Psych - W1_lecture20

Lecture 20, the reconstruction of memory. The complimentary effects of encoding and retrieval processes are also illustrated by the relation between episodic memory and semantic memory. Remember from our discussion of the constructivist view of perception, that perception requires the perceiver to draw on his or her world knowledge -- knowledge, beliefs, and expectations about the world -- in order to form a mental representation of a stimulus given the vague, fragmentary, and ambiguous nature that so often is characteristic of stimulus information. We characterize this cognitive contribution to perception in terms of schemata -- organized knowledge structures that the person brings to bear on the problem of perception. Well, it turns out that these same kinds of schemata also play a role in memory, in the encoding and retrieval of memory traces. Schemata provide the cognitive basis for memory as well as the cognitive basis for perception, guiding encoding and retrieval, facilitating processing and filling in information that's missing from the memory trace. So that leads us to the next question in understanding the processes of memory, which is, what's the relationship between our memory for particular events and our knowledge of the world in general? The effects of cognitive schemata on memory are nicely illustrated in a variant on the verbal learning paradigm known as the person-memory experiment, in which subjects study and remember information about specific individuals. And the idea behind the person memory experiment is that we have two different kinds of knowledge about the people we know. We have episodic knowledge about the things they've done and the experiences they've had, and we have semantic knowledge about what they're like in general. And it turns out that our knowledge about what people are like in general, our semantic knowledge about people, affects our memory for what specific experiences they've had, our episodic knowledge of people. In the person-memory experiment subjects study, and are asked to remember, information about specific individuals. We present them first with some generic or semantic information about the individual's general traits and attitudes. And then we present the subject with some episodic information about specific behaviors and experiences that the individual has engaged in, or had. So, for example, in the first phase of the experiment we present the subject with a trait ensemble -- that is, a list of personality traits that ostensibly characterize some target person. So we might describe a person, Judy, as intelligent, as intellectually sophisticated, as artistically sensitive, refined, imaginative, and witty. Now do you have a mental picture what Judy is like in general? Good. That's the point. The purpose of presenting the trait ensemble is to convey a unitary impression of the target person. This unitary impression is going to serve as a schema -- an organized knowledge structure representing our knowledge and beliefs and expectations concerning some aspect of the world. Schemata -- or you can call them schemas if you want -- schemata are generalized and abstract: they don't concern particular behaviors or particular experiences. As was the case for perception, they provide the cognitive basis for encoding and retrieving memories. Cognitive schemata are critical to what F.C. Bartlett, the British psychologist, called the person's effort after meaning as he or she is trying to perceive and remember a world that is vague and fragmentary and ambiguous. In the next phase of the experiment, we present the subject with a list of specific experiences that that target person has had, or specific behaviors that that target person has engaged in. Some of these behaviors are schema-congruent, in that they have a high probability of having occurred, given the impression that we have of the person -- such as Judy won the chess tournament, and Judy attended the symphony concert. These are the kinds of behaviors we'd expect from an intellectually sophisticated, artistically sensitive, refined person. Other behaviors are schema-incongruent, in that they have a low probability of occurring, given the impression we have of the person. For example, Judy made the same mistake three times in a row, or Judy was confused by the day time television show. These are not the kinds of behaviors we'd expect from a person who is intelligent, and intellectually sophisticated, and artistically sensitive, and refined. And still other behaviors are schema-irrelevant, in that they don't bear on the schema at all one way or the other. They're not the kinds of things that would be predicted, or contra-predicted, by what we know about the individual -- like, Judy ordered a sandwich for lunch, which is something anybody can do, and Judy took the elevator to the third floor, which you can do whether you're intellectually sophisticated or not. Expressed in mathematical terms, for schema-congruent behaviors the probability of the behavior given the schema is greater than chance, say greater than .50. Given that a person's intelligent, it's more likely that she won the chess tournament than not. For schema-incongruent behaviors, the probability of the behavior given the schema, is less than .50 approaching zero. Given that Judy's very intelligent, it's very unlikely that she made the mistake, the same mistake three times in a row. And for schema-irrelevant behaviors, the probability of the behavior given the schema is .50, just chance. Given that she's smart Judy might have ordered a sandwich for lunch, or she might have ordered a salad. Now when we plot memory for the behaviors as a function of their congruence