9.6 Memory as the Process of Remembering

There is more to memory than “memories.” Memory also refers to the processes of encoding, storing, and retrieving information, and it is to that aspect of memory that we now turn.

Encoding Information into Long-Term Memory

As you read a book, or attend to a conversation, or admire scenery, some of the sensory information that reaches your conscious mind enters your long-term-memory store, allowing you to recall it later. Why does some but not all of the information that reaches the short-term store get encoded into the long-term store?

As we discussed earlier, verbal information can be maintained in working memory simply by repeating it over and over. This is not, however, a good way to encode information into long-term memory. People who participate in a digit-span test, holding a list of digits in mind by reciting them over and over, rarely remember the digits even a minute after the test is over.

Thinking through the lines It Is not exactly clear how performers commit long passages to memory, but, as suggested by the thoughtful expression on this actor’s face, they apparently do much more than repeat the passages over and over. Instead, they likely engage in a form of elaborative rehearsal.

Keith Morris/Alamy

Everyday life also has many examples of the failure of repetition to promote long-term memory. Long ago a psychologist named Edmund Sanford (1917/1982) illustrated this point by describing his own poor memory of four short prayers that he had read aloud thousands of times over a 25-year period as part of his daily religious practice. He found, when he tested himself, that he could recite, on average, no more than three or four successive words of each prayer from memory before having to look at the printed text for prompting. I myself (Peter Gray) have looked up certain telephone numbers dozens of times and held them in working memory long enough to dial them, without ever encoding them into long-term memory. Yet I remember some other numbers that I have looked up only once or twice. The ones I remember are the ones I thought about in some way that helped me encode them into long-term memory.

Cognitive psychologists today distinguish between two kinds of rehearsal. Maintenance rehearsal is the process by which a person holds information in working memory for a period of time, and encoding rehearsal is the process by which a person encodes information into the long-term store. The activities that are effective for maintenance are not necessarily effective for encoding. Research suggests that some of the most effective activities for encoding involve elaboration, organization, and visualization.

Elaboration Promotes Encoding

Most of what we learn and remember in our everyday lives does not come from consciously trying to memorize. Rather, we remember things that capture our interest and stimulate our thought. The more deeply we think about something, the more likely we are to remember it later. To think deeply about an item is to do more than simply repeat it; it is to tie that item to a structure of information that already exists in long-term memory. Psychologists who study this process call it elaboration, or elaborative rehearsal. The immediate goal of elaboration is not to memorize but to understand, yet attempting to understand is perhaps the most effective of all ways to encode information into long-term memory.

Memory techniques centering on elaborative rehearsal capitalize on the human tendency to remember things that conform to some sort of logic, even if the logic is fictional. There is no obvious logic to the fact that stone formations hanging down in a cave are called stalactites while those pointing up are called stalagmites. But you can invent a logic: A stalactite has a c in it, so it grows from the ceiling; a stalagmite has a g, so it grows from the ground. Memory of a person’s name can be improved by thinking of a logical relation between the name and some characteristic of the person. Thus, you might remember Mr. Longfellow’s name by noting that he is tall and thin or, if he is actually short and stout, by recalling that he is definitely not a long fellow. Students may remember the name of a Professor Gray or Professor Brown by relating it to the color of the professor’s hair.

Laboratory Evidence for the Value of Elaboration In a classic experiment demonstrating the value of elaboration, Fergus Craik and Endel Tulving (1975) showed subjects a long series of printed words, one at a time, and for each word asked a question that required a different form of thought about the word. In some cases, the question was simply about the printing of the word (“Is it in capital letters?”). In other cases, the question asked about the word’s sound (“Does it rhyme with train?”). In still others, the question referred to the word’s meaning (“Would it fit in the sentence, The girl placed the__on the table?”). As you can see by looking at Figure 9.15, subjects remembered many more words when they had been asked questions that focused on meaning than they did in the other conditions.

Figure 9.15: Superior memory resulting from meaningful elaboration Subjects were shown a long sequence of words and, for each, were asked questions that required them to focus on the way the word was printed, how it sounded, or what it meant. The type of question dramatically affected the subjects’ later ability to recognize words as ones that had appeared in the sequence.

(Based on data from Craik & Tulving, 1975.)

Many subsequent experiments have confirmed the idea that thinking about meaning promotes long-term memory. In one experiment, for example, the best memory for a list of objects occurred in the group of subjects who were asked to think about how each object might help them to survive if they were stranded in the grasslands of a foreign country (Nairne et al., 2008). In another experiment, on memory for lines from a play, the best memory was shown by those who were asked to study the lines in the manner that professional actors study their lines—by thinking about the meanings that each line is meant to convey and how best to convey those meanings in reading the lines (Noice & Noice, 2006). In both of these experiments, those given the thought instructions performed better on a subsequent test of memory than did those who were asked to memorize the words or lines deliberately. This was despite the fact that the thought groups, unlike the memorize groups, did not know that the experiment was concerned with memory and that they would be tested later.

