blood to bring oxygen to this part of the brain. The increase in oxygenated blood in this area changes its magnetic properties. We can detect these localized changes in the magnetic field using a powerful magnet, several thousand times the magnetic pull of the earth and guaranteed to render your credit cards useless if you bring them too close. In this way, we can map activation by identifying the brain circuit recruited for each particular capacity.
In the 2009 study, functional MRI allowed researchers to record the brain activity of a twenty-two-year-old engineering student as he recited the first 540 digits of pi. The student used the method of loci to remember the digits in order, and functional MRI scans conducted while he performed this feat showed that his retrieval processes elicited robust activation in particular areas in his prefrontal cortex. These areas are known to support working memory and attention, suggesting that the student engaged cognitive-control processes to rattle off the well-learned sequences of pi.11
To gain insight into how people acquire such vast amounts of information in the first place, the researchers asked the engineer to learn a novel series of one hundred random digits as he was being scanned. The result was impressive: after three six-minute scans, the student had encoded all one hundred digits in the correct order, using his own variation on the method of loci. During early stages of encoding, the functional MRI images showed greater engagement of cortical areas specialized for visual processing, emotion-related learning, motor planning, task scheduling, and working memory than during the retrieval of pi digits. These areas were activated because the task required a great deal of mental effort, and to succeed the student had to harness numerous resources, including visual processes in the back of his brain and cognitive control processes in the front.
The student explained that in his particular method of loci, he relied heavily on color, emotion, humor, vulgarity, and sexuality to build his memory palace—“the more emotional and gruesome the scene, the easier the recall.” The researchers linked his consistent use of highly emotional images to a structural difference in his brain—increased volume of an area in his cingulate gyrus, part of the limbic system. This student was brilliant at memorizing groups of digits, but he did not necessarily possess a superior intellect or memory; he had worked hard and had developed highly effective cognitive-control circuits for retaining information. As one psychologist previously put it, “exceptional memorizers are made, not born.” Phenomenal memory performance is within your reach, whenever you want to apply it. When you try to memorize names, numbers, words, pictures, and so on, you will remember the items better if your brain is optimally active during your initial exposure to the material.12
Encoding is the gateway to memory formation, with consolidation and storage following closely behind. Henry could encode the information presented to him and register it briefly, but then his processing broke down. He could not consolidate and store that information.
In 1995, when functional MRI was in its infancy, we had the opportunity to observe Henry’s encoding processes in action. In this experiment, we asked him to look at pictures of scenes and indicate whether they were indoors or outdoors. These questions were intentionally easy. He answered them correctly, so we knew he was looking at and processing the pictures. The corresponding MRI images showed increased activity in his frontal lobes as he performed the picture-encoding task. Subsequent experiments in other labs with healthy research participants have expanded on this finding, and the results indicate that two separate areas in the left frontal lobe and an area in the right frontal lobe are normally active during encoding. Henry could fire up his frontal cortex to encode the objects he perceived, but after that, he was stopped in his tracks—the process of memory formation broke down, severely impeded by his inability to consolidate and store information.13
When the brain receives and encodes new information, the content must be processed further to make it available for future use. The initial transmissions are not immediately encased in long-term storage. The longer process by which memories become fixed, consolidation, is a lasting change that happens in individual neurons and their molecular components. Connections between adjacent cells become stronger or weaker in response to learning experiences. Henry’s defunct hippocampus was unable to initiate and complete the active processes required for consolidation.
