Psychology_ frontiers and applications-286-323.pdf

Final PDF to printer CHAPTER 8 CHAPTER OUTLINE Memory MEMORY AS INFORMATION PROCESSING A Three-Component Model Research Foundations: In Search of the Icon ENCODING: ENTERING INFORMATION Effortful and Automatic Processing Levels of Processing: When Deeper Is Better Exposure and Rehearsal Organiza...

Final PDF to printer CHAPTER 8 CHAPTER OUTLINE Memory MEMORY AS INFORMATION PROCESSING A Three-Component Model Research Foundations: In Search of the Icon ENCODING: ENTERING INFORMATION Effortful and Automatic Processing Levels of Processing: When Deeper Is Better Exposure and Rehearsal Organization and Imagery How Prior Knowledge Shapes Encoding STORAGE: RETAINING INFORMATION Memory as a Network Types of Long-Term Memory Why Do We Forget? Applications: Improving Memory and Academic Learning Amnesia Forgetting to Do Things: Prospective Memory MEMORY AS A CONSTRUCTIVE PROCESS Memory Distortion and Schemas The Misinformation Effect and Eyewitness Testimony The “Recovered Memory” Controversy: Repression or Reconstruction? Frontiers: How Accurate Are Young Children’s Memories? THE BIOLOGY OF MEMORY Sensory and Working Memory Long-Term Memory RETRIEVAL: ACCESSING INFORMATION The Value of Multiple and Self-Generated Cues The Value of Distinctiveness Context, State, and Mood Effects on Memory FORGETTING The Course of Forgetting HOW ARE MEMORIES FORMED? Synaptic Change and Memory Long-Term Potentiation Focus on Neuroscience: The Neuroscience of Retrieving Accurate and Fabricated Memories The charm, one might say, the genius of memory is that it is choosy, chancy, and temperamental. —Elizabeth Boren What are the issues here? What do we need to know? Where can we find the information to answer these questions? pas77416_ch08_266-303.indd 266 What were you doing during the afternoon of October 25, 2006? You probably cannot remember, unless it was an important date for you such as a birthday. But Aurelien Hayman (pictured) can tell exactly what he was doing on that date. He remembers what he what he had for lunch, who he was with, what he was wearing, what the temperature was, and if anything happened in the news that day. Aurelien, who was born in Cardiff, Wales, notes that his memory is somewhat limited. While he can tell you almost any fact about his life, his performance in university 15/11/13 10:15 AM Final PDF to printer does not seem to be helped at all. He notes that his memory is like a visual file drawer that he can access, but only if he experienced the events himself. There are only about 20 people in the world who are known to have this memory enhancement, including actress Marilu Henner. In almost every case, the superior memory started when they were about 14 years old. M tasks that he has never seen them before? In this chapter, we explore these and other fascinating questions about memory. emory refers to the processes that allow us to record and later retrieve experiences and information. Memory is precious and complex, as illustrated by the case of H.M. H.M. had most of his hippocampus and surrounding brain tissue surgically removed in 1953 to reduce severe epileptic seizures. The operation succeeded, but it unexpectedly has left H.M. with amnesia, or memory loss. He can discuss his childhood, teens, and early 20s, but has forgotten some events that occurred within the two years prior to surgery, and has lost the ability to form new memories. Typically, once an experience or fact leaves his immediate train of thought, he cannot remember it. Spend the day with H.M., depart and return minutes later, and he will not recall having met you. He forgets that he has recently eaten and reads magazines over and over as if he has never seen them before. What prevents H.M. from recalling new experiences, while leaving most of his pre-1953 memories intact? Why is it, as Figure 8.1 shows, that H.M. can learn and remember how to perform new tasks, yet swear each time he encounters these MEMORY AS INFORMATION PROCESSING Psychological research on memory has a rich tradition, dating back to late 19th century Europe, when Hermann Ebbinghaus (1885) studied the rate at which new information is forgotten and Sir Francis Galton (1883) investigated people’s memories for personal events. Decades later, the cognitive revolution within North American psychology and the advent of computers ushered in a metaphor that has influenced memory research since the 1960s: the mind as a processing system that encodes, stores, and retrieves information (Bower, 2000). Encoding refers to getting information into the system by translating it into a neural code that your brain processes. Encoding is a little like what happens when you type on a computer keyboard, as your keystrokes are translated into an electrical Number of errors per trial 40 Day 3 30 20 10 1 (a) Day 2 Day 1 1. In what ways is memory like an informationprocessing system? 5 11 1 5 Trials 10 1 5 9 (b) FIGURE 8.1 (a) On this complex task, participants trace a pattern while looking at its mirror image, which shows their hand moving in the direction opposite to its actual movement. (b) H.M.’s performance rapidly improved over time, indicating that he had retained a memory of how to perform the task. Yet, each time he performed it, he stated that he had never seen the task before, and had to have the instructions re-explained. Adapted from Milner, 1965. pas77416_ch08_266-303.indd 267 15/11/13 10:15 AM Final PDF to printer 268 CHAPTER EIGHT 2. What is sensory memory? How did Sperling assess the duration of iconic memory? code that the computer can understand and process. Storage involves retaining information over time. Once in the system, information must be filed away and saved, as happens when a computer stores information on a hard drive. Finally, there must be a way to pull information out of storage when we want to use it, a process called retrieval. On a computer, retrieval occurs when you give a software command (e.g., “Open File”) that transfers information from the hard drive back to the screen where you can view it. Keep in mind, however, that this analogy between human and computer is crude. For one thing, we routinely forget and distort information, and may “remember” events that never occurred (Laney & Loftus, 2010; Morris et al., 2006; Pickrell et al., 2003). Human memory is highly dynamic, and its complexity cannot be fully captured by any existing information-processing model. Encoding, storage, and retrieval represent what our memory system does with information, and they could not take place without memory having some type of organization or structure. Thus, before exploring these processes in more detail, let us examine some basic components of memory. A Three-Component Model 3. Describe the limitations of shortterm memory, and how they can be overcome. Our encounter with H.M. suggests an interesting possibility regarding how memory might be organized. If you told H.M. your name or read him a series of numbers, he could recall the information for a short time. Yet he could not form a lasting memory; once his train of thought changed, that information would be lost forever. Could it be, as William James (1890) suggested long ago, that memory has distinct yet interacting components, one temporary and the other more long-lasting? The model shown in Figure 8.2 incorporates this assumption. Originally developed by Richard Atkinson