DE PE SITE-UL APA.

ORG

Memory illusions wobble, but they don't fall down

False recollections are not easily dismantled, research shows. By SIRI CARPENTER Monitor Staff October 2000, Vol 31, No. 9 Print version: page 38 In the aging mind, memory falters not just in terms of what people forget, but also in terms of what they remember, according to recent research. And even among the younger set, studies indicate that false memories--recollections of stimuli that were never presented--can be quite robust, said Washington University psychologist Henry (Roddy) Roediger, PhD, speaking at APA's 2000 Annual Convention in Washington, D.C., Aug. 48. In 1995, Roediger and colleague Kathleen McDermott, PhD, described research on what they termed an associative memory illusion. In Roediger and McDermott's paradigm, participants study lists of 15 associated words. For example, one such list includes the words bed, rest, awake, tired, dream, wake and snooze, among others--all words that are associated with the word sleep, which the researchers call a "critical item." The critical item is never presented in the list. Next, participants are asked to recall the words that were on the list, without guessing. In their first published account of such an experiment, Roediger and McDermott reported that participants are about as likely to mistakenly remember the word sleep as they were to correctly remember words that were presented in the middle of the study list. That finding has now been replicated numerous times. At the APA convention, Roediger described efforts to reveal why such a false memory effect arises and the circumstances necessary to strengthen or weaken it. "We believe that the concept of 'sleep' is aroused through activation spreading through a semantic network while people listen to the list words," Roediger said. "During the test, subjects must monitor their memories to distinguish events that actually occurred from those that were only activated internally. While we're perceiving things from the outside world, we automatically make inferences that form a coherent knowledge structure." Roediger's group has learned that false memories are least likely to occur when study words are only weakly associated with the critical item, when critical items are distinctive and when

participants have greater opportunity to encode detailed information about the items on the list. Likewise, allowing participants to study word lists multiple times and warning them about the false memory effect in advance each boost accuracy. "The bottom line is that we can diminish the effect, but we've had a terrible time making it go away," Roediger said. False recall is further heightened, the researchers have found, when critical items rhyme with or look similar to list items. For example, the word sleep is more likely to be falsely recalled when a study list includes not only semantically related words such as bed and rest, but also rhyming words such as weep and keep. "The more sources of activation you can get to converge, the greater is false recall," Roediger explained. Recently, Roediger and a team of colleagues led by psychologist David Balota, PhD, have examined how false memory is associated with normal aging and Alzheimer's disease. They have found that older participants and Alzheimer's patients remember fewer list words than do younger participants and wrongly recall critical items more often. "Aging and Alzheimer's disease are a kind of double-edged sword," said Roediger. "You are less likely to remember things that really did happen to you but you're more likely to remember things that never happened to you." Roediger pointed out that these recent results agree with the observations of another Missourian 100 years ago, who was "an acute student of the human condition." Late in life, Mark Twain remarked that, "When I was younger, I could remember anything, whether it had happened or not; but my faculties are decaying now, and soon it shall be so I cannot remember any but the things that never happened. It is sad to go to pieces like this, but we all have to do it."

Imagine that you are studying neuroanatomy one afternoon in the library. You are trying to learn the specific shape and location of the hippocampus. As you stare intently at an image of the brain, you suddenly hear a loud gunshot. You immediately stand and see several confused, shocked people, one of whom catches your eye because she is wearing a vibrant orange dress. The gun is nowhere in sight. After a few moments of commotion you learn that someone just committed suicide in the stacks. Later, as you are packing up your books to leave, you see your notes on emotional memory from a recent psychology class. Hmm, you think, Based on previous research, how should my memory for this afternoon differ from my typical study sessions at the library? Well, past research shows that you will be more likely to remember this afternoon than most other

afternoons in the library, and that you will have a vivid memory for the gunshot itself. But what would you predict about your memory for the information you were studying just before the gunshot, or what you saw immediately after? Will you remember the image of the hippocampus in your textbook any more or less clearly next week than you would have otherwise? Would your memory for people seated around you be better or worse than if you had not heard the gunshot?