with the schema, we see an interesting pattern of results -- a pattern known as a U-shaped function, because that's what it looks like. Schema- relevant behaviors, whether they're schema-congruent or schema-incongruent, are remembered better than schema-irrelevant behaviors. But among the schema-relevant behaviors, schema-incongruent behaviors are actually remembered better than schema-congruent behaviors. Surprising, unexpected schema incongruent behaviors are remembered better than schema congruent behaviors. This U-shaped relationship between schema-relevance and memory illustrates the schematic processing principle, which says that memory for specific events -- episodic memory -- is a function of the relationship between that event and preexisting schemata, generalized knowledge, expectations and beliefs, recorded in semantic memory. We can understand the role of schemata in memory with reference to our earlier stage analysis of memory because it turns out there are two different processes contributing to the schematic processing principle -- one operating on the encoding end, the other operating on the retrieval end. Schema-congruent events fit right into our prevailing schemata, our prevailing knowledge, expectations, and beliefs about the target person. And so the schema provides extra cues at the time of retrieval -- which is going to make retrieval more likely to succeed. Remember the cue-dependency principle. If you want to know what a person has done, it helps to remember what kind of person he or she is, because chances are the person behaved in a manner that's consistent with these general traits and attitudes. So, knowing that Judy is intelligent, intellectually sophisticated, and artistically sensitive will lead us to search memory for intelligent, intellectually sophisticated, artistically sensitive, kinds of behaviors and experiences. Schema-incongruent events, however, don't fit into our prevailing schemata. They are surprising, they are not predicted by what we know, and therefore we have to explain them. And this explanatory activity, in turn, is going to result in more elaborate processing at the time of encoding -- and therefore better recall at the time of retrieval. If I tell you that this intelligent, intellectually sophisticated person had difficulty understanding the day-time television show, you're likely to try to figure out why that occurred. Maybe she was distracted by trying to solve Fermat's last theorem. Or maybe she was on the phone with Pablo Picasso -- something that explains why she did this unexpected thing. And the very process of explaining this unexpected behavior creates a more elaborate memory trace, which favors that information at the time of retrieval. Schema-irrelevant events, however, don't get either of these advantages. They're not unexpected, so they don't receive much elaboration at the time of encoding, nor does the schema provide informative cues to their retrieval. They get a kind of double whammy, and because of this double disadvantage, schema-irrelevant events are poorly recalled. So here's an expanded statement of the schematic processing principle. Episodic memory is a function of the relationship between the event you're trying to remember and pre-existing schemata that comprise part of semantic memory. Schema-relevant events are remembered better than schema-irrelevant events. That's the basic schematic effect on memory but among schema-relevant events, schema-incongruent events are remembered better than schema-congruent events. That is because schema-incongruent events get a lot more processing at the time of encoding because they're unexpected and they have to be explained. So now we have a set of basic operating principles for episodic memory and, it turns out, surprisingly, it's a pretty short list of principles. We've got two principles governing the encoding of memory, elaboration and organization. We've got one principle governing the storage phase of memory processing, time dependency. And we now understand that the basic mechanism underlying time-dependent forgetting is interference, proactive interference and retroactive interference, competition among available memories. And then we have a set of principles governing the retrieval process. We have a principle of cue-dependency, which basically says that you gain access to available memories depending on the amount of information provided by the retrieval cue. This is qualified by a principle of encoding specificity, which says that it's not so much the sheer amount of information provided by the retrieval cue as the right kind of information. The information that's guiding retrieval has to somehow match the information that was processed at the time of encoding. And then we have a principle of schematic processing, by which schema-incongruent behaviors get a memory advantage by virtue of extra elaboration, and schema-congruent behaviors get an extra memory advantage by virtue of cue-dependency, because the schema provides extra retrieval cues. That's six principles: elaboration, organization, time dependency, cue dependency, encoding specificity, and schematic processing. But, you know that paper on the magical number 7, plus or minus 2. It'd be really nice if there were seven principles of memory. And it turns out that there is. So far, our analysis of memory has been guided by what has been called the library metaphor of memory. In this metaphor, memories are represented by memory traces, which are like books -- which contain information. Memories are encoded -- just as books are purchased and catalogued and shelved by librarians. And memories are stored -- just as books remain on the shelf until the information in