128 Cognition ‘modal model of memory Armemory medelproposed byAtkinson and Shifrin (9681 consisting of Sensory mamory short-term emer) analong tom memory iconic and echoic ‘sensory memory “The visual and auditory sencory memory sytem, respectively Sensory memory nas te abit 19 regsteralarge amount of information athouga it typical decays quickly icone memory has 2n Upper limit of one second; echoie memory has lit cof two seconds ‘decay “The tem used toveler to the time courseot forgetting DISCOVERY tab FIGURE 5.2 | Some of the many memory systems that memory researchers investigate The Modal Model of Memory Sensory Memory Although they are now outdated, the memory systems proposed by James (1890) pro- vided an excellent framework for later research. Atkinson and Shiffrin’s (1968) modal model of memory was so influential that most people probably know of it even if they haven't studied psychology. In its most basic form, the modal model contains three in- terconnected memory systems: sensory, short-term, and long-term. Information from the outside world is initially processed by our senses. For example (as you may recall from. Chapter 3), light enters through the lens of the eye and is then transduced by cells in the retina before the information it carries is sent to the occipital lobe for further processing. Inthe visual modality, this information is stored fora very brief moment (likely less than. a second) in the form of a visual or iconic memory; the auditory equivalent is known, as echoic sensory memory. Essentially, sensory memory can be understood as a buffer system for stimuli received through the senses. You have probably had the experience of seeing (or hearing) something for only a moment and yet being able to remember exactly what it was and where it was located (or where it came from). This experience is exemplified in seminal research by George Sperling (1960, 1963), in which partici- pants were flashed a matrix of nine letters (in rows of three) fora brief moment (50 ms). As Figure 53 shows, in a partial report task participants were required to immediately5| Memory Systems 129 FIGURE 5.3 | Participants in Sperling's (1960) sensory memory experiment were briefly flashed a matrix of nine letters In the whole report condition they were required to recall all the letters In the partial report cancition atone indicated which row of laters they should report. The tone could occur atthe moment when the max of letters disappeared. or after some predetermined delay ie. stimulus onset asynchrory, oF SOA: see Chapter recall only one row of letters, and they did so with sur- prising accuracy. By contrast, performance in a whole re- port task—where participants were required to recall all, nine letters—was very poor. Of crucial interest to memory research, Sperling was also able to use this paradigm to determine the time course of sensory memory, essentially asking how long it lasted. To do so he would wait until the matrix had disappeared before indicating which of the three rows participants were to recall, varying the length ofthe delay. As you can see in Figure 5.4, the rate of decay for visual sensory memory is relatively fast: after merely a second’s delay, participants’ ability to recall three let- ters was no better than their ability to recall all nine letters immediately. ‘These seminal findings are important for several rea- sons, First, they showy that our sensory system has the abil- ity to store a large amount of information at a single glance. Although Sperling's experiments involved only the visual modality, an analogous sensory memory system appears to 10 Whole report performance level Number of letters recalled po 6 3 5 t Delay of signal (seconds) FIGURE 5.4 | Typical results from Sperling's experiment Notice that performance is relatively good when the stimulus (Gone) is presented immediately: however, it quickly declines when the stimulus is delayed,130 Cognition Res eage sess Is your memory better than a chimpanzee’s? If you are confident that itis, you might want to reconsider. ‘yuma (the chimp in Figure 5.5) is part of a project led by Japanese researcher Telsuro Matsuzawa at Kyoto University's Primate Research Institute. The Ai Project (named after Ayuma's mother] is designed to explore chimpanzee cognition as a way of improving our un- derstanding of human cognitive capabilities It’s true that you would be much better than Ayuma at remembering the answers to questions on 2 cognitive psychology exam, but his memory is also capable of some impressive feats, The first step in this intriguing research project was to teach Ayuma how to count from 2 to 9 This was accomplished by rewarding him each time he touched the correct order of numbers on a touch screen Starting with easy arrays of only two numbers (1 and 2), he gradually learned to count all the way t0 9. At this point in the experiment the researchers used a limited-hold memory task that involved flashing all nine digits in ran dom locations on the screen: then the numbers were re placed by white boxes, and Ayuma was required to tap the boxes in the correct order. When the numbers were displayed for around 700 ms and then removed from the screen, there were no performance differences between ‘Ayuma and the university students who also participated in the experiment: everyone scored around 80 per cent, However, when the display time was decreased to about 200 ms, human performance plummeted to about 49 per cent, while Ayuma still scored around 80 per cent, ‘Thus it appears that chimps’ visual sensory memory (at least Ayuma's) is vastly superior to ours (see Inoue & Matsuzawa, 2007), FIGURE 5.5 | Tetsuro Matsuzawa and Ayuma, one of the participants in this intriguing sensory memory experiment Ayumma performed much better than his human counterparts when recalling the lacations of the digits on the sereen in the correct numerical order. exist forall six senses. For example, auditory or echoic sensory memory lasts approximately two seconds: most people are able to repeat something they have just heard nearly word for word. Despite its very short duration, sensory memory may help us to see the world in a continuous manner, without interruptions, each time we blink or make a saccade (a quick eye movement), While sensory memory can register a large amount of information, not all of it enters short-term memory—the second stage of the modal model of memory. DISCOVERY LAB Short-Term Memory The second major component of Atkinson and Shiffrin’ (1968) modal model of memory is short-term memory, which receives information from both sensory memory and long- term memory. As previously discussed, sensory memory is capable of registering a large quantity of information. However, most of that information fades from memory (decays) unless itis given attention, In its simplest form, the modal model claims that attended5| Memory Systems 134 information enters short-term memory and is retained there for a short time, typically no more than 18 seconds (Peterson & Peterson, 1959). Beyond this upper limit, information once again decays unless itis adequately rehearsed. Suppose someone told you a phone number to call and you didn't have a smart phone {@ situation the authors ofthis book lived with for most of their lives). You would probably rehearse (repeat) the number over and over until you were sure you had it stored in your long-term memory. Interestingly, the development of the phone number dovetails with seminal research conducted by Harvard psychologist George Miller. Miller (1956) demon- strated that the capacity of human short-term memory is somewhere between five and nine items. In fact, you might know that when it comes to memorizing items, the “magical number” is “seven plus or minus two” That saying comes from Miller, and it means that holding more than nine items in short-term memory is a very tough task, The introduction of the seven-digit phone number is often credited to the fact that Bell Laboratories looked to Miller for advice in the 1950s. While rehearsal is one common way of promoting the consolidation of lang.term mem- oties, there are other strategies for increasing the span of short-term memory. One of them, called chunking, reflects the idea that you can increase the capacity of your short-term, memory by grouping pieces of information together. Thus you can remember a seven-digit sequence (say, 5-6-4-3-2-47) much more easily if you chunk it into two groups (564-3247). There are many ways of reducing the total number of items to be remembered. Just think of acronyms such as NHI. for the National Hockey League, or ROYGBIV for the colours of the spectrum of light. Despite the influence that Miller’ (1956) work has had, both in the scientific world and in the public realm, recent research suggests that seven might not be the magic num- ber after all. Determining exactly how much information can be stored in a single chunk is a challenge. However, given its limited capacity, it seems unlikely that the short-term memory system can handle much more than four chunks of information at a time (see Cowan, 2001). Long-Term Memory ‘The final component of the modal model is long-term memory: information that is stored and brought back to short-term memory for immediate usage. There are many different divisions of longterm memory, and we will explore them in detail shortly. First, however, it is important to introduce a concept that the modal model does not take into account: working memory. Working Memory Before we begin to discuss working memory, try this quick exercise: read the rest of this paragraph while keeping in mind the numbers 0 - 5 - 2- 1- 4. The concept of working memory has been at the centre of Alan Baddeley’ (1986, 1989, 2000a, 2003, 2002a, 20026; Baddeley & Hitch, 1974; N. Morris & Jones, 1990; Parkin & Hunkin, 2001) influential re- search program. Working memory “involves the temporary storage and manipulation of in- formation that is assumed to be necessary for a wide range of complex cognitive activities” rehearsal ‘The process through whieh infotmationin short-term ‘memory mantaine. consolidation The process through which memory traces ‘ace stabilizes to form long-term memores. See Chapter 6 fora ful ‘chunking A strategy used to increase ‘arranging elements ‘ups (chunks hat ‘can be more easily remembered working memory ‘The system that allows forthe temporary storage and manipulation of Information requestor various cognitive activities,132 Cognition central executive ‘The component of working memory that coordinates information ram the tee phonological loop Temporary store of linguistic information visuo-spatial sketchpad Temporary sore of information episodic buffer reves information to an from long-texm memory DISCOVERY LAB fluid systems Cognitive processes that ‘manilate information crystallized systems Cognitive systems that accumulatelong-term knowlecee, Baddeley, 2003, p. 189). While this may sound similar to short-term memory, there are many differences. In fact, the shortterm memory model turns out to have a number of shortcomings. Now, without returning to the beginning of this section, what were the num- bers that you were supposed to keep in mind? Chances are you were right, or nearly so. Yet according to the modal model, the cap- acity limitations of short-term memory would have made it impossible to store the number and at the same time read the text, Baddeley recognized this and concluded that, under some ciscumstances, the short-term memory system must in fact be capable of carrying out two tasks at once. ‘As conceptualized by Baddeley and Hitch (1974), working memory is the system that pulls all the other memory systems together, enabling us to work with different types of information in a dynamic fashion. It consists of four distinct subsystems (Figure 5.6), one of which (the central executive) coordinates information from the other three (the phono- logical loop, the visuo-spatial sketchpad, and the episodic buffes). The phonological loop represents the entirety of shortterm memory as conceptualized by the modal model of memory. Quite simply, anything that is auditory in nature or language-elated is said to have obligatory access to the phonological loop (c.e., it must be processed and temporarily held in the phonological loop). However, not all the information that we keep in memory is auditory or language-based. If you've ever moved from one home to another, you will have faced the problem of manoeuvring a large piece of furniture down a staircase or through a narrow doorway. In that situation, most of us would first step back and try to imagine how to navigate without damaging anything, This example highlights the visuo-spatial sketchpad: a separate component of working memory that we use for non-verbal infor- mation, Both the phonological loop and the visuo-spatial sketchpad interact with long- term memory, while the episodic buffer is used to move information to and from long-term memory. Its most important function is to organize information “from the phonological and visuo-spatial subsystems of [working memory] with information from {long-term memory}" Baddeley, 2001, p. 1349). All three of these subsystems have limited capacities and hold information only temporarily (Baddeley, 2000b, p. 421). The central executive selects and integrates information from across the three subsys- tems. Itis intimately associated with consciousness, constituting a workspace within which solutions are formulated (Baars, 2002). Finally, notice the distinction in Figure 5.6 between fluid systems and crystallized systems. The former are cognitive processes that manipu- late information but are ‘themselves unchanged by learning,” while the latter are “cognitive systems capable of accumulating long,term knowledge” (Baddeley, 2000b, p. 421). To illustrate how some of the components of working memory interact, Baddeley (1989, p. 36) used a simple example (previously used by Shepard, 1966, and Neisser, 1970). Suppose you were asked to recall the number of windows in your house or apartment, You would probably form a mental image of the building, using the visuo-spatial sketchpad, and imag- ine walking around it, counting the windows as you go. The counting is done by the phono- logical (or asticulatory) loop, and the entise process is coordinated by the central executive. Baddeley (2003) suggested that the phonological loop evolved as an aid in the acqui- sition of language, facilitating the learning of words by allowing us to temporarily store and rehearse them. Once learned, speech becomes a powerful tool, capable of influencing the behaviour both of others and of ourselves. Sub-vocal speech can be used to articulate our plans and is an important aspect of self-control. Baddeley (20036) also suggested that5| Memory Systems 133 /s BB Aiud systems 1) Crystallized systems: FIGURE 5.6 | Working memory model Source: Donald T.Stuss and Robert T. Knight. (2002) Principles of Frontal Lebe Function Oxford. Oxford University Press) Alan Baddeley. "The episodic bulfer:@ new component of working memory?” Trends m Cogntive Sciences, Vol. 4, ssue 13, pp. 417-25, the visuo-spatial sketchpad evolved in order to facilitate the representation of things and their relations. Thus it aids in tasks as diverse as planning a route (e.g, how to get to a novel location) or figuring out how to put the parts of something together (e.g,, assembling fur- niture components). Processes ofthis kind will be explored in the chapters on imagery (7), problem-solving (so), reasoning (23), and creativity (22). Working Memory and the Brain Baddeley (2002a) observed that working memory is a complex system and hence unlikely to map in a simple way onto an anatomical structure such as the frontal lobes. However, it is clear that the frontal lobes play an important role in integrating information from many other areas of the brain, and are crucially involved in its manipulation for purposes such as learning, com- prehension, and reasoning, ...