Chapter 6: Memory Processes

Having discussed the general phenomenon of memory in the previous chapter, the author now turns to more specific operations related to memory"

The concept of representation is significant: when you recall a sensory stimulus you do not experience it again. When you remember seeing something, you do not actually see it again. On rare occasions, you might close your eyes to re-envision it. But in most instances the memory does not require conjuring an imagined representation.

Of note, encoding, storage, and retrieval is a linear process and the steps are done in that order each time. A memory cannot be erased or changed, but a new memory created that takes precedence in the retrieval process. That is to say that if we base our representation of "lamp" on the first lamp we saw, and later see give more lamps that are slightly different, it can change the representation when we recall it in future, but it does not erase the memory of the first lamp.

Encoding and Transfer

Encoding of memory is not as straightforward as it might seem. For example, one experiment (Conrad 1964) presented subjects with a visual display of a sequence of letters, asking them to memorize them and write them down. When researchers considered the errors that participants made, most were swapping letters that were phonetically similar (V, B, C, P, etc.) in spite of the fact that they perceived the list visually, they were clearly encoding the letters according to the way they sound rather than the way they look.

As such, memory cannot be rightly regarded as the storage of sensory information in its raw form, but as a more cognitive phenomenon that involves processing sensory data before it is encoded, even in short-term memory.

Further research tested the ability of subjects to recall lists of words, considering whether the similarity of words (big, long, large, wide) as opposed to dissimilarity (pig, hat, tree, radio) and found there to be no statistically significant difference in the results. And likewise, errors in which the wrong word was recalled generally resulted in the reporting of a word that rhymed with the original.

(EN: The author does not mention any results for key-value testing, but I found a few references to studies online [chiefly Seidenberg 1976] that reported improved accuracy and speed in recalling rhyming pairs like coat-goat and non-rhyming pairs like coat-hat.)

Evidence suggests the encoding of long-term memory is semantic. When long-term memory is examined (Grossman 1970), it is found that mistakes are often a synonym of the correct response, by nearly a two-to-one margin when compared to mistakes that were unrelated. Other experiments (Bousefield 1953) related that, given a list of words to remember in random order, participants recalled them in lists of related concepts (animals, plants, professions, names) far more often than would be expected by chance

A separate experiment (Frost 1972) also added the element of visual orientation. Participants were show 16 drawings of objects in four categories, which were also presented as images, some of which were upright, some angled to the left, some to the right, and some ups-de down. When asked to recall them freely, they were often organized into categories, but a statistically significant number of participants grouped them by the original visual orientation.

As such, it is suggested that while long-term memory involves a process of grouping or categorization, the manner in which items are categorized seems entirely arbitrary.

Given that short-term memory and long-term memory encode information differently, the means by which information is moved from one to the other often depends on whether the information is declarative or nondeclarative, and evidence suggests that nondeclarative memory is more volatile, decays more quickly, and is not as efficiently transferred into long-term memory as readily.

(EN: Reading this, I wonder if the practice of transference has ever been questioned. That is, the common assumption seems to be that short-term memory is transferred to long-term - but it also seems entirely possible that they are two different things. That is, sensory information may be received into short-term and long-term memory at the very same time, rather than one before the other, and encoded by separate processes into each. I'm not aware this theorem has been considered. - but it seems plausible given that people are often surprised by details that they remember, and claim not to have noticed at the time.)

It is suggested that a person's memory can be improved by making a conscious effort to remember something - that is we have some degree of voluntary control over the amount of attention we give to certain information.

There are various methods for doing so in the moment of perception:

In addition to the perception methods, there are also cognitive methods used to improve memory:

It is noted that these various methods for memory are things that we may choose to do intentionally, or they may be things we happen to do unconsciously.

Learning is often a process of committing information to memory through a conscious process - whether it is learning information for its own sake (study), or learning information in context of a task (practice). Study and practice are often done by repetition, whether it is massed practice ("cramming" to learn it quickly) or distributed practice (re-reading information a few times over a longer period of time).

Studies (Glenberg 1977, Leicht 1987) have shown that distributed practice is more effective in long-term learning, which substantiates the advice of teachers to study slowly over the course of a semester rather than trying to cram before examinations.