them is needed by some reader. And then memories are retrieved -- just as we look books up in a catalogue, find them on the shelf, check them out, and read them for the information they contain within. The six principles of memory that I've just outlined are analogous to the processes involved in processing a book through the library. This idea of memory as a storehouse of information like a library is very old: it goes back at least as far as the ancient Greeks. And it's the view that dominated modern scientific studies of memory, from the time of Ebbinghaus until relatively recently. The idea here is one of memory as reproduction. Just as perceptual processes produce a mental representation of some event in the present, the idea is that memory processes re-produce a mental representation of some event from the past. Encoding creates a record of our knowledge about some event, and retrieval gains access to that knowledge -- much as if you want to know something about some topic, you pick up a book and you read about it. From this point of view, remembering really entails reproducing some mental representation of some past event. And the mechanisms of forgetting boil down to basically two different kinds of processes. Either you forget because the information in the memory trace is unavailable to you -- perhaps because it was poorly encoded; or you forget an event because the information about the event was inaccessible to you -- it's in memory, but you simply can't retrieve it. This reproductive view of memory has taken us a very long way toward understanding how memory works, but it turns out that it's not the entire story. An alternative view of memory was first proposed by Frederick C. Bartlett, a British psychologist -- we've talked about him before -- in a book he wrote in 1932. Bartlett vigorously objected to the techniques of verbal learning that were used by Ebbinghaus and those who followed in his tradition. And he was especially upset about Ebbinghaus' invention of the nonsense syllable, in an attempt to develop stimulus materials that were stripped of meaning. For Bartlett, the whole point of perception and memory is that the subject is engaged in effort after meaning -- and you can try all you want to strip your materials of meaning, in an effort to study what Ebbinghaus thought of as pure memory. But that effort is going to fail for the simple reason that subjects will find meaning in even meaningless material -- because that's what they have to do. It is our nature as cognitive beings that we try to find meaning in our experience. And it's just a mistake to study perception and memory in a way that tries to get around that fundamental fact of human nature. So Bartlett said look, memory isn't like reproducing rote associations that you've memorized one syllable after another. Instead, remembering an event is more like telling a story. The things we've done, the things we've experienced, occur in a particular kind of context. They have a beginning, and a middle, and an end. And you've really got to take account of that if you really want to understand how memory works. So what Bartlett proposed to do was to study memory for stories, not for lists of nonsense syllables or randomly selected words, or anything like that, because memory for stories captured what Bartlett thought was the essential feature of memory. Memories are not things like books, that are encoded and stored and retrieved and reproduced as verbatim accounts of some past event. The problem for memory is not to reproduce the past as it occurred and forgetting isn't simply a failure of reproduction. By contrast, Bartlett argued that memories are narratives about the past. They are reconstructions of past events rather than reproductions of them. And he proposed to study this reconstructive aspect of memory by changing the basic paradigm for the study of memory. Instead of asking subjects to memorize lists of nonsense syllables or words or whatever, he asked them to listen to and then retell stories. In his method of repeated reproduction, Bartlett would tell a subject a story and then ask the subject to recall the story on several occasions, one time after another, spread out over a period of hours or days. In the method of serial reproduction, he'd tell one subject a story, and then have that subject tell the story to a second subject, who in turn would tell the story to a third subject, and so on. In this way, Bartlett hoped to capture this essential feature of remembering as telling a story about the past, and he wanted to see how these narratives about the past changed over time and across people. Now in selecting his stimulus materials, Bartlett tended to rely on unfamiliar, unusual stories. Bartlett was an anthropologist as well as a psychologist. And he was particularly interested in Native American stories because they had contents, and structures, and details that would be relatively unfamiliar to his English university subjects. And Bartlett, always interested in the perceiver's effort after meaning wanted to watch his subjects try to make sense of this very unusual material. One of Bartlett's favorite stories for this purpose was a Native American folktale called "The War of the Ghosts," originally collected by the pioneering anthropologist Franz Boas. Here's the story. Just listen. One night, two young men from Egulac, went down to the river to hunt seals and, while they were there, it became foggy and calm. Then they heard war cries and they thought, maybe this is a war party. They escaped to the shore and hid behind a log. Now canoes came up and they heard the noise of paddles and saw one canoe coming up to them. There were five men in the canoe and they