The Value of Elaboration for School Learning In a study of fifth graders, John Bransford and his colleagues (1982) found that students who received high marks in school were far more likely to use elaborative rehearsal than were those who received lower marks. The researchers gave the children written passages to study for a later test and asked them to explain what they did as they studied each passage. For example, one passage described two different kinds of boomerangs, a returning kind and a nonreturning kind, each used for different purposes. Academically successful students often reported that they rehearsed the material by asking themselves questions about it. They might wonder what a nonreturning boomerang looks like or why it is called a boomerang if it doesn’t return, and this caused them to think deeply about what a boomerang really is and about the information in the passage. Less successful students, in contrast, usually studied the passages simply by rereading them.

Bransford’s study was correlational in nature, so it does not prove that elaborative study caused better test performance; it shows only that the two tended to go together. But other research suggests that elaborative study can improve students’ grades. In one long-standing program aimed at helping students perform better in college, students are taught to write down questions about every textbook section that they read as they read it and about the lecture notes they take as they take them. The process of generating these questions and trying to answer them presumably deepens students’ thought about the ideas and facts that they are reading about or hearing and thereby improves both understanding and memory. In a series of field experiments, students who were taught these techniques subsequently achieved higher grades in their college courses than did otherwise comparable students who received either subject-matter tutoring or no special help (Heiman, 1987).

Such findings are compatible with the following advice for studying this or any other textbook:

• Don’t passively highlight or copy passages as you read, for later rereading. Focus on the ideas, not the author’s exact words.
• Constantly ask yourself questions such as these: Do I understand the idea that the author is trying to convey here? Do I agree with it? Is it relevant to my own life experiences? (Research has shown that people remember new information better when they relate it to themselves, Symons & Johnson, 1997). Has the author given evidence supporting it? Does the evidence seem reasonable? How is this idea relevant to the larger issues of the chapter? (In this textbook, the numbered focus questions that appear in the margins give you a start on this task.)
• As you ask such questions, jot down notes in the margins that bear on your answers, such as “This idea seems similar to …,” or “I don’t understand what he means by ….”

Through this active process, you will encode the material in a far richer and more lasting way than you could accomplish by simple rereading. In the process, you will also generate questions that you might want to ask other students or your instructor.

Organization Promotes Encoding

As a memory strategy, organization is closely tied to elaboration. Organizing items to be remembered is itself a means of elaboration; you must think about the items, not just repeat them, in order to organize them. Moreover, organization can improve memory by revealing or creating links among items that would otherwise be perceived as separate.

One way to increase the number of items one can retain in the short-term store is to group adjacent items that are at first perceived as separate, thus making them a single item. This procedure, known as chunking, decreases the number of items to be remembered and increases the amount of information in each item (Miller, 1956). As a simple illustration, if you had to memorize the series M D P H D R S V P C E O I H O P, your task would be made easier if you saw the series not as a string of 16 independent letters but as a set of five common abbreviations—M.D., Ph.D., RSVP, CEO, and IHOP. You could make your task still easier if you then chunked these five abbreviations into one sentence: “The M.D. and Ph.D. RSVPed to the CEO of IHOP.” In developing such a story, you would not only be chunking but also elaborating—making the information more meaningful by adding some new information of your own.

Beginning music students find it easier to remember the notes on the lines of the treble clef as one sentence, “Every Good Boy Does Fine,” than as the senseless string of letters E G B D F. Similarly, physiology students can recall the seven physiological systems (skeletal, circulatory, respiratory, digestive, muscular, nervous, and reproductive) by matching their first letters to the consonants in SACRED MANOR. Both devices involve chunking. In the first example, the five notes are chunked into one meaningful sentence, and in the second the seven systems are chunked into two meaningful words. By reducing the number of separate items, and by attaching more meaning to each item, chunking provides an advantage both for maintaining information in working memory and for encoding it into long-term memory.

The Role of Chunking in Expert Memory We are all much better at forming long-term memories for information that is within rather than outside our realm of expertise. For example, master chess players can look for just a few seconds at a chess game in progress and form a long-term memory of the locations of all the pieces on the board (de Groot, 1965). Similarly, football coaches have excellent memories for diagrams of football plays (Garland & Barry, 1991), architects have excellent memories for floor plans (Atkin, 1980), and physicians have excellent memories for information gained in diagnostic interviews of patients (Coughlin & Patel, 1987).