The challenge of understanding arousals selective effects on memory

Researchers have investigated how experiencing an emotionally arousing event such as watching a comedy skit, a video of oral surgery, or a stress induction like immersing ones hand in ice water might affect memory for what is seen just beforehand. These studies address the question of whether hearing the gunshot will affect your memory for the picture of the brain that you were carefully studying. But the challenge is that findings across studies differ and in some instances appear to contradict one another. In some studies an emotionally arousing event enhances later memory for neutral information presented beforehand (Nielson & Powless, 2007; Nielson & Bryant, 2005). In other studies manipulations of stress or emotional arousal either disrupt or have no effect on memory for neutral stimuli presented beforehand (Buchanan & Lovallo, 2001; Cahill, Gorski, & Le, 2003; Smeets, Otgaar, Candel, & Wolf, 2008). Other experiments find that even just the presentation of an emotional picture or word can affect memory for neutral items shown just beforehand but again, whether the emotional stimuli enhance or impair memory for information around them depends on the study, with many studies showing impairment (e.g., Detterman & Ellis, 1972; Knight & Mather, 2009; Strange, Hurlemann, & Dolan, 2003) but a couple demonstrating enhancement (Anderson, Wais, & Gabrieli, 2006; Knight & Mather, 2009). Other findings related to the way that emotional arousal leads to selective memory also present some apparently conflicting findings. For instance, there is some evidence that emotional arousal enhances memory for generalized forms of information that lack detail, a type of memory known as gist memory (Adolphs, Tranel, & Buchanan, 2005), whereas other studies show that emotional arousal enhances memory for the specific perceptual details of an item (Kensinger, Garoff-Eaton, & Schacter, 2006, 2007a, 2007b; Mather & Nesmith, 2008). How can we account for these seemingly contradictory effects of arousal on memory? Across studies, one theme that emerges is that arousal has selective effects on memory. Whether arousal enhances or impairs memory depends on the experiment, and the type of information that the participant is tested on. But what determines what arousal will enhance in memory and what it will impair? One previous hypothesis is that emotional arousal leads to memory narrowing, in which arousal enhances memory for central details at the cost of peripheral details (Burke, Heuer, & Reisberg, 1992; Christianson, Loftus, Hoffman, & Loftus, 1991). However, this account cannot explain how arousal sometimes enhances and sometimes impairs memory for information that precedes or follows an arousing stimulus, as reviewed above.

Arousal-biased competition
To address the puzzling discrepancies that have been observed across studies, we outlined a new theory of arousal and memory (Mather & Sutherland, 2011). This arousal-biased competition (ABC) theory builds on the notion that during perception and thought, our active mental representations compete with each other (Beck & Kastner, 2009; Duncan, 2006). Whichever representation becomes dominant tends to suppress the less prominent representations. ABC theory proposes that arousal leads to both winner-take-more and loser-take-less effects, resulting in stronger biased competition in the brain. Thus, arousal modulates ongoing competitive processes of mental representation. However, the key to understanding arousalbiased competition lies in the concept of priority, that determines which mental representations will be enhanced versus suppressed by the process of selective attention. Two primary factors determine prioritytop-down goals and bottom-up perceptual salience (Fecteau & Munoz, 2006). Top-down goals refer to the subjective importance assigned to, and the expectations one has for what is perceived. Bottom-up perceptual salience refers to the degree to which a stimulus attracts attention as a result of its perceptual properties, such as when a brightly lit object attracts more attention than a dimly lit one. Thus if a stimulus in an experiment is relevant to the participants task (top-down goal), or it is more perceptually salient (bottom-up perceptual salience) than other stimuli in view, it will have priority. According to ABC theory, perception and memory for high priority stimuli are enhanced by arousal, while low priority stimuli are suppressed.