said, what do you think? We wish to take you along. We are going up the river to make war on the people. One of the young men said, I have no arrows. "Arrows are in the canoe", they said. "I will not go along, I might be killed. My relatives do not know where I've gone. But you," he said, turning to the other, "may go with them." So one of the young men went but the other returned home. And the warriors went up the river to a town on the other side of Kalama. The people came down to the water and they began to fight and many were killed. But presently the young man heard one of the warriors say: Quick, let us go home, that Indian has been hit. Now he thought, Oh they are ghosts. He did not feel sick, but they said he had been shot. So the canoes went back to Egulac, and the young man went ashore to his house and made a fire. And he told everybody and said, behold, I accompanied the ghosts, and we went to fight. Many of our fellows were killed, and many of those who attacked us were killed. They said I was hit and I did not feel sick. He told it all and then he became quiet. When the sun rose he fell down. Something black came out of his mouth. His face became contorted. The people jumped up and cried. He was dead. Now, just take a minute or two to write down the story of the war of the ghosts as you remember it and when you're finished, proceed to the next slide. Here's the first half of the War of the Ghosts. And you can turn to the next slide for the second half. When you've checked your memory against the original, just go on to the next slide and we'll talk about what Bartlett found. Whether Bartlett employed the method of repeated reproduction or the method of serial reproduction, he found pretty much the same things. Of course, one of his observations was that his subjects progressively forgot minor details, though they always retained the gist of the stories -- that there were Indians and ghosts and something about canoes and somebody dying. But Bartlett also observed other kinds of changes which he thought were part and parcel of normal remembering and forgetting. First his subjects tended to omit features of the story that didn't conform to their expectations. If there was something really unusual, really unfamiliar about the story, that tended to disappear. But not only did they omit things from their stories, they actually added things to the stories, committing errors of commission, as well as errors of omission. For example, in rationalization, subjects would import details to explain puzzling passages. They'd bring things into the story that really hadn't been in the original. They would also transform details from being relatively strange and unfamiliar to be relatively more familiar. They might not talk about canoes; they might talk about boats instead. And they'd also transform the order of the story in order to provide a more coherent structure to the narrative -- so it had a clearer beginning, middle, and end just like you are familiar with in Western literature. Over time, or over storytellers, Bartlett observed that the story got shorter, but it also got more coherent. Bartlett's conclusion was that remembering involves reconstructive, not merely reproductive, activity. By analogy to the perceiver, in Bartlett's view, the rememberer is trying to build up a representation of a past event, just the way the perceiver is trying to construct, build up, a representation of a current event. Just as the perceiver extracts prominent features from the stimulus, in Bartlett's view the rememberer retrieves the dominant details of the story, maybe combines that with something about his or her attitude toward the event -- Oh, this is a positive thing or this is a negative thing or whatever -- and then builds up the rest from that foundation, resulting in a narrative that may well be coherent but may not at all be accurate in terms of being an accurate representation of the event that took place. Just as perceivers make schema-based inferences about the stimulus in the current environment, so rememberers make schema-based inferences about what must have happened in the past -- inferences that are based on their attitudes, their expectations, and their knowledge of the world. This idea that the rememberer creates, builds up, a, mental representation of some past event, much the same way the perceiver creates, builds up, a mental representation of some current event, is the essence of Bartlett's reconstructive view of memory, the alternative to the Ebbinghausian view of memory as a matter of reproducing the past. Bartlett's work illustrates what we'll call the reconstruction principle in memory, which simply says that memory reflects a blend of information contained in memory traces with knowledge, expectations, and beliefs that are derived from other sources. The reconstruction principle reminds us that remembering is not simply a matter of retrieving encoded information from passive storage. If memory were merely a byproduct of perception, then all we could do is either remember what happened to us or forget it. But reconstructive processes allow us to remember events differently from the way they happened -- and even allow us to remember things that did not happen at all. This is because memories are not actually reproductions of the past. They're really beliefs about the past -- beliefs that, like any other belief, are consistent with what we know and believe about ourselves and the world; but they may not be entirely accurate. For the most part, then, we recreate events and experiences in our minds each time we remember them. We begin with fragmentary details, perhaps suggested by the retrieval query, perhaps retrieved from the