When most people think of memory, they tend to think of a place in which information is put and stored until it is needed, much like a library. Unfortunately, this metaphor is quite misleading in that it implies a static, veridical process. Nothing really happens to library books while sitting on the shelf; they may grow musty, they might be miss helved, but once one has the book, the contents are identical to the last time the book was consulted. In contrast, human memory is a dynamic, fundamentally reconstructive set of processes that enable previously encoded information to affect current and future performance.
Construction And Reconstruction
Memory works like perceptual and other cognitive processes: people use whatever cues and information are available to achieve a sensible interpretation of the information processed. Consider the case of recalling what happened at the college football game last week. The first time a retrieval attempt is made, there are three sources of information: (1) from the event itself, (2) from similar events, and (3) from general knowledge of what happens at football games. All three sources of information are involved in the construction of a memory. The spectator might remember a specific play, which most likely comes from memory of the event itself. But if asked whether he remembers the coin toss that starts the game, he might mistakenly recall a coin toss from the previous week’s game, or he might base his response on his general knowledge that a coin is tossed to determine which team goes on offense first. Note that the information from general knowledge or from a different event might be accurate, even though it has been retrieved from a different source.
The second time a retrieval attempt is made, there is an additional source of information: memory of the previous recollection. Memory is said to be reconstructive in that each time a particular event is recalled, a new version is constructed based on the cues and information available at that particular time. The constructed version is then a potential source of information for subsequent recollections.
Remembering Things That Did Not Happen
One consequence of the reconstructive nature of memory is that it is quite easy to create a situation in which people recall events that did not happen. Unfortunately, this has been given the name “false memory.” There are two problems with this term. First, it suggests a dichotomy between those memories that are true and those that are false. The problem is that because all memories are a combination of details recalled about the event and information from other sources, at least part of every memory is likely to be “false” because at least some details are not of the original event. Second, it suggests that the memory system stored false information. This obscures the important point that the memory system is not like a library; rather, memory is dynamic and continuously changing, as the examples below demonstrate.
There are now so many laboratory demonstrations, that show that people are easily led to remember events that did not happen that it is beyond doubt that this also happens with great frequency in everyday life. For example, people see a list of items (“molehill,” “valley,” “summit,” “goat,” “ski,” “bike,” “peak,” and so on) that are all related to a critical word. When people try to recall such lists, they usually recall the word “mountain.” More important than the fact that they recalled a word that was not presented is the finding that they are convinced that this word was in fact on the list. The explanation is that all the words in the list are related to “mountain,” and presumably the person thinks of this word often during list presentation. At test, the person has the same sort of evidence that “mountain” was presented as any of the other words.
Memories can be created (or implanted) in everyday settings also. One common paradigm is to get family members to play a game of “remember when.” Here, an older sister, for example, might ask a younger brother whether he remembers getting lost at a mall when he was about 5 years old. At first, he says no for the reason that this never happened. About a week later, however, he will now have some “memories” of this event. This is known as an attribution error: the memory of being lost is attributed to a real event rather than to a discussion held with a sibling a week ago. Like the laboratory demonstrations, this memory has the same characteristics to the brother as genuine memories. Simply showing a photograph in which a person’s face has been digitally implanted (such as a photograph of the person taking a hot-air balloon ride) can lead the person to begin to recall aspects of the event.
Eyewitness Memory
Because memory is so malleable, and because eyewitness memory can have such important consequences, there has been much research on how to assess whether a particular memory is accurate or inaccurate. The general conclusion is that in the absence of objective evidence, there is no aspect of a person’s recollection that predicts whether the memory is likely to be accurate or inaccurate. This conclusion comes from both laboratory studies and “real-world” studies. In particular, factors that have been shown to be unrelated to the accuracy of a memory include: the emotionality of an event; the duration of an event; the importance of the event to the person; the amount of detail also recalled about the event; the speed at which the event is recalled; and the accuracy of other related memories. Because of this, many countries do not allow for criminal prosecution if the only evidence is from an eyewitness.
If an eyewitness to a car crash is asked how fast the cars were going when they “hit” each other, the eyewitness will report a much lower estimate than if asked how fast the cars were going when they “smashed into” each other. People act as if they are making an inference: for cars to “smash” into each other they must be going fast, therefore the estimate is higher. Answers to subsequent questions are also affected by these initial answers. When asked if there was broken glass, people who receive the “smashed” question are more likely to report there was broken glass than those who receive the “hit” question.
It is important to note that in laboratory studies, it is assumed that the witness is trying to be as honest as possible and is trying as hard as possible to recall details. In the legal system, this may not be a valid assumption, which makes interpretation even more difficult. It is also important to note that this does not mean that all eyewitnesses are recalling information inaccurately or even that memory of an event cannot be accurate. What it does mean is that there is no way to assess the accuracy of a particular memory: It could be very accurate, it could be very inaccurate, or it could be somewhere in between.