[TJhese are precisely the roles attributed to working memory (p. 258). One frontal area that has been singled out as particularly important for working memory (Figure 57) is the dorsolateral prefrontal cortex (DLPFC). When we reviewed Stroop research in Chapter 4, on attention, we noted that the DLPFC was believed to play a role in selecting between alternative response tendencies. This is an important function, of working memory, in particular of the central executive. Curtis and D'Esposito (2003) have suggested that the DLPFC is an integral part of working memory, acting to monitor and contzol alternative courses of action,134 Cognition FIGURE 5.7 | The dorsolateral prefrontal cortex Divisions of Long-Term Memory Ken Jennings holds the record for the longest winning streak on the television game show: ee Jeopardy! He won an amazing 74 games in a row, amassing more than 3 million dollars declarative memory (One of two major divisions fof memory also known asexglich memory, te memory system that contain knovledge that canbestates episodic memory ‘The subdivision of declarative memory ‘concerned with personal experience semantic memory ‘The subdivision of declarative memory concerned wath generat knowledge (ea, facts, words, and concepts. along the way. Apparently his ability to access information in long-term memory is vastly superior to the average person’. Research has determined that long-term memory, like working memory, consists of several subsystems. To begin with, there are two main divisions of long-term memory: declarative and non-declarative. Declarative Memory Declarative memory (also known as explicit memory) is the system in which knowledge that can be stated (eg, factual information) is stored. According to Endel Talving (1972), who played a major role inthis research, it has two subdivisions: episodic memory handles “temporally dated, spatially located, and personally experienced events or episodes,’ while semantic memory handles “knowledge about words and concepts, their properties and interrelations” (Tulving & Thomson, 1973, p. 354). In short, episodic memory deals with events that are personally experienced, while semantic memory is concerned with general knowledge.5 | Memory Systems The Episodic-Semantic Distinction Among the examples of episodic memories given by Tulving (1972) are the following: remember seeing a flash of light a short while ago, followed by a loud sound afew seconds later. Last year, while on my summer vacation, I meta retired sea captain who knew more jokes than any other person I have ever met. remember that I have an appointment with a student at 9:30 tomorrow morn- ing (Tulving, 1972, p. 386). By contrast, examples of semantic memories include the following (Tulving, 1972): remember that the chemical formula for common table salt is NaCl. Tknow that the name of the month that follows June is July (p. 387) As Tulving (1972) observed, memories of this kind differ from episodic memories in that they are general knowledge, not “personally experienced unique episodes" (p. 387). How- ever, it’s important to note that episodic and semantic memories are not mutually exclu- sive. For example, you might remember the day in grade two when Mrs Butterworth taught you that Ottawa is the capital of Canada. It was only after rehearsing your episodic mem- ory of the lesson that you were able to store the fact in your semantic memory. In other words, episodic memory can serve as a gateway for the formation of semantic memory. Neuropsychological Evidence for the Episodic-Semantic Distinction Studies of people with brain injuries have provided compelling evidence for the theory that episodic and semantic memory are independent systems. A good example is a study by Klein, Loftus, and Kihlstrom (1996) of a patient known as W], who suffered retrograde ammesia—inability to recall events prior to the injury—following a closed head injury (one in which the skull remains intact) that impaired her episodic memory. The issue that this case addresses is whether episodic memory is necessary in order to have a sense of personal identity. “Is it possible for someone who cannot recall any personal experiences—and there- fore cannot know how he or she behaved—to know what he or she is like?” (Klein, Loftus, & Kihlstrom, 1996, p. 250. Distinguishing between these two types of personal knowledge is actually quite simple. For instance, semantic personal knowledge “might include the facts that the person is kind, outgoing or lazy... Episodic personal knowledge, by contrast, consists of memories of spe- cific events involving the self [and] could inchude memories of instances in which one was kind, outgoing o: lazy” (Klein, Loftus, & Kiblstrom, 1996, pp. 250-52). If semantic personal memory and episodic personal memory were truly independent of one another, then dam- age to one system should not affect the other system. Thus someone might recall her person- ality as outgoing without being able to recall personal instances demonstrating this trait. 135136 Cognition WJ was an 18-year-old undergraduate who had sustained a concussion in a fall. When she was initially tested, about five days after the injury, she had no episodic memories for the preceding six or seven months. However, her general knowledge was good. She knew which classes she was enrolled in, although she could not remember attending any of them. She also knew the names of teachers and friends, although she could not recall any per- sonal experiences involving them. Essentially, her semantic memories for that time period ‘were more or less intact: only her episodic memories were absent. In most cases retrograde amnesia resulting from closed head injury is temporary and the patient recovers within a few weeks. Fortunately, three weeks after the event, W's memory for events prior to the accident had returned to normal. ‘technique invented by Crovitz and Schiffman (1974) was used to evaluate W)' episodic ‘memory during her amnesia, She was given a list of 24 nouns, each of which could easily be represented by a picture (eg, alarm clack), and asked to recall a personal event in relation to each word from any time in the past, and then date the memory in terms of when it oc- curred. Thus someone might recall hitting the snooze button repeatedly on the alarm clock eet this morning. When tested five days after her injury, W] produced a very different pattern recency bias vs primacy bias tendency torecal experiences from the recent past versus a tendency t recat relatively cstant past. Percentage of memories FIGURE 5. ti Data from Alen, SB. Loftus, J, & Kihlstrom, 2F. (996) p. 255. Copyright © 1996 by the American Prycnolagical Association Adaates vith permission ‘The use of APR informanion does not imply endorsement ay APA 70 1e periods 12 | Percentage of personal memories for different of results than did a control group of three undergraduate women of approximately her age. ‘As Figure 58 shows, the control participants had a recency bias in that they tended to recall experiences from the previous 12 months. By contrast, W] showed a primacy bias in that she tended to recall experiences in the relatively distant past. Four weeks later, however, after she had recovered, her pattern of episodic memory was similar to that of the control group. To test her semantic personal memory while she was still amnesic, W] was asked to tate herselt in terms ot 80 personality traits (i.e., to indicate the extent ta which she was agreeable, dominant, and so on). She then repeated this exercise after her amnesia had lifted. There TH Wo Micontrois was strong agreement between the two sets of ratings—a level of consistency similar to that of control participants. While amnesic, W) was also asked to rate her personality as if she were in high school. The fact these ratings differed from her current ratings indicated awareness that she had changed since starting college. “W] knew something about what she had been like at college, which was different from what she was like in high school; but she knew this despite the fact that she could not recall anything from her time in college" (Klein, Loftus, & Kihlstrom, 1996, p. 256). These results are consistent with the hy- pothesis that episodic and semantic memory are 15-60 61-120 2120 two distinct systems, and that we can have access ‘Age of memories (months) to semantic knowledge without having access to episodic memory. Another case of amnesia, which Tulvi 4985, 2002) reported on, also supports the disso- jation between episodic and semantic memory. ‘The patient known at the time as NN (KC in laterCan Amnesiacs Learn? Earlier we discussed the tragic story of Lucy, who fell in love with Henry in the movie 50 First Dates. While the type of memory loss that Lucy was said to suffer was a bit far-fetched, there was a scene at the end of the movie {spoiler alert) in which she cid seem to have some mem- cory of Henry, based on the numerous paintings that she had made of him. Is it possible for someone with severe ‘amnesia to actually learn some form of declarative mem- cory? Recent research suggests that this might in fact be the case, albeit in a very limited sense. Much of what we know about amnesia comes from Henry Molaison, who until his death in 2008 was known only as HM (Figure 59). As a young man he had a severe form of epilepsy, which was alleviated by a bilateral me- lobectomy, The surgery removed a number of brain structures in the medial ternporal lobe strongly associ- ated with memory, including most of the hippocampus. (Figure 5.20). Without a functioning hippocampus, Henry was unable to form new semantic memories: you could, spend hours with him, butif you left the room for a couple. of minutes and then retumed, he would have no recollec- tion that he had ever met you. At the time it was widely, thought that people with severe amnesia, like Henry, were completely unable to form any type of declarative memory, semantic or episodic. However, more recent research suggests that it may be possible for such peo- ple to retain some form of declarative memory after all Henry was the participant in one of these studies (O'Kane, Kensinger, & Corkin, 2004) FIGURE 5.9 | Henry Gustav Molaison (HM, 1926-2008) Henry underwent his surgery in 2953. Thus in theory he should not have any memory, semantic or episodic, of anything attr that date. However, when O'Kane etal (2004) used recall and recogrition tasks to test whether he had been able to acquire knowledge about 35 people who became famous after 1953, Henry was able to recall the last names for 12 of them. For example, when given the name “Bille Jean’ he suoplied the last name “King” {Bille Jean King was a famous tennis player inthe 19708) And he performed nearly twice as well when he was given semantic cues about the people in questionalthough stil much worse than someone with an intact hippacam- Us would, Ths pointed tothe existence of a network of semantic knowledge despite the absence of the brain steuetures thought necessary to form such memories So what might account for Henry’ ability to form new se- mantic memaries, however ti ited, about certain people when ive actucoten ne he consistently failed to form es such memories about otner things? One possibilty is simple repetition: in the case of doha F. Kennedy, for instance, he would almost certainly have heard many discussions of the 1963 assassination, both on the news and in his presence. tn ary event, the re- sus ofthis study and others ike, suggest that brain tuc- ‘ures other than the hippocampus may have the capacity [albeit limited! to encode long-term semantic memories hippocampus Asien the brain that FIGURE 5.10 | A hippocampus looks like a sea horse 137138 Cognition non-declarative memory The other major avsion emo, aso known asimplict memory, tne memory system aseociated with behaviour that coe: nat require conscious tought FIGURE 5.21 | Patient KC/NN publications; see Tulving, 2002a) was born in 1951 and suffered a closed head injury when he was 30 asa result of a motorcycle accident (Figure 5.11). With “extensive brain lesions in multiple cortical and subcortical brain regions? (Tulving, 2002, p. 130), he never recovered his episodic memory. However, his other intellectual skills remained intact. For example, he was able to outline a standard restaurant script (cripts will be discussed in the next chaptet)—evidence that his semantic memory was still functioning. However, he could neither recall individual events from his past nor imagine what he might da in the future. Tulving (1985) reported a fragment of an interview that he (ET) conducted with NN: ET: Let’ try the question again about the future. What will yoube doing tomorrow? (There is a 15-second pause) NN: smiles faintly, and says, “I don't know.* ET: Do you remember the question? NN: About what I'l be doing tomorrow? ET:Yes, How would you describe your state of mind when you try to think aboutit? (A second pause) NN: A blank I guess (p. 4). At various points NN described his blank state of mind as resembling “being asleep,” “being in a room with nothing there and having guy tell you to go find a chair, and there's nothing there’ o: ‘swimming in the middle of a lake [ith] nothing there to hold you up or do anything with’ (Tulving, 1985, p. 4). While NN retained his semantic memory systems, chen, his episodic memory system was drastically impaired. Non-Declarative Memory ‘The other major division of longterm memory is non-declarative memory, also known as implicit memory. Unlike declarative memories (episodic and semantic), non-declarative memories do not require conscious thought and are often difficult to put into words.MBean es let ty Many people believe that their sense of smell is in- timately connected with memory. In this case the people are right: even though our perceptual experi- ences tend to be dominated by sight and sound, smell can often trigger a memory that is decades old. It’s likely that many of you have had the experience of a distinctive scent sparking a memory (fond or otherwise} ‘of something from years past. Perhaps, for instance, the smell of apple pie takes you back to your grandmother's kitchen How is it that smell can trigger a memary? Address- ing this question is relatively easy. After an odour mol- ecule enters your nose and is recognized by olfactory sensors, this information is sent first to the olfactory bulb and ultimately to the cortex and limbic system of your 5 | Memory Systems brain. As you may have learned in other classes, the lim- bic system includes a number of structures associated with memory. The hippocampus is especially important far the consolidation of long-term memories. (On the other hand, the sensory neurons in the ep- ithelium (where smells are first registered) survive for only about 60 days. So how can a smell trigger a mem= ory of something that happened years ago? It turns out that once a particular connection has been established between neuronal units in the epithelium and the hip- Pocampus, it remains in place even as new olfactory neurons are generated to replace those that have died Because the connections of the neurons always go to the same place in the hippocampus, the memories asso- ciated with a particular smell survive. 139 ‘There are many situations in which we may remember something without being aware of doing so (eg, Jacoby & Dallas, 1981; Jacoby & Witherspoon, 1982). It is possible to distinguish the effects of memory for prior episodes or experiences on a person’ current behaviour from the person's awareness that he or she is remembering events of the past’ (Eich, 1984, p. 105). Schacter (1987, 1992) proposed the term “implicit memory" for such phenomena, Implicit memory comes into play when “information that was encoded during a particular episode is subsequently expressed without conscious or deliberate recollection” (Schacter, 1987, p. 501). ‘A demonstration ofthe practical consequences of participating in priming experiments isa study by Jacoby and Hollingshead (1990) in which participants read incorrectly spelled words. This experience subsequently increased the likelihood that the participants would unintentionally make the same spelling mistakes themselves. As Jacoby and Hollingshead observed, this kind of research may be unethical, because it can impair participants’ per- formance in an importantarea (pelling) without their knowing it;in the same way, reading students’ poor spelling “may be hazardous” to professors (p. 345). Jacoby and Hollingshead (1990) also noted that the latter herself “lost confidence in her spelling accuracy [and could] no longer judge spelling accuracy on the basis of a word ‘looking right. The word might look right because it was one of our incorrectly spelled words” (p. 356). (This finding also ‘makes us wonder about the wisdom of relying on all the texting shortcuts we use today.) Implicit memory has also been studied using a famous name paradigm (Jacoby & Kelley, 1992; Relley & Jacoby, 2000). Participants were divided into two groups and given a series of non famous names (eg, Sebastian Weisdorf) to read. As the names appeared on a computer screen, one group read them with full attention (1e, no distractions) and one with divided attention (achieved by adding a task such as listening to a series of numbers and pressing140 of non-famous names judged to be famous FIGURE 5.12 | Fame judgment task (full Cognition 0.35 os 0.25 oz ous o1 005 HIFul Mi Divided Old names New names versus divided attention) Data from: Jacoby, LL, & Kaley, .. (1994) %of stems completed with words from first ist 07: 06 as. Full BH inctusion i exclusion Dividea FIGURE 5.23 | Inclusion versus exclu- sion task (full versus divided attention) The simplest way to understand non-dectarative memory isin terms of its two major subtypes the perceptual representation system (eg. priming) and procedural memory. Data from: Jacoby, LL. & Kelley, CM. (1994), .275. Copyright © 1994 by (SAGE! Reprinted by Permission of SAGE Publications ne a key if there were three odd numbers in a row). Then names from the first trial (old nonfamous names) were mixed with new names, some famous and some not, and the participants were asked to judge whether or not they were famous. Participants were also told thatall thenames on the first list were non-famous. Consequently, anyone who recognized a name as coming from the first ist was aware that it was not famous. Now examine Figure 5,12, in which the vertical axis is the percentage of non- famous names judged to be famous. The higher the number, the more non-famous names that group judged to be famous. Under conditions of full attention people tended to recognize names from the frst list, and +0 say that fewer names from