memory trace, perhaps both. Then through reconstructive activity, we fill in the gaps, and flesh out the details, making inferences from our knowledge of ourselves and the world. The resulting memory, then, is a kind of educated guess about what might have happened, rather than a more or less veridical representation of what actually did happen. Bartlett used memory for stories in his experiments because that's what he thought memory was like: telling a story. But his basic point, that memory is reconstructive in nature, not merely reproductive, can be made with studies that employ variants on the traditional verbal learning paradigm. For example, Elizabeth Loftus and her colleagues employed a variant on verbal learning in their studies of eyewitness memory. In these studies, subjects view a series of slides or a short film, depicting an accident or a crime, and they're later asked questions about what they saw. The subjects thus take the role of bystanders or eyewitnesses to some event. In one experiment the subjects viewed a series of slides depicting an auto-pedestrian accident. I'm going to show you these slides now, and I want you just to take the role of a bystander watching the following scenes. Set your pointer so that you can advance to the next slide when I tell you to, and then stop when I tell you to. It's a fairly long series of slides. Alright, ready, go. Okay, now you should be on slide 50 and now I'm going to ask you some questions about what you just saw just as we would in an experiment. Just take out a sheet of paper and write down your answers: you can check them later. First, did anyone pass the couple while they were walking down the driveway? Just write down your answer. Next, did another car pass the red Datsun while it was stopped at the stop sign? Just write down your answer. And next, was there another car in the street when the man entered the crosswalk? Just write down your answer. And finally was the man in the red shirt the driver of the Datsun? Just write down your answer. And now for a different kind of test, involving recognition rather than recall. Which of these slides did you actually see in the slideshow? Slide A or Slide B? Which of these slides did you see in the slide show, Slide A or Slide B? Here you can check your answer. You actually saw slide B with the pedestrian in the cross walk not outside of it. And you probably got that one right, but you may well have gotten this one wrong. You actually saw slide A with the triangular yield sign, but you may have very well have remembered seeing slide B with the octagonal stop sign. You can go back and check: it's Slide 36. If you made that mistake, recognizing a stop sign when you actually saw a yield sign, you fell into what's now known as the post-event misinformation effect. You watched a traffic accident and then engaged in a kind of interrogation in which you were asked questions about what you saw. The critical question was about the red Datsun while it was stopped. Half the subjects in the experiment got a non- leading form of that question: Did you see another car pass the red Datsun was at the yield sign? -- because after all, it was stopped at a yield sign. But the other half of the subjects got a leading question. Did you see another car pass the red Datsun while it was stopped at the stop sign? Now, you didn't actually see a stop sign, but the leading version of the question presupposes that there was a stop sign. That information, which is misinformation encoded after the event affects subsequent recognition -- so that subjects who got the leading question are more likely to falsely recognize the stop sign, when in fact they saw the yield sign. Here's the actual design of the experiment, and there have been a lot more experiments like this since this one. Half the subjects saw a version of the slide show with a yield sign and the other half saw a version with the stop sign. Half the subjects received a question referring to a yield sign, and the other half received a question referring to a stop sign. For the subjects who actually saw a yield sign, the question assuming a yield sign is not a leading question, because there really was a yield sign. But the question assuming a stop sign is a misleading question because it assumes something that wasn't true. And the reverse holds for those subjects who saw the stop sign. And here are the results of the study. Subjects who got the non-leading question were pretty accurate in their recognition, more than 70% correct. But for subjects who got the misleading question, their recognition accuracy was cut almost in half. Again, this phenomenon is known as the post-event misinformation effect, because information, misinformation, acquired after an event is incorporated into the person's memory for the event itself. The legal implications of this effect are important, obviously, because it shows that eyewitnesses can incorporate information from leading questions into their memories, leading them to remember things differently than they otherwise would have had the leading questions been omitted. But in this case, the post-event misinformation effect shows how subjects take all the knowledge that's available to them, not just knowledge from the memory trace, and use all that knowledge to build up a mental representation of some past event. The memory trace, such as it is, was of a yield sign. But the leading question suggested that there was a stop sign. You might not have paid too much attention to the sign as you were looking at the slides, and so when trace information is vague, fragmentary, and ambiguous, the person is naturally going to draw on other things that he or she knows, or believes, to fill in