Cognitive Perspectives on Memory PDF

# CHAPTER SIX INTRODUCTION TO COGNITIVE PERSPECTIVES ## Table 6.1 Key characteristics of the three components of the dual-store model of human memory | Component of the Human Memory System | Means Through Which It Receives Information | Capacity | Forms in Which Information Is Stored | Duration o...

# CHAPTER SIX INTRODUCTION TO COGNITIVE PERSPECTIVES ## Table 6.1 Key characteristics of the three components of the dual-store model of human memory | Component of the Human Memory System | Means Through Which It Receives Information | Capacity | Forms in Which Information Is Stored | Duration of Contents | |---|---|---|---|---| | Sensory Register | Various senses | Very large | Neurological equivalents of the forms in which information has been received | Typically less than a second for visual information; can be 2-3 seconds or a bit longer for auditory information | | Working Memory | Attention to certain contents of the sensory register | Quite limited | Various sensory modalities; sometimes encoded in a different modality than that in which it was originally received (e.g., verbal information is often encoded in a subvocal auditory form, even when presented visually) | Often only a few seconds unless it is regularly refreshed (e.g., through maintenance rehearsal) | | Long-term Memory | Cross-communication with working memory | Virtually boundless | Several possible ways, including language-based representations, sensory images (e.g., visual images), and nonverbal abstractions | Indefinitely long | ## Challenges to the Dual-Store Model Up to this point, we've been talking about the dual-store (three-component) model of memory almost as if it were Ultimate Truth. But not all psychologists agree that this model accurately represents how human memory functions. I myself sometimes wonder whether the three components are the distinctly different entities that the model portrays. For example, recall my earlier point that auditory information in the sensory store typically lasts for only a few seconds. Also recall how Baddeley's phonological loop can maintain a list of spoken words in working memory only to the extent that the list is short and consists of quickly pronounceable words—a list that can be repeated in, say, 2 to 4 seconds. In fact, Baddeley (2007) has suggested that auditory information stored in working memory tends to last only a few seconds unless it's rehearsed. I'm guessing that a sensory register and working memory might be overlapping mechanisms in this situation. Furthermore, a number of theorists have argued that working memory and long-term memory are actually different aspects of a single (rather than dual) storage mechanism. Others have challenged a related idea: They wonder whether active, conscious processing in working memory is really necessary for storage in long-term memory. We now look at some of the evidence related to each of these issues. ## Are Working Memory and Long-Term Memory Really Different? Earlier in the chapter I described a serial learning task—a task in which an experimenter presents a list of items (e.g., "giraffe, Jason, cabbage,... ")—and asks people to recall the list in its original presentation order. A serial learning curve is usually observed when such tasks are given: People learn the first few items and last few items more quickly and easily than they learn the middle items (J. F. Hall, 1971; McCrary & Hunter, 1953; Roediger & Crowder, 1976). If we were to graph the speed with which the various items in a serial list are learned, we might obtain results similar to what you see in Figure 6.8. A common example is the way in which most children learn the alphabet: They learn the first letters (A, B, C, D) and the last letters (X, Y, Z) before they learn the middle letters (e.g., J, K, L, М). Using a dual-store model of memory to explain this curve, we might say that people process the first few items sufficiently to store them in long-term memory, and they continue to hold the last few items in working memory after the entire list has been presented (e.g., Baddeley et al., 2009; Norman, 1969). People lose many of the middle items because they don't have enough time to process them adequately before later items "bump them out” of working memory. Supporting this interpretation is the finding that people remember a greater number of words at the beginning of the list when presentation rate is slowed down—allowing for more processing in working memory—and may not remember any early items at all if processing is prevented (Glanzer & Cunitz, 1966; L. R. Peterson & Peterson, 1962). In contrast, recall for words at the end of the list seems to be more affected by the recall interval: If people have to wait a while before they can recall the list—decreasing the chances that anything is still in working memory—they have trouble remembering items at the end of the list (Glanzer & Cunitz, 1966; Horn, 2008; Postman & Phillips, 1965). Yet other research studies have cast doubt on the idea that the serial learning curve necessarily reflects the use of a working memory separate from long-term memory (R. G. Crowder, 1993; R. L. Greene, 1986; Öztekin, Davachi, & McElree, 2010; Wickelgren, 1973). For example, in a study by Thapar and Greene (1993), college students viewed a series of words presented two at a time on a computer screen; they also performed a 20-second distractor task (mentally adding a series of digits) after each pair of words. The students remembered the last few words in the list much better than the middle words, even though—thanks to the distractor task—none of the words could possibly have still been in working memory. Considering results such as these, some psychologists have suggested that the serial learning curve can be explained as easily by a single-store model as by a dual-store model. One possible explanation is that items in a list are easier to remember if they're distinctive in some way. Items near the end of the list might be more memorable because of their positions: A learner may specifically identify a word as "the last one" or "the next-to-last one” (R. L. Greene, 1986; Unsworth, Heitz, & Parks, 2008; R. K. Wagner, 1996). A second possibility is simply that forgetting occurs rapidly at first and then slowly tapers off—a pattern that has been observed for many different species and many