Arousal-biased competition in perception

ABC theory makes specific predictions about what arousal will do to mental representations once the arousal is experienced, and the influence it will have on these representations in memory. When a person becomes emotionally aroused, any high priority information that is perceived will be represented in attention and working memory more strongly. Often the high priority information is also the stimulus that induced the arousal, and a wide range of experiments have shown that arousing stimuli are enhanced in attention and memory (Mather, 2007). However other experiments have revealed that being exposed to an emotionally arousing stimulus can enhance visual processing of neutral stimuli that are presented next. For example, briefly viewing an image of a face with a fearful expression, rather than a neutral expression, can increase a persons ability to identify features of line gratings shown next (Phelps, Ling, & Carrasco, 2006). Similarly, hearing or seeing taboo words that evoke strong emotional responses can enhance ones ability to identify briefly displayed non-emotional words that immediately follow (Bocanegra & Zeelenberg, 2009; Zeelenberg & Bocanegra, 2010). It is important to note that in each of these studies the non-emotional stimulus that followed the arousing stimulus had priority because it was goal-relevant. Thus once the emotional stimulus was removed and no longer competing for attention, the lingering arousal enhanced visual processing of the neutral item because that item had priority and was the focus of attention.

To investigate these effects more closely we created a working memory task that allowed us to compare attention to high and low priority stimuli (Sutherland & Mather, under review). We wanted to see how emotional arousal would influence attention to more than one neutral object competing for attention. So we had people listen to short sound clips that varied in emotional intensity. After the sound played eight letters were briefly flashed onto a computer screen (Fig. 1A). Some of the letters were printed in a dark font, giving the letters a high contrast with the white background. Other letters were printed in a lighter font, giving them a relatively low contrast with the white background. The first finding we observed was that regardless of what type of sound was played, high contrast letters were more often correctly recalled compared to low contrast letters, which is evidence that the high contrast letters had priority. However, the key finding was that when participants heard an emotionally arousing sound, they became even more likely to report the high contrast letters and even less likely to report the low contrast letters, compared to when they heard neutral sounds (Fig. 1B). This increase in the influence of salience under arousal was small, but statistically significant. What this shows is that emotional arousal modulates the effects of priority by enhancing representations of high priority letters and weakening representations of low priority letters. Thus, to return to the opening scenario, a woman wearing a bright orange dress in the library might attract more attention than usual, and

might be more memorable in the short term if she is seen immediately after a frightening gunshot.

Arousal-biased competition in memory

In addition to accounting for arousal effects on visual processing, the ABC perspective also explains many of the discrepancies across studies looking at how arousal affects memory. For instance, ABC theory predicts that if you experience a sudden brief increase in arousal (such as when hearing a gunshot), this should increase the degree to which you remember high priority neutral information presented shortly before or after the arousing event. Thus, studies that asked participants to try to learn a list of neutral words for a later test, but then had them watch arousing or nonarousing videos afterwards found better memory for words seen before the arousing video, compared to words seen before the neutral video (Nielson & Powless, 2007; Nielson & Bryant, 2005). Because participants were purposefully trying to learn the words, the words were given priority, meaning their representations were strong, and the arousal experienced afterwards enhanced memory for these words (compared with memory for words learned before watching the neutral video). But the effects of a post arousal or post stress induction change when emotional items are included in the material that is learned beforehand. In studies that had participants view a mixed list of emotional and neutral stimuli, followed by a stress manipulation (Buchanan & Lovallo, 2001; Cahill et al., 2003; Smeets et al., 2008), only memory for the emotional items was enhanced, as they received priority due to their arousing nature. In this case neutral items lose the competition for mental representation when pitted against emotional items, and thus receive no benefit from arousal or stress experienced shortly after the items are learned. In studies that showed that an emotionally arousing picture or word impairs memory for neutral items seen just before and after it (e.g., Detterman & Ellis, 1972; Strange et al., 2003), participants were shown lists of neutral items with one emotional or non-emotional oddball included in the list. Whereas a study showing the opposite effect of this, namely better memory a week later for neutral images shown before emotional pictures (Anderson et al., 2006) had participants learn only one neutral image before the arousal-inducing picture was shown. The key difference between these studies is that the one in which arousal enhanced memory presented only a single neutral stimulus prior to the arousing stimulus on each trial (with a time gap in between each trial), meaning the neutral stimulus did not have to compete for attention with other preceding stimuli. Thus it received priority at the time it was processed and was enhanced in memory if arousal was experienced shortly after. Studies showing the opposite effect (retrograde impairment) had participants learn multiple neutral items that directly competed for mental representation with other neutral items before the emotional oddball items appeared (e.g., Detterman & Ellis, 1972; Strange et al., 2003). None of the neutral items in the oddball lists were strong winners in the competition for representation and thus they were inhibited (loser-take-less) as the emotional oddball stimuli were enhanced (winner-take-more). ABC theory accounts for these conflicting findings, as it predicts that emotionally arousing items will lead to retrograde amnesia when the preceding information was not something already