and Richard Shiffrin (1968), and subsequently modified, it proposes that memory has three major components: sensory memory, shortterm or “working” memory, and long-term memory. The model does not assume that each component corresponds to a specific structure within the brain. Rather, the components may involve interrelated neural sites, and memory researchers use these terms in a more abstract sense. Sensory Memory Sensory memory holds incoming sensory information just long enough for it to be recognized. It is composed of different subsystems, called sensory registers, which are the initial information processors. Our visual sensory register is called the iconic store, and in 1960, George Sperling conducted a classic experiment to assess how long it stores information (see this chapter’s Research Foundations feature). As Figure 8.3 illustrates, the time course for visual sensory memory is very brief. Indeed, it is difficult, perhaps impossible, to retain complete information in purely visual form for more than a fraction of a second (Figure 8.4; Barsalou, 1992). The auditory sensory register, called the echoic store, is studied by asking participants to recall different sets of numbers or letters that are simultaneously presented to their left and right ears via headphones. Echoic memory lasts longer than iconic memory. A nearly complete echoic trace may last about two seconds and a partial trace may linger for several more (Winkler et al., 2002). Short-Term/Working Memory Because our attentional capabilities are limited, most information in sensory memory simply fades away. But through selective attention, a small portion enters short-term memory, which holds the information that we are conscious of at any given time. Short-term memory also is referred to as working memory, because it consciously processes, codes, and “works on” information (Atkinson & Shiffrin, 1968; Baddeley, 2003). Rehearsal Encoding Sensory input Sensory registers Encoding Attention Working (short-term) memory Long-term memory Retrieval FIGURE 8.2 In this model, memory has three major components: (1) sensory registers, which detect and briefly hold incoming sensory information; (2) working memory, which processes certain information received from the sensory registers and information retrieved from long-term memory; and (3) long-term memory, which stores information for longer periods of time. Adapted from Atkinson & Shiffrin, 1968. pas77416_ch08_266-303.indd 268 15/11/13 10:15 AM Final PDF to printer Memory 269 Research Foundations IN SEARCH OF THE ICON Discussion Introduction Sperling argued that some kind of memory trace must remain after the visual stimulus is removed. This trace is very shortlived (less than one second) but is available for scanning and, thus, any line in the matrix can be accurately recalled. However, when asked for a total report (recall as many letters as possible without the tone cue), the trace has faded by the time one line is reported. This memory trace (referred to as an icon; Neisser, 1967) is a purely visual representation of the stimulus array. It is subject to interference by additional visual information, and its strength is affected by visual factors such as contrast and intensity. This notion of a sensory storage mechanism was quickly integrated into many models of memory and Sperling’s 1960 paper remains one of the most cited studies in psychology. How does information from some sensory input get translated into memory? Are we able to attend to all the information or is only some of it available? These questions were of central importance in George Sperling’s pioneering work on iconic memory. Method Sperling (1960) had participants view matrices of letters such as the one shown in Figure 8.3. The matrix was presented for a very brief time (about 50 milliseconds). When asked to report what they had seen, participants could, on average, correctly identify only 4.5 letters (typically from the first row). Even if the presentation time was increased to 500 milliseconds or the number of letters was reduced, the results remained the same. Thus, it would appear that the memory span for a visual stimulus was quite limited—only about 33 percent of the display could be reported. Sperling devised a method of partial report to demonstrate that much more information was actually available. In Study 2, the same matrices were presented, but when the visual stimulus was removed a tone was presented. For a high-pitched tone, participants were to report the letters in the first row. If the tone was low-pitched, the bottom row was to be reported. A medium-pitched tone called for a report of the middle row. Results Results indicated that approximately 75 to 90 percent of the letters could be correctly reported, regardless of the line they appeared in. Since the pitch of the tone was determined randomly, participants could not predict which line they needed to attend to until the stimulus display was gone. Design Question: Can participants report the letters from each row of a brief visual display if you cue the row for them to attend to? Type of Study: Experimental Independent Variable Dependent Variable Tone, three types • low-pitched • medium-pitched • high-pitched Number of letters reported Source: George Sperling (1960). The information available in brief visual presentations. Psychological Monographs: General and Applied, 74(11), 1–30. Fixation Display (1/20 s) plus tone Report S F C B Pitch signals row to report High D L H P Medium A K R G Low DLHP FIGURE 8.3 After a participant fixates on a screen, a matrix of letters is flashed for 1/20 of a second. In one condition, participants do not hear any tone and must immediately report as many letters as they can. In another condition, a high-, medium-, or low-pitched tone signals the participant to report the top, middle, or bottom row. If the tone occurs immediately, participants typically can report three or all four letters, no matter which row is signalled. pas77416_ch08_266-303.indd 269 15/11/13 10:15 AM Final PDF to printer 270 CHAPTER EIGHT (Conrad, 1964). Likewise, given word lists such as (1) man, mad, cap, can, map; (2) old, late, thin, wet, hot; and (3) big, huge, broad, long, tall, people become most confused recalling the first list, in which the words sound similar (Baddeley, 1966). Such findings suggest that phonological codes play an important role in short-term memory. FIGURE 8.4 The arc of light that you see traced by a fiery baton, or the lingering flash that you see after observing a lightning bolt, results from the brief duration of information in iconic memory. Because of a slow camera shutter speed, this photo captures more arcs of light than you could actually see: Because your iconic memory stores complete information for only a fraction of a second, the image would quickly vanish. Memory codes. Once information leaves sensory memory, it must be represented by some type of code if it is to be retained in short-term and eventually long-term memory. For example, the words that someone just spoke to you (“please buy some gum”) or the phone number that you just looked up must somehow become represented in your mind. Such mental representations, or memory codes, can take various forms (Jackendoff, 1996). We may try to form a mental image (visual encoding), code something by sound (phonological encoding), or focus on the meaning of a stimulus (semantic