Two ambitious experimental psychologists at the University of Göttingen, Georg Elias Müller and Alfons Pilzecker, introduced the concept of consolidation in 1900. Ever since, scientists have struggled to understand the mechanisms by which the brain consolidates memories. This quest has inspired thousands of experiments in many species from insects to humans, and has fueled healthy controversy about how different kinds of memories are anchored in our brains.14
Müller and Pilzecker made the novel discovery that declarative learning, consciously retrieving facts and episodes, does not immediately lead to enduring memory. Instead, consolidation depends on changes in the brain that occur gradually over time. During this time, the newly learned material is susceptible to interference. The German researchers came to this conclusion after eight years of experiments performed on a small group of participants that included their students, colleagues, family members, their wives, and themselves. First, they created 2,210 nonsense syllables, assembled them into pairs such as DAK-BAP, and generated lists of six pairs. They tested one participant at a time, with the arduous training and testing taking place over twenty-four days. During training, the participants read the list of six pairs aloud and tried to mentally associate the two nonsense syllables that formed each pair. Then came the memory test: the participants saw a cue—the first syllable of each pair, for example, DAK—and were asked to say the second syllable of the pair, BAP.15
The researchers’ key insight came when they focused their analysis on the participants’ intrusion errors. An intrusion error was scored if participants, while recalling the items in one list, inserted a nonsense syllable that was in an earlier list, thinking that it belonged to the current one. In this experiment, if participants had previously learned the syllable JEK, they might pair that with DAK instead of the correct answer, BAP. The psychologists reasoned that these intrusion errors resulted from the persistence of the just-learned material in the participants’ recent memory. These errors were most numerous when the test was performed twenty seconds after training, when the brain was still encoding information, and became less frequent as the time between the training and the test increased from three to twelve minutes, the period during which the brain was consolidating information. Intrusion errors did not occur twenty-four hours later; by then, the subjects had successfully consolidated the information. The experiments revealed that consolidation is an active process that takes time. The associations were easily broken right after encoding but became stronger from minute to minute.16
Subsequent research with animals supported Müller and Pilzecker’s hypothesis. In 1949, a physiological psychologist at Northwestern University trained groups of rats to avoid a grid that delivered a mild shock. They then administered electroconvulsive shock (ECS) to the rats’ brains at various times after the end of each learning trial. The group that received ECS after twenty seconds was the most impaired, and the deficit became progressively less as the time between learning and ECS increased to forty seconds, one minute, four minutes,
and fifteen minutes. The rats that received ECS after one hour or longer were unimpaired. The longer the interval between encoding and ECS, and the more time for consolidation, the better the memory. These results show that for a limited time after training, disrupting brain activity blocks the mechanisms underlying consolidation. In Henry’s brain, the critical cellular events in the hippocampus and cortex that occur for minutes or hours after encoding never activated, and thus new declarative information could not be secured.17
From these and many other experiments, neuroscientists learned that the neural infrastructure of memories, the physical traces that exist within the brain, are initially frail and strengthen gradually. They can be disrupted by behavioral manipulations in the laboratory and by more direct insults to brain physiology in the form of drugs, alcohol, or head injury. A too-familiar example of the vulnerability of memory formation is prevalent in North American football. In the fall of 2012, a linebacker playing his first varsity high-school game went to tackle the opposing team’s running back who was carrying the ball. They collided—helmet to helmet—and both players went down but soon got up and went back to their positions. After two more plays, the linebacker’s teammates went to the sideline and told the coach that the injured kid was playing out of his position and, when told to move into his position, did not do so. He was immediately examined by the team’s physical therapist who determined that his neurological status was normal, he was not nauseous, and he did not have a headache. The player’s striking symptom was that he had no memory of the hit or anything that transpired after the hit. The blow to his head had derailed the consolidation of these fragile memory traces.
Most of the memory tests administered in clinics and laboratories assess declarative, episodic memory—the ability to form associations among words, elements in a story, or details in a picture. Before Henry, memory researchers did not know for sure which brain structures were responsible for establishing these connections. The crucial lesson Henry taught us is that the hippocampus is necessary for building associations. Without a functioning hippocampus, Henry found it impossible to form associations between familiar words; he could not link them in his memory. His long-term memory stores for new information were always empty.
Association, a basic concept in both animal and human learning, is the essence of episodic memory; it enables us to characterize a unique event (reading this chapter) by integrating its context in time (3 p.m.) and space (in the kitchen with light coming in the window). The context may be rich, including who else is in the room, whether music is playing, and specific thoughts about each sentence read.
In everyday life, associations develop and strengthen over time when particular items occur together repeatedly. When we move to a new neighborhood, we gradually get to know the people who live in our community—our neighbors and the people who work in the coffee shops, pharmacies, and restaurants that we frequent. Eventually, we get to know some of these people well, as we gather information about their personal lives bit by bit. We learn, for example, that the gentleman behind the espresso machine, who always asks about our dog, is a student who has been working on his degree for five years and who dreams of being a journalist, and that the older man at the convenience store who always has a cheery greeting lost his granddaughter to cancer. We experience what it is like to live in this neighborhood in the spring, summer, fall, and winter, and we store the sights, sounds, and smells that characterize the environment. Over time, our brains build up an elaborate representation of our neighborhood, in which lots of individual facts and events have become connected to one another. After we have lived there for a few years, we can provide a vivid, detailed picture of what the neighborhood is like.