As a step toward explaining the expertise advantage in memory, K. Ander Ericsson and his colleagues have posited the existence of a special kind of long-term memory, called long-term working memory (Ericsson & Delaney, 1999; Ericsson & Kintsch, 1995). They conceive of this as memory for the interrelated set of items (such as a patient’s case history or the pieces on a chess board) that is crucial for solving the problem or completing the task at hand. Such memories are encoded into long-term storage in a manner that makes the entire structure of information easily accessible to working memory, at least until the problem is solved or the task is finished. Such memories allow a physician to puzzle over a particular patient’s symptoms as she drives home from work or a chess master to mull over the possibilities inherent in a particular set of chess positions while he is away from the game. Such memories are not lost as a result of interruptions, as short-term working memories are, so they allow the person to go back to a previous task after time spent on another task (Oulasvirta & Saariluoma, 2006).

Chunking plays a major role in the formation of long-term working memories. In order to form such a memory of, say, a particular arrangement of pieces on a chess board, a person must already have in long-term storage a great deal of well-established information about possible and likely ways that such items might be arranged. This knowledge provides a foundation for the efficient chunking of new items of information. Chess games normally progress in certain logical ways, so logical relationships exist among the pieces, which experts can chunk together and remember as familiar formations rather than as separate pieces. If the chess pieces are arranged randomly rather than in ways that could occur in a real game, masters are no better, or little better, than novices at remembering their locations (Gobet et al., 2001; see Figure 9.16). Experts in other realms also lose their memory advantage when information is presented randomly rather than being grouped in ways that make sense to them (Vicente & Wang, 1998; Weber & Brewer, 2003).

Figure 9.16: Chess memory The top drawings exemplify the types of chess layouts used in chess memory research. On the left is a board taken from a master’s game, and on the right is one with randomly arranged pieces. The graph shows the memory performance (averaged over 13 experiments) of chess players of various skill levels who studied either a game board or a random board for 10 seconds or less. The average number of pieces per board was 25. Skill level is measured in ELO points: 1,600 is Class B, 2,000 is Expert, and 2,200 is Master.

(With permission from Gobet, F., Lane, P. C. R., Croker, S., Cheng, P. C-H., Jones, G., Oliver, I., & Pine, J. M. (2001). Chunking mechanisms in human learning. Trends in Cognitive Sciences, 5, 236-243. Copyright © 2001, with permission from Elsevier.)

Hierarchical Organization Promotes Encoding The most useful format for organizing some kinds of information is the hierarchy. In a hierarchy, related items are clustered together to form categories, related categories are clustered to form larger (higher-order) categories, and so on.

In an experiment demonstrating the advantage of hierarchical organization for long-term memory, Andrea Halpern (1986) gave subjects a chart listing 54 well-known song titles to be memorized. In some cases the chart was organized in a hierarchical manner, with songs arranged according to meaningful categories and subcategories. In other cases a similar chart was used but organized randomly, with no systematic relation among categories, subcategories, and song titles. When tested later for their memory of the song titles, subjects who had studied the organized chart recalled accurately many more titles than did those who had studied the disorganized chart. During the test the former group of subjects would first recall a category name and then the songs that had been listed under that name.

As was pointed out in Chapter 1, the information in this textbook (like that in nearly all textbooks) is hierarchically arranged: Each main heading refers to a set of related ideas, and each subheading refers to a smaller subset of ideas within the larger set. An efficient way to summarize the information in almost any textbook chapter is to sketch it out in a manner that preserves the author’s hierarchical organization. As an illustration, a hierarchical sketch of the section you are now reading appears in Figure 9.17. Notice that the top node indicates in key words the theme of the section, the nodes subordinate to it indicate the main ideas pertaining to that theme, and the comments under them indicate the examples or evidence provided for each idea. You could summarize the whole chapter with six such sketches, one for each of the chapter’s main sections. Such a summary would be a far more efficient aid in helping you commit the information to memory for a test than would a string of terms and names that does not preserve the connections of ideas to each other or of ideas to evidence. The section reviews of this textbook also preserve the hierarchical organization of ideas and evidence within each section (as discussed in Chapter 1).

Figure 9.17: Hierarchical summary of a textbook section Summarized here are the ideas and evidence presented in this section (pp. 349-357) of the textbook. Such a chart is an aid to thought and memory because it reveals the logical connections among the items of information. Notice that some of the boxed items correspond to headings within the section.