dominating the competition for mental resources, but retrograde enhancement when it was the highest priority mental representation at that moment. The last set of conflicting findings that ABC theory can explain has to do with arousals influence on memory for gist and memory for detail. When viewing individual pictures one by one accompanied by a narrative that makes a coherent story or theme out of the pictures, if that theme is emotionally arousing, people have better gist memory for what went on in the story (Adolphs et al., 2005). However, other studies report that when pictures are shown individually people have better memory for the details of the picture if it is emotionally arousing (Kensinger et al., 2006; Mather & Nesmith, 2008). While these findings appear to conflict, it is important to note that when narratives are combined with visual stimuli to simulate a thematic story or event, the theme becomes the most salient, and thus has priority. Therefore, when the theme of the story turns emotional, it enhances gist memory for the thematic information of the story because this information has priority. On the other hand, when an individual emotionally arousing picture is shown, it captures attention, leading any benefits of arousal to be concentrated on the details of that picture. Thus the concept of priority allows ABC theory to account for both of these seemingly contradictory effects. Finally, some additional evidence for ABC theory comes from Knight and Mather (2009), whom used an oddball-list paradigm to examine a number of factors to see which ones determine whether emotionally arousing stimuli create retrograde amnesia or enhancement for the neutral non-oddball items in the list. They found that reducing the number of neutral items on the list reduced the retrograde impairment effect, consistent with the idea that arousal is most likely to impair memory for information that was already somewhat lost in the crowd. Also, they found that when participants were trying to learn the neutral items on the list for an upcoming memory test (recall condition), the emotional oddballs led to retrograde enhancement more than when they were just rating whether the neutral items were natural or artificial (no-recall condition; Fig. 2).

In other words, arousal enhanced memory for what happened just beforehand if (and only if) that preceding information was the dominant focus of attention. Thus returning to the opening scenario in the library, if the brain image you were studying before the gunshot was your

dominant mental representation at that moment, then your memory for the picture of the hippocampus would be stronger than it would be had nobody fired a gun in the stacks. In contrast, your memory for things that happened before the gunshot that you were not paying attention to, such as someone nearby coughing, should be even worse than it would be if there had been no gunshot. In summary, the findings we reviewed here indicate that emotional arousal makes things that are perceptually salient stand out even more and makes any high priority information even more memorable. At the same time, arousal reduces processing of low priority information. This increase in selectivity under arousal is likely to be adaptive in many situations, and can explain why sometimes arousing stimuli impair memory for nearby stimuli and sometimes they enhance it.

Acknowledgements
The work described here was supported by grants RO1AG025340, K02 AG032309 and T32AG000037 from the National Institute on Aging.

Cover Story

Feelings' sway over memory

New research suggests that emotions can strengthen and shape memory. By SADIE F. DINGFELDER Monitor Staff September 2005, Vol 36, No. 8 Print version: page 54 Where were you during the terrorist attacks of 9/11? For most, recalling this information is easier than remembering, for example, the details of Wednesday morning last week. This phenomenon-along with more than two decades of experimental evidence--has led many psychologists to posit that emotions can enhance memory and recall. (See May Monitor , page 10.) However, new research suggests a more complicated picture, says Daniel Reisberg, PhD, a psychology professor at Reed College in Oregon. Some psychologists are finding that the stress associated with an emotion may affect how deeply an event is etched into memory, he notes. Additionally, the type of emotion felt may determine what details people recall from an event, says University of California, Irvine, psychology professor Linda Levine, PhD.