One theory to explain this suggests that distributed learning gives us different contexts in which to remember the information: reading a textbook in a library, hearing a lecture in a classroom, reviewing notes in a dormitory cell, give us three environments, three different somatic states, and three different internal states to which the new information can be associated.

There has also been a "total time" hypothesis that indicates that whether the amount of time studying something is significant, and it does not matter whether a person studies it for thirty minutes at once or for ten minutes in three separate sessions, but this theory ahs been largely disproven. There is evidence that there is only so much that can be learned by repetition in a single sequence - i.e., whether something is repeated four times or twelve times does not have a significant impact when it is done in a single session, but repeating it four times in three sessions does seem to embed it in memory better than repeating it twelve times in one sitting.

Another theory of learning contrasts maintenance rehearsal in which facts to be remembered are merely repeated and elaborative rehearsal in which facts are remembered in a fabricated context, with elaborative rehearsal being the more effective. An example would be learning a new word by repeating its definition five times (maintenance) versus learning the word by hearing in used in five different sentences (elaborative)

There is also strong coordination between deep sleep (the REM stage) and long-term memory, which supports the notion that information gathered during the day is consolidated into long-term memory by night, as we lie sleeping. Various studies suggest that information experienced in the last few hours of the day is better recalled than information gathered earlier in the day, by virtue of its proximity to sleep. It is also well documented that individuals who remain awake for extended periods of time are less able to remember information than when they are following their normal sleep patterns.

Additional studies link the hippocampus to long-term memory, which supports the notion that memory of sense experience exists in the different regions of the brain related to sensory perception, but the coordination and association of data gathered by different senses is managed by this central structure. It is also observed that the hippocampus remains active during periods of sleep, further supporting the notion that long-term encoding occurs during an unconscious state, when that region of the brain is not processing stimuli that are being received from the external environment, and can therefore focus on managing the internal one.

Another way in which researchers have examined the way in which information is stored in the brain has been through free recall exercise, such as those which request subjects to speak or write as many words as they can remember, and considering the way in which those words are reported. In some instances, researchers could readily observe the logical connections between one word and the next in sequence - and even when there was no logic obvious to the researcher, a participant could readily explain the linkage when asked. In general, words were strung together in a specific category of knowledge, and then there would be a bridge to a different category, and the pattern would repeat.

The author then turns to mnemonic devices, which are specific techniques that have been developed over time to improve memory skills. A list of several such techniques is provided:

(EN: The author provides no scientific evidence that any of these asinine practices actually work, and I've long suspected that most of them are just someone trying to be clever after the fact, who doesn't actually use the technique he advocates to others.)

There is a brief mention of common practice, also without scientific evidence, of mnemonic devices such as taking notes, compiling lists, and setting alarms to remind us to do things. (EN: Though I do wonder if this is really the function. That is a shopping list does not cause me to remember, but substitutes for having to remember because I can refer to the list rather than keeping the items in mind. And an alarm is there to interrupt whatever I happen to be doing so that I can remember to do something else - the alarm itself doesn't remind me of what it is. As such I think that these devices serve functions related to memory, but do not actually serve memory itself.)

A distinction is also drawn between retrospective memory (the desire to remember things that occurred in the past) and prospective memory (the desire to remember things we need to do in future). Various research shows that we are much better at retrospective than prospective, and we remark with some sense of irony at very intelligent people who have a great deal of knowledge but cannot seem to remember to do basic things (the absent-minded professor syndrome).

Finally, it must be noted that the understanding of encoding is highly flawed, in that the act of encoding cannot be effectively isolated. That is to say that what is often being assessed is not the way in which memory is stored, but the way in which it is retrieved (discussed in the next section).

We can say with certainty that a person who remembers something has encoded it, but we cannot be certain that a person who fails to remember something has also failed to encode it - they may well have encoded it but are unable to remember it at the time they are asked (and may remember it later on). There is simply no getting around this.


Retrieval of information from memory is considered to be a separate process from encoding, but still suffers from some overlap, in that when a subject fails to retrieve information, it begs the question of whether the problem was in the act of retrieval or that of encoding - though if additional prompting enables them to remember it, this indicates a retrieval rather than encoding error.

In regard to short-term memory, there are tests of serial processing (recalling items in a specific order) and parallel processing (recalling an item from a set in no particular order). It's generally accepted that items are remembered in parallel, but placing it into an order requires a separate task: recalling not only the identity of a stimulus, but its temporal relationship with other stimuli received.