the gaps. The memory's not a reproduction of what the person saw. It's a reconstruction of the event. When a subject who has seen a yield sign falsely recognizes a stop sign, the memory is, in some sense, illusory. The subject was remembering something that didn't actually occur. Another illusion of memory can be induced by asking such subjects to study a list of words which are all semantic associates of another word, which is omitted from the list. For example, subjects might study this list of words, all of which are associatively related to the word needle. On later tests of recall and recognition, subjects will often falsely remember that the semantic associate, "needle," also occurred on the list when, in fact, it did not. And under certain circumstances, false memory for the critical target, "needle," can be as strong as true memory for actual list items like thread and sewing and knitting. This phenomenon is known as the associative memory illusion and it can be very strong. In this study correct recognition was very good for items that were actually on the study list -- more than 80% correct. But false recognition of the associatively related item, needle, was also very strong -- more than 80% of subjects falsely recognized needle when in fact it didn't occur on the list. And this was not because they were saying "yes" to just anything, because false recognition of unrelated words, like the word sleep, occurred at a very low rate. This is an illusion in which people have false memory for items that are associatively related to items that they did study. Here's how the illusion occurs. All the items in the study list are strong associates of the word needle. If I gave you the world needle and asked you to tell me the first word that comes to mind, you're very likely to give me words like thread, or pin, or eye, or thimble, or haystack, or cloth, or knitting. These are known as forward associations. I give you the word needle and you respond with thread, or whatever. But semantic associations can also work backwards. If the word needle makes you think of the word thread, the word thread can also make you think of the word needle. And the more this happens -- the word pin making you think of needle, the word sewing making you think of needle, the word thimble making you think of needle -- the more you start to believe that needle was on the list as well. It's an illusion of memory. It's an illusion produced by the associative relations among items. The associative memory illusion is just one of many illusions of memory, including the post-event misinformation effect. They are illusions, because in each case the person is remembering an event that never occurred -- just as in perceptual illusions, the person sees something or hears something that isn't really there. In the same way that we have perceptual illusions that lead us to perceive the present environment inaccurately, we can also have memory illusions that lead us to remember past events inaccurately. And just as perceptual illusions are produced by the misapplication of the perceiver's preexisting knowledge and beliefs, including unconscious inferences, so memory illusions are a product of preexisting knowledge, including knowledge of the semantic associations among words that a person brings to encoding and retrieval. Illusions occur in perception, but they also occur in memory. In perception, illusions are a product of constructive activity, in which the perceiver goes beyond the information given by the stimulus. In memory, illusions are a product of reconstructive activity. In which the rememberer goes beyond the information given in the memory trace. The reconstruction principle has extremely important implications for how we think about memory. Memories are not records of the past that are simply reproduced from passive storage. Rather, they're beliefs about the past which we construct and reconstruct based on all the knowledge that's available to us. In the ordinary course of everyday living, when we remember events and experiences, we recreate them every time we remember them. We begin with fragmentary details, either suggested by some question or retrieved from the memory trace. And then through reconstructive activity we fill in the gaps, flesh out the details, making inferences from our knowledge of ourselves and the world. The resulting memory then is not so much a reproduction of some indelible record of the past, but rather is a reconstruction. It's a belief; it's an educated guess about what might have happened. So there you have it: a full set of principles governing how the memory system works. We have principles of elaboration and organization that guide the encoding of new memories, making them available for later use. Storage is governed by a principle of time dependency -- which, in turn, is -- at least as far as long term memory is concerned -- produced by the phenomenon of interference. The retrieval process by which we gain access to information that's available in memory store is governed by principles of cue-dependency, encoding specificity, and schematic processing. But retrieval is also governed by a principle of reconstruction. Remembering an event from the past is not really like pulling a book off the library shelf and reading something off a page. Rather, remembering is more like writing a page afresh, each time, based on fragmentary notes, at least so far as conscious recollection is concerned. And that's what we've been discussing so far in these lectures. Whether we're talking about short-term memory, or long term memory, or working memory, the memory in question is conscious. It's something that we're aware of. But it turns out that memories