PSYC20007 False Memory & Memory Errors Lecture Notes PDF

PSYC20007 False Memory and Memory Errors Adam Osth  So we know that memory is lost over time But are memories merely lost, or do they change with time? What do people think? Simons and Chabris (2011): surveyed the general public about how they think memory works 2/3 of people...

PSYC20007 False Memory and Memory Errors Adam Osth  So we know that memory is lost over time But are memories merely lost, or do they change with time? What do people think? Simons and Chabris (2011): surveyed the general public about how they think memory works 2/3 of people think that memory works like a “video camera” Probably the single greatest consensus among memory researchers… This is simply not true Memory errors can accumulate over time Memory Errors Recall: AB-AC lists of paired-associates More errors in AB-AC lists than AB-CD lists In AB-AC lists, there are more intrusion errors AB-AC: when given the A cue in list 2, participants recall either B or C AB-CD: when given the C cue in list 2, participants mostly recall D What does this mean? We use cues to search our memories Memory errors can be caused by common cues Memory Errors Sometimes intrusions don’t even require a shared cue! In recall tasks, participants often recall items from prior lists Drewnowski and Murdock (1980): most intrusion errors come from the immediately preceding list Intrusions from more distant lists (e. g., ones that were studied a long time ago) are much less likely What does this mean? We are most likely to confuse things that were studied nearby in time Similar results from Ecker et al. from last lecture Greater memory errors when lists were studied closer together in time Memory Errors What about similarity of the learned content? Underwood (1965): similarity between words in a recognition memory task Subjects presented with a list of items During the test phase, presented with studied items and novel items Subjects have to say “YES” to the studied items and “NO” to the novel items Underwood found higher levels of false alarms (erroneously saying “YES” to novel items) if the novel items were associates of studied items E.g., if ”dog” was a studied word, there are higher false alarms to words such as “canine” or “cat” Lower levels of false alarms to unrelated words, such as “broom” or “couch” Memory Errors What does this mean? Underwood argued that his results were due to implicit associative responses When you learn a word, you might think of other related words E.g., when studying “BOY”, you might think of the word “GIRL” ”GIRL” gets associated to the current situation, which makes you more likely to false alarm to it when presented with it later In other words… Even events we imagine or are reminded of can be a source of false memories! The Deese-Roediger-McDermott (DRM) paradigm (Roediger & McDermott, 1995) Most popular lab-based method of eliciting false memories – originally used by Deese (1959) Participants study a list of words that are all associates of a given word “robber”, “crook”, “burglar” are all associates of the non-presented word thief “door”, “glass”, and “pane” are all associates of the word non- presented word window At test, participants had high rates of falsely remembering the non-presented words This applied to both recall and recognition Recognition of the critical lure was often with very high confidence  Very high levels of false recall! Critical lures recalled almost as often as actual words! Data from Roediger and McDermott (1995)  High levels of false recognition (falsely recognizing the critical lure) despite the fact that participants are very good at rejecting Data from Roediger and McDermott unrelated (1995) words! The Deese-Roediger-McDermott (DRM) paradigm (Roediger & McDermott, 1995) DRM errors in the original paradigm were semantic… can perceptual errors be observed as well? Yes! Sommers and Lewis (1999): DRM lists where participants study phonologically similar words Studied “fat”, “that”, ”cab” – all similar to the non-presented word cat (critical lure) Studied “buff”, “put”, “bet” - all similar to the non-presented word but (critical lure) Findings resemble those from semantic false memory Very high levels of false recall and recognition of the criticial non-presented words! What happens if you warn participants about false memories? They still occur! Gallo, Roberts, and Seamon (1997): inclusion of a forewarning condition Participants were told that the lists were specifically designed to induce false memories and given examples of the effect False memories were reduced but not eliminated Data from Gallo, Roberts, and Seamon (1997)  Reduced levels of false recognition of the critical lure in the forewarning condition… But still substantial levels – and much higher than false recognition of the unrelated lures! Data from Gallo, Roberts, and Seamon (1997) What moderates the effects of false memory? Note: DRM lists often contain a large number of associates of the critical word What happens if you study less of them? Robinson and Roediger (1997): manipulated the number of associates on study lists False recognition increases rapidly as the number of studied associates is increased In other words: the more similar content we learn, the easier it is to falsely remember something that is similar to what we experienced  What moderates the effects of false