encoding). For physical actions, such as learning sports or playing musical instruments, we code patterns of movement (motor encoding). Study of memory codes and their underlying neural mechanisms may provide a key to understanding how the brain represents and makes sense of information received through the senses (Tsien, 2007). Note that the form of a memory code often does not correspond to the form of the original stimulus. For example, as you read these words (visual stimuli) you probably are not storing images of the way the letters look. Rather, you likely are forming phonological codes (saying the words silently to yourself) and, as you think about the material, semantic codes that represent their meaning (Lee, 2009). When people are presented with lists of words or letters and asked to recall them immediately, the errors that they make often are phonetic. They might recall a V instead of a B because of the similarity in how the letters sound pas77416_ch08_266-303.indd 270 Capacity and duration. Short-term memory can hold only a limited amount of information at a time. Depending on the stimulus, such as numbers, letters, or words, it is believed that most people can hold no more than five to nine meaningful items in short-term memory, leading George Miller (1956) to set the capacity limit at “the magical number seven, plus or minus two,” though others suggest that the number may in fact be as few as four (Cowan, 2001). To demonstrate this, try administering the digit-span task in Table 8.1 to some people you know. If our short-term memory capacity is so limited, how can we remember and understand sentences as we read? To answer this, read the line of letters below (about one per second), and then cover it up and write down as many letters as you can remember in the order presented. BIRCYKAEUQSASAWTI Did you have trouble remembering even half of these 17 letters in order? Now we rearrange (reverse) the letters and again ask you to write them down in order. Here are the 17 letters: “It was a squeaky crib.” No doubt, you find this task much easier. The TABLE 8.1 Digit-Span Test Directions: Starting with the top sequence, read these numbers at a steady rate of one per second. Immediately after saying the last number in each series, signal the person to recall the numbers in order. Most people can recall a maximum sequence of five to nine digits. 8352 43931 714937 5469236 15248584 932658214 6813194735 42469521743 379846172495 15/11/13 10:15 AM Final PDF to printer Memory limit on short-term memory capacity concerns the number of meaningful units that can be recalled, and the original 17 letters have been combined into five meaningful units (words). Combining individual items into larger units of meaning is called chunking, and it can greatly aid recall. Short-term memory is limited in duration as well as capacity. Have you ever experienced rapid forgetting, such as being introduced to someone, starting a conversation, and then suddenly realizing that you don’t have the foggiest idea what her or his name was? Without rehearsal, the “shelf-life” of information in short-term memory is indeed short, perhaps lasting about 20 seconds. Lloyd and Margaret Peterson (1959) demonstrated this by presenting participants with three-letter syllables (all consonants), such as BSX, followed by a three-digit number, such as 140. Upon seeing the number, participants counted backwards by threes, which prevented them from rehearsing the letters. As Figure 8.5 indicates, after counting backwards for as little as 18 seconds, few syllables were recalled. By rehearsing information, we can extend its duration in short-term memory indefinitely. This occurs when you look up a telephone number and keep saying it to yourself, either out loud or silently, while waiting to use a phone. This simple repetition of information is called maintenance rehearsal. In Percentage of syllables correctly recalled 100 80 60 40 20 3 6 9 12 15 Retention interval (seconds) 18 FIGURE 8.5 Participants who were prevented from rehearsing three-letter syllables in working memory showed almost no recall of the letters within 18 seconds, illustrating the rapid forgetting of information in short-term memory. Based on Peterson & Peterson, 1959. pas77416_ch08_266-303.indd 271 271 contrast, elaborative rehearsal involves focusing on the meaning of information or relating it to other things we already know. Thus, you could rehearse the term iconic memory by thinking about examples of iconic memory in your own life. Both types of rehearsal keep information active in short-term memory, but elaborative rehearsal is more effective in transferring information into long-term memory, which is our more permanent memory store (Gardiner et al., 1994; Mäntylä, 1986). Putting short-term memory “to work.” Picture the seemingly endless stacks of a library (representing long-term memory) and a tiny loading platform (representing short-term memory) outside the building. New books (pieces of information) rapidly arrive and, because there isn’t enough space, knock other ones off the platform. According to the original three-stage model, items that remain on the short-term loading dock long enough—such as through maintenance rehearsal—eventually get transferred into the long-term library. The original three-stage model of memory focused on short-term memory primarily as a loading platform or holding station for information along the route from sensory to long-term memory. Many cognitive scientists now reject this view of short-term memory as too passive and too sequential. Instead, they view short-term memory as a working memory—a “mental workspace” that actively and simultaneously processes different types of information and supports other cognitive functions, such as problem solving and planning, and interacts with long-term memory (Baddeley, 2010). Metaphorically, rather than a loading platform, working memory “is instead more like the office of a busy librarian, who is energetically categorizing, cataloging, and cross-referencing new material” (Reisberg, 1997, p. 139). To illustrate how working memory stores information, processes it, and supports problem solving, add the numbers 27 and 46 “in your head.” Your working memory stores the numbers, calls up information from long-term memory on “how to add,” keeps track of the interim steps (7 + 6 = 13, carry the 1), and coordinates these mental processes. One model, proposed by Alan Baddeley (1998, 2007; Repous & Baddeley, 2006), divides working memory into four components. First, we maintain some information in an auditory working memory (the “phonological loop”), such as when you repeat a phone number, name, or new vocabulary terms to yourself mentally. A second component, 4. Why do researchers refer to short-term memory as working memory? 5. Identify three components of working memory. 