Visualization Promotes Encoding

Our discussion of encoding so far has centered mostly on memory for verbal information. But we can also encode pictures or visual scenes into long-term memory, apparently in a nonverbal form, which can be recalled later into the visuospatial sketchpad of working memory. Visual and verbal memories interact and supplement one another in our everyday experience. If we asked you to describe your living room, you would probably summon a pictorial memory of that room and then find the words to describe it.

A good deal of research suggests that people can improve their memory for verbally presented information if they encode that information visually as well as verbally. One study, for example, demonstrated that memory for information in news stories improves if relevant pictures accompany the stories (Prabu, 1998). Lacking pictures, people can improve their memory for verbally presented information by constructing visual images to represent that information (Paivio, 1986).

Visualization may improve memory through several different means. It may provide a distinct visual memory trace to supplement the verbal memory trace, thereby increasing the chance that the memory will be recalled in the future (Paivio, 1986). It may also provide an efficient way to chunk several verbally presented ideas together. For example, a verbal description of a person you haven’t met may contain many separate items of information that you can combine into a single visual image. In addition, visual imagery may improve memory by linking newly learned information to information that is already well encoded in long-term memory. An example of that is found in a memory technique called the mental walk, used by nearly all contestants in the annual World Memory Championships when they are asked to memorize, in order, the names of a long list of objects (Maguire et al., 2003). As they hear the list of objects, they imagine themselves taking a walk along a familiar route and leaving each object next to a familiar landmark on the route. Then, during recall, they mentally walk the route again, inspect each landmark, and “see” each item that they had previously deposited there.

Memory Consolidation

According to the information-processing model we’ve been following in this chapter, once information has been attended to in short-term memory and encoded, it becomes part of permanent, long-term memory. However, the process is not quite as simple as that. Initial long-term memories are quite fragile and must be consolidated. As you’ll see, some memories quickly fade while others are strengthened or even modified over time.

Retrograde Amnesia as Evidence for Gradual Consolidation of Long-Term Memories

Earlier in this chapter we discussed the disorder called temporal-lobe amnesia, suffered by the much-studied patient H. M. As we noted then, destruction of the hippocampus resulted in an inability to encode new episodic memories, although such patients can still form new implicit memories. This most dramatic type of amnesia observed in H. M. and other patients with temporal-lobe damage is anterograde[ăn´-tƏ-rō-grād´] amnesia, the loss of capacity to form long-term memories of events that occur after the injury. However, these patients also show a considerable degree of retrograde amnesia, loss of memories of events that occurred before the injury.

Retrograde amnesia is generally time graded; it is greatest for memories acquired just before the injury and least for those acquired long before (Manns & Buffalo, 2012; Wixted, 2004, 2005). H. M., for example, lost all his memories of events that occurred within several days of his surgery, some of his memories of events that occurred within a few years of his surgery, and essentially none of his memories of events that occurred in his childhood, many years before the surgery (Eichenbaum, 2001). Such time-graded retrograde amnesia is also seen in people who have suffered a severe blow to the head.

The time-graded nature of retrograde amnesia suggests that long-term memories are encoded in the brain in at least two forms: a labile, easily disrupted form and a stable, not easily disrupted form. Long-term memories appear to be encoded first in the labile form. Then, gradually over time, they apparently are either re-encoded in the stable form or lost (forgotten). The process by which the labile memory form is converted into the stable form is referred to as consolidation.

A prominent theory today is that the labile form of long-term memory involves neural connections in the hippocampus and that the stable form involves neural connections in various parts of the cerebral cortex, without dependence on the hippocampus (Eichenbaum, 2001; Medina et al., 2008). This theory is supported by the time-graded retrograde amnesia observed after loss of the hippocampus, and it is also supported by neuroimaging research with people who have intact brains and normal memories. When people recall memories that were acquired relatively recently, neural activity in the hippocampus increases; but when they recall memories that were first acquired many years earlier, increased activity occurs in parts of the cerebral cortex but not in the hippocampus (Haist et al., 2001).

Role of Retrieval in Memory Consolidation and Modification

Nobody knows just how memory consolidation occurs, though it apparently involves modification of existing synaptic connections and growth of new synaptic connections in the brain. It is also not clear why some memories appear to consolidate relatively quickly (within days) and others much more slowly (over years) or why many memories are forgotten within minutes, hours, or days and never become consolidated. There is, however, evidence that memories that are recalled and used repeatedly, over relatively long time periods, are the ones most likely to be consolidated into a form that resists disruptions (Altmann & Gray, 2002).