memory? Data from Robinson and Roediger (1997) Why do false memories in the DRM paradigm occur? Several possibilities! One possibility: Generation of the critical lure during learning Participants may think of the critical lures while studying the list E.g., when you see words like “sheets”, “dream”, ”slumber”, etc., you might be reminded of the word “sleep” This makes it more likely that you will falsely recall or recognize the word later This is referred to as a source monitoring error False memory occurs because we were unable to distinguish between a real event (a presented word on a list) and an imagined one (generating the word ourselves in our own minds) This is similar to the Underwood (1965) idea discussed earlier Generation of the critical lure during study How can this idea be tested? Overt rehearsal procedure: have participants rehearse the words out loud as they’re studying the words Subjects often rehearse words to preserve their memory for the material If the critical lure is generated during study, it should appear in their rehearsals As it turns out, participants do rehearse the critical lure during study! …but false memory is still present even when they don’t rehearse the critical words (Marsh & Bower, 2004; Seamon et al., 2002)  Generation of the critical lure during study Lower levels of False recognition false recall when still occurs at the critical lure substantial rates was not regardless of rehearsed during whether the the study phase, critical lure was but still rehearsed at substantial levels study of false recall! Data from Seamon et al. (2002) Why do false memories in the DRM paradigm occur? Generation of the critical lure during learning might contribute to false memory, but cannot be the only factor responsible What remains? One possibility: false memory occurs because we use the “gist” of a learning episode in conjunction with real memories to reconstruct the event Fuzzy trace theory (Brainerd & Reyna, 2002) “Verbatim” traces – real memories that are incomplete or have encoding errors “Gist” traces – an overall sense of what was learned in an event or study list How does fuzzy trace theory explain false memories? Verbatim Gist memory Study Memory List Traces Web Web Insect Insect Something Bug Bug about Fright Fly Fly Arachnid spiders Arachnid Crawl Crawl and Tarantula Tarantula insects! Poison Bite Bite Creepy How does fuzzy trace theory explain false memories? At retrieval, both verbatim and Verbatim Gist memory gist memory traces are used Memory to reconstruct the list Traces memories Web Insect Something Recalls Bug about Insect Fly spiders Bug Arachnid Web Crawl and Fly Tarantula Spider insects! Arachnid Bite More on fuzzy trace theory Why doesn’t forewarning reduce false memory? Gist is still a useful or necessary cue to reconstruct an event Why does increasing the number of associates increase false memory? Stronger extraction of gist traces Why do false memories still occur even if they are not generated at study? Gist is used at retrieval! That’s not the only possibility! Another candidate theory: the critical lure gets activated via a process of spreading activation This is the basis of activation monitoring theory Words we learn become activated – this makes it easier to retrieve them later Words associated with what we learn are also activated This activation of associated memories leads to false memory at retrieval Spreading activation models Even less activation from ROSES to FIRE Less activation from due to being ROSES to RED due to separated by a node their larger distance (RED) Very strong activation from ROSES to FLOWERS due to close Graph from Collins & proximity in the Loftus (1975) network More on activation monitoring theory Why doesn’t forewarning reduce false memory? Activation spreads through the network involuntarily Why does increasing the number of associates increase false memory? Each associate activates the critical lure, so studying more associates increases the activation of the critical lure even more Why do false memories still occur even if they are not generated at study? Activation may not be high enough for the words to be rehearsed at study, but high enough at retrieval to generate false memory Another explanation of false memory Global similarity accounts (Arndt & Hirshman, 1998) A number of computational models have been developed where retrieval is determined by the similarity to all memories Recognition memory: Similarity is calculated to each item in memory These similarity values are summed together to produce an index of global similarity – how similar a probe item is to the entire contents of memory If this is value is sufficiently high, the probe word is endorsed Global Similarity ”Web” is likely to be endorsed Memorie because it matches its own s memory and other items Web Bug Test item Fly Web Tarantula Poison Similarity values are summed Creepy together – if they are high, the word is endorsed Global Similarity “carpet” is dissimilar to all of the Memorie memories and likely to be s rejected Web Bug Test item Fly Carpet Tarantula Poison Similarity values are summed Creepy together – if they are high, the word is endorsed Global Similarity “Spider” has very high similarity Memorie to many items and is very likely s to be accepted as studied even Web though it was not on the list! Bug Test item Fly Spider Tarantula Poison Similarity