different tasks (J. R. Anderson, 1995; Wickelgren, 1973; Wixted & Ebbesen, 1991). From this perspective, easy recall of the last few list items may be the result of the fact that the last items of a list haven't yet undergone that rapid decay. Another body of evidence that has been used on both sides of the debate comes from people who have undergone certain brain injuries or neurosurgical procedures. Sometimes these individuals show an impairment of one kind of memory without a corresponding loss of function in the other (R. C. Atkinson & Shiffrin, 1968; Baddeley et al., 2009; Linden, 2007). Some individuals can recall events experienced before a brain trauma but are unable to retain new experiences. Such a disability may suggest either (a) a problem with working memory while long-term memory remains intact (a dual-store explanation) or (b) a problem in general storage processes (a single-store explanation). Other individuals with brain injuries can recall new experiences long enough to talk briefly about them but can't remember them a few minutes later or at any point thereafter. These might be cases in which (a) working memory is functioning but new information seemingly can't be transferred into long-term memory (a dual-store explanation) or (b) general retrieval processes have been impaired (a single-store explanation). To some degree, working memory processes and long-term memory processes do seem to depend on different parts of the brain (Nee, Berman, Moore, & Jonides, 2008; Zola-Morgan & Squire, 1990). And different areas of the brain are active when people are trying to recall items from the beginning versus end of a serial list (Talmi, Grady, Goshen-Gottstein, & Moscovitch, 2005). Even so, as noted in Chapter 2, most learning and thinking tasks—even very simple ones—tend to involve many parts of the brain. Certainly different parts of the brain specialize in different functions, but a human learner is apt to rely on many parts regardless of the specific learning or memory task at hand (Logie, 2011; Nee et al., 2008). ## Is Conscious Thought Necessary for Long-Term Memory Storage? In the dual-store model, information must go through working memory before it can be stored in long-term memory. Working memory is, by definition, an active, conscious mechanism. It would seem, then, that a learner would have to be actively involved in storing virtually anything in long-term memory. This isn't always the case, however. Some kinds of information—once they've captured a person's attention in some minimal way—seem to be automatically stored in long-term memory even if not specifically selected for further processing (Bocanegra & Hommel, 2014; Frensch & Rünger, 2003; Greenwald & Banaji, 2017; Hintzman, 2011). For example, consider the following question: Which word occurs more frequently in the English language—bacon or pastrami? You probably had no trouble answering correctly that bacon is the more frequently occurring word. Hasher and Zacks (1984) found that people could easily answer such questions about the frequency of events even though they'd never been concerned about counting them. Similarly, people could answer questions about where various events occurred without having intentionally processed this information. Such automatic storage of frequency information and locations begins quite early in life and may help to establish a knowledge base on which future learning can build (Bocanegra & Hommel, 2014; Siegler & Alibali, 2005). Much of this seemingly nonconsciously processed information becomes implicit (rather than explicit) knowledge. Quite possibly, the brain learns—and stores information in long-term memory—in at least two distinctly different ways. One is a very conscious way in which working memory plays an active role. Another is a more basic, “thoughtless” way that involves formation of simple stimulus-stimulus and stimulus-response associations, as well as the accumulation of general “statistical” information about how often various objects and events occur in the environment (Bachevalier, Malkova, & Beauregard, 1996; Frensch & Rünger, 2003; Mudrik, Lamy, & Deouell, 2010; D. J. Siegel, 2012). Complicating the picture even further is the possibility that thinking itself may sometimes occur outside the confines of working memory. Sometimes people can more effectively address a complex problem—one involving much more information than working memory's limited capacity can handle—when they don't actively think about the problem for a period of time (Dijksterhuis & Strick, 2016; Hassin, 2013; Strick, Dijksterhuis, & van Baaren, 2010). Even when not consciously mulling over a complex problem, people might be slowly analyzing the problem, determining which aspects of the problem are more and less important to take into account, imprecisely estimating particular quantities, and integrating problem-relevant information into an overall summary. The result is that a complex problem is sometimes better solved when it remains outside of working memory's limited-capacity limelight for a while. The products of such nonconscious thinking are often implicit and hard to put a finger on. For instance, people might describe them as “intuition” or a “gut feeling” they can't easily explain (Bargh & Morsella, 2008; Dijksterhuis & Nordgren, 2006, p. 105). ## Alternative Views of Human Memory In attempts to address weaknesses of the dual-store model, some theorists have offered alternative models. Here we'll look at a levels-of-processing model that gained temporary prominence in the 1970s and at an activation model that is still quite popular today. Both of these theories emphasize cognitive processes involved in human memory more than the possible structures of which memory might be comprised. ### Levels of Processing The levels-of-processing model of human memory (Cermak & Craik, 1979; Craik & Lockhart, 1972) was the first major theoretical alternative to the dual-store model. According to this view, incoming information is processed by a central processor at any one of a number of different levels of complexity. This central processor has a limited capacity and can process only a small amount of information at any one time. However, according to this model, the more deeply an item is processed (the more it is analyzed, the more it is related to existing knowledge), the better it will be remembered.