can also affect us unconsciously, outside of our awareness. Modern cognitive psychology draws a distinction between two expressions of episodic memory, explicit and implicit. Explicit memory refers to, conscious recollection of some past event, as exemplified by performance on things like recall and recognition tests. But if you think about it, there are lots of other ways in which we can express our memory for some past event. Implicit memory refers to any change in the person's experience, thought or action that is attributable to a past event. The fact that some past event has affected our current behavior implies that we've retained some memory for that past event -- even if we can't consciously remember it. And it turns out that these two expressions of memory, explicit and implicit, can be dissociated, such that a person's explicit memory, conscious recall and recognition, can be grossly impaired, but, the person's implicit memory, these changes in behavior that are caused by these past events, can be completely unaffected. Implicit, unconscious expressions of memory can be spared, even when explicit, conscious expressions of memory are severely impaired. Here's an example. In one experiment, amnesic patients, patients like patient H.M., were asked to study a list of words, and later they were tested for their memory for the list of words by means of a standard test of recall or recognition. But later, they were asked to play a guessing game, in which the experimenter gave them the first three letters of the word and asked them to fill in the blanks. This is known as word-stem completion or another game, known as word-fragment completion in which the investigator gave them some random letters from a word and then asked them to fill in the blanks and complete the word. It turns out that these patients tended to fill in the blanks with words from the list that they had studied -- even though they didn't remember any of the words from the list that they had studied. Here, a past event, studying the list of words, affected their performance on some current task, word-stem completion or word-fragment completion -- even though they had no conscious recollection of the past event. That's what implicit memory is all about: the unconscious influence of past events on present behavior. Here's the study. It's now a classic in the neuropsychological literature on memory. Compared to non-amnesic control subjects, the amnesic patients showed very poor memory for the list, based on tests of free-recall and recognition -- standard tests of explicit or conscious memory. But despite the fact that they couldn't recall or recognize the words that they had just studied, they were just as likely as the non-amnesic controls to fill in the blanks with words from the study list. Implicit memory, as expressed in word-stem completion and word-fragment completion performance, is dissociated from explicit memory, as reflected in recall or recognition. This is a phenomenon known as priming, which is a standard measure of implicit memory. In priming, performance of one task, like studying a list of words, affects, usually facilitates, performance of another task, like performance on word stem or fragment completion tests. There are lots of different kinds of priming. In positive priming, which is what we've been talking about here, the first task facilitates performance of the second task. In negative priming, the first task can actually impair performance of the other task. But either way, whether the priming is positive or negative, the fact that priming occurs means that some memory for the first task has got to be preserved in memory. And this is true even in the case of the amnesic patients; they don't remember the first task at all. Here's another example of priming, a quick example: a study from my laboratory of priming in retrograde amnesia caused by electroconvulsive therapy. You'll remember from our discussion of consolidation, that a dose of ECT typically renders the patient amnesic for events that occurred in the period of time immediately prior to the treatment -- retrograde amnesia. Well, in this study we asked a group of patients to study a list of words right before they received a dose of ECT. After recovering from the treatment we tested their memory first by an explicit memory test, a test of cued recall, on which they performed very poorly. But when we gave them a completion test, they performed as well as controls -- in fact a little bit better. They showed priming on the completion test even though they didn't have very much recall of the words to begin with. Implicit memory, represented by priming effects on word completion, was spared even though explicit memory, performance on the recall test, was grossly impaired. The principles governing implicit or unconscious memory may differ somewhat from those that govern explicit or conscious memory. But even so, I think you can see that we can offer a unified view of perception and memory. Perception draws on memory; and perception changes memory. But both perception and memory reflect constructive activity in conscious cognition. Through perceptual construction, we build up a mental representation of our current experience. Through memory reconstruction, we build up a representation of our experiences of the past. In the same way that we can have unconscious perception, in the form of subliminal perception and other manifestations of implicit perception, we can have unconscious memory, in the form of priming effects in amnesia. Memory and perception go together. Memory and perception both involve reasoning, judgment, problem solving, and decision making. And it's to those topics that we turn next.