15/11/13 10:15 AM Final PDF to printer 272 CHAPTER EIGHT visual-spatial working memory (the “visuospatial sketchpad”), allows us to temporarily store and manipulate images and spatial information, as when forming mental maps of the route to some destination. A third component, the episodic buffer, provides temporary storage space where information from long-term memory and from the phonological loop and/or visuospatial subsystems can be integrated, manipulated, and made available for conscious awareness. For example, after reading or hearing someone say, “How much is 87 plus 36?” your phonological loop initially maintains the acoustic codes for the sounds of 87 and 36 in working memory. Your visuospatial sketchpad also might maintain a mental image of the numbers. But to do this task, the rules for performing addition must be retrieved from long-term memory and temporarily stored in your episodic buffer, where they are integrated (i.e., applied to) information from the phonological and visuospatial subsystems. This creates the ingredients for the conscious perceptions that you experience as you perform the mental addition (e.g., “7 + 6 = 13, carry the 1 . . .”). The episodic buffer also comes into play when you chunk information. Finally, a control process, called the central executive, directs the action. It decides how much attention to allocate to mental imagery and auditory rehearsal, calls up information from long-term memory, and integrates the input. Research suggests that the prefrontal cortex, the seat of “executive functions” described in Chapter 3, is heavily involved in directing the processing of information in working memory (Nelson et al., 2000; Tsujimoto et al., 2004). Long-Term Memory As already noted, long-term memory is our vast library of more durable stored memories. Perhaps there have been times in your life, such as periods of intensive study during finals, when you have felt as if “the library is full,” with no room for storing so much as one more new fact inside your brain. In reality, barring brain damage, we remain capable of forming new long-term memories until we die. And, as far as we know, long-term storage capacity essentially is unlimited. Once formed, a long-term memory can endure for up to a lifetime (Bahrick et al., 1994). Are short-term and long-term memory really distinct? Case studies of amnesia victims, such as H.M., support this distinction, but another source of evidence comes from laboratory experiments in which participants with normal memory learn lists of words. Suppose that we present you with a series of unrelated words, one word at a time. The list might contain 10, 15, 20, or even 30 items. Immediately after the last word is presented, you will recall as many words as you can, in any order you wish. As Figure 8.6 illustrates, most experiments find that words at the end and beginning of the list are the easiest for participants to recall. This U-shaped 80 Tested immediately Test delayed by 30 seconds Primacy 70 Proportion correct Recency 60 50 40 30 No recency 20 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Position in original list FIGURE 8.6 Immediate recall of word lists produces a serial position curve, in which primacy and recency effects are both evident. However, even a short delay of 30 seconds in recall (during which rehearsal is prevented) eliminates the recency effect, indicating that the later items in the word list have disappeared from short-term memory. Adapted from Glanzer & Cunitz, 1966. pas77416_ch08_266-303.indd 272 15/11/13 10:15 AM Final PDF to printer Memory pattern is called the serial position effect, meaning that recall is influenced by a word’s position in a series of items. The serial position effect has two components, a primacy effect, reflecting the superior recall of early words, and a recency effect, representing the superior recall of the most recent words. What causes the primacy effect? According to the three-stage model, as the first few words enter short-term memory, we can quickly rehearse them and transfer them into long-term memory. However, as the list gets longer, short-term memory rapidly fills up, and there are too many words to keep repeating before the next word arrives. Therefore, beyond the first few words, we cannot rehearse the items and they are less likely to get transferred into long-term memory. If this hypothesis is correct, then the primacy effect should disappear if we can prevent people from rehearsing the early words, say by presenting the list at a faster rate. Indeed, this is what happens (Glanzer, 1972). As for the recency effect, the last few words have the benefit of not being “bumped out” of short-term memory by any new information. Thus, if we try to recall the list immediately, all we have to do is “read out” the last words while they linger in shortterm memory. In sum, according to the three-stage model, the primacy effect is due to the transfer of early words into long-term memory, whereas the recency effect is due to short-term memory. If this explanation is correct, then we should be able to wipe out the recency effect—but not the primacy effect—by eliminating the last words from short-term memory. This happens when the recall test is delayed, even by as little as 15 or 30 seconds, and you are prevented from rehearsing the last words. To prevent rehearsal, we might briefly ask you to count a series of numbers immediately after presenting the last word (Glanzer & Cunitz, 1966; Postman & Phillips, 1965). Now, by the time you try to recall the last words, they will have faded from short-term memory and been “bumped out” by the arithmetic task (six . . . seven . . . eight . . . nine . . .). Figure 8.6 shows that, indeed, under these delayed conditions, the last words are recalled no better than the middle ones, while a primacy effect remains. Having examined some of the basic components of memory, let us now explore more fully how information is encoded into long-term memory, how it is stored, and factors that affect our ability to retrieve it. you wish to retrieve it. In a library, new material is assigned a call number before it is placed in storage. As noted earlier, our “call numbers” come in various forms—semantic, visual, phonological, and motor codes—that later enable us to activate information in long-term memory and access it. The more effectively we encode material into long-term memory, the greater the likelihood of retrieving it (Van Overschelde et al., 2005). Effortful and Automatic Processing Think of the parade of information that you have to remember: names, phone numbers, computer passwords, and mountains of schoolwork on which you expect to be tested. Learning such information involves effortful processing, encoding that is initiated intentionally and requires conscious attention. Rehearsing, making lists, and taking class notes illustrate effortful processing. In contrast, have you ever been unable to answer an exam question, and said to yourself, “Why can’t I answer this? I can even picture the diagram; it was on the upper portion of the left page”? Here incidental information about the diagram’s location on the page (that you were not trying to learn) appears to have been transferred into long-term memory through automatic processing, encoding that occurs without intention and requires minimal attention. Information about the frequency, spatial location, sequence, and timing of events often is encoded automatically (Gallivan et al., 2009). For example, if someone asks you what you did yesterday, you probably will have little trouble remembering your sequence of activities, despite the fact that you never had to sit down and intentionally memorize this information. Some processes (e.g., reading) are so automatic that we have difficulty switching to a more effortful style. 6. What is the serial position effect? Under what conditions do primacy and recency effects occur? 7. According to the three-component model, why do primacy and recency effects occur? 