values are summed Creepy together – if they are high, the word is endorsed More on global similarity accounts Why doesn’t forewarning reduce false memory? False memory is an involuntary consequence of the similarity between the memories and the cues – forewarning won’t change this! Why does increasing the number of associates increase false memory? More similar items in memory = greater global similarity, and greater levels of false memory! Why do false memories still occur even if they are not generated at study? Rehearsal is completely unnecessary for false memories – it falls directly from the similarity between the items So which account is right? Good question! There have yet to be strong tests of each of the theories Each of them make very similar predictions There is very little computational modeling in false memory research! Computational modeling is a useful tool when theories make similar predictions Each theory can be applied directly to data and we can evaluate how well they succeed at accounting for all of the data Real world false memories and their implications Real world false memories Previous findings had to do with laboratory tasks Are these for real, or just examples of how people approach laboratory tasks in strange ways? In the real world, people don’t often use rehearsal or other strategies when we experience events in our lives Reconstructive Memory Bartlett (1932): the “War of the Ghosts” story Subjects read a story and had to repeatedly recall the story across longer intervals Reconstructive Memory Bartlett observed that as the time between reading the story and recall of the story increased, errors increased While there were omissions of details, there were also intrusions of other details Participants were more likely to intrude with details from their own cultural life stories “War of the Ghosts” is a Native American folk tale but Bartlett conducted his studies in England Reconstructive Memory Memory isn’t like a recording we can view Instead, memory is actively reconstructed every time we remember Analogy: when we remember, it’s like organizing a play Actors forget their lines Actors play to an audience Actors age! Similar in spirit to fuzzy trace theory Reconstructive Memory Kintsch et al. (1990): different rates of forgetting for content in memory from sentences Memory for underlying meaning unaffected by delay interval (immediate to four days) Memory for surface form and text (e.g., meaning what the words were in the sentences, passive versus active voice) declines rapidly text What does this mean? Even as we forget the specific words or surface phrases, we can use our memory for the meaning to reconstruct a sentence or story Memory is subject to misinformation – the misinformation effect “What you get out of memory depends on the question you ask” What happens if you ask a misleading question? (Loftus & Palmer, 1974) Subjects saw a video of a car stop at a stop sign, then drive ahead and hit a car Two groups: Control group asked: “How fast was the car going when it hit the other car?” Experimental group asked: “How fast was the car going when it smashed the other car?” Experimental group gave higher speed estimates, also were more likely to claim they saw something they didn’t (broken glass)  How does the timing or context of misinformation impact memory? Lindsay (1990): Subjects watched slides of a maintenance person stealing a computer and some money, two days later tested on memory Difficult condition: Same narrator immediately gives misinformation Easy condition: Different narrator gives misinformation two days later Less misinformation in reports in the easy condition Misinformation presented long after the event is easier to disregard, or possibly not part of the memory False childhood memories! Loftus and Pickerell (1995): “Lost in the mall” study Participants were asked about a number of actual childhood memories Told to describe a story in which subjects were “lost in a shopping mall” Many participants described a story in detail even though it may never have occurred False childhood memories! Wade et al. (2002): Hot air balloon study Researchers solicited photos from the participants’ families and stories about what the participants did as children Received confirmation that the participants had never been in a hot air balloon Subjects presented with photos, along with a photoshopped picture of the participant as a child in a hot air balloon For each picture they were told to remember the event, and if they couldn’t remember it, to picture the event Most subjects did not initially remember the event, but after repeated interviews, as many as 50% of participants described the event Implications of the misinformation effect When would we be asked repeated misleading questions in the real world? Police interviews and legal proceedings! Wells and Bradfield (1998): when witnesses pick a suspect from a lineup, post-identification feedback from police can influence their confidence Positive feedback example: “Good – you picked the right person” Witness’s confidence increases significantly relative to no feedback …but surely some memories are immune to errors, right? “Flashbulb” memories (Brown & Kulik, 1977) Possible exception to erroneous normal memories Highly confident and vivid recollections of major life events, such as: Assassination of John F Kennedy The Challenger explosion