9.7 Retrieving Information from Long-Term Memory

Once we encode an item into long-term memory, we may or may not be able to find it there when we need it. Why are we sometimes unable to remember a name or fact that we “know” that we know? Why do we at other times remember a name or fact that we thought we had long since forgotten? Why, in some cases, do we remember an event differently each time we retell it, or remember an event that didn’t really happen as if it did? These are some of the questions that cognitive psychologists have addressed in research and theories concerning the retrieval of information from long-term memory.

The Internet contains literally billions of pages of textual information, sitting in hundreds of thousands of servers throughout the world. All that information would be useless to you were it not for search engines, such as Google, that can rapidly scan the pages and find what we are looking for. The long-term-memory store of your brain likewise contains vast quantities of information (though not as much as the Internet), which is useful only to the degree that we can call information forth at the moment we need it.

Retrieval of specific items from any information-storage system depends on how the stored information is organized. Books in a library are organized by topic, making it easier for us to find all there is about, say, the Civil War or flower gardening. Words in a dictionary are organized alphabetically, so we can find them on the basis of their spelling. Web pages on the Internet contain links to other pages that deal with related issues, and efficient search engines take advantage of those links to rank pages in the order of their relevance to the search terms that we have typed in. In the human mind, long-term memories are stored not in isolation, but in networks in which each item is linked to many others through connections referred to as associations, somewhat analogous to the links among websites (Griffiths et al., 2007). When any one memory is activated by an appropriate stimulus or thought, other memories associated with it become temporarily activated, or primed, to become more easily retrievable. A stimulus or thought that primes a particular memory is referred to as a retrieval cue for that memory.

The evidence for these ideas about memory organization and retrieval comes not from knowledge of how memories are stored physically in the brain, but from behavioral studies.

Mental Associations as Foundations for Retrieval

Speculation about mental associations goes back at least to the time of Aristotle. Aristotle considered two concepts to be associated if the thought of one tends to evoke (call forth from long-term memory) the thought of the other, and he proposed several principles of association, the most central of which are contiguity and similarity.

According to Aristotle’s principle of association by contiguity, some concepts are associated because they have occurred together (contiguously) in the person’s previous experience. Thus, napkin and plate might be associated in your mind because you have frequently seen napkins and plates together. When you see the face of someone you know, his or her name leaps to your mind because you have often experienced that face and that name together in the past. The contiguity principle also accounts for our ability to bring quickly to mind the various properties of an object when we hear its name. If you hear apple, you can immediately think red, round, sweet, tart, grows on trees, good in pies because you have experienced all those properties of apples contiguously with apples themselves and with the word apple.

According to the principle of association by similarity, items that share one or more properties in common are linked in memory whether or not they were ever experienced together. Your thought apple might evoke the thought rose because both are red, even if you have never seen an apple and a rose together.

In his great introductory psychology textbook, written more than a century ago, William James (1890/1950) pointed out that association by similarity can be understood as a derivative of association by contiguity, the latter being the more primitive and fundamental principle. Contiguity allows us to think of the properties of any given object and then allows us to think of other objects that have those same properties, leading to associations by similarity. Thus, your thought apple leads to your thought red (because of contiguity), and your thought red leads to your thought rose (again because of contiguity), with the result that your thought apple leads to rose (similarity). James suggested that the ability to separate mentally the various properties of objects and events from their concrete referents, and to use those properties to link objects and events that were never experienced contiguously, represents a basic difference between the human mind and that of other animals. “Thoughts [in dogs] will not be found to call up their similars, but only their habitual successors,” he wrote. “Sunsets will not suggest heroes’ deaths, but supper-time. This is why man is the only metaphysical animal” (p. 353).

Why Elaborative Rehearsal Creates Easily Retrievable Memories

Earlier in this chapter we discussed network models of memory organization, describing how concepts such as red, roses, cherries, apples, sunsets, and fire are organized in semantic memory. How you place new information into your network of associations has a big effect on your subsequent ability to retrieve it. This brings us back to the role of elaborative rehearsal in memorization, discussed earlier. The more mental associations you create in learning a new item of information, the more ways will be available for you to retrieve it later.

Figure 9.19: Value of self-generated retrieval cues Subjects generated either one or three singleword properties related to each of 500 nouns. Later they were tested for their ability to recall each noun, using either their own self-generated properties or another subject’s self-generated properties as retrieval cues.