In recent decades, thanks to contributions from thousands of labs in many countries, scientists have come to understand the cognitive processes and neural representations that support these kinds of associations. The cortical neighbors of the hippocampus, the parahippocampal areas, flood the hippocampus with complex perceptions, ideas, and contexts, and the hippocampus associates this wealth of information in three ways. First, the hippocampus links distinct objects with one another and with the time and place we encountered them—for example, all the objects and people we saw, the sounds we heard, and the aromas we smelled in our neighborhood coffee shop this morning at 7:55. Second, it links events in time to record the flow of experiences that comprise a unique episode—for example, the sequence of entering the coffee shop, getting in line, reading the menu, ordering a large cappuccino, waiting for the server to brew our drink, picking up our order, and rushing out the door to get to work. Third, it links many events and episodes in terms of their common features to form a network of relationships—for example, connecting this morning’s coffee-shop memory to memories of meals in other coffee shops and restaurants that we frequent, thus composing our general knowledge of eating out.18
Each morning, when we encode the details of an experience in a coffee shop, this new learning reactivates many separate events from the past, resulting in an updated, rich associative representation that transcends the individual events. To establish this inclusive representation of eating out, we depend on cooperative interactions in our brains between our hippocampus and regions in the midbrain, a two-centimeter-long structure that connects the cortex and striatum to areas lower in the brain. Cross-episode integration, connecting separate experiences that have common characteristics, guides decision making in everyday life. (Should I go to the coffee shop that has the best cappuccino or to the one that has fabulous pastries?) This intricate cognitive and neural infrastructure was not available to Henry.19
When Milner tested Henry for the first time in 1955, she examined his ability to form word associations by reading aloud eight word pairs. Some of the pairs were considered easy to remember because their meanings were readily associated, while others were hard because the words were unrelated.
Metal–Iron (easy)
Baby–Cries (easy)
Crush–Dark (hard)
School–Grocery (hard)
Rose–Flower (easy)
Obey–Inch (hard)
Fruit–Apple (easy)
Cabbage–Pen (hard)
Five seconds after reading the list of word pairs, Milner asked Henry, “Do you remember what went with Metal? Baby? Crush?” and so on, through the list. On the first try, he had one correct answer, Iron. Milner reread the list of word pairs, and then tested him again. The second time he recalled Cries, Iron, and Flower, all belonging to the easy pairs. The third, final round, he remembered Apple, Cries, and Iron. He could not consolidate the difficult pairs. A half hour later, Henry retained the one association that he had correct on each of the three trials: Metal–Iron. The other associations had melted away because Henry’s brain lacked the medial temporal-lobe infrastructure required to consolidate and store them.20
In their groundbreaking 1957 paper detailing Henry’s operation and his psychological test results, Scoville and Milner launched the modern era of memory research. Although studies of previous patients, especially F.C. and P.B., had suggested that the hippocampus was critical for the establishment of long-term memory, Henry’s case clinched the connection. As a result of his consistently poor performance on numerous and diverse memory tests, the hippocampus became the focus of thousands of memory researchers all over the world.21
We now know that consolidation is an essential aspect of memory, but how exactly does it occur? What are the underlying processes in the brain? Answering these questions is essential to understanding Henry’s memory impairment.
Memory consolidation depends on dialogues among brain circuits, coupled with cellular changes within networks of cells, specifically those in the hippocampus. It requires intense conversations between the hippocampus and areas in the temporal, parietal, and occipital cortices where bits and pieces of memories are stored. These communications between neurons reorganize and strengthen connections among memory-processing regions to ensure that i
nformation is preserved in the cortex.22
Messages travel from each neuron to its neighbors by way of a long tail, an axon. At the end of the axon, the message, coded in electrical and chemical signals, crosses the gap between the neurons, known as a synapse. The synapse contains a synaptic cleft, a corridor through which molecules travel from one cell to the next. On the other side of the synapse, dendrites, treelike branches of an adjacent neuron, receive messages and pass them securely to the body of that neuron for processing. Each neuron has its own output, the axon, and many inputs, the dendrites (see Fig. 8).
Permanent Present Tense Page 15