8. Provide some examples of effortful and automatic processing in your own life. Levels of Processing: When Deeper Is Better Imagine that you are participating in a laboratory experiment and are about to be shown a list of words, one at a time. Each word will be followed by a question, and all you have to do is answer “Yes” or “No.” Here are three examples: ENCODING: ENTERING INFORMATION 1. POTATO “Is the word in capital letters?” 2. horse “Does the word rhyme with course?” 3. TABLE “Does the word fit in the sentence, ‘The man peeled the _____’?” The holdings of your long-term memory, like those of a library, must be organized in terms of specific codes if the information is to be available when Each question requires effort but differs from the others in an important way. The first question requires superficial structural encoding, since you pas77416_ch08_266-303.indd 273 273 15/11/13 10:15 AM Final PDF to printer 274 CHAPTER EIGHT 9. Explain the concept of “depth of processing.” have to notice only how the word looks. Question 2 requires a little more effort. You must engage in phonological (also called phonemic) encoding by sounding out the word to yourself and then judging whether it matches the sound of another word. The last question requires semantic encoding, because you must pay attention to what the word means. In this experiment, every word shown to you will be followed by a question similar to one of these. Unexpectedly, you will then be given a memory test. Which group of words will be recognized most easily: those processed structurally, phonologically, or semantically? According to the levels of processing concept developed by Fergus Craik and Robert Lockhart (1972, 2008) of the University of Toronto, the more deeply we process information, the better it will be remembered. In the study just mentioned, semantic encoding involves the deepest processing because it requires us to focus on the meaning of information. Merely perceiving the structural properties of the words (e.g., capitalized versus lowercase) involves shallow processing, and phonemically encoding words is intermediate. You can see in Figure 8.7 that the results of a study conducted by Craik and Endel Tulving (1975) in Toronto support the value of deeper, semantic encoding. Although many experiments have replicated this finding (Gabrieli et al., 1996), at times the concept Percentage correctly recognized 90 80 70 60 50 40 10. How effectively do maintenance and elaborative rehearsal process information into long-term memory? Structural Phonemic Semantic (shallow) (deeper) (deepest) Type/depth of encoding FIGURE 8.7 Depth of processing facilitates memory. Participants were shown words and asked questions that required superficial structural processing of a word, somewhat deeper phonemic processing, or deeper semantic processing. Depth of processing increased later recognition of the words in a larger list. Data from Craik & Tulving, 1975. pas77416_ch08_266-303.indd 274 of “depth of processing” can be difficult to measure. Suppose that some randomly assigned students study a chapter by creating hierarchical outlines and notes. A second group creates flash cards, jumbles them up, and rehearses them. Which study method represents deeper processing? If the first group performs better on a test, should we assume that they must have processed the information more deeply? To do so, warns Alan Baddeley (1990), is to fall into a trap of circular reasoning. Still, the levels of processing model has generated much research (Craik, 2002; Froger et al., 2008). Then again, there are situations in which few would argue with at least a broad distinction between shallow and deep processing. The following section discusses one of them. Exposure and Rehearsal Years ago a student came into my (M.W.P.) office after failing the first exam in introductory psychology. He told me he had been to all the lectures, completed the chapters ahead of time, and reread each chapter twice more just before the exam. Yet when I looked through his textbook, not a word or sentence had been underlined or highlighted. I asked whether he took notes as he read or paused to reflect on the information, and he said, “No.” Instead, he read each chapter quickly, much like a novel, and assumed that merely by looking at everything three times the information would somehow “sink in.” Unfortunately, this student’s approach stood little chance of success. To learn factual and conceptual information presented in most academic or job settings, we need to employ effortful, deep processing. Simple repeated exposure to a stimulus without stopping to think about it represents shallow processing. To demonstrate this, try drawing from memory a picture of a Canadian penny, accurately locating all the markings. Few of our students can do this. Thus, even thousands of shallow exposures to a stimulus do not guarantee long-term retention (Jones, 1990; Nickerson & Adams, 1979). Rehearsal goes beyond mere exposure because we are thinking about the information. Of course, not all thinking is created equal, and neither is all rehearsal. As noted earlier, maintenance rehearsal involves simple repetition, as when silently repeating an unfamiliar phone number while waiting to use the phone. Maintenance rehearsal is most useful for keeping information active in short-term, working memory, and it may help to transfer some information into long-term memory (Naveh & Jonides, 1984; Wixted, 1991). However, it is an inefficient method for bringing about long-term transfer. 15/11/13 10:15 AM Final PDF to printer Memory In contrast, elaborative rehearsal focuses on the meaning of information—we elaborate on the material in some way. Organizing information, thinking about how it applies to our own lives, and relating it to concepts or examples we already know illustrate such elaboration. According to Craik and Lockhart (1972, 2008), elaborative rehearsal involves deeper processing than maintenance rehearsal and should be more effective in transferring information into long-term memory. In contexts as varied as university students learning word lists to Grade 6 students learning CPR (cardiopulmonary resuscitation), experiments support the greater effectiveness of elaborative rehearsal (Gardiner et al., 1994; Mäntylä, 1986; Rivera-Tovar & Jones, 1990). Even thinking about examples of concepts that other people provide for us facilitates later recall (Palmere et al., 1983). Psychologists K. Anders Ericsson and Peter Polson (1988), who studied J.C., found that he invented an overall organizational scheme to aid his memory. He divided his customers’ orders into four categories (entrees, temperatures, side dish, dressing) and then used a different system to encode the orders in each category. For example, he represented dressings by their initial letter, so orders of Thousand Island, oil and vinegar, blue cheese, and oil and vinegar would become TOBO. Imposing organization on a set of stimuli is an excellent way to enhance memory. An organizational scheme can enhance the meaningfulness of information and also serve as a cue that helps to trigger our memory for the information it represents, just as the word TOBO jogs J.C.’s memory of the four orders of salad. Organization and Imagery Organizing material in a hierarchy takes advantage of the principle that memory is enhanced by associations between concepts. Gordon Bower