Another distinction to be made with serial memory is recalling items that were given to the subject one at a time (shown one by one, with order being determined by time) or all at once (shown in a list, with order being determined by top-to-bottom or left-to right, which is presumed to correspond to the temporal order in which they were read)

When asked to recall items in a list, there is also a distinction between exhaustive and self-terminating processing. That is, if an item is the third in a list of five, does the subject have to remember all five to recall it, or do they test against the first, second, and third only, stopping because they have found a match?

The resolution to these questions is proposed to be that short-term memory is serial and exhaustive. That is, we remember items in a specific order and consider the full list in remembering any of the items. This is evident by an experiment (Sternberg 1969) that first tested memory of one item versus a lit of items, then comparing the amount of time taken to recall an item near the beginning to a list to one near the end of it. Since there was an increase in response time for a list versus a single item, it is believed the items in the list were recalled serially; and since there was no difference in response times for items at the start of a list versus the end, it is believed that the search was conducted exhaustively. Granted, this proof is not infallible or definitive, but it does seem to support these assumptions.

Long-term memory appears to be categorical, as evidenced by the experimentation considered earlier, in which subjects asked to memorize a list of items tended to repeat them in categories. Participants also preformed better when given a categorized list and asked to recall the items in each category than they did in recalling a list of items that was not categorized.

Categorization dramatically improves memory when subjects are given a detailed hierarchy and taxonomy. For example, if animals are classified into mammals, reptiles, amphibians, fish, and birds, subjects can recall the names of more animals than they can in a list without a clear structure. The difference is quite dramatic. In one such experiment (Bower 1969) those who had a categorization schema recalled 65% of words in a list whereas those who were not given a schema recalled only 19%.

Given this difference, cognitive psychologists have made a distinction between availability an accessibility. Availability concerns itself as to whether the data is resident in memory at all, and accessibility determines whether it can be remembered given certain cues.

That is, a fact that is available in memory may not be remembered if the subject is give none prompt, but it may be remembered if they are given a different prompt. Therefore we cannot definitely say that a piece of information is not available in memory (unless we have tried every possible cue to retrieve it) and can say only that it is inaccessible given a specific method of trying to prompt the subject to remember it.

Forgetfulness and Distortion

of greater concern for practical purposes is not assessing our ability to remember, but addressing our tendency to forget, particularly when forgetting things prevents us from functioning as well as we should.

A distinction is drawn between interference and decay. Interference maintains that something specific occurs that causes us to be unable to remember something, and decay is the loss of memory over the passage of time. These are entirely separate factors that impair recollection.

Different forms of interference are identified. Retroactive interference is an event that occurs after something has been learned but before we are able to recall. For example, when a person is asked two questions, one right after the other, the second question may interrupt their ability to recall the answer to the first. Proactive interference is an event that occurs during the learning process that prevents us from encoding information properly, such as hearing a sequence of random numbers while trying to commit a telephone number to memory. Both of these are said to play a significant role in short-term memory.

Distortion of memory is also common when the amount of information to be remembered is significant. This is particularly evident in asking a person to recount the details of a story he recently heard. Invariably, he will forget some of the details of the story - and often he will invent new details to fill in the gaps in his memory. It is suggested that when this occurs, subjects are altering the details of the story in order to make it easier for them to remember - and it is further theorized that the experience and knowledge that a person has from their life experience will make certain details easier or more difficult for them to recall. That is, they are making changes when they correlate new knowledge into their existing knowledge.

Its also noted that there is a serial-position curve in memory, which causes people to remember the first and last items in the list, but forget the ones in the middle. This phenomenon is considered to be a combination of the recency effect (we remember better what happened most recently) and the primacy effect (we remember better the things that seem more significant). That is, the words at the beginning are memorable because they occurred at the beginning of a task and seemed important, the ones at the end are memorable because they occurred most recently, and the ones in the middle fall into a sort of slush for lack of primacy or recency.

Moving on to decay, it is theorized that memory gradually disappears over time unless something is done to maintain it. The knowledge is not merely disused or misplaced, but gone entirely from memory. This theory is described as "exceedingly difficult to test" because of the nature of memory. A person might appear to have forgotten something for a time, but then suddenly remember it. The examples in the previous chapter of elderly patients who clearly recall incidents and details from their childhoods would seem to indicate memory is indefinite.