(Based on data from Mäntylä, 1986.)

Suppose you are learning for the first time that the capital of Vermont is Montpelier. You might notice, as you encode this new fact, that the syllable mont appears in both the capital name and the state name, and you might then think about the fact that mont is French for mountain, that Vermont is known as the Green Mountain state, and that many of the early settlers were French. More imaginatively, you might notice that Montpelier sounds like mount peeler and you might think of a strong wind peeling the snow off one of Vermont’s mountains. Through such observations and thoughts you are setting up many possible retrieval cues for remembering, later, that the capital of Vermont is Montpelier. The last syllable of Vermont, or the thought of Vermont’s mountains, or of French settlers, or of wind, or of snow, or of anything that can be peeled, could prime the name Montpelier in your memory store.

In an experiment that demonstrated the value of retrieval cues generated by elaborative encoding, Timo Mäntylä (1986) presented 500 nouns one by one in a single very long session. He did not ask the subjects to memorize the nouns but asked them to write down either one or three words that they regarded as properties of the object named by each noun. For example, for the word barn a subject might write large, wooden, red. He then surprised the subjects with a test of their ability to recall all 500 nouns. As cues, he gave them either their own self-generated properties or those generated by a different subject in the same experiment. Subjects who received three self-generated properties for each word were able to recall correctly more than 90 percent of the 500 nouns. When only one property was available or when the properties had been generated by someone else, recall was much poorer (see Figure 9.19).

Contextual Stimuli as Retrieval Cues

The environmental context that we find ourselves in at any given moment provides retrieval cues that prime our memories for our past experiences in that context. Such priming is generally adaptive because our past experiences in a given context are often relevant to our future experiences in that context. Sitting behind the steering wheel of an automobile primes my memories relevant to driving, the sight and smell of a gymnasium prime my memories of basketball rules and strategies, and standing at a podium primes my memories of lecture techniques that were helpful in the past.

Many experiments have shown that people who are given facts or word lists to memorize in a particular context do better at recalling that information if tested in that same context than if tested in a different context (Smith & Vela, 2001). The context might be a particular room, or it might be a single very noticeable stimulus that seems to be incidental to the learning task. In one series of experiments, for example, people who studied a list of words in a room that smelled of chocolate or cinnamon/apple or mothballs performed better on a recall test if that same smell was present than they did if a different smell, or no smell, was present (Schab, 1990). Other experiments have shown that even instructions to imagine the context that was present during learning can facilitate recall (Smith & Vela, 2001).

Memory Construction as a Source of Distortion

Remembering is not just a process of retrieving traces that were laid down during the original encoding. The brain is not a tape recorder, video camera, or CD burner that records information at high fidelity; and remembering is not a matter of finding the right cassette or disk and playing it back. Instead, remembering is an active, inferential process guided by a person’s general knowledge and intuitions about the world and by cues in the present environment. When you hear a story or experience an event, your mind encodes into long-term memory only some parts of the available information. Later, when you try to recount the story or event, you retrieve the encoded fragments and fill in the gaps through your logic and knowledge, which tell you what must have happened even if you can’t quite remember it. With repeated retelling, it becomes harder to distinguish what was present in the original encoding from what was added later. Thus, memory of the story or experience is not a simple readout of the original information but a construction built and rebuilt from various sources. Our ability to construct the past is adaptive because it allows us to make logical and useful sense of our incompletely encoded experiences. But the process can also lead to distortions.

Effects of Pre-existing Beliefs: Fitting Memories to Schemas and Scripts

One of the first psychologists to call attention to the role of people’s general knowledge or beliefs in their more specific memories was the British psychologist Frederick Bartlett. Bartlett (1932) used the term schema to refer to one’s generalized mental representation, or concept, of any given class of objects, scenes, or events. He used the term especially in relation to concepts that may vary from culture to culture and that involve spatial or temporal relationships among the individual units of the object, scene, or event. For example, in our culture today people might share a relatively common schema of a living room, perhaps including a couch, an easy chair, and a rocking chair, all oriented around a television set, with a coffee table in front of the couch. When we enter a new living room, we recognize it as a living room and assess its unique features by comparing it with our already-existing schema. Schemas that involve the organization of events in time, rather than of objects in space, are commonly called scripts by today’s cognitive psychologists (Schank & Abelson, 1977). The typical children’s birthday party is a good example: The celebration involves the presentation of the cake, the singing of “Happy Birthday,” and the blowing out of the candles; birthday presents may also be opened.