the firs ist were famous. However, partici pants from the divided attention group had implicit memories of the old names without explicitly recognizing them as coming from the fist list, “Thus they were more likely to categorize names from the frst list as famous because they seemed familiar. Itwasasiftthey said to themselves, “Oh, Ive heard that name before, but I dont remember where—it must be famous.” In this way Sebastian Weisdorf became famous. Another way of showing unconscious influences in memory is through studies using the method of opposition (Curran, 2001), which pits conscious (explicit) and unconscious (implicit) tendencies against one another (eg,, Jacoby, 1998; Jacoby & Kelley, 1992; Toth, 2000). Par- ticipants are shown a number of words (e.g, motel) under conditions of either full or divided attention. They are then given a number of word stems (eg,, mot- -). Half the participants in each condition (full and divided attention) are asked to complete the word stems using either a word from the list previously seen or, if they could not remember one, the first word they thought of. The other half are asked to complete the word stems by not using a word from the list previously seen, Thus if ‘motel was on the previous list, they should choose an ending otherthan 1 to complete the stem (e.,,-or to form motor) The frst instruction is the inclusion condition, while the second is the exclusion condition, Now examine Figure 513. The vertical axis isthe percentage of stems completed with words from the firs list. Under the full attention condition, participants performed well on both the inclusion and exclusion tasks. ‘When told to include words from the fist list, they completed the word stems with words from the fist lst 61 per cent ofthe time, When they were teld to exclude words from the first list, the number of stem completions using words from the first ist dropped to 36 per cent. Tis difference indi- cates that the participants had some degree of conscious control over the process of stem completion. However, in the divided attention condition there was no difference between the inclusion and exclusion task. These participants completed the stems with words from the first list equally often, whether they were trying to do soor trying not to. Thus they did not demonstrate any conscious control over their behaviour in these tasks. method of opposition Pits conscious explicit) and uncorscious (amie) tendencies against one Conscious control is reflected in differences between performance when one is trying to do something, and performance when one is trying not to do it (Jacoby & Kelley, 1992, -177)-A good analogy to the memory experiments we have been reviewing might be the5| Memory Systems 144 various real-world attempts we make not to do something. For example, when we diet we try ————— not to eat as much as usual, or when we quit smoking we try not to smoke. If there is a dif- eee ference between our behaviour when we are trying not to do something and our “ordinary” Syefon (ee) behaviour, then we are demonstrating that we can consciously contral that behaviour. How: A memory apse ever, if our behaviour when we are trying not to do something is the same as italways was, <2Ntsrirgveryspectic representations of events that is hypothesized to be responsi for ping then we are demonstrating that we are unable to consciously control that behaviour. The Perceptual Representation System os priming ‘The results of experiments on implicit memory are consistent with the notion that there is The unconscious process a memory system called the perceptual representation system or PRS (Hayman & Tulving, _"ovghwhichrecognsion 1989; Tulving & Schacter, 1990; Schacter, Wagner, & Buckner, 2000) that is responsible for pated Gries priming effects (discussed below) The important distinction is between ‘systems concerned posure toanidentcalor with explicit recollection of past events” as opposed to “primed identification of previously felaed tem encountered objects” (Tulving & Schacter, 1990, p. 302). The PRS contains very specific repre- prime sentations of previously encountered events, Thus ifyou had been shown the word fragment Theitem hat saresertes M-S-OU-I, only that fragment would be in your Ps: the complete word (MISSOURD would ..,,,.mumarines not. The episodic memory system and the PRS would be driven by different processes. The" Umertcumoreies episodic memory system operates with a deeper understanding of information, whereas the __iemsare generally faster pS deals with information on a more superficial evel. Ammnesiacs have an impaired episodic Haneeee ‘memory system, but their PRS may be relatively intact, Precisely how the PRS relates to other the seconditen pase ‘memory systems, such as semantic memory, is still unclear. As Berry and Dienes (1991) ob- nr apiming experiment, served, there may be similarities between the processes responsible for implicit memory and» e,dsntet roots those underlying implicit learning—a topic we will consider in Chapter 8, on concepts. Priming is the unconscious process through which our response to a given stimulus is ___‘xical decision task facilitated by previous exposure to a related (or identical) stimulus, making ous response both araneeaatl quickerand more accurate than it would otherwise be. Priming can be consideredasa process parcioartstosetermine that occurs in the PRS: that is, one that functions without conscious awareness (ie, implicit). __hsther «presented sting Experiments conducted by Dehaene ct al. (1998) demonstrated that implicit presenta. Ss Beworeorre tions (below the threshold for awareness) nonetheless facilitated participants’ responses to related information. This finding, and others demonstrating the effects of priming, can be inferred by certain testing paradigms, resting on the assumption that prior expo sure does indeed facilitate processing. Specifically, the experimentation requires both a prime (i, the item that is presented first) and a probe or target (i.e, an item presented after the prime that may be identical, related, or unrelated to it). Response rates are faster 4 DISCOVERY vwhen the prime and target are related than when they are not. LAB One means of testing the effects of priming isa lexical decision task (LDT), in which participants are presented with a letter string and asked to indicate whether or not it constitutes a word (eg, house or boesu). Studies (e.g, Meyer & Schvaneveldt, 1971; Tweedy, Lapinski, & Schvaneveldt, 1977) have found that the recognition of lexically correct words was faster when participants had already been exposed to words typically asso: Giated with the target. For example, the target doctor would be recognized as an actual word much faster if it had been preceded by the prime nurse. Similarly robust priming effects are found with word fragment completion tasks, where participants are exposed to an incomplete word and asked to complete it (see for example, Sloman, Hayman, Ohta, Law, & Tulving, 1988; Tulving, Schacter, & Stark, 1982); image completion tasks, in which142 Cognition procedural memory The memory system cconcemed with knowing how to do things tacit knowledge Knowwng now to 0 something without being able to say exactly what itis thatyou know. participants are asked to name an incomplete image that slowly becomes more and more complete over time (see for example, Matsukawa, Snodgrass, & Doniger, 2005); and decision-making paradigms (ee for example, Klein, 1993). Whatever the task, such experiments consistently show that primes are stored in memory and facilitate process- ing of the related items without the participants awareness. Procedural memory Another type of non-declarative memory, procedural memory (JR. Anderson, 1976), wn- derlies rote skills, such as riding a bicycle. Like priming, it operates without our conscious awareness. One way of thinking about the distinction between procedural and other forms of memory is in terms of the distinction between explicit and tacit knowledge. Polanyi (4958) pointed out that "the aim of a skilled performance is achieved by the observance of a set of rules which are not known as sth to the person following them’ (p. 49) Even if you don't remember learning to balance on a bicycle, you still know how to doit; and even if you haven't ridden for years, you have no problem doing it (though your legs may tire out fast). However, you would likely find it difficult to explain what youre doing. Indeed, most procedural skills are very hard to put into words, regardless of a person's expertise. An expert pianist or golfer would have similar trouble explaining their complex motor skills;and their performance would suffer if they had to think consciously about what they were doing Physical skills are not the only ones stored in procedural memory:so are many cognitive skills,including the ability to read. The factthat you can read this sentence without (we hope!) any conscious mental effort demonstrates that reading is a type of procedural memory. If you were to outline the principles that underlie reading, or riding a bike—that is, to convert your tacit knowledge into explicit knowledge—you would be using your semantic memory. And if you were to recall a particular experience of reading or bicycling—as in “Remember that time we rode from Vancouver to Abbotsford and it rained all day, and we ran out of food ... "you would be drawing on episodic memory. Procedural knowledge is tacit (implicit) in that we aren necessarily aware of what it is that we know. Further Developments in Memory Systems Theory Tulving’s distinction between episodic and semantic memory has continued to stimulate research for more than four decades, and the field has developed in response to this work. Tulving himself published a number of modifications to his basic theory (eg, Tulving, 1983, 1984, 1985, 1986, 2000, 2001, 2001b, 20022, 20028), and many other researchers have explored directions suggested by his findings, We will look at some of their studies now. Episodic Memory and Development According to Tulving (1985), children acquire episodic memory relatively late compared with other kinds of memory. In this connection, Tulving cited the work of authors as diverse as Neisser (1978a) and K, Nelson and Gruendel (1981), who suggested that very5| Memory Systems 143 young children do not experience anything that adults would call episodic memory. The hypothesis is that episodic memory develops out of semantic memory (Kinsbourne & Wood, 1975). Although small children are prodigious learners, most of their learning seems to involve the acquisition of general knowledge, rather than the accumulation of individual experience. Perner (2000, p. 01) suggested that episodic memory does not emerge until roughly four to six years of age. It is then that children are able to discriminate between things they have known for a long time and things they have learned recently. Ina similar vein, four-year-olds are able to discriminate between events they have observed and events they have been told about, while younger children can seldom make that distinction. Wheeler, Stuss, and Tulving (1997, p. 345) conclude that “the self knowledge necessary for episodic remembering is not reached until around age 4 or later.” The Butcher-on-the-Bus Phenomenon As we have seen, Tulving distinguished between episodic “remembering” and semantic “knowing” Thus “even when a person does not remember an event, she may know something about it” (1985, p. 6). For example, I may know that I have eaten at a particular restaurant ‘without remembering the experience of eating a specific meal on a specific occasion, The feeling of knowing in the absence of episodic memory has been called the butcher-on-the- bus phenomenon (Yovel & Paller, 2004). The term refers to the experience of running into someone who looks familiar (the butcher) in an unexpected place (on the bus) and not being, “able to remember the circumstances of any previous meeting or anything else about’ him (Yovel & Paller, 2004, p. 789). ‘Tulving’s distinction has been made in various forms by other memory researchers (eg, Gardiner, 2001; Gardiner & Richardson-Klavehin, 2000; Squire, 2004) and has stimu- lated quite a bit of ingenious research. Tip-of-the-Tongue Phenomenon ‘As we have seen, Tulving (1972) compared semantic memory to a mental thesaurus containing words, concepts, and their relations, but basically the term refers to gen- eral knowledge—a category that includes names. When (as often happens) you can’t quite come up with a particular name, you experience the tip-ofthe-tongue phe- nomenon (TOT). James (1890), who described this experience as a “gap,” noted that it is highly specific: when incorrect names are suggested, they are immediately re- jected because “[t}hey do not fit into its mold.” Further, the “gap of one word does not feel like the gap of another. ... When I try vainly to recall the name of Spalding, my consciousness is far removed from what it is when I vainly try to recall the name of Bowles’ (p. 251). R, Brown and McNeill (1966) famously gathered data on the properties of TOT. First, they observed this state when it occurred in themselves. For example, one of them was try ing to remember the name of the street on which a relative lived. He kept coming up with names like Congress, Corinth, and Concord. When he looked up the street name, it turned out to be Cornish. This example illustrates several properties of the TOT state that Brown and McNeill subsequently observed in an experiment. butcher-on-the-bus ‘phenomenon ‘The feoling ot knowing 1a person without being oie to remember the Circumstances of any previous meeting oF anything ese about him tip-of-the-tongue phenomenon (TOT) Krewing that you know something without quite being abletorecalt144 Gognition Participants in that experiment were given the definitions of 49 low-frequency (i.e, uncommon) words, such as apse, neporism, cloaca, ambergris, and sampan, and asked them. to identify the words. When participants found themselves in a TOT state, they were often able to identify some aspects of the target word, such as its first letter or the number of syllables it contained, In addition they were often able to make judgments about words that came to mind while they searched for the target word. Sometimes they knew that their incorrect guesses were similar to the target either in sound or in meaning, For example, while searching for sampan, some participants knew that Siam and sarong had a similar sound, and that barge and bouseboat had a similar meaning. Thus they had access to quite a bit of information about the target word before they were actually able to recall it. Generic recallis the term used by Brown and McNeill for this ability to recall parts and attributes of a word without explicitly recalling the word itself, If you want to experience a TOT state yourself, try naming the seven dwarfs from the film Snow White and the Seven Dwarfs (Meyer & Hilterbrand, 1984). In a study exploring the connection between TOTs and stress, A.S. Brown (i991) surveyed 79 undergraduate psychology majors and found that 75 per cent of them said they experienced TOTs more often under stress (¢g,, during exams). Studies of the frequency with which TOTS occur find that adults generally experience about onea week, although the frequency tends to be somewhat greater in older people. Several studies have replicated Brown and McNeill’ finding that when people are ex- periencinga TOT, they are likely to recall words that are similar either in sound orin mean- ing. They are often able to guess the frst letter of the desired word with a high degree of accuracy (¢.g,, Rubin, 1975). They may also know the last letter, though to a lesser extent than the first (A.S. Brown, 1991, p. 212). One of the most intriguing aspects of the TOT phe- nomenon is the often-reported experience of recalling the desired word only after we have stopped trying to recall it: “1 hour and 39 minutes after the start of the recall attempt, the word came... . hours prior to the solution, there was no recollection of thought on the topic” (Norman and Bobrow, 1976, p. 116). Burke, McKay, Worthley, and Wade (1991) believe that TOs occur mainly with words that the person concerned has not used very often or has not used very recently, with the result that the link between its meaning and its pronunciation may have atrophied be- cause of disuse. Consequently, other words that have a similar sound and/or meaning may be elicited along with the correct word. For example, consider charity and chastity. These words not only sound similar but may be considered related in that both are associated with virtue; furthermore, neither of them is used very often. Thus they may interfere with one another, causing a TOT state. Burke et al. (1991) also reported the interesting result that the names of famous people are particularly likely to lead to TOTs in older people. It's possible that this effect comes about because older people learned these names longer ago than younger people did. As a result, the names may be fresher in the memory of younger people and less prone to interference. Brown and McNeill (1966) suggested that memory for words and their definitions (usu- ally considered a central part of semantic memory) is organized like a dictionary. However, they realized that the