As such, the decay theory is difficult either to prove or disprove: when a piece of information cannot be recalled, there is no proof whether it has been permanently eliminated by decay or the subject simply can't recall it at the moment, given whatever prompt.

The Constructive Nature of Memory

Reflecting back to the previous example about the way in which the details of a story are distorted - with some omitted and others invented to fill in the gaps - it is clear that memory is not like a machine that captures raw sensory information and replays it exactly as it was perceived. But instead, memories are constructed from details by the subject.

The construction of memory is based on a limited amount of sense data, which is generally what the subject considers to be salient detail rather than incidental. This is evident in experiments (Bransford 1972) in which an ambiguous narrative is recounted in different ways by different subjects, as the mind attempts to consolidate details into a sensible construct. Similar experimentation (Bower 1975) presented subjects with nonsensical pictures, in which participants discerned different shapes as a method of remembering them.

Autobiographical Memory

Memory is deemed to be autobiographical in three senses:

  1. Reality is perceived by an individual person who has limited perception even the raw data from which memory is constructed is limited to that which a person was capable of perceiving.
  2. Each person interprets events as they relate to themselves, both in terms of their self-interest as well as to their existing knowledge and experience.
  3. Similarly, events are remembered based on when they occurred in the timeline of a person's life. What happened (to them) before an after a remembered event impacts the way in which they remember it.

Studies of autobiographical memory have found no clear indication of factors that create persistence. One such study (Linton 1982) asked participants to use a dairy to record two significant experiences a day, indicating not only the date but the level of emotional impact over a six-year period. Subjects' recall of these events was not related to the recency of events (older events were just as likely to be recalled as newer ones) nor to the self-reported emotional portent of the events.

A similar experiment (Sehulster 1989) asked long-term attendees of an opera house about performances that were staged over a 25-year period, and these followed the more expected pattern of showing stronger recall for recent performance, slightly stronger recall for early ones, and weakness in between.

Memory Distortion

Aside of losing access to some memories, people also have the tendency to distort their memories - or more accurately, to encode or retrieve them in a way that differs from the actual sensory experience. This differs from intentionally changing facts - the subject honestly believes the information he is presenting to be correct. Schacter (2001) defined seven different ways in which memory may be distorted:

  1. Transience - Memory fades quickly. We consider certain facts to be salient and others the be incidental, and while the latter fade quickly from memory, we can even lose salient detail.
  2. Absent-mindedness - Recall of memory is imperfect when we are not focused on a given task. A person may wash their hair a second time in the shower, having forgotten they did so just moments ago, or may walk into a room and immediately forget the reason they entered.
  3. Blocking - A memory block occurs when we are unable to recall a detail that we believe ourselves to know. For example, being unable to recall the name of another person you previously met.
  4. Misattribution - The source of a piece of information is an incidental detail rather than a salient one, and as such we may remember what he heard or read, but not be able to identify the person who said it.
  5. Suggestibility - People may indicate they remember something that did not actually happen if someone else suggests it. One example was the experiment in which a research asked Dutch moviegoers about a scene that was not in a film they recently watched, and surprisingly many indicate they remember it and confabulate details.
  6. Bias. People have a set expectation of the way things ought to be, and their memory suggests that this is the way that things were, even when they actually were not.
  7. Consistency. People likewise develop expectations of consistency in themselves and others, and will report that something happened when it did not. A person who cannot remember where they put their keys considers that they always put them in a certain place, even though they put them somewhere else the last time.

One of the more serious consequences of memory is the problem of eyewitness testimony - studies of false convictions point to incorrect identification as "the single largest factor" in false conviction, which has sparked research in the accuracy of memory in legal testimony.

For example, one study (Loftus 1978) showed subjects a video of a car striking a pedestrian at an intersection where there was a stop sign. Some were asked if another car passed it at the yield sign (purposefully changing that detail), and when later asked to describe the scene, participants indicated there was a yield sign at the intersection. In other studies, it has been found that eyewitness testimonial was considered credible by jurors even when the accounts given by witnesses were plainly wrong and even inconsistent to one another.

