Henry’s deficit was in consolidating, storing, and later retrieving new facts and events. Still, his amnesia could not necessarily be blamed on a deficit in retrieval: he could still recall facts he had consolidated and stored before his operation. He loved to talk to our team at the CRC about his preoperative experiences and family life. When he retrieved these memories, however, he could not integrate the old memory traces with information in his current life. For instance, when he talked about his gun collection, he could not update this narrative by saying what had happened to it. Henry had consolidated information before his operation but could not reconsolidate it after, so his childhood memory traces remained engraved in his brain, unrefreshed.
Think of reconsolidation as a memory-updating process. If you unpack a suitcase and then repack it, the clothes will be arranged slightly differently than they were before, and you may leave out some items and add others. Retrieval and reconsolidation of old memories suddenly makes them labile, a state in which they are again vulnerable to distortion and interference. At this time, memories can be modified by new information.
If I ask you when you last had Chinese food, you initiate a mental search based on internal thoughts of food, dinner, leftovers, chopsticks, and so forth. You might be driving down a street in Chinatown, and that environment could help revive your memories too. When you recall the meal I asked you about, the process of retrieval reactivates consolidation mechanisms similar to those that occurred at the time of the original meal, even if it was years ago. The memory will be altered by your current thoughts. For instance, finding out a year after you ate the Chinese meal that you were allergic to MSG modified the memory of the original experience by incorporating the cause of that night’s splitting headache. Also, having a similar meal, such as celebrating your last birthday at a Korean restaurant, could interfere with your memory. Anything you think about at the same time as you mentally revisit the Chinese meal helps shape the new memory. But while you may expose your memory of the event to distortion each time you retrieve it, you also make the edited version more likely to stick in your mind. Reactivating an old, previously consolidated memory creates new memory traces, and this mushrooming of traces makes older memories more resistant to interference from other brain activity. If you reactivate the memory of that meal every month for the next six months, the new memory will be more robust and more likely to survive over time, although its resemblance to the initial episode may diminish.
Retrieval is a reconstructive process—a more complex operation than simply activating the appropriate memory traces. Memories change every time you retrieve them. No two memories are alike because the process of calling them up and engaging them changes the content of the memory that you put back in storage. Each time you recall your last birthday, the details are slightly different—some are deleted and others are added. Consolidation occurs anew. The realization among neuroscientists that consolidation happens again, during and after retrieval, has been named the reconsolidation hypothesis. The basic idea is that memories are taken out of storage—retrieved—and then put back again—reconsolidated. Remembering is a mixture of the information previously stored in long-term memory with the information in an immediate situation.38
Whenever a memory—good or bad—is retrieved, the new information in the retrieval environment must be integrated into the existing, well-established network underlying that particular memory. When a memory is reconsolidated, new content is interleaved with existing content, leaving the reconsolidated memory embellished and changed. False memories are a striking example. Neuroscientists at the University of California, San Diego, questioned college students as to how they heard about the O. J. Simpson trial verdict, whether from radio, television, or a friend, three days after the announcement. Some of the students were tested again fifteen months later and gave relatively accurate answers; others were retested thirty-two months after and were less accurate despite being more confident in their answers. At both time periods, the students misremembered how they heard the information, supporting the view that consolidated memories are unstable and can be edited.39
In 1997, two neuroscientists at l’Université Pierre et Marie Curie in Paris found physiological evidence to show that memory can be changed. They trained rats on a maze that had eight arms radiating from a small central platform. At the end of three arms, they placed Cocoa Puffs cereal to entice the rats to choose those three arms and not the five unbaited ones. Once a day, the researchers placed each rat in the center of the maze and allowed it to visit the alleys as it wished. The trial ended after it visited the three alleys with the treats. After a few days, the rats learned this trick perfectly, entering the three alleys with the Cocoa Puffs immediately and not bothering with the five empty ones.40
To test the strength of the rats’ newly consolidated memories, the researchers reactivated the memory by allowing the animals to perform a single errorless trial. They then immediately injected the rats with Dizocilpine, a memory-degrading drug that blocked the activity essential for consolidating their experience on the single errorless trial. Twenty-four hours later, the rats did not remember which arms contained the Cocoa Puffs.
The researchers demonstrated reconsolidation using a clever experimental strategy. They gave the rats one additional test session, and that is all the practice they needed to regain their pre-drug proficiency in the maze. Their memory for the maze had been disrupted but not totally lost. The extra test session gave the rats information that they reconsolidated with the established memory, enabling them to perform once again at their pre-drug level.41
This experiment established that when a memory is reconsolidated, at least some of the cellular events that occurred during the initial consolidation are reenacted. The evidence that long-term memories are changed each time they are retrieved reinforces the view that memory is an ongoing dynamic process driven by life’s events. Further biological confirmation supports the hypothesis that recapitulation of a long-term memory makes it stronger and more stable.