and his colleagues (1969) demonstrated this experimentally by presenting some participants with a logically organized list of words, based on a hierarchical tree like the one in Figure 8.8a. Other participants received the same words placed randomly within the tree. As Figure 8.8b shows, participants presented with a meaningful hierarchy remembered more than three times as many words. Notice that the hierarchy in Figure 8.8a does not reduce the amount of information to be 275 Hierarchies and Chunking Dining at the restaurant where J.C. is a waiter can be an awe-inspiring experience. Perhaps you would like a filet mignon, medium-rare, with a baked potato, and Thousand Island dressing on your salad? Whatever you choose, it represents only one of over 500 possible options that can be ordered (seven entrees × five serving temperatures × three side dishes × five choices of salad dressing). Yet, you and 20 or so of your best friends can place your selections with J.C., and he will remember them perfectly without writing them down. How does he do it? 11. Why do hierarchies, chunking, mnemonic devices, and imagery enhance memory? 100 Encoding 90 Automatic processing Effortful processing Elaborative rehearsal (deeper processing) Maintenance rehearsal (shallower processing) Percentage recalled 80 70 60 50 40 30 20 10 Links to your life and existing knowledge Meaning of information Organization (a) Imagery 0 Random hierarchy Meaningful hierarchy (b) FIGURE 8.8 Words presented in a logically organized hierarchical structure (a) are remembered better than the same words placed randomly in a similar-looking structure (b). Source: Bower et al., 1969. pas77416_ch08_266-303.indd 275 15/11/13 10:15 AM Final PDF to printer 276 CHAPTER EIGHT remembered. With or without it, there are the same number of words to learn. Rather, a logical hierarchy enhances our understanding of how these diverse elements are related, and as we proceed from top to bottom, each category can serve as a cue that triggers our memory for the associated items below it. Because the hierarchy has a visual organization, there also is a greater possibility of using imagery as a supplemental memory code. Chunking refers to combining individual items into a larger unit of meaning, and it widens the information-processing bottleneck caused by the limited capacity of short-term memory (Gobet et al., 2001; Miller, 1956). To refresh your memory, read the line of letters below to yourself (about one per second) and try to recall as many as you can, in the same sequence. CTVYMCAIBMKGBFBI If you remembered four to eight of the letters in order, you did quite well. Now we can reorganize these 16 individual bits of information into five larger, more meaningful chunks: CTV, YMCA, IBM, KGB, and FBI. This rearrangement is easier to keep active in short-term memory and, should you be so motivated, to rehearse and transfer into long-term memory. A common example of chunking in everyday life is the way we encode and later retrieve phone numbers from long-term memory. Thus, if you periodically call someone who lives far away, you probably encode the number as a set of three chunks (e.g., 905-430-5147) rather than as 10 individual numbers. Mnemonic Devices The search for memory aids dates back thousands of years. In fact, the term mnemonics (neMON-iks), which refers to “the art of improving memory,” derives from the name Mnemosyne, the Greek goddess of memory. A mnemonic device is any type of memory aid. Hierarchies and chunking represent two types of mnemonic devices. So do acronyms, which combine one or more letters (usually the first letter) from each piece of information you wish to remember. For example, many students learn the acronyms HOMES and ROY G. BIV to help remember the names of the five Great Lakes of North America (Huron, Ontario, Michigan, Erie, Superior) and the hues in the visible spectrum—the “colours of the rainbow” (red, orange, yellow, green, blue, indigo, violet). Acronyms are one of the most popular mnemonic techniques among university students (Soler & Ruiz, 1996; Manolo, 2002). Keep in mind that when you are learning new material, mnemonic devices do not reduce the pas77416_ch08_266-303.indd 276 amount of raw information you have to encode into memory. Rather, they reorganize information into more meaningful units and provide extra cues to help you retrieve information from long-term memory. When chunking seven digits into 4305147, you still have to encode seven digits. And the acronym HOMES is useful only when you have also encoded the names of the Great Lakes into longterm memory. Thus, some researchers argue that acronyms—DAM—don’t aid memory, or at least do so only when you are already familiar with the material (Carney et al., 1981, 1994). Visual Imagery How many windows are there in your home? Can you tell us, in as much detail as possible, what your bedroom looked like during your high school years? To answer these questions, you might try to construct and scan a series of mental images in your working memory, based on information that you draw out of long-term memory. Allan Paivio (1969, 2006) proposes that information is stored in long-term memory in two forms: verbal codes and non-verbal (typically visual) codes. According to his dual coding theory, encoding information using both codes enhances memory, because the odds improve that at least one of the codes will be available later to support recall. In short, two codes are better than one, though dual coding is harder to use with some types of stimuli than others. Try to construct a mental image for each of the following: (1) fire truck, (2) light bulb. Now construct an image for these words: (1) jealousy, (2) knowledge. You probably found the second task more difficult, because the latter words represent abstract concepts rather than concrete objects (Sadoski et al., 1997). Abstract concepts are easier to encode semantically than visually. Memory improvement books often recommend using imagery to dual-code information, and research supports this approach (Tye, 1991). The ancient Greeks developed an effective and wellknown imagery technique called the method of loci (loci is Latin for “places”). To use this technique, imagine a physical environment with a sequence of distinct landmarks, such as the rooms in a house or places on your campus. In one psychology class, students rapidly learned to use the 40 locations on the Monopoly game board as their visual reference (Schoen, 1996). To remember a list of items or concepts, take an imaginary stroll through this environment and form an image linking each place with an item or a concept. To remember the components of working memory, you might imagine walking into the administration building 15/11/13 10:15 AM Final PDF to printer Memory (central executive), then watching a band rehearsal in your gym (phonological loop), visiting an art class (visuospatial sketchpad), and finally, the offices of the campus newspaper (episodic buffer). Many studies support the method of loci’s effectiveness (Massen et al., 2009). How Prior Knowledge Shapes Encoding Long-term memory is densely populated with semantic codes that represent the meaning of information. Typically, when we read, listen to someone speak, or experience some other event, we do not precisely record every word, sentence, or moment. Rather, we form a mental representation that captures the essential