Following the script Years from now, a child who attended this birthday party may construct a memory of the event largely from his or her knowledge of the typical birthday-party script.

Stockbroker/Alamy

According to Bartlett and the results of many studies since his time, schemas do not just help us recognize and label the objects, scenes, and events that we encounter in daily life; they also affect the way we remember them later. We tend to remember any particular living room or birthday party as being more like the standard living room or birthday party than it really was. That is because we fill gaps in our memories for specific scenes and events with information drawn from our more general schemas and scripts.

In a classic demonstration of the effect of general knowledge on memory for the specific, Bartlett (1932) asked British university students to listen to a Native American story entitled “The War of the Ghosts” and later asked them to retell the story from memory. He found that the story often changed in the retelling, and he found certain consistencies in those changes. Details not essential to the plot tended to drop out, and those that were essential were often exaggerated. Also, points in the story that were consistent with Native American beliefs but not with the students’ own beliefs were often changed to be more consistent with the latter. For example, the protagonist’s obligation to certain spirits—a key component of the original story—tended to be transposed into an obligation to his parents. The changes were not deliberate; the students were trying to retell the story accurately, but they inevitably used their own ways of understanding things—their own schemas—to fill the gaps in their memory.

False Eyewitness Memories: Effects of Suggestion

Memory construction is affected not just by pre-existing schemas, but also by events that occur after the event being remembered was encoded. Such constructions assume more than academic importance in courtrooms or psychotherapists’ offices, where they can have serious consequences.

Perhaps the most egregious of false memories are those that send innocent persons to prison or, sometimes, to death. In the 1990s, when DNA testing began to identify dozens of people who had been convicted of serious crimes they did not commit, the U.S. Attorney General’s office ordered a study to discover the causes of those injustices. That study, and similar studies since then, concluded that the great majority of those convictions came about because of highly confident eyewitness identifications (Wells et al., 2002, 2006). In many cases the eyewitness was not confident early in the investigative process, but became so over time. An example is the case of Larry Mayes, who spent 21 years in prison for a rape he did not commit (Loftus, 2004). The victim failed to pick Mayes in two initial police lineups, but then became confident that he was the rapist after police helped her “recover” her memory of him through hypnosis. By the time of trial, the face of Larry Mayes was clearly embedded in her mind, and she was certain he was the rapist.

Hypnosis is a state of high suggestibility; psychologists have shown repeatedly that false memories can be created relatively easily through suggestions or encouragement made in that state (Lynn et al., 1997; Steblay & Bothwell, 1994). People who hold steadfastly to truly bizarre memories—such as having been abducted by aliens from another planet—often first recalled or became confident about those memories when questioned under hypnosis by investigators who believe in such phenomena (Newman & Baumeister, 1996). Many experiments—some involving simulated crimes—have shown that memories can also be altered or created through suggestions and encouragement without hypnosis (Spinney, 2008). In one study, simply saying, “Good job, you are a good witness,” dramatically increased witnesses’ confidence in their memories of who committed the simulated crime, whether or not the memories were true (Wells & Bradfield, 1999). Other studies have shown that leading questions can alter people’s memories not just of who was involved in an incident, but also of what happened.

In a classic experiment, Elizabeth Loftus and J. C. Palmer (1974) had adults view a film depicting a traffic accident. Later, the researchers asked some of the subjects how fast the cars were going when they hit each other, and they asked others how fast the cars were going when they smashed into each other. The question with the word smashed elicited estimates of faster speed than did the question with the word hit. Moreover, when the subjects returned a week later and were asked to remember the film and say whether there was any broken glass in the accident, those who had heard the word smashed were more likely to say they saw broken glass (though actually there was none in the film) than were those who had heard the word hit. In subsequent studies, Loftus and others showed that such memory distortion is especially likely if the misinformation is introduced in just the manner that a clever but biased detective or cross-examiner is likely to introduce it—repeatedly but subtly, in the context of recounting aspects of the event that really did occur (Loftus, 1992; Zaragoza & Mitchell, 1996).