structure of a mental dictionary was unlikely to be the same as that ofa standard dictionary. Since the 1960s a great deal of work has gone into trying to deter- mine how the mental store of words is organized and how we go about searching through it to find the information we need.5| Memory Systems 145 Spreading Activation An important notionto emerge fromthe study ofsemanticmemoryistheconceptofspreading —____ activation. Spreading activation was proposed by Quillian (1969) and elaborated by Collins _ spreading activation and Loftus (1975). The idea is that when you search a semantic network, you activate the paths The Seathat activation of where the search takes place. This activation spreads from the node at which thesearch begins. Mays neve sh? “The spread of activation constantly expands, firsttoall the nodes linked tothe firstnode,then fromthe node twch he to ll the nodes linked to each of these nodes, and so on” (Collins & Loftus, 1975, p. 408). The search begins. moreactive a nodeis, the more easily its information can be processed. Thus information from active nodes can be retrieved more quickly than information from less active ones. Now let's see how the idea of spreading activation can help to explain how priming works. Several experiments on priming have been done within the framework of the study of semantic memory; for a review of some of the best known, see Meyer and Schvaneveldt (976). One experiment on word recognition followed the procedure outlined in Figure 5.14 Participants looked at a screen with two fixation points, one above the other. Then a string of leters appeared at the top point, Sometimes the string was an English word such as wine and sometimes a non-word such as plame, The participants were required to complete a lexical decision task—that is, to decide whether or not the letter string was a word—and to respond “yes” if it was a word and “no” if it was a non-word. The time it took to make this response was recorded. Then another string of letters appeared and participantshad to jy DISCOVERY decide if this second string of letters was a word. LAB Sometimes the two leter strings were semantically related words: for example, bus and tuck, or sunset and sunrise. These are pairs of words that you might expect to be close to- gether in a semantic network. By contrast, if the first word was sunset and the second word ‘was truck, then this pair would be semantically unrelated. You would not expect these two. DISCOVERY ‘words to be close together in a semantic network. LAB This study found that the time it took to correctly recognize the second word was partly determined by the nature of the first word. If the second word was semantically. related to the first, then the time it took to recognize it was less than if it was semantic- ally unrelated. That is, ithe two words were semantically related (sunset~sunrise), then the first word primed recognition of the second. This priming effect did not occur if the two words were semantically unrelated (sunset-truck) Fixation First First Second Second points: letter string response —elter string response Reaction_,|, Delay ' time ie Foreperiod re" interval Reaction_,! \ y FIGURE 5.24 | Meyer, Schvaneveldt, and Ruddy’s priming procedure ‘Meyer, DF, Schvaneveiat. RW. and Ruddy, MG. (975). Loc! of contextual elects on visual recognition, In PALA, Rabbitt S. Dorie [Eds Attention and performance V |p. 100). London, Academic Press.146 Cognition involuntary semantic memory ("mind popping”) ‘semantic memory ‘nat popsinto your ming without episode context One way of understanding this finding is in terms of the spreading activation theory outlined above. Consider the fragment of a semantic network in Figure 5.15, In this dia- gram, the greater the distance between any two concepts, the less closely they are related toone another. Thus cherries and apples are directly connected, but the connection between street and flowers is indirect, running through several paths. The closer together concepts are, the more easily activation will spread from one concept to another. If clouds is acti vated, it will in turn activate (or prime) sunrises and sunsets, but several more activations would be required to prime vehicle. Now let’s consider Meyer and Schvaneveldts results in relation to Figure 5.15, Suppose that bus and rruck and sunset and sunrise are connected in a semantic network in the way de- picted in that figure. When the word buss seen, activation quickly spreads to the truck node, ‘The nodes for sunrise and sunset are much farther away in the network and so will be less act vated. Consequently, truck will be primed by bus much more than by either sunset or sunrise. Conversely if the participant sees sunset first, then sunrise will be primed much more than either bus or truck. In this way, the semantic network model can explain the priming effect. ‘The precise mechanisms governing spreading activation still need to be worked out (Bodner & Masson, 2003; Chwilla & Kolk, 2002; McRae & Boisvert, 1998). However, the concept of spreading activation has proven to be quite durable (McNamara, 1992), and continues to bea useful explanatory tool for studies of semantic memory. A good example is a study of involuntary semantic memory by Kvavilashvili and Mandler (2004). FIGURE 5.15 | Fragment of a semantic network Collins, AM, and 6 Loftus, EF, 1975). Copyright 3975 by the American Paychological Association, Reprinted with permission. Use af APA information dass nc imply endarssment by APA5| Memory Systems 147 Involuntary Semantic Memories ‘An involuntary semantic memory occurs whenever a semantic memory (e.g, a tune) pops into your mind without any episodic context. That is, you dont recall any autobiograph- ical information that might have triggered the semantic memory; it just pops up by it self and appears to be irrelevant to what you are currently thinking about. Kvavilashvili and Mandler (2004) call this mind popping, Kvavilashvili and Mandler (2004) reported on diary and questionnaire studies designed to probe the mind-popping phenomenon. Kvavilashvili kept two diaries of her semantic “mind pops’: the first for 19 weeks, at the age of 35, and the second for 18 weeks, two years later. She logged a total of 428 memories, which tended to be either words (eg, rummage) or images (eg, a view of a road and a small church in Cardiff). She had no episodic information accompanying these involuntary semantic memories. Most of the mind pops occurred while she was engaged in routine activities not requiring a lot of attention, and at first they appeared unrelated to the cur- rent activity. In retrospect, however, Kvavilashvili was often able to find cues that had triggered the memories without her awareness. For example, one pop-up was lichy and Scratcby, the names of two characters from The Simpsons television show. Kvavilashvili recalled she was scratching her back when the pop-up occurred. Examples like this sug- ‘gest that involuntary semantic memories are primed by events of which we are typically unaware. Kvavilashvili and Mandler’s (2004) study shows how ecologically valid research can complement and extend laboratory work. Connectionist Models of Memory When we introduced connectionist models in Chapter 2, we noted that they were designed to represent neural networks as they might exist in the brain. McClelland (2009, p. 583) observed that from a connectionist viewpoint individual items are not ‘stored in memory": rather, memories consist of certain patterns of activity. Nor are copies of particular experi- ences stored as memory traces; rather, neuron-like units representing each of the properties ofan experience are connected to other neuron-like units. McClelland etal. (1986) pointed out that some experiences will have the same properties, and so the unit for a particular property will tend to be connected with several different experiences. Every time one prop- erty is activated, all the units to which it is connected will tend to be activated as well. In order to accurately recall previous experiences, therefore, the system needs to excite some connections and inhibit others. McClelland (1981) and McClelland er al. (1986, pp. 274) have demonstrated how a simple connectionist system might work. Figure 5.36 lists the members of two gangs, the Jets and the Sharks, with some of their properties: name, gang affiliation, age, ed- ucation, marital status, and occupation. Some of the units that would be required to represent these individuals in memory are then presented in Figure 5.17. In the centre of the diagram aze units representing shared properties for each of the persons listed in Figure 5.16. These individual units are connected to the appropriate property units. For convenience, the property units are grouped within different ‘clouds,’ and units within a particular cloud inhibit each other. Thus an individual cannot be called both Lance and Art at the same time. excitatory and inhibitory connections ‘Connections that either enhance or ain nish the sspeiations netween the unis that make up eneura network,148 Cognition ‘The Jets and the Sharks Name Gang Age Edu. © Mar. Occupation Ar Jets. 40s. JH. Sing, Pusher Al Jets 30s. JH. «Mar. Burglar Sam Jets 205. JH. Sing. Burglar Clyde Jets 40s «JH, ~— Sing. Bookie Mike Jets 30s. JH, ~— Sing, Bookie Jim Jets 20s. JH. (Dv, Burglar Greg Jets 20s HS. Mar. Pusher John Jets 20s. JH. Mar. Burglar Doug Jets 30s «HS. Sing. Bookie lance Jets 20s + COL. Mar. Bookie George Jets 20s JH. Div. Burglar Pete Jets 20; +-HS. Sing. Bookie Fred Jets 20._:«S HS. Sing. Pusher Gene Jets 20s COL. Sing, Pusher Raloh Jets 30s. JH, ~— Sing. Pusher Phil Sharks. 30s COL. Mar. Pusher Ike Sharks. 30s. «JH. Sing. Bookie Nick Sharks 30s HS. Sing, Pusher Don Sharks 30s COL. Mar. Burglar Ned Sharks 30s COL. Mar. Bookie Kor Sharks. 403 «HS, Mar. Bookie Ken Sharks. 205 HS. Sing. Burglar Earl Sharks. 40s HS. Mar. Burglar Rick Sharks 30s HS. «Div. Burglar Ol Sharks. 30s COL. Mar. «Pusher Neal Sharks 30s HS. Sing, Bookie Dave Sharks. 30s HS. (Div. Pusher FIGURE 5.26 | Some properties of gang members ‘aaptes trom McClelland, JL. Retrieving generat and specine krowieage trom stores knowledge of species, Proceedings of the Third Annual Conference of the Cognitive Science Society. Berkeley, Cali . 27. Copyright 1988 by JL, MeCleland, Imagine that you have met all these individuals at one time or another. Imagine also that you find yourselfin a conversation about the individual whose name is Arr. When you try to remember what Artis like, what happens? Initially, when you hear the name Art, the name unit for Art will be activated. This name unit is connected to the individual unit for Art. The individual unit for Artis connected to all the property units that Art possesses. ‘Activation of all of these property units corresponds to remembering Art. This model can be seen as an extension of the spreading activation model. Ofcourse, the act of remembering is not always so straightforward. If you hear some- one talking about a gang member who is single, that information is not enough to enable you to identify the person in question, since several individual units are connected to the “single” property. However, a combination of properties might serve to specify the indi- vidual more or less completely. ‘Thus if you also hear that the single person in question is5 | Memory Systems FIGURE 5.27 | Aconnectionist model of the information in Figure 5.16 From: meCleliand. JL. 982) p. 28 Copyrignt © 2881 By I. McClelland. Reprinted by permission. a burglar in his twenties with only a junior-bigh education, all four of those properties will point to the Sam unit, causing more activation there than in any of the other units, Activat- ing the Sam unit will also activate the property units associated with Sam, and make you think that the person being talked about is a member of the Jets. In this memory model, information about individual experiences is derived from the state of the entire system at a particular time, The pattern of excitation and inhibition in the system asa whole determines what you will remember. Aging and Memory Disorders Memory and Aging Folk wisdom says that memory declines with age, but in fact some forms of memory are relatively unscathed by advancing years. It is episodic memory in particular that shows a strong age effect (Craik & Grady, 2002, p. 529). A good illustration can be found in a study by Mitchell (1989) that explored differences in the ways episodic and semantic memories decline with age. Remembering that I put salt on my food at lunch yesterday is an episodic memory; remembering that the formula for salt is NaClis a semantic memory. When Mitchell tested the memories of two groups of people, one aged 19-32 and the other aged 63~8o, the 149150 Cognition associative deficit hypothesis ‘Te ypotnesisthat older agutsnavea aencleney in éreating and reviewing inks between single nits of young people clearly outperformed the older ones on the episodic memory tasks, but this ‘was not the case on the semantic memory tasks. Although older people may sometimes have trouble recollecting recent personal experiences, their memory for general knowledge can continue to function well. One frustrating aspect of getting older is a deterioration in the ability to recognize people and/or recall their names. In one study by Schweich, van der Linden, Bredart, Bruyer, Nelles, and Schils (1992), three groups of participants kept diaries of occasions on which they had experienced such difficulties. Group 1 contained young university stu- dents (19-25 years old) who reported no previous difficulties in recognizing faces; Group 2 contained young university students (19-25) who claimed they were often embarrassed by their inability to recognize faces; and Group 3 consisted of older people (54-73) also attending university. Over a one-month period, Group 2 reported the greatest incidence of difficulties, while Groups 1 and 3 reported significantly fewer; in fact, the incidence of difficulty was identical for the two latter groups. Group 2, not surprisingly, had the greatest difficulty with recognizing faces, while most of Group 3 problems centred on attaching names to faces they recognized. The inability to consciously retrieve a name, given the face, is similar to the inability to directly remember episodic associations. Given enough time, however, most of the older participants in this study were able to recall the names they were looking for. Naveh-Benjamin (2000; Naveh-Benjamin, Hussain, Guez, & Bar-on, 2003) has shown, that older people’s problems with names and faces may have the same source as their more general difficulty with episodic memory. According to his associative deficit hypothesis, older adults have a “deficiency in creating and retrieving links between single units of in- formation” (Naveh-Benjamin, Guee, Kilb, & Reedy, 2004, p. 541). In one experiment, the ability to recall face-name associations was examined for two groups of men, one with a mean age of about 21 and the other with a mean age of about 72. All participants were shown 4o pairs of names and faces for three seconds per pair and were told to try to learn not only the faces and names, but also their pairings. They were then given a name recog- nition test, a face recognition test, and a face-name association test. In the frst test, they were shown a name they had seen in the first part ofthe experiment, paired with one they had not seen, and were asked to identify the one they had seen before. They were shown 16 such pairs in all. The face recognition test was the same, except that the pairs were faces. Finally, in the face-name association test participants were shown eight names they had seen before, each one paired with two faces, one they had seen previously and one they hhad not. They were also shown eight faces they had seen before, each paired with two names, one seen previously and one not. The task was to identify the names and faces they had seen before. In general, the older participants were less able than the younger ones to correctly identify names or faces when they were presented by themselves; however, the difference was quite small. The big difference associated with age became apparent when partici- pants had to correctly identify the name that went with a face, or the face that went with a name (Figure 5.1). In this task the older participants were considerably less successful than the younger ones. Thus the problem was not so much that older people don't recognize names or faces as it was that they don't bind them together as easily as younger people do. Older adults have trouble in situations requiring the “merging of different aspects of an ep- isode into a cohesive unit” (Naveh-Benjamin, Guez, Kilb, & Reedy, 2004, p. 542). Recalling5 | Memory Systems Cosme MCR Ce coed Many readers will be too young to remember it, but there was a time not too tong ago when humans’ access to information was much more limited than it is today. When the authors of this book were growing up they had to resort to books, libraries, or experts with spe- cific knowledge just to complete homework or figure ‘out which movie won the Oscar for Best Picture in 1976 (Google says it was Rocky—which demonstrates the point). Today information is never farther away than the nearest computer or smart phone. Indeed, the company