meaning or gist of that event. For example, in the two preceding paragraphs we described the method of loci. Can you recall those paragraphs word for word? More likely, what you have encoded is the gist—the general theme—that the method of loci involves forming images that link items to places. Schemas: Our Mental Organizers The themes that we extract from events and store in memory are often organized around schemas. A schema (plural: schemas, or schemata) is a “mental framework”—an organized pattern of thought about some aspect of the world, such as a class of people, events, situations, or objects (Bartlett, 1932; Koriat et al., 2000). We form schemas through experience, and they can strongly influence the way we encode material in memory (Tse et al., 2007). To demonstrate this, read the following paragraph: The procedure is actually quite simple. First you arrange things into different groups. Of course, one pile may be sufficient depending on how much there is to do. If you have to go somewhere else due to lack of facilities, that is the next step; otherwise you are pretty well set. It is important not to overdo things. That is, it is better to do too few things at once than too many. In the short run this might not seem important, but complications can easily arise. A mistake can be expensive as well. . . . After the procedure is completed, one arranges the materials into different groups again. Then they can be put into their appropriate places. Eventually they will be used once more, and the whole cycle will have to be repeated. However, that is part of life. (Bransford & Johnson, 1972, p. 722) Asked to recall as much as you can of the preceding paragraph, you would probably have difficulty remembering much of it. Certainly, participants pas77416_ch08_266-303.indd 277 277 in the original experiment did. However, suppose we tell you that the paragraph is about a common activity: washing clothes. Now if you read the material again, you will find that the abstract and seemingly unrelated ideas suddenly make sense. Your schema—your mental framework for “washing clothes”—helps you organize these ideas and recall a great deal more. This example illustrates that how we perceive a stimulus shapes the way we mentally represent it in memory. Essentially, schemas create a perceptual set, which is a readiness to perceive—to organize and interpret—information in a certain way. Schemas, Encoding, and Expertise When people who have never learned to “read notes” look at a musical score, they see an uninterpretable mass of information. In contrast, musicians see organized patterns that they can easily encode, eventually learning to play a piece “from memory.” In music as in other fields, acquiring expert knowledge can be viewed as a process of developing schemas—mental frameworks—that help to encode information into meaningful patterns. William Chase and Herbert Simon (1973) demonstrated the relation between expertise, schemas, and encoding in a classic study. Three chess players—an expert, an intermediate player, and a beginner—were allowed to look at a chessboard holding about 25 pieces for only five seconds. Then they looked away and, on an empty board, attempted to reconstruct the placement of the pieces from memory. This procedure was repeated over several trials, each with a different arrangement of pieces. On some trials, the chess pieces were arranged in meaningful positions that actually might occur in game situations. With only a five-second glance, the expert typically recalled 16 pieces, the intermediate player eight, and the novice only four. What may surprise you is that when the pieces were in random positions there was no difference in recall between the three players. They each did poorly, accurately recalling only two or three pieces. What explains these results? We have to reject the idea that the expert had better overall memory than the other players, because he performed no better than they did with the random arrangements. But the concepts of schemas and chunking do explain the findings (Chase & Simon, 1973; Gobet & Simon, 1998). When the chess pieces were arranged in meaningful positions, the expert could apply well-developed schemas to recognize patterns and group pieces together. For example, he would treat as a unit all pieces that were positioned to attack the king. The intermediate player 12. What is a schema? Explain how schemas influence encoding. 13. In what sense are schemas and expert knowledge related? 15/11/13 10:15 AM Final PDF to printer 278 CHAPTER EIGHT STORAGE: RETAINING INFORMATION Thinking critically WOULD PERFECT MEMORY BE A GIFT OR A CURSE? If you could have a perfect memory, would you want it? What might be the drawbacks? Think about it, and then see the Answers section at the end of the book. and especially the novice, who did not have welldeveloped chess schemas, could not construct the chunks and had to try to memorize the position of each piece. However, if the pieces were not in positions that would occur in a real game, they were no more meaningful to the expert than to the other players. In this case, the expert lost the advantage of schemas and had to approach the task on a piece-by-piece basis just as the other players did. Similarly, football coaches show much better recall than novices do after looking at diagrams of football plays (patterns of Xs and Os) only when the plays are logical (Figure 8.9). You may not be an advanced chess player, but there are many areas in which you possess expert knowledge. You have used language for most of your life and have years of experience about how the world works. As the washing machine example illustrates, your own “expert schemas” strongly influence what you encode and remember. 14. Explain the concepts of associative networks and priming. Mean number of elements recalled 25 Experts (coaches) Novices 20 15 10 5 Logical plays Illogical plays FIGURE 8.9 Diagrams of football plays were shown to football coaches (experts) and people who had played football but were not coaches (novices). Coaches, allowed to see each play for just five seconds, displayed excellent memory—but only when the plays were logical. Their well-developed football schemas were of little use when the patterns of Xs and Os were illogical. The findings are very similar to those obtained when expert and novice chess players tried to reproduce meaningful and random arrangements of chess pieces. Data from Garland & Barry, 1991. pas77416_ch08_266-303.indd 278 After information is encoded, how is it organized and stored in long-term memory? Consider the following statements, indicating as quickly as possible whether each is true or false: 1. 2. 3. 4. 5. 