Reconsolidation occurs constantly in your everyday life. Say that, after an absence of ten years, you visit the community where you grew up. It will strike you that what you see does not exactly match what you remember; the neighborhood may look, sound, and even smell different. As you retrieve long-term memory traces of your old stomping grounds, you update them by incorporating new features of the landscape and evaluating your roots from an adult perspective. In doing so, you are taking advantage of the lability of the old traces to establish more current and stronger memory traces. This updating process happens because a mismatch occurred between your old consolidated memory and the new information. You can adjust your opinion of people in the same way: a colleague who made a bad first impression may turn out to be a respected and valued partner.
The concept of reconsolidation was first proposed in the late 1990s, and it has the potential to bring about important breakthroughs, for instance in the treatment of post-traumatic stress disorder (PTSD). Many Iraq War veterans have been crippled by PTSD to the extent that they cannot resume their prewar lives; they persistently re-experience traumatic events, with symptoms including sleeplessness, irritability, anger, poor concentration, and constant anticipation of danger. A team of researchers has been testing a method to reduce the pain of PTSD without wiping out the memory of the distressing incident altogether. Their approach is to give patients Propranolol, a drug that dampens the activity of the sympathetic nervous system, the body’s tool for physically expressing emotion. This drug selectively blocks the reconsolidation process for the emotional content of the event, but not for the facts themselves. After even a brief treatment with Propranolol, patients may feel better; they can still recall the details of the traumatic events, but without extreme mental anguish. This work suggests that during retrieval and reconsolidation of a traumatic memory, Propranolol reduces activity in emotion-related areas, but does not interfere with function in the hippocampus where the basic facts are r
econsolidated.42
In our daily lives, memory for unique events, episodes, also benefits from reconsolidation. Reactivating one element of a complex episodic memory also reactivates many other memories linked to that event. For example, I remember that on my first day at MIT, the administrative officer in our department showed me how to use the Xerox machine. He told me to put my hand down on the glass and pressed a button, and after a few seconds, out came a piece of paper displaying an image of my hand. I was impressed and walked away from the marvelous machine with my first Xerox copy. When I called up my Xerox machine memory from 1964, I unleashed a barrage of recollections related to that day at MIT. If you frequently retrieve the details of a particular event in your life, like your first day on the job, then your memory of this event will be more reliable than if you had allowed time to pass without retrieving the event at all.
Studying how Henry forgot gave us a better understanding of how we remember. More than a hundred years ago, Müller and Pilzecker first proposed that memories consolidate over time and that partially consolidated memory traces are vulnerable. In 2004, a psychologist at the University of California, San Diego, took this observation further and proposed a coherent theory of forgetting based on converging evidence from three disciplines—psychology, psychopharmacology, and neuroscience. Forgetting occurs because we are constantly forming new memories that interfere with other memories still undergoing consolidation. During this probationary period, memories can be degraded by mental exertion of any sort, whether it is related or unrelated to the incompletely consolidated memories.43
Imagine a forty-five-year-old professional embarking on a one-week vacation at a tennis resort. On the long drive from his home, he thinks back to the meeting he had with his supervisor the afternoon before, in which she asked him to lead a new project in his area of expertise. Recalling her specifications, he starts to plan out his approach to the new project, making a mental outline of his proposal and then, as the miles fly by, pondering ways to implement each stage of the venture. At last, he arrives at the resort, and his mental work time comes to a halt. He is greeted by his hosts and immediately gets caught up in the fun. He sees unfamiliar faces, hears new names, and tries to absorb instructions about what he will do when, where, and with whom. He happily wheels his suitcase and tennis bag to his room, unpacks quickly, and heads for the pool. He settles into a different world and lets his everyday life fade. For the next seven days, he focuses on his forehand, two-handed backhand, serve, overheads, and net game. All the time, unbeknownst to him, his hippocampus has been actively recording the novel sights, sounds, conversations, and places, and creating a mental map of the resort. A week later, when he gets back in his car to drive home, he likely will have forgotten many details of his meeting with his supervisor and his brilliant plans for the new undertaking. The barrage of information from his vacation interrupted consolidation of the information he had encoded the week before. Newly acquired memories are easily altered, and interference from subsequent events can erase them partially or completely.
Forgetting with the passage of time is normal; remembering an event that happened ten years ago is more difficult than remembering one that happened last week. We forget the past because newer activities and thoughts push those old memories aside. Loss of memories over time may also occur when the initial episode was poorly encoded. Having Henry as a willing research participant gave my lab a unique opportunity to address these topics and determine whether damage to structures in the medial temporal lobe exacerbated forgetting.