"People don't just feel 'emotional.' They feel sad, scared, angry, happy," says Levine. "Those emotions have different functions, and they influence information processing and memory in different ways." For example, people generally feel anger when something is keeping them from reaching their goals, she notes. As a result, angry people tend to focus on what they perceive to be the obstacle and may retain obstacle-related information particularly well, Levine says. In contrast, happiness signals that all is well, and happy people will perceive--and recall--a scene broadly without focusing in on particular details, found Levine and Susan Bluck, PhD, a University of Florida psychology professor, in a recent study in Cognition and Emotion (Vol. 18, No. 4, pages 559574.) In addition to adding to psychologists' understanding of the inner workings of memory, such research may also help judges and juries evaluate the testimony of eyewitnesses--who typically experience high levels of emotion when watching a crime, says Reisberg. Tunnel memory People who witness an armed robbery often demonstrate how negative emotion can narrow attention and memory, says Reisberg. Such witnesses tend to recall the gun in great detail, but not the particulars of the perpetrator's appearance. "You focus your attention on the weapon because, quite obviously, whether or not it is pointing at you is very important," says Reisberg. Even when not in immediate peril, people experiencing negative emotions tend to focus in on specific details, while happy people take in a situation more broadly, found Levine and Bluck in their 2004 study. To test their hypothesis, the researchers took advantage of an unusual situation: the televised announcement of the 1995 O.J. Simpson murder trial verdict. The event offered a unique opportunity to study the effect of different emotions on memory because a large number of people witnessed the same footage, says Levine. Moreover, many people experienced strong positive or negative emotions, depending on whether they deemed the defendant guilty or innocent, she says. Seven days after the verdict, the researchers asked 156 undergraduate students how they felt about the trial's result. About half of the students were angry or sad about the verdict, a quarter were happy and a quarter did not care. Fourteen months after the verdict, the researchers tested the participants' memory of the announcement by asking them which items on a list of events occurred during the announcement. Half of the events listed actually happened--such as Simpson mouthing the words "thank you" to the jury. The other half were made up by the researchers-such as the defendant giving a "thumbs up" to his lawyer. As the researchers expected, the students who felt happy about the verdict tended to recall the entire scene better than the sad, angry or neutral students. However, the happy students also

tended to make more errors of commission--saying that events happened that did not. Students flooded with negative emotions tended to recall less about the verdict announcement overall, but they also made fewer errors in which they recalled details that did not happen. "The happy people and the people who felt negative made opposite types of errors, so there was no overall difference in accuracy," says Levine. However, the happier or angrier the person felt about the event, the more vivid their memory of it, she says. The results suggest that happiness works like a broad-tipped highlighter, illuminating an event in memory and capturing many details, says Levine. However, unlike a highlighter, happy memories also can include events that did not occur but seem plausible, Levine says. Negative emotions tended to act like a narrow highlighter, accentuating particular details at the expense of others, she notes. Memory illusions Other researchers, such as David Rubin, PhD, have found that intensity of emotion matters more than an emotion's kind. In one study, published in Memory & Cognition (Vol. 32, No. 1, pages 1,118-1,132), Rubin--a psychology professor at Duke University in North Carolinia--and his collaborators found that when recalling episodes in their own lives, people tended to recall emotional memories equally vividly regardless of whether they were happy, sad, angry or fearful at the time. However, he notes, the detailed nature of such memories could be illusory. "After an important event, you tell a story about it, and you eventually come to believe your own story," he says. For example, many people talked for days or months after 9/11 about where they were and how they felt at the time of the attacks. As people fill in missing details, it can lead to a false sense of accuracy about a memory, notes Rubin. However, if future studies support Levine's theory that strong emotions--happiness in particular-can lead to broadly remembered events, people may be able to harness their emotions to aid memory, says Bluck. However, such practices could go against memory's primary function, she says. "People in the classroom and the courtroom are of course concerned with complete accuracy, but it is not clear that the memory system has accuracy as its primary goal in everyday life," she notes. Specifically, memory helps people use their experiences to inform their future actions, says Bluck. By highlighting important information or even including things that did not happen, emotion-bound memory may allow us to make better decisions than a picture-accurate memory would, she notes.