The September 11 attacks on the World Trace Center People often report that they remember where they were, who they were with, and are highly confident Brown and Kulik argued that these memories are preserved due to the high level of emotional content Are flashbulb memories really more accurate? Critical limitation of the Brown and Kulik work: they could not verify the accuracy of the reports Neisser and Harsch (1973): longitudinal study on flashbulb memories Recruited participants the day after the Challenger explosion and collected reports of what they had done Participants given surveys up to 3 years after the event Huge changes in the details of the reports as time progressed! Examples: Some participants who were at work later claimed they were at home Some participants who claimed they heard it from a friend later claimed they heard it on TV Are flashbulb memories really more accurate? Several flashbulb memory studies have replicated the Neisser and Harsch findings with other major life events Talarico and Rubin (2003): Data from Talarico and Rubin (2003) direct comparison between ordinary memories and flashbulb memories Accuracy is similar for both types of memories! Vividness and confidence is much higher for flashbulb memories Are flashbulb memories really more accurate? Why are flashbulb memories inaccurate? We’re not entirely sure Multiple possibilities: When we retell events, we embellish details to others and these become part of the memory For instance, you might describe a few people at a scene as a “massive crowd” to make a story more interesting Later you might forget that you embellished this – it becomes part of the memory itself We incorporate other people’s retellings/other sources into our own memories Neisser and Harsh observed that later recollections more frequently incorporated hearing about the events on TV than earlier ones Are flashbulb memories really more accurate? What do these results imply about flashbulb memories? They honestly might not be any more accurate than any other kinds of memories! These are subject to reconstruction in the same way as other memories – they are not recordings! Are flashbulb memories really more accurate? How do we incorporate other memories into our own? Recall: interference in AB-AC designs When encountering an AC pair in List 2, there is interference from the two lists When given the A cue, both B and C come to mind! Similar mechanisms can be at work with flashbulb memories Initial Event: Hearing Later Event: Seeing about the 9/11 attacks many reports of the at home from family 9/11 attacks on members television Memory cue: How did you hear about the 9/11 attacks? Stronger associations to the recollections of TV due to the frequent The “memory wars” The ”memory wars” Began with the question of whether memories are repressed Sigmund Freud argued that traumatic memories that are too troubling for the conscious mind are repressed in the mind These memories are unable to be consciously accessed, but still cause anxiety Freud claimed that many people who have psychological difficulties but don’t know the causes have repressed memories AND that clinicians can unlock/uncover these memories Even today, many clinicians and therapists believe that memories can be repressed and that these memories can be recovered by a trained clinician The ”memory wars” Memory repression became a controversial topic because there were a number of abuse cases entering the courtroom where the victim made the allegations on the basis of recovered memories assisted by a therapist Beth Loftus and others were skeptical of these memories for a number of reasons: Traumatic memories tend to be well remembered – even *over— remembered People with PTSD report that memories of trauma are intrusive and distracting The types of suggestion that therapists use has been demonstrated to produce false memories Therapists often continue to probe about memories of abuse even if the patient cannot remember any This resembles the techniques used in the “lost in the mall” study But are there cases of recovered memories of abuse? Yes, actually Schooler (2001) discussed seven cases of individuals who recovered memories of abuse decades later where there was independent evidence that the abuse had occurred However, these were not in therapy – often this was in response to an event or trigger E.g., a film depicting an act of sexual abuse, seeing the abuser in person, discussions about sexual acts Are there cases of recovered memories of abuse? Was this actually memory recovery? In some cases, the victims had actually discussed the events previously – in other words, the event was not totally forgotten It is possible that this was an *emotional* revelation, finally grappling with the emotional significance of what happened to them Memory recovery is nonetheless still possible In some instances, the victims were very young and did not classify the event as sexual abuse Young children do not always have a category or understanding of “sexual abuse” These instances can be painful or uncomfortable, but they do not have a category for the event Reminders of event details or perpetrators may prompt retrieval of the event …but is this recovery due to memory repression? Not necessarily In each case, there was no evidence that the memory was blocked by the unconscious mind to protect the conscious mind Memories