name Google is naw an official word in the dictionary, and is regularly used as a verb However, recent research has suggested that even though itis vitwally indispens- able in our daily ves, the ability to “google” whatever information we need could have consequences for our memory. Sparrow, Liu. and Wegner (2011) devised an inter- esting experiment to explore whether having easy ac- cess to information could have an impact on how such information is encoded. In one of the experimental conditions participants were given a number of trivia statements that they had to type into a computer to verity. For example, a participant would read “An os- trich’s eye is bigger than its brain’ and then make that statement the subject of an Intemet search. The key manipulation was that half of the participants be- lieved the computer was saving whatever they typed, ‘whereas the other half were led to believe it was not, and therefore that the statements would not be ac- cessible at a later time. In a subsequent test of their ability to remember the trivia statements, participants performed better if they had thought that their search queries were being deleted. This could indicate that we will put less effort into ‘encoding and storing information it we think we will hhave easy access to it at a later date than we will if we think we won't have that access. In fact, in another ex- periment Sparrow, Liu, and Wegner found that memory for where to find the information (e.g.. which website was best) was better than memory for the information itself. By extension, although “googling it” can often give us instant answers, we may not remember those answers as well as we might wish, Google FIGURE 5.18 | Google search engine previous episodes becomes difficult because the parts of a previous experience have not been bound together to form a coherent whole. While older people may have difficulty consciously recal g recently experienced events, Howard, Fry, and Brune (1991, Experiment 2) found that they can still demon- strate knowledge of those events when tested more subtly. Younger (18-24 years) and older (62-75) people were asked to learn new associations. They were shown pairs of words (eg, queen-stairs, author-project) and asked to make up a sentence containing them. When participants were given the first member of a pair and asked to recall the 151152 Cognition second member, the younger people tended to do better than the older ones. However, older people did just as well as the younger ones on a more indirect test of memory. The indirect test was a word fragment completion task similar to those we mentioned ear- lier in the section on implicit memory. Participants were shown. the first word of a pair plus a fragment of the second. Sometimes the word-fragment pairs corresponded to pairs shown in the first phase of the experiment (e.g,, queen-sta—, author-pro—) but some- times the pairs were mixed up (eg, queen-pro—, author-sta—. Participants were asked to complete the stem with the first word that came to mind. Notice that in each case there are several words that could come to mind besides the one given (star, process, etc.) However, if the pair-stem combination corresponded to one they had seen before (e.g, queen-sta—), then both younger and older participants tended to choose the word originally shown (stairs), an FIGURES.a9 | Young andold meeting _effect that was not present if the word fragment pair had not been Winich one is betterat remembering names? seen before (e.g,, queen-pro—). This demonstrated that the older Korsakoff’s syndrome form of amnesia afecting te ably to Ferm ew memories attributes te thamine defciency and often though net exclusively seen in chronic aicohaiies participants had implicitly learned the new associations, even though they may not have realized it. Howard and colleagues concluded that you can teach old dogs new tricks—though only if they are given as much time as they want to learn those tricks. Self-pacing appears to be very important. Ina review of studies such as that of Howard, Fry, and Brune (1991), Mitchell and Brus (2003) confirmed that older adults do seem to be able to form implicit memories just as casily as younger people. They also respond to priming just as readily as younger people. Implicit memory appears to he stable across age. Thus memory deficits in older people tend not to be general (Rabbitt, 1990). Not only may there be great individual differences in the rate at which memory declines, but the extent of a deficit may depend on the context in which itis tested. Rabbitt (1990, p. 230) suggested that repeated testing may improve the performance of older people to the point that age differences are eliminated The Amnesic Syndrome Both Schacter (1987) and Baddeley (19872) pointed out the relevance of Korsakof’s syndrome for the study of memory. A form of amnesia affecting the ability to form new long-term memories (see Figure 5.20), Korsakoff’s syndrome is attributed to the atrophy of brain tissue resulting from malnutrition, particularly thiamine deficiency (Brokate, Hildebrandt, Eling, Fuchtner, Runge, & Timm, 2003) and is often (though not exclusively) seen in chronic alcoholics. Edouard Claparéde (1873-1940) was a pioneer in the investigation of this syndrome (Kihlstrom, 1995). Here is Claparéde’s (1911/1951) famous description of a 47-year-old Korsakoffs patient: Her old memories remained intact. She could correctly name the capitals of Europe, make mental calculations, and so on. But she did not know where she was, though she had been at the asylum five years, She did not recognize the doctors whom she saw every day, nor her nurse who had been with her5| Memory Systems 153 for six months. When the latter asked the patient whether she knew her, the patient said: “No Madame, with whom have I the honor of speaking?” She for- got from one minute to the next what she was told, or the events that took place. She did not know what year, month, and day it was, though she was being told constantly (p. 68) Schacter (1987) also reviewed research on the so-called amnesic syndrome. This is a disor- der produced by brain lesions, and is seen in patients with Korsakoffs syndrome, as well as Tulving’s patient NN, whose case we mentioned earlier in this chapter. Amnesic patients may be able to operate normally in many areas, but unable to remember events that have occurred since the beginning of their affliction. Talland (1968) described this kind of patient: FIGURE 5.20 | Korsakoff’s syndrome, or Wernicke-Korsakoff encephalopathy Note the pigmentation of grey matter, which occurs with thiamine (vitamin B,) deficiency, most often in chronic alcoholics University of Texas (Heuston) IE time has come to a stop for the amnesic patient itis because he remembers virtually none of the events that he has witnessed since the onset of his illness. ‘The days go by and none seems to be different from the others. Staff members and fellovr patients reappear looking no more familiar than complete strang- cers. A story gives as much satisfaction on the tenth asit did on the fist reading, its novelty never seems to wear off. Ifthe patient recognizes a new figure in his environment, his doctor for example, as someone familiar, he may easily con- fuse him with another figure encountered in the same environment or name him correctly but as diffidently as if it were a wild guess. In the literature there are several accounts of the medical examination that had to be interrupted for afew minutes, in which the patient greeted the doctor on his return as someone he had not met for along time (p. 123). Warrington and Weiskrantz (1982) suggest that these amnesic patients have a discon- disconnection nection syndrome: they may be able to acquire new information and yet not be aware that, syanres learning has taken place. It's asif there are at least two memory systems (Tlving, 1985) that ee normally interact but have become disconnected. This interpretation is reinforced by stud- oration na yet not ies using the fartous name paradigm described earlier. Squire and McKee (1992) showed be aar=thoternng nas lists of names to people with amnesia and a control group. Some of the names were famous Serpe (cg, Olga Korbut, a celebrated Olympic gymnast) while others were not (eg, Ema Lekovich ‘Then all participants were shown another set of famous and non-famous names, some of which had been on the previous list and some of which had not, and asked to rate them as famous or non-famous. Both the control group and the people with amnesia tended to rate a name as famous ifit had appeared on the first list, even if it was not actually famous. ‘Thus simply having been exposed to a name tends to influence participants’ judgments.154 Cognition However, while the participants in the control group were usually able to recognize a name as one that had appeared on the first list, people with amnesia were much less able to do so. They could judge a non-famous name to be famous because they had just seen it on the previous list, and yet not know that they had seen it Together, several studies reviewed by Warrington and Weiskrantz (1982) strongly sug- gest that amnesic patients do poorly on tasks requiring explicit memory, but much better on those requiring implicit memory. Graf and Schacter (1985) were able to further demon- strate this by giving a word-fragment completion test to both people with amnesia and a control group. Participants were presented with pairs of words, some of which were related by existing associations (e.,, buttoned~sbirt) and some of which were not (eg, window~ shirt) In one part of the experiment, participants were required to make up a sentence that related the words “ina meaningful manner” Thus, given the word pair ripe-apple, a partici- pant might generate the sentence “He ate the ripe apple.” Participants were then allowed to study each pair of words once. Each participant was given two tests: a word completion test for implicit memory, and acued recall test for explicit memory. In the former, participants were shown the first member of a word pair and the first three letters of the second member of the pair (the “word fragment”) and asked to complete the word. The interesting question was whether they would complete the fragment with the word they were shown initially. Both groups tended to do so, However, on the cued recall test, in which participants were given the first half of a word pair and had to recall the second, the people with amnesia were much less successful than the control group. Levy, Stark, and Squire (2004) also observed that people with amnesia showed similar performance to the control group on an implicit memory task, even though their perfor- mance on an explicit memory task was far below the control level. People with amnesia may be able to form associations, and thus learn new material (Schacter & Graf, 1986; Kihlstrom, Schacter, Cork, Hurt, & Behr, 1990). However, this learning would be available to them only in implicit, not explicit, form. Korsakoff himself (1899, pp. 512, 518, cited in Schacter, 1987, p. 503) described this phenomenon as follows: Although the patient was not aware that he preserved traces of impressions that he received, those traces . .. probably existed and had an influence in one way or another on the course of ideas, atleast in unconscious intellectual activity, We notice that a whole series of traces which could in no way be restored to consciousness, neither actively or passively, continue to exist in unconscious life, continue to direct the course of ideas of the patient, suggesting to him some or other inferences and decisions. That seems to metto be one of the most. interesting peculiarities of the disturbance about which we are speaking. Alzheimer’s Disease Alzheimer's disease is among the most feared memory disorders (see Figure 5,22). In an American survey conducted in 2002, 95 per cent of respondents agreed that Alzheimer’s ‘was a “serious problem facing the whole nation,’ and 64 per cent of those between 35 and 4g years of age said they were aftaid of getting it themselves (Halpern, 2002, p. 16). “Four5 | Memory Systems Although @ definitive diagnosis of Alzheimer’s stil requires an autopsy, the symptoms are better understood today than they were when this form of dementia was first identified, more than a century ago. y of In anormal brain, cells called neurons transport electrical messages using chemical transmitters (neurotransmitters). In Alzheimers patients, areas ofthe brain tissue are damaged or lost, and as a result some electrical messages do not transmit. FIGURE 5.22 | Alzheimer's disease From: wow alzheimersmemorylossinfolalzheimers-disease jpg million Americans already have the disease, a number that is expected to grow to fourteen nillion by mid-century” (Halpern, 2002, p. 16) Similar rates are likely in other industrial- ized countries (Di Carlo et al, 2002) ‘The criteria for diagnosis of Alzheimer's disease have become much clearer since the first case was recorded by Alois Alzheimer in 1907. The disease is progressive, beginning with a deterioration of episodic memory. A decline in the ability to retain recently acquired information is characteristic of the early stages (Hodges, 2000, p. 443). An ex- ample given by Jacoby (1999) illustrates how frustrating Alzheimer’s can be for caregiv- ers, He describes a woman diagnosed with Alzheimer’s who was taken to visit a nursing home before moving there. She was introduced to the customs of the nursing home, one of which was that there was no tipping in the dining room. This fact was repeated several times, Later, however, when the prospective resident was asked if she had any questions, she asked if she should tip in the dining room. “[Rlepeated asking of questions is one of the most striking and frustrating symptoms of memory impairment resulting from Alzheimer's disease’ (Jacoby, 1999, p. 3). 155156 Cognition Not everyone who is unable to learn new material will go on to develop Alzheimer’s. “It is... necessary to follow such patients for years in order to determine that they do indeed have” it (Hodges, 2000, p. 445). As the disease progresses, Alzheimer’s patients will show impaired semantic memory (Glosser & Friedman, 1991; Hodges, 2000, p. 445). For example, Hodges, Salmon, and Butters (1992) gave a group of Alzheimer’s patients a battery of semantic memory tests: asking them to name as many exemplars of a category (eg, animals) as possible; to identify the objects in drawings; to generate definitions of words (eg, alligator) that could be understood by “someone from a different country who has never seen or heard of such a thing’; and so on (p. 305). On all tests the Alzheimer’s patients performed less well than normal controls. Moreover, performance on one test was correlated with performance on others: for example, failure to identify the animal in a drawing as an alligator went along with an inability to define an alligator. This suggests that what Alzheimer’s disease involves is not so much the inability to retrieve existing knowledge as the deterioration of knowledge that once existed. Hodges, Salmon, and But- ters (1992, p. 312) observe that the definitions given by the Alzheimer's patients were par- ticularly instructive. They included very general, non-specific characteristics (defining a land animal as four-legged), as well as inappropriate intrusions (defining a rhinaceros in terms more appropriate for an elephant). Salmon, Butters, and Chan (1999), in a review of a large number of studies, concluded that the evidence supported the hypothesis of semantic memory deterioration in Alzheimer's patients: ‘The normal organization of semantic memory is disrupted by this loss of se- ‘mantic knowledge and the semantic network appears to deteriorate as the dis- ‘ease progresses. Although the neuroanatomical basis of the deterioration of semantic memory in patients with [Alzheimer’s is currently unknown, it is likely that results from synapse loss, neuron loss, and other neurodegenera- tive changes in the association cortices that presumably store semantic repre- sentations (p. 125) The Retraining of Memory Efforts to rehabilitate memory have usually heen based on intuition rather than adequate theories of memory disorders. B.A. Wilson and her colleagues (Kapur, Glisky, & Wilson, 2002; Wilson, 2002; Wilson & Moffat, 1984; Wilson & Patterson, 1990) have tried to identify treatment approaches that have a history of working andjor are based on a sound theor- etical foundation, Environmental Adaptations Minimizing the number of situations requiring memory can be helpful. A written time- table, located where the patient can’t miss it, can serve as a guide, helping him oz her move from one activity to another. Simply posting signs that tell patients which room they are in can be valuable (Giles & Clask-Wilson, 1988). In general, environments should be designed so that they elicit the desired behaviour.S| Memory Systems 157 External Memory Aids Prospective memory comes into play when we need to remember to do something at some future time, Older people in general, as well as those with memory disorders specifically, may forget not only when to do something, but also whether or not they have done it. For example, in addition to forgetting to take their medication on time, they may forget that they have done soand