6. A raccoon has wings. Moscow is in Russia. A bat is a fish. Coca-Cola is green. An apple is a fruit. Some fire engines are red. Chances are, you were able to respond to each statement almost instantaneously. Considering their diversity, it is remarkable that you could access the information so quickly. The fact that we are able to perform such tasks routinely—that we can recall an incredible wealth of information at a moment’s notice—has influenced many cognitive models of how knowledge is stored and organized in memory. Memory as a Network We noted earlier that memory is enhanced by elaborative rehearsal, which involves forming associations between new information and other items already in memory. The general principle that memory involves associations goes to the heart of the network approach. Associative Networks One group of theories proposes that memory can be represented as an associative network, a massive network of associated ideas and concepts (Bower, 2008; Collins & Loftus, 1975). Figure 8.10 shows what a small portion of such a network might be like. In this network, each concept or unit of information—fire engine, red, and so on—is represented by a node somewhat akin to each knot in a huge fishing net. The lines in this network represent associations between concepts, with shorter lines indicating stronger associations. For simplicity, Figure 8.10 shows only a few connections extending from each node, but there could be hundreds or more. Notice that items within the same category—types of flowers, types of fruits, colours, and so on—generally have the strongest associations and therefore tend to be clustered closer together. Alan Collins and Elizabeth Loftus (1975) theorize that when people think about a concept, such as “fire engine,” there is a spreading activation of related concepts throughout the network. For example, when you think about a “fire engine,” related 15/11/13 10:15 AM Final PDF to printer Memory Vehicle Street Car Truck Bus Ambulance House Fire engine Fire Orange Yellow Red Apples Green Cherries Violets Pears Roses Flowers Sunsets Sunrises Clouds FIGURE 8.10 A network of concepts in semantic memory. The lines in the semantic network represent associations between concepts, with shorter lines indicating stronger associations. Adapted from A.M. Collins and E.F. Loftus, 1975. For an interesting three-dimensional look at an associative network, check out the visual thesaurus at www.visualthesaurus.com. concepts, such as “truck,” “fire,” and “red,” should be partially activated as well. The term priming refers to the activation of one concept (or one unit of information) by another. Thus,“fire engine” primes the node for “red,” making it more likely that our memory for this colour will be accessed (Chwilla & Kolk, 2002). The notion that memory stores information in an associative network provides one possible explanation for why hints and mnemonic devices help to stimulate our recall (Reisberg, 1997). For example, when someone says, “Name the colours of the rainbow,” the nodes for “colour” and “rainbow” jointly activate the node for ROY G. BIV, which in turn primes our recall for “red,” “orange,” and so forth. Neural Networks The neural network approach provides a different and increasingly popular model of memory and cognition (Chappell & Humphreys, 1994; McClelland & Rumelhart, 1985). A neural network has nodes that are linked to one another, but these nodes are physical in nature and do not contain individual units of information. There is no single node for “red,” for “fire engine,” and so on. Instead, each node is more like a small informationprocessing unit. As an analogy, some proponents pas77416_ch08_266-303.indd 279 would say: Think of each neuron in your brain as a node. A neuron processes inputs and sends outputs to other neurons, but as far as we know, the concepts of “red,” or “fire engine,” or your mental image of an elephant are not stored within any single neuron. Where, then, is the concept “red” stored? In a neural network, each concept is represented by a particular pattern or set of nodes that becomes activated simultaneously. When node 4 is activated simultaneously (i.e., in parallel) with nodes 9 and 42, the concept “red” might come to mind. But when node 4 is simultaneously activated with nodes 75 and 690, another concept enters our thoughts. Looking across the entire network, as a multitude of nodes distributed throughout the brain fire in parallel at each instant and spread their activation to other nodes, concepts and information are retrieved and thoughts arise. For this reason, neural network models are often called parallel distributed processing models (PDP). Researchers in many fields are using the neural network approach to model learning, memory, language disorders, and other cognitive processes (Botvinick & Plaut, 2006; Joanisse, 2009; Vogels, Rajan, & Abbott, 2005). 279 15. How do neural network models differ from associative network models? Types of Long-Term Memory Think back to the nature of H.M.’s amnesia. Since his brain operation, H.M. has been unable to consciously recall new facts or personal experiences once they leave his short-term memory. Each time he meets you he will believe it is the first time. Yet with practice, H.M. learned new tasks, even though he would never remember having seen them before (Milner, 1965). Based on research with amnesia patients, brainimaging studies, and animal experiments, many cognitive scientists believe that we possess several long-term memory systems that interact with one another (Squire & Zola-Morgan, 1991; Tulving, 2002). This view is consistent with the concept, described in Chapter 6, that the mind involves distinct yet interrelated modules. Declarative and Procedural Memory Declarative memory involves factual knowledge, and includes two subcategories (Figure 8.11). Episodic memory is our store of factual knowledge concerning personal experiences: when, where, and what happened in the episodes of our lives. Your recollection that you ate pizza last night is an episodic memory. Semantic memory represents general factual knowledge about the world and language, including memory for words and concepts. You know that Mt. Everest is the world’s tallest peak and that e = mc2. Episodic and semantic memories 16. Use the concepts of declarative versus procedural memory, and explicit versus implicit memory, to explain the pattern of H.M.’s amnesia. 15/11/13 10:15 AM Final PDF to printer 280 CHAPTER EIGHT Long-term memory Declarative Personally experienced events (episodic memory) Facts— general knowledge (semantic memory) Procedural (nondeclarative) Skills— motor and cognitive Classical conditioning effects FIGURE 8.11 Some theorists propose that we have separate but interacting declarative and procedural memory systems. Episodic and semantic memories are declarative; their contents can be verbalized. Procedural memory is nondeclarative; its contents cannot readily be verbalized. 17. Describe some ways to measure explicit and implicit memory. are called declarative because, to demonstrate our knowledge, we typically have to “declare it”—we tell other people what we know. H.M.’s brain damage severely impaired both components of his declarative memory, but this is not always the case. Some brain-injured children with amnesia cannot remember their daily personal experiences but can retain general factual knowledge, enabling them to learn language and attend mainstream schools (Vargha-Khadem et al., 1997). In contrast to declarative memory, whose contents are verbalized, procedural memory (nondeclarative memory) is reflected in skills and actions (Cohen et al., 2005). One component of procedural memory consists of skills that are expressed by “doing things” in particular situations, such as typing, riding a bicycle, or playing a musical instrument. Classically conditioned responses also reflect procedural memory (Gabrieli, 1998). After a tone was repeatedly paired with a puff of air blown toward H.M.’s eye, he began to blink involuntarily to the tone alone (Woodruff-Pak, 1993). Although H.M. could not recall undergoing this procedure, his br