were not “unlocked” by a therapist This type of memory recovery can occur even for non- traumatic memories E.g., suddenly remembering events from childhood like visiting a foreign country, meeting a famous person whom you didn’t know the name of at the time, or that you had seen a particular movie So what is happening with recovered memories of trauma? Some recovered memories may be false Otgaar et al. (2017): DRM study with emotional lists, such as associates of “sad” Individuals with depression or PTSD had higher false memory rates for these lists Some recovered memories may be genuine, but were recovered with the correct retrieval cues Roediger and Bergman (1998): we know from lab studies that the correct retrieval cues can recover forgotten memories Tulving and Pearlstone (1966): providing the names of categories of studied lists can dramatically improve recall rates If somebody did not categorize an event as sexual abuse, they may say “no” when asked about whether this happened to them Still no evidence of active memory repression …but isn’t memory repression consistent with retrieval inhibition? Recall: inhibition as measured by retrieval induced forgetting (Anderson, Bjork, & Bjork, 1994) Several problems with this line of thinking: Inhibition theory is still controversial – Raaijmakers and Jakab (2012) and other researchers argue that these results can be explained by interference Inhibition does not require trauma or negative events to occur, which is different from the concept of repression Inhibition also cannot be unlocked or recovered by a clinician Can we trust our memories? Can we trust our memories? Huge legal implications: can we trust eyewitness memory? Many say no – memory is too unreliable! Elizabeth Loftus has often testified in defense trials and used her expertise to undermine the accounts of eyewitnesses Can we trust our memories? Recent evidence, however, has generally found that memories endorsed with high confidence can be highly accurate Mickes et al. (2011): accuracy in lab-based item recognition tasks is virtually perfect for high confidence identifications Data from Mickes et al. Can we trust our memories? But what about eyewitness memory paradigms? Similar findings! High confidence identifications are highly accurate Wixted and Wells (2017) re- analyzed several datasets and found that high confidence identifications are associated with extremely high accuracy Can we trust our memories? Diamond, Armson, and Levine (2020): Long-term recall (2 years later) and recognition of two major events: a guided tour around a university campus, and a training protocol Overall frequency of details recalled declined with time However, the accuracy of the recalled details was extremely high – close to perfect! Diamond et al. (2020) study Diamond et al. (2020) study  Diamond et al. (2020) study Many Close to details perfect omitted accuracy when for the recalling details the events that two years *were* later! recalled! Hold on a minute… If flashbulb memories are so inaccurate, how can memories for a training procedure and a guided tour be so accurate? Hold on a minute… Flashbulb memories are often memories which we retell in groups Very easy for other people’s recollections to intrude into our own Recall: this is like AB-AC learning – we acquire many associations to the same cues The events in the Diamond et al. study are less likely to be events that we share or revisit with time Less contamination from other events There was still considerable forgetting evident, ~20% of details were recalled from the original event This is more like AB-CD learning! The events are less likely to be re-experienced, so they don’t share cues with other events What about eyewitness memory? John Wixted has a similar take on eyewitness memory Highly confident memories can be accurate under a limited set of circumstances When confidence is recorded at the initial identification, not after feedback has been given When the suspects are presented to the witnesses before they have heard descriptions of the event or recollected it to others When the police officers do not provide feedback on their identification To close… Accuracy of memory likely depends on circumstances! False or intruded memories become more likely when there are misleading questions False memories become more frequent when we learn very similar events – this frequency increases with the number of similar events (Robinson & Roediger, 1997) False memories are more likely when there are shared cues between situations (AB-AC learning) Memory can be highly accurate for situations that are relatively distinct in memory Learning Outcomes Understand what types of errors occur on memory tasks Understand the DRM paradigm and how false memory is produced, and what conditions increase or decrease the rate of false memories being observed Understand the different theories of false memory: fuzzy trace theory, activation monitoring theory, and global matching theory Understand the reconstructive nature of memory and how it relates to false memory Understand the misinformation effect Understand flashbulb memories and their degree of accuracy Learning Outcomes Understand the ideas of memory repression and critically evaluate whether recovered memories are due to memory repression Understand the conditions under which memory can be accurate or trusted Next week: Language!