thus take it again (Einstein, McDaniel, Smith, & Shaw, 1998). An electronic diary that sounds an alarm when it’ time to perform a particular task and keeps track of the patient’s behaviour may help with this problem (Harris, 1984). Electronic organizers such as a smart phone may be useful too, serving as ‘prosthetic memories” with ‘a built in camera, handwriting recognition system and diary” (Abraham, 2004). The patient can take pictures of people as an aid to future recognition. These devices can also be programmed with a week's “to do’ list and prompt the user to “remember” on each occasion. Of course, teaching patients how to use the device is along and painstaking process, because most must rely on implicit rather than explicit memory. Sheer repetition makes a difference, and it’s important that the teacher help the patient avoid errors. Errorless learning is widely believed to max- imize patients’ ability to use whatever memory resources they still have, Patients learn to do only what should be done and never learn to do things incorrectly (Wilson, 2002, p. 667). Teaching Domain-Specific Knowledge Learning in amnesiacs is unlikely to be generalizable to contexts very different from the one in which the original learning took place. Indeed, there is no evidence that attempting to restore “general memory ability” through practice actually accomplishes anything for the patient (Kapur, Glisky, & Wilson, 2002, p. 772). A more realistic goal is to teach amne- siacs specific skills that might be useful to them. To this end, Glisky and Schacter (1986) reported on what they called the method of vanishing cues. Amnesic participants were taught the meaning of computer commands by being presented with definitions of the commands and fragments of their names (e.g, S~—for the command SAVE). Additional letters were presented until the participant guessed the word. Then letters were progres- sively removed until the patient was able to give the name of the command upon being presented with its definition. Glisky and Schacter report that this technique allowed am- nesic patients to successfully perform basic computer operations. In such situations it’s important that the material to be learned be concrete and specific and that the patient not be required to generalize very far from the original learning context Summary In this chapter we have examined memory systems, The modal model of memory has been especially influential in the study of memory. This approach to memory outlines how infor- mation arriving at our senses is frst stored in sensory, and then by shortterm memory, even- tually being transferred to long-term memory should sufficient processing be conducted, ‘The classification of the various systems of long,term memory reflects the influential work of Endel Talving, who was a pioneer in distinguishing between episodic and semantic memory. Episodic memory provides long-term storage for personal experiences and events, whereas seriantic memory stores factual information and general knowledge about the world prospective memory ‘The intention to remember tede something at some errorless learning Paticipants in alearning situation ae aught in such ‘a way tat they never have te opportunity to make ‘method of vanishing Away of teaching amnesic patients the meaning of ‘computer commands by presenting them with destin ofthe ‘commands and fragments ‘ofthe commands’ names ‘Adaliona letters ae presented untithe patent ‘uesses the word. Then letters are progressively remaved unt the pallets able ta give the ‘name ofthe command ‘nen presented with iederinion158 Cognition ‘The independence of these two memory systems is supported by studies of brain injured per- sons, especially those with amnesia. Tulving’s distinction between remembering and knowing is reflected in the butcher-on-the-bus phenomenon. The last 4o years have seen a lot of research oon phenomena associated with semantic and episodic memory, including tip-of the-tongue phenomenon, spreading activation, and involuntary semantic memory. The more recent literature suggests that there may be as many as five memory systems in all, Procedural memory contains our knowledge of how to do things that involve a sequence of operations, such as riding a bicycle. Such knowledge is tacit (implicit) rather than explicit because we arent necessarily aware of how we perform those operations. The perceptual representation system (the system associated with priming) contains specific representations of previously encountered events. It is responsible for priming implicit memory—that is, unintentional or unconscious recollection ofa prior episode (for example, experiments by Jacoby and Holling- shead). Working memory—the system that allows for temporary storage and manipulation of theinformation necessary for many cognitive activities—pulls all the other memory systems together. Baddeley’s working memory model has four components: the central executive, the phonological loop, the visuo-spatial sketchpad, and the episodic buffer. The connectionist model of memory provides an alternative to the idea of memory as a stored memory trace. In this model neuron-like units representing the various properties of an experience are connected to other neuron-like units. ‘The excitation or inhibition of specific connections between units creates a pattern of activity that constitutes memory. In evolutionary terms, procedural memory is considered the oldest system and epi- sodic the most recent. In terms of development, Tulving argues that we begin to develop episodic memory relatively late, at around 4-6 years. Do all forms of memory decline with age? A large body of research suggests that they do not, Studies by Howard, Fry, and Brune (4991) and Mitchell and Bruss (2003) show that even though older adults often have diffi culty recalling recent events, they form implicit memories just as readily as younger people do. Similarly, amnesic patients do not differ from control participants on implicit memory tasks; however, their performance on explicit memory tasks is below normal. Patients with Alzheimer’s disease experience a loss of semantic knowledge and a disruption in the organi: zation of semantic memory. For the elderly and some amnesiacs, the rehabilitation of mem- ory typically involves environmental adaptation, external memory aids, and instruction in domain-specific knowledge. CASE STUDY Case Study Wrap-Up At the beginning of this chapter we briefly discussed the strange memory deficit of Lucy Whitmore, the young woman played by Drew Barrymore in the movie 50 First Dates. Alter reading this chapter you should have a better idea of what Lucy's problem was. If you recall a car accident had supposedly left her unable to retain new memories from one day to the next: however, all the memories she had formed before the accident were intact. and she was able to form new memories from moment to moment. In fact, all the forms of memory discussed in this chapter (semantic. episodic, tacit, explicit, implicit) appeared to operatenormally for Lucy throughout any particular day: the only problem was that the new semantic and episodic memories disappeared overnight. Presumably Lucy ‘would have been able to retain new implicit memories, as patients with disconnection syndrome do, although this wasn’t demonstrated in the movie. What type of real-world amnesia most closely re- sembles Lucy Whitmore's memory deficit? Earlier in the chapter we discussed Korsakoff's syndrome, a severe form of anterograde amnesia in which the in- dividual is unable to form new memories. While this appears to describe exactly the deficit suffered by Drew Barrymore's character, there is a key difference that separates fact from fiction, Someone who suffers from Korsakoff’s syndrome (or an equivalent anterograde amnesia) is unable to form new long-term memories. The components of In the Know: Review Questions 5 | Memory Systems ‘working memory can often be intact, in which case the individual may be able to retain new information for 30 seconds or so. However, as soon as that information leaves working memory it is lost, apparently forever. Thus the real-life equivalent of Lucy Whitmore’s mem- ‘ory problem is much more severe than the movie sug- gests. While the fictional character's memory functions normally until she falls asleep. in real life people with Korsakoft's syndrome (or an equivalent anterograde amnesia) have no normal moment-to-moment mem- ‘ory function. If you introduced yourself to a Korsakoft's patient and then left the room for 30 seconds, he or she might not recognize you when you came back. If you enjoyed 50 First Dates you might want to check out Memento, which presents a much more ac- curate (though still embellished) picture of this par- ticular memory dysfunction. 1. Quiline Tulving’s approach to memory. emphasizing the distinctions between different memory systems 2. Discuss the role of spreading activation in semantic to illustrate memory, using relevant experiments 3. What do older people, people with the amnesic syndrome, and people with Alzheimer’s disease tell us about the nature of memory? What are the most effective ways of helping such people? Why? Key Concepts associative deficit hypothesis butcher-on-the-bus phenomenon central executive chunking consolidation crystallized systems decay declarative memory disconnection syndrome episadic buffer episodic memory (Tulving) errorless learning excitatory and inhibitory connections fluid systems hippocampus iconic and echoic sensory memory involuntary semantic memory ("mind popping! Korsekoff’s syndrome lexical decision task (.o7) method of opposition method of vanishing cues modal model of memory non-declarative memory perceptual representation system (prs) Phonological loop prime prime memory 159160 Cognition priming semantic memory (Tulving) probe or target spreading activation procedural memory tacit knowledge prospective memory tip-of-the-tongue phenomenon recency bias vs primacy bias, visuo-spatial sketchpad rehearsed. working memory secondary memory Links to Other Chapters episodic memory Chapter 9 (social context of language, Chapter 13 (memory and consciousness) evolution of language) Chapter 10 (flexblity-riaiity ano the brain) araintic sao Chapter 11 (natural deduction systems) Chapter 7 (synesthesia) ‘Chapter 12 working memory and g) working memory fluid and erystallized systems Chapter 6 (Are memory traces permanent?) Chapter 12 fluid intelligence and 9) Further Reading For additional angles on memory systems see Gaffan (2003), Moscovitch (2000), Roediger (1990), and Roediger and Blaxton (1987) Evidence from functional neuroimaging studies illustrating the variety of areas in the brain involved in semantic memory is discussed in Maguire and Frith (2004) and Thompson-Schill (2003). Tulving and his colleagues have presented a model suggesting that the left prefrontal cortex is more involved than the right prefrontal cortex in acquiring episodic memories, but that the reverse is true for recalling episodic memories. See Nyberg, Cabeza, and Tulving (2996) and Habib, Nyberg, and Tulving (2003), For a critique of this model see Owen (2003). It’s difficult for "normals’ to imagine what learning is like for amnesiacs. An analogy might be the kind of learning that takes place while a normal person is unconscious. For ex- ample, Kihlstrom, Schacter, Cork, Hurt, and Behr (1990) investigated memory for events that occur while a (non-amnesic) patient is anesthetized. They pointed out that following surgi- cal anesthesia, patients are typically not able to recall anything that took place while they were unconscious. However, there is also some evidence suggesting that events that take place during surgery can affect patients’ subsequent behaviour, although these findings are controversial For a thorough review that supplements our discussion of the rehabilitation of cognitive deficits see Park and Ingles (2001), Gaffan, D. (2003). Against memory systems. In A. Parker, A. Derrington, & C. Blakemore (Eds). The physiology of cognitive processes (pp. 234-51). Oxford: Oxford University Press, Habib, R.. Nyberg, L.. & Tulving, E, (2003). Hemispheric asymmetries of memory: The HERA model revisited, Trends in Cognitive Science, 7, 241-2445 | Memory Systems Kihlstrom, JF, Schacter, D. Cork, R., Hurt. C., & Behr, S. (2990) Implicit and explicit memory following surgical anaesthesia. Psychological Science, 1, 303-306. Maguire, EA. & Frith, C.D. (2004). The brain network associated with acquiring semantic. knowledge. Neurolmage, 22, 171-178 Moscovitch, M. (2000). Theories of memory and consciousness. In E. Tulving & FLM. Craik (Eds), The Oxford handbook of memory (pp. 609-25). New York: Oxford University Press. Nyberg, L., Cabeza, R., 8 Tulving, E, (1996). PET studies of encoding and retrieval: The HERA model. Psychonomic Bulletin and Review, 5, 135-148. ‘Owen, A. (2003). HERA today, gone tomorrow? Trends in Cognitive Sciences, 7, 383-384 Park, N.W,, & Ingles, J.L (2003). Effectiveness of attention rehabilitation after an acquired brain injury: A meta-analysis. Neuropsychology, 15, 199-220. Roediger, H.L. (1990). Implicit memory: Retention without remembering, American Psycholo~ gist. 45, 1043-1056. Roediger, H.L., & Blaxton, TA. (1987). Retrieval modes produce dissociations in memory for surface information. In D. Gorfein & R.R. Hoffman (Eds.), Memory and cognitive processes. The Ebbinghaus centennial conference (pp. 349-377], Hillsdale, NJ: Erlbaum. Thompson-Schill, S.L. (2003). Neuroimaging studies of semantic memory: Inferring “how” from "where." Neuropsychologia, 42, 280-292 164Memory Traces and Memory Schemas Chapter Contents Case Study: When Memory Falls 164 Intraduetion 165 The Trace Theory 165 Schema-based Theories of Memory 272 Scripts 182 Autobiographical Memory 283 Levels of Processing 187 ‘Two Approaches to Memory Research 230 To distinguish between memory traces and memory schemas. To outline the concept of flashbulb memories. To examine schema theory and review experimental evidence supporting it To understand the concept of encoding specificity. To review research into eyewitness testimony and false memories, To identity the strengths and weaknesses of the “levels of process- ing’ framework.Dy When Memory Fails The previous chapter highlighted some basic ideas about the various systems of memory. In this chapter we will focus on different types of long-term mem- ory, but first let's turn our attention to a case study that highlights the importance of memory and the dire consequences that can follow when memory fails In July 1984 a college student named Jennifer Thompson woke up and saw a man with a knife be~ side her bed. When she screamed, he held the knife to her throat and said he would kill her if she didn’t keep quiet. She offered him her wallet, even her car, but he told her he didn’t want those things. Realizing what he was there for, she made a conscious effort to study his face, looking for details she could remember later and use to identify him, Eventually she escaped to aneighbour’s by persuading the rapist to let her get hhim a drink Thompson worked closely with the police to cre ate a composite sketch, When she was shown a photo lineup she studied it carefully before deciding she rec- ‘ognized the man that the police had identified as the suspect. She picked out the same man, Ronald Cotton, in a physical lineup, and when the case went to court she swore that he was the one who had raped her. Cotton, 22, was sentences to life, In prison Cotton met someone who looked so much like him that other prisoners sometimes mis- took one for the other. Bobby Paole had been tiv- ing in the same North Carolina town as Cotton and Thompson, and was serving time for a series of rapes. When Cotton asked if he had raped Thompson. he denied it, but another inmate informed Cotton that Poole told him he had. Finally, in 1995 a ONa test confirmed Poole’s guilt, Cotton was exonerated, and FIGURE 6.1 | Jennifer Thompson and Ronald CottonThompson was overwhelmed with remorse. How could she have made a mistake that cost an innocent man more than 10 years of his life? In fact. eyewitnesses are very often mistaken. As of September 2015, ONA tests have led to the exoneration of 330 convicted men in the US alone, and in approx- imately 75 per cent of those cases faulty eyewitness testimony had been involved (Innocence Project 2015). Why are eyewitnesses so often wrong? How are false memories created? How is it that the people who hold them believe them to be true? These are among the many questions that this chapter will explore. 