In 1986, we designed a series of experiments with Henry, then fifty-eight years old, to clarify the circumstances under which forgetting occurs. Conventional wisdom held that amnesic patients forgot information more quickly than healthy individuals. To test this theory, we showed Henry and a control group 120 slides, each with a different complex, colorful magazine picture, depicting animals, buildings, interiors, people, nature, and single objects. Henry viewed each picture for twenty seconds and, as instructed, tried to remember them. In subsequent testing, we asked him to look at two pictures side by side, one studied and one new, and to say which one he thought he had seen before. We call this kind of memory retrieval recognition memory—consciously choosing between two possible answers, one of which is right.44
We tested Henry’s memory for the 120 pictures in four sessions. The tests occurred at different intervals after his initial exposure to all 120 pictures. Henry saw thirty pictures after ten minutes, a different thirty after one day, a different thirty after three days, and the final thirty after one week. For us to compare Henry’s rate of forgetting with that of the control participants, his scores after ten minutes had to be identical to theirs. We achieved this critical parity by allowing Henry, in the initial learning phase, to see the pictures for twenty seconds, whereas control participants had only one second to encode them, thus compensating Henry with nineteen extra seconds to encode each picture.
To everyone’s surprise, the information that Henry encoded in twenty seconds allowed him to recognize the pictures as well as, if not better than, the controls after one day, three days, and one week. More astonishing still was the discovery that Henry’s recognition memory was normal six months after his initial exposure to the pictures. Crucially, when compared with healthy adults of his age and education level, he did not show faster forgetting of complex, colorful pictures.45
How could Henry achieve normal picture-recognition scores when most other indices of his long-term memory pointed to failure? He remembered virtually nothing about his everyday life and bottomed out on every other test of new declarative memory. I was at a loss to explain the picture recognition result theoretically. Describing the experiment to colleagues, I speculated that Henry just had a vague feeling in the pit of his stomach that helped him decide in one of two ways. On each trial, he had to choose between two pictures—one that was in the study list and a new one. Thus, he based his response either on accepting one of the pictures as familiar or on rejecting one of the pictures as familiar and choosing the other one. This process occurred automatically and was based on the strength of the memory for one picture compared to another. We could not ask Henry to describe what he was thinking because constant interruptions would have compromised the experiment. But it is likely that Henry was automatically engaging cortical areas in the back of his brain that are known to have a massive storage capacity for visual material.
Around the time we were performing these tests, researchers in mathematical psychology capitalized on information-processing theory to shape a solid theoretical framework for recognition memory. This framework was rooted in a model that Richard Atkinson and James Juola proposed in 1974. Their long-term memory tasks required participants to memorize as many as sixty words. In the test that followed, participants saw one word at a time and had to decide whether the word was one they had memorized. If the participants answered quickly, their response, based only on a feeling of familiarity, they had a high probability of being wrong. If they took additional time, however, they had the opportunity to explicitly recall that the word was in the study list, with a high probability of their being correct. Based on their results, the researchers described the process of recognition in normal individuals as consisting of two independent retrieval processes. These ideas about recognition memory were formalized in 1980 when cognitive psychologist George Mandler introduced the now-popular dual-process model of recognition. He formalized two kinds of recognition memory—familiarity and recollection. Subsequent studies by another cognitive psychologist elucidated a fundamental distinction between these two uses of memory, consistent with the 1974 model—familiarity relies on fast, automatic processes, whereas recollection is a slower, intentional use of memory that demands a person’s attention.46
We have all had the experience of seeing someone on the street whose name we do not know, but with a vague sense that we have met him before. This kind of recognition is based on familiarity. It is not atten
tion demanding, and it happens automatically. Henry engaged this process when he identified the complex magazine pictures he had seen, a task with a relatively small cognitive demand. Conversely, when we meet an old friend on the street, we easily recall in great detail the good times we have shared. This kind of recognition is based on recollection, a process that demands effort and attention for searching our memory stores. Because this process depends on intact hippocampi, Henry was unable to engage it in his daily life and on most formal tests of his memory.
Henry’s unexpected test results in the 1980s revealed a crucial fact about the division of labor in the brain—familiarity and recollection are managed by independent processes in separate brain circuits, one preserved in Henry’s brain and the other destroyed. This observation was later clarified by behavioral evidence from hundreds of sources linking recollection with processes in the hippocampus and familiarity with processes in the perirhinal cortex. Although the hippocampus and perirhinal cortex are close, interconnected neighbors, they each make unique contributions to memory retrieval.47
This distinction helps us understand Henry’s ability to recognize complex pictures up to six months after he encoded them. His operation removed part but not all of his perirhinal cortex, so Henry possibly recognized the pictures in our experiments by engaging his partially spared perirhinal cortex in collaboration with other normal areas in his cortex. Still, this partnership was not sufficient to support his long-term memory in everyday life. Our picture-recognition finding was a remarkable exception, in stark contrast to his performance on other declarative memory tests where his recognition memory was consistently deficient.48
Permanent Present Tense Page 17