Recent research, mostly with nonhuman animals, suggests that every time a memory is recalled and put to use, the neural trace for that memory enters temporarily into a new labile stage—that is, a stage when it can be modified (Lee, 2008; Tronson & Taylor, 2007). Depending on what happens when the memory is recalled, that memory may be strengthened, or weakened, or changed by the addition of some new content to it. From this point of view, long-term memories are not static, like words on a page, but are dynamic entities, changing in some ways every time they are used. Recent research using fMRI in humans has been able to observe brain-related changes when a memory is reactivated, and although this research has yet to show convincingly that reactivation of a memory enhances the strength of that memory, it provides a way of assessing such questions not easily afforded by behavioral methods (Levy & Wagner, 2013).

From a functional point of view, the modifiability of memories during retrieval makes sense. There is no need to clutter up my permanent collection of memories with information such as where I parked my car this morning—information that I will not need after the end of the day. It is better to reserve that memory space for information that might always be useful. Cues that my brain might use to consolidate memories into a very long-lasting form include the frequency and time course of my retrieving them. At the end of today, when I go to my car, I will retrieve for the last time the information about where I parked this morning. There may even be something about my mental attitude in retrieving it that signals my memory system that I won’t need this memory again, so it can fade. I will, however, continue for a long time to retrieve memories of more enduring value, such as information about what my car looks like, so those memories are likely to consolidate into forms that resist disruption and are not dependent on my hippocampus. Moreover, each time that I retrieve my memory of what my car looks like, in the presence of my car, that memory may be updated to include any modifications in the car’s appearance. Researchers are only barely beginning to learn about the kinds of cues that the long-term memory system uses to strengthen, weaken, or modify memories in useful ways.

The Role of Sleep in Memory Consolidation

Have you ever found that your memory for, and maybe even your understanding of, some newly learned information was better after a period of sleep than it was before? Many recent experiments have shown that sleep, shortly after learning, helps to consolidate newly acquired memories, making them more easily retrievable and less susceptible to disruption than they were before the sleep (Rasch & Born, 2008; Stickgold, 2005). A similar period of wakefulness does not have this effect. The effect seems to occur no matter what time of day the sleep occurs, as long as it occurs within a few hours after the learning experience. Some of these experiments used paired-associate tasks, in which subjects were presented with pairs of words and then were tested for their ability to recall the second member of each pair after seeing just the first member (Backhaus et al., 2008). But improved memory after sleep has been shown with many other tasks as well.

Recall from Chapter 6 that sleep occurs in a series of stages, each associated with certain brain wave patterns. For paired-associate tasks and other tasks involving conscious recall, the type of sleep that seems most valuable is slow-wave, non-REM sleep (Rasch & Born, 2008). The improved learning correlates positively with the amount of slow-wave sleep, not the amount of REM sleep. There is also evidence that the hippocampus becomes activated at various times during slow-wave sleep. One prominent theory is that the hippocampal activity represents activation of the memory trace, which allows consolidation of the memory into a new, more stable form.

Figure 9.18: Sleep after training improved insight Subjects were trained, in the morning or in the evening, to follow a seven-step procedure for solving a type of mathematical problem. Then, they were tested, either immediately after training or 8 hours after training. The crucial measure was the percentage who discovered an easier, two-step way to solve the problem (achieved “insight”). As the graph shows, the 8-hour interval increased insight only if subjects slept during that period. An 8-hour interval spent awake, whether it was during the day or during the night, did not increase insight.

(Based on data from Wagner et al., 2004.)

Sleep may improve not just the durability of new memories, but also their quality, in a manner that helps to achieve new insights. In one experiment demonstrating this, Ullrich Wagner and his colleagues (2004) trained people to solve a certain type of mathematical problem by following a series of seven steps. Unbeknownst to the subjects, the problems could also be solved by a simpler method, involving just two steps. In the experiment, the subjects were given a small amount of training on the task and then, 8 hours later, were tested with a long series of the same type of problems to see how quickly they could solve them. Some subjects were trained in the morning and then tested in the evening, after no sleep; others were trained in the evening and tested in the morning after a night’s sleep; and still others were trained in the evening but kept awake during the night before testing in the morning. For comparison, two other groups received their training and testing all in one block, occurring either in the morning after sleep or in the evening after a period awake.

The results of the study are shown in Figure 9.18. As you can see, the subjects who had a period of sleep between training and testing were more than twice as likely to discover the simpler way of solving the problem than were any of the other subjects. Notice that the improvement is not simply the result of sleep, but is the result of sleep occurring after initial training. Those who were both trained and tested in the morning, after sleep, did not do any better than any of the other groups. The experiment suggests that there is some validity to the adage that the best thing to do if you have a problem to solve is to “sleep on it”—but only if you have first spent some time working on the problem.