Hypnosis And Memory
In studies of the effect of hypnosis on memory where there exists objective information that allows the researcher to assess the accuracy of recall, there is no evidence that hypnosis per se facilitates recall. On the contrary, people under hypnosis are easier to mislead than non-hypnotized people, and they have difficulty distinguishing between what happened under hypnosis and what happened when not hypnotized. Moreover, there is evidence that whereas hypnotized people are no more accurate than nonhypnotized people, they often express far greater confidence in their recollections.
Related to the finding that people have difficulty distinguishing between events that happened while hypnotized and those that happened when not hypnotized are compelling results from a slightly different paradigm. People are asked to perform a task or are asked to imagine performing the task; in another version, they are shown pictures of objects or asked to imagine a picture of the object. At test, they are asked to categorize the test items as either real (they performed the action or saw the picture), imagined (they imagined performing the action or imagined the picture), or new (the item was not part of the experiment). The judgment of real versus imagined is much more difficult than the judgment of whether the item is new or not. Items are more likely to be judged as “real” when they were imagined than the other way round.
The Encoding Retrieval Interaction
One of the primary determinants of recollection is the relation between the conditions at encoding (when the information was acquired) and the conditions at retrieval. If a person is happy when studying a particular set of information, more information will be recalled if the person is happy at test than if unhappy. More information will also be recalled if a person is unhappy at both encoding and retrieval than if the emotional state differs. Similar results are found with environmental context and with pharmacological state. People taking scuba-diving lessons need to learn decompression tables that tell them how to ascend to avoid decompression sickness; they will do much better if they learn the information underwater, the same environment in which they will need to recall the information, than if they study only on land. In general, information acquired under a particular pharmacological state (e.g., alcohol intoxication) will also be better recalled under the same state than when sober.
The reason is that items and events are not processed in isolation but rather as part of ongoing mental processes. People who are happy (or underwater or intoxicated) will process words and events differently than if they are unhappy (or on land or sober). Thus, when trying to access information originally processed in a different state, people will generally be trying to process the information inappropriately.
There is a very important consequence of acknowledging that whether something will be remembered depends on the relation between the conditions at encoding and the conditions at test: one cannot make absolute statements about the mnemonic properties of an item, a process, or a cue. This means, for example, that one cannot say recognition is easier than recall, or pictures are easier to remember than words, or one particular cue is better than another kind of cue. It is always possible to change the retrieving conditions to render the absolute statement false.
Theoretical Accounts Of Memory
There are two basic theoretical accounts of memory. One views memory as a set of different memory systems (the systems or structuralist account) whereas the other views it as a single system and emphasizes the role of processing (the processing or proceduralist perspective).
The Multiple Systems View
Among proponents of the systems view, most hold that there are five different memory systems. Working memory (also known as short-term memory) is used for the temporary storage and manipulation of information. Long-term memory is made up of two systems, episodic memory (also called autobiographical memory) and semantic memory (also called generic memory and general knowledge). The difference between the two lies in whether the rememberer recollects just the fact itself or whether there is also awareness of the context in which the information was learned. For example, most people know that Paris is the capital of France, but they have no memory of when they learned this. In contrast, imagine that you are just told that the capital of Samoa is Apia. In addition to knowing the fact itself, you have information about when you learned it (you read it in the International Encyclopedia of Communication, you were sitting in a library, and so on). In short, you can place yourself back at the learning situation. Episodic memory is sometimes described as having the property of mental time travel: you can project yourself backward into particular episodes. Thus, there is no difference in the type of information stored in the episodic and semantic systems; the sole difference is whether the information is accompanied by awareness of the learning episode.
The last two memory systems differ from the preceding three in that they are not part of the declarative group of memory systems. A rule of thumb is that if one can say that one knows that something (e.g., one knows that 2 + 2 is 4, one knows that the capital of Assyria was Nineveh), it is in a declarative memory system. If, on the other hand, one knows how to do something, then it is in a procedural memory system. A mother may know how to ride a bicycle, but this information cannot be usefully communicated to her son. She can say, “Balance, peddle, and steer,” but he will most likely fall off. The two nondeclarative memory systems are procedural memory and the perceptual representation system. Both of these are sometimes referred to as implicit memory.
Implicit memory is the term used to refer to situations in which memory is affected by prior events but the rememberer has no awareness of this influence. People with anterograde amnesia, who have difficulty remembering new information, often demonstrate unimpaired implicit memory. For example, such people might not remember the name of their physician even though the same physician has treated them for months. However, they can acquire and use information implicitly. The Tower of Hanoi (also called the Tower of London and the Tower of Babylon) problem has three pegs and a number of different-sized disks that must be moved from the first peg to the last peg. A larger disk cannot be put on a smaller disk, and only one disk can be moved at a time. People with anterograde amnesia demonstrate that they can learn this task: each time, it takes them fewer trials to succeed. However, they often claim to have never performed the task before and to not know the rules.