6 | Memory Traces and Memory Schemas We may not know everything there is to know about how memary operates, but one thing we do. know is that memory doesn't work the way a video camera does. You can't turn it on and off, it doesn’t simply record whatever is going on in front of its lens, and it can't be rewound and replayed at wil Rather, it seems that we capture the gist of incoming messages, assimilate that information with schemas that are already stored in our long-term memory, and then use those schemas to “fill in the blanks" when we're missing important bits and pieces of information, 165 Introduction We will begin this chapter by considering the distinction between a memory trace (@ con- cept introduced in Chapter 3) and a memory schema. Both Paul (1967) and Erdelyi (1985) suggest that Freud’s (1925/1961) mystic writing pad model might be a good place to start. ‘The mystic writing pad is a common children’ toy consisting of a sheet of waxed paper sandwiched between layer of black wax and a sheet of clear plastic (celluloid). When you write ‘on the plastic with a stylus, you can see what you've written because the waxed paper sticks to the wax wherever you have pressed on it. When you lift the waxed paper and plastic, the writing disappears from the overlay, but the imprint of the stylus remains visible on the black wax. The overlay is like our perception of an event. Such perceptions are transitory; we pass from one experience to the next. Memory traces are similar to what remains on the wax tablet after we lft the plastic—afier-effects of perception. The problem is that aver time they tend to run together. Thus if you examine the black wax layer after using the pad fora while, you will see numerous lines overlapping, as in Figure 6.2. How can you tell what was orig- inally written? You can't all you can do is reconstruct it on the basis of whatever evidence is available. In this case, given the traces that you can make out—a clear p followed by a slightly less clear e, two straight descending lines, what looks like another ¢, and a clear r—you would probably decide that the word writ- ten was pepper. Although the mystic writing pad analogy is imperfect—it’s highly un- likely that memory has the truly haphazard structure of the fragments that remain on a wax tablet—it may help to clarify the significance of the sche- ma-based theories that we will examine a little later in this chapter. First, however, we will take a look at how the trace theory has evolved. mystic writing pad model A model of memory based ona toy writing tablet that tans ragments of ld messages even after they have been ‘erasea” In time, these fragments accurate and begin tecoveriap. so that ney become increasingly har to reo. FIGURE6.2 | The mystic writing pad From: Etdely, MM. (1985). p. 129. Copyright 1985 by Matthew Hugh Erdelyi Used with permission DISCOVERY LAB The Trace Theory For a long time, memory traces were assumed to be permanent and complete copies of past events, and remembering was thought to be like re-experiencing the past; a modern166 Cognition reappearance hypott Neisse's term forthe now rejected ieea that ine same memory can reappeat, Unchanged, again an again flashbulb memories Vive, detailed memories of signicart events Now Print! theory ‘The theory tat especially significant experiences tre immediately "photocopied, preserved Intong-term memory and resistant to change, analogy might be a video recording that can be preserved indefinitely and replayed over and over. This notion is what Neisser—who rejected it—called the reappearance hypothe- sis: “that the same ‘memory’... . can disappear and reappear over and over again’ (1967, p. 282). To the contrary, he followed Bartlet (1932) in arguing that there are no ‘stored cop- ies of finished mental events” and that memory is schematic, relying on “fragments ... to support a new construction’ (p. 286). Today, as we will see, virtually all memory theorists agree. Itnow seems clear that even the most vivid memories are subject to change. Flashbulb Memories Even as the idea of the schema was gaining ground, some psychologists argued that one particular type of memory was in fact permanent. For example, a great many people who ‘were of school age or older in 1963 still claim to remember exactly how they learned of John F. Kennedy's assassination. These memories were first investigated by R. Brown and Kulik (4977), who in the mid-1970s asked 80 Harvard undergraduates to recall the circumstances under which they heard the news. In addition to writing a free recal? account of what they remembered, they estimated how consequential they felt the event was at the time, and how frequently they had talked about it. A similar procedure was used to study memories of the assassinations or attempted assassinations of several other prominent figures, such as Martin Luther King, Ir The students’ accounts typically included information on five specific subjects: Where they were when they learned of the assassination; What they were doing at the time; The person who told them; ‘Their affect (how they felt at the time); and 5. Theaftermath: what they did immediately after hearing the news. Fee Almost every participant had what appeared to be vivid, detailed memories not only ofthe Kennedy assassination, but often of the others, and the more consequential they felt the event to have been, the more often they had rehearsed it (.e, discussed it with others). ‘The term that Brown and Kulik (1977) coined to refer to such unusually vivid and detailed accounts was flashbulb memories. To explain how such memories are produced, they expanded on Livingston's (1967) Now Print! theory to create the model illustrated {in Figure 63. The Now Print! theory, as originally envisioned by Livingston (1967), resem- bled the production of a photocopy’ just press the Print button and your brain will store a faithful reproduction of everything in the scene, including the context in which the experience occurred. In short, flashbulb memories would be examples of highly detailed memory traces. The sequence that Brown and Kulik proposed has five stages. First, the stimulus event is tested for ‘surprisingness."Ifit’s completely ordinary, we will pay no attention toit. However, if the event is extraordinary (as assassinations generally arg), then we will pay very close at tention. In the second stage the event is tested for consequeatiality. Events that fal this test vill be forgotten, but those that we consider important as well as surprising will move on to the third stage, in which flashbulb memories are formed. Flashbulb memories will vary in vividness and completeness depending on how surprising and consequential the events are.6 | Memory Traces and Memory Schemas Stimulus event FIGURE 6.5 | Brown and Kulik’s model of flashbulb memories 167 From: Brown, R. & Kulik, J, (2977). Copyright 1977 by Elsevier Science Publishers. Reprinted by permission ‘The fourth stage is rehearsal, in which we think about those memories and develop verbal accounts of hem, Finally, in the fifth stage we tell and retell those accounts to other people. Investigating the Flashbulb Hypothesis Brown and Kalik’s work attracted a great deal of attention, and a number of historically important events were eventually investigated as possible sources of “flashbulbs.” Among them was the Challenger space shuttle explosion (Figure 6.4) in January 1986 (Bohannon, 1988; Bohannon & Symons, 1992; McCloskey, Wible, & Cohen, 1988). Three days after the disaster, McCloskey and colleagues asked 45 people the following questions: 1. Where were you when you frst learned of the Challenger explosion? 2. What were you doing? 3, Did you see the event as it was actually happening, or did you learn about it later? 4. What were your first thoughts on hearing the news? Then, approximately nine months later, they had 27 of the original participants, along with 31 new people, complete the same questionnaire. All participants remembered something about the circumstances in which they had heard about the disaster. However, comparison of the immediate and nine-month168 Cognition FIGURE 6.4 | Flashbutb memories: The Challenger explosion questionnaire data showed that quite a bit of information had been lost over the interval, and that the details were not always consistent, Although seven of the nine-month accounts ‘were mote specific than the immediate accounts, 20 were more general (less specific). For example, someone who had initially named the person he heard the news from might only refer to “a friend” nine months later. In addition, seven of the nine-month accounts were inconsistent with the original reports on matters such as where the respondents were when they heard the news (eg, sitting at the desk versus walking out the office door). Even so, it important to note that none of the nine-month accounts were wildly in- consistent with the easlier versions. Rather, the inconsistencies were “the same sort of re- constructive errors that seem to occur frequently for ‘ordinary’ memories.” In both cases, “inaccuracies may be introduced when information that cannot be retrieved from memory is filled in through inference or guesswork” (McCloskey et al, 1988, p. 175) In addition, McCloskey et al. concluded that there is no need for any special mech- to account for so-called flashbulb memories: if such memories seem easier to recall in vivid detail than ordinary memories, it’s only because we have replayed them so often and thought about them so much, Recall that in Brown and Kulik’s model, flashbulb mem- ores are associated with events that are sufficiently significant not only to attract intense attention, but also to become the subject of frequent rehearsal, both in the media and in discussions with others. These factors in themselves could explain why we might feel that our own flashbulb memories are particularly strong, A number of studies have explored flashbulb memories concerning the destruction of the World Trade Center in New York City on 11 September 2001 (see Figure 6.5) (e.g, Greenberg, 2004; Neisser, 2003). Talarico and Rubin's (2003) study was particularly well designed. On 12 September 2001, they had 54 Duke University students complete an open-ended questionnaire (similar to Brown and Kulik’s) asking for descriptions not only6 | Memory Traces and Memory Schemas DAWN eee CAML e es Coto kee ced Imagine that you have just been attacked on the street by three people. Fortunately, in the midst of the attack ‘you see a police officer racing towards you. But he runs right past! The officer later says he was chasing some- one else and did not recall seeing what was happening to you. Do you believe him? Remember your answer. This very scenario played out in Boston in 1995 Around 200 am, police officer Kenneth Conley re- sponded to a call about a shooting and ultimately gave chase to the suspect. During the chase he ran right past three fellow police officers who were beating a man they mistakenly believed to be the suspect (in fact. he was an undercover police officer). Conley claimed he never saw the attack, but the jury rejected his story. He was later convicted of perjury and obstruction of justice and sen- tenced to 34 months in jail As unlikely as this scenario seems, it raises a serious question: could Conley really not have seen a brutal assault that was taking place right In front of his eyes? This real-world example should remind you of our earlier discussions of the use of cell phones while driv- ing (Chapter 2) and inattentional blindness (Chapter 4) In both cases, the point was that people whose atten- tion is distracted can sometimes fail to consciously perceive things that are right in front of their eyes This obviously has a direct impact on their ability to remember things. If Conley never had any conscious perception of the assault, how could he have had any recollection of it? Conley’s conviction was eventually overturned on. technical grounds. Recently, though, a group of re- searchers at Union College in New York set out to deter- mine if his story could in fact have been true (see Chabris, Weinberger, Fontaine, & Simons, 2011}, In thelr study they asked participants to run behind a confederate (some ‘one helping with the experiment), while three other con- federates staged a fight along the route, Amazingly, a post-experiment questionnaire revealed that only 35 per cent of the participants recalled noticing the fight when the experiment was conducted at night (the Conley inci dent took place in the dead of night). And although the percentage jumped to 56 per cent in daylight, almost half the participants stil failed to notice the fight! This ‘experiment highlights the importance of attention and conscious perception for memory. If the event was not consciously perceived, then memories of it if they exist at all) will be degraded, After reading the previous three chapters you should be aware of how interconnected many cognitive processes are. That is it should be ap- parent that sensation, perception, and attention all play significant roles in how and what we remember. Now recall your answer to the question we asked in the first paragreph above. Would you have believed Conley? Do you think he should have been convicted? of the momentous events of the previous day, but also of an ordinary event, such as a party, that each participant had recently experienced. In addition, Talarico and Rubin used other measures to test additional aspects of the flashbulb phenomenon, such as the intensity of the emotion felt when the events were recalled. They then divided the 54 participants into three groups of 18 each and se-tested each group once. The first group was tested 1 week later; the second, 6 weeks later; and the third, 32 weeks later. The ma- jor variable of interest was the consistency of the accounts given at the three different intervals. For example, if a participant said on 12 September that “Fred” was with him when the event occurred and later said that “Alice” was with him, but not “Fred,” that response was scored as inconsistent. Each participant's recall was given a consistency 169170 Cognition consolidation theory ‘The classic treory thet memory taces of an event ae not fully formed irmmeitely attr that event, but take some tine FIGURE 6.5 | Flashbulb memories: Terrorist attacks on the World Trade Center, New York City score based on the number of details consistently recalled, as well as an inconsistency score. Figure 6.6 shows the change in consistency and inconsistency scores as a function of time. Notice that both flashbulb and everyday memories show a decline in consistency and an increase in inconsistency. Although the flashbulb memories had more emotion associated with them, in terms of their actual content they were no more accurate than “ordinary” memories. However, participants erroneously believed that their flashbulb memories were more accurate than their “ordinary” memories, Talarico and Rubin con- cluded that although a flashbulb event ‘reliably enhances memory characteristics such as vividness and confidence,” people should not put too much faith ‘in the accuracy of their flashbulb memories” (2003, p. 460) Are Memory Traces Permanent? ‘These flashbulb memory experiments have led many to question the idea that memory traces persist unchanged over time. Asa result, the trace theory has undergone consid: erable modification (Dudai, 2004; Nader, 2003; Wixted, 2004a). According to the classic consolidation theory (Woodworth, 1938, p. 51), memory traces of an event are not fully6| Memory Traces and Memory Schemas 4171. [i Fashbulb Il Everyday SB Consistent items Week 6 Weeks 32 Weeks Time [Ei oshbulb ll Everyday Inconsistent items Week 6 Weeks 32 Weeks Time FIGURE 6.6 | Properties of flashbulb memories Notice that the y-axes of the two graphs are not to the same scale (ata trom: Taarico, 0M. & Rubin, D.C, |2003), formed immediately after that event, but take some time to consolidate. This process of —_2_s consolidation can be disrupted by events that occur after the event to be remembered; retroactive such disruption is called retroactive interference. Woodworth’s (1938, p. 227) review of the interference classic literature (eg, Jenkins & Dallenbach, 1924) concluded that “rest immediately after. Serine nrerabot one earning... allows for full consolidation of the traces, while strenuous mental work just at event this time... leaves the traces weak. This conclusion is echoed in the recent literature. For example, Wixted (2004a, p. 247) argued that “even if the intervening study material is not related to the original learning in any obvious way, the new learning draws on a limited pool of resources that may have otherwise been available to consolidate the original learn- ing. Asa result, memory for the original material suffers.” Not surprisingly, sleep depriva- tion has also been associated with deficits in memory (see Box 65, later in this chapter) It is known that the hippocampus is a crucial site for the consolidation of memory traces, converting immediate memories into long term memories. “If the hippocampal for mation is damaged before the consolidation process is complete, recently formed memo- ries that are still undergoing the consolidation process will be impaired” (Wixted, 2004a, pp. 242) Itis likely that retroactive interference occurs because “ordinary mental exertion172 Cognition eee reconsolidation ‘The hypetnenical process ‘nereny amemary ce Is reased ane reconsoldated method of repeated reproduction One partcipantis given rmutiple opportunities to recall a story overtime method of serial reproduction Onepartcipant. A. wites co what he a she can fecal of previously read Story. Aversion given toasecend parcpant 3 wt react hen es to reproduce 85 version inturn ven tC. and and memory formation’ detract from an ongoing process of [hippocampal] consolidation” (Wixted, 2004a, p. 264). Itwas long believed that once the consolidation process was complete, then the mem- ory trace in question was fixed and permanent. However, it now appears likely that when the stored trace is re-activated, it becomes