False Memories of Childhood Experiences: Effects of Suggestion and Imagination

If an adult in psychotherapy begins to recall instances of having been abused in childhood, does the memory necessarily reflect the truth, or might it have been constructed from ideas implanted by a well-intentioned but misguided therapist? In the 1980s, a number of high-profile court cases centered on this question. Some people were suing their parents for past, horrible abuses discovered in therapy, and some parents were suing therapists for implanting false memories in their offspring. The cases led to a spate of psychological research showing that techniques of suggestion, encouragement, and imagination indeed can create false memories of childhood experiences.

False memories of childhood abuse can sometimes be developed in therapy by well-intentioned therapists.

©2004 S. Cullari Salvatore S. Cullari, Professor Emeritus of Psychology Lebanon Valley College

In one such study, Elizabeth Loftus and Jacqueline Pickrell (1995) led adults to believe that at age 5 they had been lost in a certain shopping mall and had been helped and comforted there by an elderly woman—an experience that in fact had never happened, according to the subjects’ parents and other close relatives who served as informants. Yet 25 percent of the subjects maintained—both at initial questioning and in follow-up interviews 1 and 2 weeks later—that they could remember the event. Some even elaborated upon it with details beyond those supplied by the researchers.

Figure 9.20: Effect of imagery on development of a false memory Adult subjects were told (falsely) about an Incident that happened In their childhood. They were asked to try to remember It either by thinking about It (control condition) or by vividly imagining what might have happened (imagery condition). This procedure was repeated the next day, and the subjects’ memories were again assessed in a third session, 2 days after the first. The graph shows the percentage of subjects in each condition, at each interview, who claimed to remember the incident.

(Based on data from Hyman & Pentland, 1996.)

Subsequent studies showed that false-memory construction can be abetted by imagination. In one experiment, researchers told each subject (falsely) that according to the subject’s parents, a particular embarrassing incident had occurred in his or her childhood (Hyman & Pentland, 1996). The made-up incident was one in which the subject, at age 5, had been running around at a wedding reception and had knocked over the punch bowl, spilling punch on the bride’s parents. Subjects in the imagination condition were asked, in two successive sessions, to form vivid mental images of this event to help them remember it, and others, in the control condition, were asked just to think about the event as a way of remembering it. The result was that mental imagery sharply increased reported memory (see Figure 9.20.) In an interview conducted after the second imagery session, 38 percent of the subjects in the imagery condition, compared to only 12 percent in the control condition, claimed that they could remember the punch-spilling incident.

Other experiments have shown that imagery alone, even without misleading suggestions from the researcher, can create false memories. In one, adults were simply asked to imagine a certain painful medical procedure—a procedure that in fact is never performed—and then, later, were asked to try to remember whether or not that procedure had ever been done to them in their childhood. The result was that over 20 percent of those in the imagination condition, compared to about 5 percent in the control condition, said that they could remember enduring that procedure (Mazoni & Memon, 2003). Similar studies done with preschool and early schoolage children have shown that they are even more susceptible to the effects of imagining (“Do you remember getting your finger caught in a mousetrap?) (Ceci et al., 1994) and will create memories of unusual and unlikely events (a rabbit getting loose in their classroom) simply by overhearing other children talk about it (Principe et al., 2010). Some children will even create memories about their infancy, based on experiences they had or things they hear years later. Recall my memory as a sickly baby sitting in a crib at my grandparents’ house (p. 347).

Such research should not lead us to conclude that all our childhood memories are false constructions. The research does, however, suggest strongly that childhood memories are even more subject to distortion by suggestion and imagination than are memories acquired later in life.

Source Confusion and Social Pressure as Causes of False-Memory Construction

According to some theorists, a basic cause of most, if not all, cases of false-memory construction is source confusion (Lindsay, 2008). We acquire information from various sources—including firsthand experiences, stories we have heard, and scenes we have imagined—and our minds reorganize the information in ways that may be meaningful but that obscure the ties between each item and its original source. Events that are conceptually linked but came from different sources may become confounded in memory. The memory of the actual traffic accident witnessed becomes confounded with the memory of what the cross-examiner implied about the accident. The memory of an actual event in childhood becomes confounded with a story heard from others or with an imagined scene constructed by free association in a psychotherapy session.