Episodic memory requires conscious awareness of the original learning episode. Semantic memory requires awareness of the information, but not of the original learning episode. Nondeclarative memory requires no awareness of the information at all.
The Processing View
According to the systems view, the mnemonic properties of an item depend on the memory system in which it resides. According to the processing view, the mnemonic properties of an item depend on the relation between the conditions at encoding and the conditions at test. According to the former view, one cannot make generalizations about memory as a whole because each memory system operates according to different rules. According to the latter, there are important principles of memory that apply to all memory regardless of the type of information, the type of processing, the hypothetical system supporting the memory, or the time scale.
The difference between the two approaches can be readily appreciated from Table 1, which characterizes short-term memory (STM; also known as working memory) and long-term memory (LTM; also known as episodic memory). When an item is in STM, it will be stored using an acoustic code (how it sounds), the capacity will be limited to seven plus or minus two items, and the items will decay within about 20 seconds. In contrast, when an item is in LTM, it will be stored using a semantic code (what it means), the capacity will be (almost) unlimited, and the item may never be forgotten.
Table 1 Short-term and long-term memory characteristics
A proponent of the processing view would agree with almost everything in the table, with one exception: what does the top row in the table add to our understanding? In other words, we could delete the top row and still have a complete and accurate description of memory. When words are processed on the basis of how they sound, there will be a very small capacity and the information will not be available for long. When words are processed on the basis of what they mean, however, there will be an enormous capacity, and the information will be available for a long time.
Transfer-Appropriate Processing
The leading processing account is known as transfer-appropriate processing and it states that memory will be best when the processing done at encoding is appropriate for the processing required at test. Consider two types of tests commonly used in academic settings: multiple choice and short answer. Each requires a different kind of processing at test. In multiple choice tests, students are given all of the key terms, and the task is to be able to generate definitions of those terms and concepts. Thus, studying for a multiple choice test should include practicing generating definitions of key terms and concepts. Reading and re-reading notes is not a good study technique, as this is not appropriate for the kind of processing required at test. For a short answer test, students are more likely to need to be able to generate not only information about key concepts and terms, but also those terms and concepts themselves. Therefore, studying for this kind of test should also include practice generating the key ideas.
Mnemonics
Transfer-appropriate processing naturally lends itself to helping people improve memory. Keep in mind that the goal is to anticipate the kind of processing that will be required at test, and then organize studying around it. For example, how can memory for people’s names be improved? At test, the only constant is typically the person’s face. Therefore, the face should be used as the retrieval cue, and studying should be built around that. Second, information can be recalled only if it is encoded. Get into the habit of repeating the person’s name as soon as the introduction is made. Third, form a link between the cue (i.e., the face) and the name. Fourth, use the person’s name a couple more times before moving on to the next person.
References:
- Conway, M. A. (ed.) (1997). Recovered memories and false memories. Oxford: Oxford University Press.
- Dinges, D. F., Whitehouse, W. G., Orne, E. C., Powell, J. W., Orne, M. T., & Erdelyi, M. (1992). Evaluating hypnotic memory enhancement (hypermnesia and reminiscence) using multitrial forced recall. Journal of Experimental Psychology, Learning, Memory and Cognition, 18, 1139 – 1147.
- Eich, E. (1989). Theoretical issues in state dependent memory. In H. L. Roediger III & F. I. M. Craik (eds.), Varieties of memory and consciousness: Essays in honour of Endel Tulving. Hillsdale, NJ: Lawrence Erlbaum, pp. 331–354.
- Foster, J. K., & Jelicic, M. (eds.) (1999). Memory: Systems, process, or function? New York: Oxford University Press.
- Loftus, E. F. (1997). Creating false memories. Scientific American, 277, 70 –75.
- Morris, C. D., Bransford, J. D., & Franks, J. J. (1977). Levels of processing versus transfer appropriate processing. Journal of Verbal Learning and Verbal Behavior, 16, 519 –533.
- Neath, I., & Surprenant, A. M. (2003). Human memory: An introduction to research, data, and theory, 2nd edn. Belmont, CA: Wadsworth.
- Neath, I., & Surprenant, A. M. (2005). Mechanisms of memory. In K. Lamberts & R. L. Goldstone (eds.), Handbook of cognition. London: Sage, pp. 221–238.
- Schacter, D. L., Wagner, A. D., & Buckner, R. L. (2000). Memory systems of 1999. In E. Tulving & F. I. M. Craik (eds.), The Oxford handbook of memory. New York: Oxford University Press, pp. 627– 643.
- Smith, S. M., & Vela, E. (2001). Environmental context-dependent memory: A review and metaanalysis. Psychonomic Bulletin and Review, 8, 203 –220.
- Stadler, M. A., & Frensch, P. A. (eds.) (1998). Handbook of implicit learning. Thousand Oaks, CA: Sage.
- Wagenaar, W. A., & Groeneweg, J. (1990). The memory of concentration camp survivors. Applied Cognitive Psychology, 4, 77– 87.