labile (.e., changeable) (Dudai, 2004; Nader, 2003). Thus recalling a previous experience places it in working memory (see Chapter 5), where it comes into contact with other experiences. For example, the context in which you recall a flashbulb event may be quite different from the context in which you originally experienced it This provides an opportunity for revision of the memory trace, although the extent of such revision is controversial. In any case, the revised trace would then undergo reconsolidation in the hippocampus (Nader, 2003, p. 66). Furthermore, there is nothing to say that this process cannot become a cycle whereby a memory trace is reactivated and reconsolidated indefinitely. Thus we have no reason to believe that a memory trace is nec- essarily a faithful rendition of the original experience. Nader (2003, p. 70) reviewed the memory reconsolidation literature, and his conclusion is extremely important: “There can be no doubt at this point that memories are funda- mentally dynamic processes, as first explicitly demonstrated by Bartlett (1932). They are not snapshots of events that are passively read out but, rather, are constructive in nature and always changing” (Loftus & Yuille, 1984; Tulving & Thomson, 1973; Schacter, 1999). In reaching this conclusion, Nader paid tribute to Frederic Bartlett (1886-1965) and in par- ticular his 1932 book Remembering, Itis to an examination of that work and its influence on cognitive psychology that we will now turn. Schema-based Theories of Memory Probably no idea is more important in cognitive psychology generally, and in theories of memory in particular, than the idea of the schema. When we first encountered the schema, in Chapter 1, we defined it in terms of what we expect to find as we explore the world, How- ever, we could also say that a schema is something that helps us to organize and categorize the information we receive. Bartlett and the Concept of the Schema It was Bartlett who made the schema concept central to the psychology of memory (Roediger, 1997; Thompson, 1997; Weiskrantz, 2000; Zangwill, 1972). His best-known. experimental techniques were the method of repeated reproduction and the method of serial reproduction. In the former, a participant, A, is given a story to read and then attempts to reproduce it, first 15 minutes later and then at longer intervals. The latter begins the same way, but A's version is then given toa second participant, B, who reads it and then writes down what he or she can recall of it. B's version in turn is given to C, and so on (if youever played “telephone” in elementary school, you'll recognize the principle). Thus each participant tries to reproduce the previous participant’ version of the original story based solely on what he or she can recall. The results showed the same pattern in both cases, but were particularly dramatic with the method of serial reproduction. The story to be repro- duced was the following North American folktale:6 | Memory Traces and Memory Schemas The War ofthe Ghosts One night two young men from Egulac went down to the river to hunt seals, and while they were there it became foggy and calm. Then they heard war- cries, and they thought: “Maybe this is a war party” They escaped to the shore and hid behind a log. Now canoes came up, and they heard the noise of pad- dles and saw one canoe coming up to them. There were five men in the canoe, and they said: “What do you think? We wish to take you along, We are going up the river to make war on the people.” One of the young men said: "Ihave no arrows.” “Arrows are in the canoe,” they said. “L will not go along. I might get killed. My relatives do not know where Thave gone. But you,” he said, turning to the other, “may go with them.” So one of the men went, but the other returned home. ‘And the warriors went on up the river toa town onthe other side of Kalama. ‘The people came down to the water, and they began to fight, and many were killed, But presently the young man heard one of the warriors say: “Quick, let us go home: that Indian has been hit.” Now he thought: “Oh, they are ghosts.” He did not feel sick, but he said he had been shot. So the canoes went back to Egulac, and the young man went ashore to his, house and madea fire. And he told everybody and said:“Behold | accompanied the ghosts, and we went to fight. Many of our fellows were killed, and many of those who attacked us were killed. They said I was hit, and I did not feel sick.” He told it all, and then became quiet. When the sun rose he fell down. Something black came out of his mouth. His face became contorted. The people jumped up and cried. He was dead (Bartlett, 1932, p. 65). To test your own recall, stop reading now and wait 15 minutes; then try to reproduce the story. In Bartlett's serial experiment, successive reproductions became increasingly different from the original. By the tenth person the story had become the following: ‘The War of the Ghosts (2) ‘Two Indians were out fishing for seals in the Bay of Manpapan, when along came five other Indians in a war canoe. They were going fighting. “Come with us,” said the five to the two, “and fight” “I cannot come,” was the answer of the one, “for I have an old mother at home who is dependent upon me.” The other also said he could not come, be- cause he had no arms. “That is no difficulty? the others replied, “for we have plenty in the canoe with us’; so he got into the canoe and went with them, Ina fight soon afterwards this Indian received a mortal wound. Finding that his hour was come, he cried out that he was about to die. “Nonsense? said one of the others, “you will not die” But he did Bartlett, 1932, p. 124). 173174 Cognition rationalization ne attempt to make memory coherent and sensibleas possible DISCOVERY LAB schema (Bartlett) anactive mats of organized past reactions tat provides a setting that ‘guides our behaviour selection The hypothesis that we select niotmation both as wereceivei and as we fecal abstraction The hypothesis thatwe tend to remember only the ist not the species, of wnat we experience interpretation The hypothesis that we interpret information by making inferences, inferences as partof the ‘origina invormation, integration Thehypothesisthatwe absvact the meaning of| an event ana then put tat ‘meaning togetner with the rest of cur knowledge te forma coherent corsistent whole, Bartlett believed that this experiment showed what happens to memory over time. Obviously, several parts of the original were dropped along the way, so that the story was simplified. Although the title was seported correctly in reproduction 10, there was no longer any mention of ghosts. Participants tended to select some material to remember and ‘omit other material. These omissions reflect a process of rationalization as each partici- pant tried to make the story as coherent and sensible as possible, from his or her viewpoint. Material that did not seem to fit tended to drop out of the narrative. Recall the death of the Indian in the original version: “When the sun rose he fell down, Something black came out of his mouth. His face became contorted.” Over successive reproductions this passage changed substantially~a “black thing rushed from his mouth’ ‘his soul fled black from his mouth,’ *his spirit fled"—until finally, by version 10, these details had disappeared al- together. Bartlett noted that unfamiliar material was transformed over time to conform to more familiar patterns. ‘On the basis of his experiments, Bartlett concluded that remembering is not the re-excitation of innumerable fixed, lifeless and frag- mentary traces. It is imaginative reconstruction, or construction, built out of the relation of our attitude towards a whole active mass of organized past re- actions or experience, and to a little outstanding detail which commonly ap- pears in image or in language form (2932, p. 213). This “active mass of organized past reactions” is what Bartlett meant by the term schema. Thus a schema in Bartlett's sense is an organized setting that guides our behaviour and memory, a standard that can be adjusted to hit changing circumstances ‘As an example Bartlett used the ability to make the proper stroke in a sport such as tennis, You must adjust your position, posture, and bodily movements (swing) to fit the current situation. You dont simply repeat a stroke that you have performed before, because the bal is not likely to be in exactly the same place twice, and neither are you. Over the years many studies have employed the schema concept in the same gen- eral way that Bartlett did. Several reviews (eg., Alba & Hasher, 1983, p. 20. Goldsmith, & Pansky, 2000, pp. 494-495) suggest that most schema theories discuss memory in terms of four processes: selection, abstraction, interpretation, and inte- gration. The schema selects information consistent with our interests at the time. We then convert that information into a more abstract form. In other words, instead of trying to preserve the specifics of the event in its entirety, we extract its gist, or mean- ing. We then interpret that information in terms of other information in our mem- ory, Finally, we integrate the information in such a way as to make it consistent with the schema. Koriat, Goldsmith, and Pansky (2000) add a fifth process, reconstruction, whereby the act of recall blends general knowledge and individual experiences in or- der to “imaginatively reconstruct” the past. Thus after you've finished reading this chapter, it’s unlikely that you will remem- ber every word of it (though we certainly hope you'll remember some of it). You will probably have selected some points to remember and let others go, depending on your interests and concerns at the time of reading. If someone asks you what the chapter is about, you're more likely to provide a selective abstract than a literal recap. Over time, you may realize that you did not simply interpret the ideas in this chapter: you also6 | Memory Traces and Memory Schemas integrated them with other ideas that you had already taken in. For example, the idea that the schema is selective may remind you of a similar point about attention that you read in Chapter 4. Moreover, the processes we are now discussing may reflect the ‘way information is encoded—another process mentioned in Chapter 4. Of course, these connections may seem somewhat vague (ie., schematic), and you may need to reread those sections in order to fully remember the concepts discussed earlier. Finally, years from now, if you set out to reconstruct what you learned from this chapter, you may have only fragmentary clues to work with, You may have trouble remembering when you took this course, what grade you got for it, who was in your class at the time, or even which textbook you used. In sum, a schema is a flexible organization, and that is what makes it useful. If our memory were just a collection of traces, it would too rigid to be useful. A schema is a more abstract and general setting within which memory traces have meaning. It's important to note that Bartlett never denied the existence of memory traces (Ost & Costall, 2002). The case of Professor Aitken (Hunter, 1977) isa very good example of the way memory traces and memory schemas can work together; see Box 6.2. Eyewitness Testimony A classic study by Loftus and Palmer (1974) investigated how eyewitness testimony can be affected by the way questions are phrased. All participants were shown the same film ofa traffic accident and then asked to estimate the cars’ speed. For some the question was phrased “About how fast were the cars going when they hit each other?” For others the basic question was the same, but the word bit was replaced by smashed, collided, humped, or contacted. The results are given in Figure 68, Notice that the estimate of the cars’ speed is a function of the intensity of the verb in the question. Ifthe question used the word smashed, then the cars were reported to have been going faster than if they had merely come into “contact” with one another. In another experiment, participants were also shown a film of collision and then asked cone of two questions: either “How fast were the cars going when they hit each other?” or “Flow fast were the cars going when they smashed into each other?” One weekelater, the same participants were asked some additional questions about the accident they had seen on film the week before. One of the questions was “Did you see any broken glass?” The participants ‘who had earlier been asked if the cars “had smashed into each other” were more likely to report seeing broken glass than were the participants who had been asked ifthe cars “had hit each other” (Figure 69). Loftus and Palmer (1974) interpreted these results as follows: We would like to propose that tw kinds of information go into one’s memory for some complex occurrence. The frst is information gleaned during the per- ception of the original event; the second is external information supplied after the fact, Over time, information from these two sources may be integrated in such a way that we are unable to tell from which source some specific detail is recalled. All we have is one “memory” (p. 588). In several subsequent publications, Loftus and others provided a great deal of evi- dence that misleading post-event information often becomes integrated with the original 175176 Cognition Cfokel MMe MW Coto MACE Scoot oC B Vea Professor A.C. Aitken (1895~1967) (Figure 6.7) was a“bril- liant mathematician ... and an accomplished violinist” whose phenomenal memory was studied for decades bby psychologists (Hunter, 1977). Often people with an extraordinary memory are said to have a correspond- ingly poor ability to think abstractly, but that was not the case with Aitken. In fact. it was precisely because of his ability to rapidly schematize information that his ‘memory was s0 prodigious. According to Hunter (1977, P. 357), Aitken would take even the most mundane ‘events and weave them into an “unusually rich densely structured gestalt of properties” that stil preserved the uniqueness of each event—a process that exemplified what Bartlett (as we noted in Chapter 4) called “effort after meaning.” Typically, Altker’s ability to remember was an unintended consequence of his desire to make things meaningful He never wanted to “memorize” anything merely by rote. Aitken felt that the most important thing was to relax and become absorbed in the material to be learned. Once you allow yourself to become inter~ ested in it, then you will begin to comprehend it in deeper and deeper ways. The more deeply you un- derstand something, the better you will remember it. (This is a theme that will recur towards the end of this chapter, when we discuss the “levels of processing” approach to memory.) Aitken observed that “the thing to do is to learn by heart, not because one has to, but because one loves the thing and is interested in it” (Hunter, 2977, p. 158. Because he was interested in his own mental pro- cesses, Aitken agreed to be a participant in a variety of experiments. In 1933 he was asked to memorize 25 un- related words. Twenty-seven years later, in 1960, he at- tempted to recall them, and succeeded. However, Aitken did not repeat even this “meaningless” list of words by rote, Rather, he proceeded by inference to reconstruct the list. Aitken’s memory operated in a schematic way even for a list of words. (We will examine some very different techniques for memorizing large amounts of ‘material in our discussion of imagery, in Chapter 7) More typical of Aitken’s memory was his knowl: edge of music, He believed that musical memory could be both rich and precise because music has so many different aspects, including “a metre and a rhythm, a tune, . .. the harmony, the instrumental colour, a parti- cular emotion, ... a meaning, ... perhaps a human in- terest in the composer’ (Hunter. 1977, p. 157). All these elements combine to create @ framework within which the act of recall can take place. You may be reminded of Aitken’s musical memory later in this chapter, when we discuss how actors learn their lines, FIGURE 6.7 | Professor A.C. Aitken6| Memory Traces and Memory Schemas 177 22 40 3a 36 3a Est, speed (mph) 32 30 Smashed Collided Bumped Hit Contacted Verb FIGURE 6.8 | Results of Loftus and Palmer's experiment From: Loftus, £8 Palmer, 3. (2974) Reprinted with permission trom Elsevier. Leading question asked —~Possible schemas activated ——_—Response of subjects asked during witness testimony cone week later, ‘Did you see any broken glass?” (There was none} ‘About how fast were the “Yes—14% cars going when they hit each other?” ‘About how fast were "Yes—32% the cars going when they smashed Into each other?” FIGURE 6.9 | One-week follow-up of Loftus and Palmer's experiment From: Lofts, €F. 6 Palmer. 3. (1974, Reprinied wih permission rom Eaever ——mwe information (eg, Cole & Loftus, 1979; Loftus, 1992, 2003; Loftus & Loftus, 1980). This has yisinformation effect been termed the misinformation effect (Loftus, 2004, p. 868; Loftus & Hoffman, 1986). The yoornes rat According t0 Loftus (2004, p. 145), recent research strongly suggests that more can be msiidingaort-evert changed than “a detail in memory for a previously experienced event” Indeed, it appears gtlsmieswihine ouaral guite possible to ‘plant an entirely false memory into the min memory ofthe eve,