Social pressure also, no doubt, figures into many cases of false memory (Loftus, 1997). In most of the experiments just described, subjects were in one way or another led to believe that they should remember the suggested events. Similar pressure occurs in crime investigations, where the investigator acts as if any normal person should remember the incident’s details. And it sometimes occurs in psychotherapists’ offices, where a misguided therapist conveys the message that certain kinds of incidents, such as childhood abuse, must have happened and that the patient’s task is to remember those incidents (McNally, 2003; Pendergrast, 1995). A person who feels pressured, even gently so, to come up with a memory is more likely than an unpressured person to identify a vague, possible memory as an actual memory; and the more often the memory is repeated, and the more praise the person receives for recalling it, the more confident the person becomes that the memory is true (Brainerd & Reyna, 2005). The opposite is probably true also: a person who feels pressured by family members or an abuser to believe that certain memories are figments of his or her imagination may come to doubt memories that are accurate.

Prospective Memory and Mental Time Travel

To this point we have been talking about remembering things in the past, called retrospective memory. But we also sometimes use memory to remember to do things in the future, termed prospective memory (Einstein & McDaniels, 2005). For example, on my way home from the office this afternoon, I need to remember to stop at the store and pick up some milk, otherwise I will have dry cereal for tomorrow morning’s breakfast. Prospective memory is a type of episodic memory, but what is retrieved is not a past event but a future intent. Researchers have distinguished two types of prospective memory. Event-based prospective memory involves remembering to perform a particular action when cued by a target event (for example, giving a message to a friend when you next see him or her), whereas time-based prospective memory involves remembering to execute an intended action after the passage of a certain amount of time or at a specific moment (for example, answering e-mails in 10 minutes). Unlike retrospective memory, in prospective memory “no one is there to put you in a retrieval mode when the target event occurs … thus, a key question in the prospective memory arena is how, in the absence of a direct request to search memory, the cognitive system supports retrieval of the intended action at the appropriate moment” (Einstein & McDaniel, 2010, p. 1082). Like retrospective memory, prospective memory is relatively poor in children, increases in adolescents and young adulthood, and declines in older adults (Kvavilashvili et al., 2009). In fact, failures of prospective memory are a frequent complaint for older adults. People of all ages complain about failures of prospective memory, and such failures can have serious consequences in many workplaces, including aviation and medicine (Dismukes, 2013).

There are three phases in prospective memory: (a) first, a subject forms an intention (buy milk on the way home); (b) second, the intention must be maintained; and (c) there must be a “switch” from the ongoing task (driving home) to execute the intention (stopping at the store and buying milk) (Ellis, 1996). The executive functioning account of prospective memory proposes that each phase requires using a limited pool of cognitive resources in an effective way (Wong & LeBoe, 2009). In particular, people need to be able plan (phase 1) and regulate their thinking and behavior, particularly during the third, “switch,” phase each of which uses cognitive resources. Support for this position comes from tasks in which subjects who are given a prospective memory task (“Remember to put the sign on the door when we leave”) perform more poorly on ongoing problem-solving tasks than subjects not given a future-oriented task to perform (Smith, 2003). Thus, like the dual tasks discussed in the section on working memory, having to remember to do something in the future reduces performance on a simultaneous task, presumably because the tasks compete for a limited set of resources in the short-term store. Further evidence for the role of executive functions in prospective memory comes from studies in which performance on the prospective memory task declines as the cognitive demands of the ongoing task increase (Marsh & Hicks, 1998).

Some researchers have referred to prospective memory as mental time travel, or the ability to anticipate the future and plan for it (Tulving, 2005; Suddendorf & Corballis, 2007, 2010). Seamlessly transporting oneself (mentally) to not only different places but also different times is a hallmark of human cognition. Similar to Tulving’s conceptualization of episodic memory, this ability requires an autonoetic (self-knowing) component, reaching beyond simple general procedural memory, permitting individuals to represent the self in the future. This has resulted in some people referring to this type of memory as episodic future thought (Quon & Atance, 2010; Szpunar, 2010).

Other species, including chimpanzees and scrub jays, display a form of prospective memory. For example, scrub jays are able to cache and later locate hundreds of hidden food morsels (Clayton & Dickinson, 1999), and chimpanzees will keep a tool for later use for up to an hour (Osvath & Osvath, 2008). However, these abilities seem not to require the autonoetic, or self-knowing, abilities characteristic of human episodic memory (Salwiczek et al., 2010; Suddendorf & Croballis, 2010), suggesting that the ability to anticipate and act on future needs or drive states sets human cognition apart from those of other species.