The author gives additional detail about lineups and eyewitness identification errors, the increased level of error in testimony by children, and other factors. Considering the findings of the various studies into the problems, it is roughly estimated that, between wrong identification of a suspect and inaccurate testimony, there may be as many as 10,000 wrongful convictions every year.

Next, the author considers the phenomenon of repressed memories - in which a trauma, often in childhood, is buried within the mind, allegedly to protect the mind from distress, and that hypnotism and other techniques can be used to cause the subject to recall them.

While the notion of repressed memories was fashionable for a time, and periodically resurfaces in pop-psychology, many psychologists doubt the veracity of the phenomenon. The "revived" memories may have been implanted, or confabulated in response to questions. There is much more compelling evidence of the ability to create a false memory than to revive an actual one.

The ability to create suggestive memory has been tested (Deese 1959) by giving subjects a list of 15 words related to a concept - such as words related to sleeping other than the word "sleep" itself - and subjects will report a memory of the target word having been spoken. This experiment has been replicated multiple times and the target evoked by various methods.

Context Effects

Studies of memory repeatedly demonstrate that sensory information is contextualized into schemas for storage, and that the process of schematization may distort and even fabricate memory in an attempt to accommodate context.

Additional studies (Chase 1973, French 1989) also support the notion that the greater a subject's knowledge and expertise, the more complex the schemas become, enabling them to better process and organize new information and to retrieve highly intricate data schema with greater accuracy.

Another factor that increases confidence in the accuracy of memory is its vividness: the intensity of sensory data and the agreement of data from different senses to create a "strong" memory that has richness in detail.

A particularly interesting form of memory is termed "flashbulb" memory, which consists of the memory of an event that is so strong and detailed that the subject can describe it in nearly as much detail as a scene in present perception. It is not uncommon for subjects who have received traumatic news to have a strong memory of where they were when the news was received: the Kennedy assassination, the collapse of the World Trade Center, and even the Challenger explosion are events for which people have intense and detailed memory of where they were when the news was received.

Various theories exist as to why this should be so: from the intensity of emotion experienced in the moment to the effects of rehearsal (people often retell the details of where they were to others, or contemplate it on their own). It's also been found that these flashbulb memories are likely to contain as much confabulation rather than fact as any other recollection.

It is further theorized (Baddeley 1989, Bower 1983) that somatic states are also associated to memory - in that we experience stimuli when we are in a given mood or state of mind, and are more likely to recall it in greater detail when we are in the same state.

There is also some evidence (Eich 1980 and 1995)that the effects of drugs and alcohol have a similar mnemonic enhancement - that we can recall incidents experienced while drunk in better detail when we ware drunk again - though the weakness of the evidence is attributed to the effect of the substance in question.

Memory also affects mood - when we recall a "happy memory" or even neutral information that was committed to memory in a state of happiness, we also revive the emotion (Matilin 1979). It is reckoned that this may also be a factor with negative moods, such that depression becomes self-reinforcing when innocuous details evoke negative mood. This does lend some scientific evidence to the folk-wisdom advice to "think happy thoughts" to improve your emotional state.

There is even evidence of contextual memory in infants, in that the way in which they interact with mobiles in their crib is specific to that environment - shown the same mobile in a different environment, they change the way they interact with it, but resume their previous interaction choices when returned to the crib.

It's also noted that recollection of words is markedly better in tests where words are paired, as opposed to tests in which words are memorized individually and without association. Further, that the ability to recall words by their key-pair is more accurate than the ability to recognize the same value-words in a list that contains distractors.

Memory Development

Considering that the phenomenon is different in children than it is in adults, and that developmental differences cause memory to function differently among individuals, the author considers some theories about the way in which the faculty of memory develops.

It is generally observed that few people have memories of early childhood, and therefore theorized that the human brain is not very efficient at encoding memory before the age of five. It is also observed that older children have greater facility tor remembering things than younger, and that the ability to remember decreases later in life.

Other research suggests that while the ability to create episodic memory and to place new information within a context is rather poor at first, young children have a significant ability to create semantic memory: to remember things, but without a context.

Of particular interest is the seeming inability of children to distinguish between memory and perception when describing things. In one experiment (Flavell 1983), children were given the opportunity to react with an object, then the object was placed behind a pane of glass that distorted the shape and color of the item. When asked to describe to object, their responses showed a random assemblage of the remembered qualities of the object and its present appearance through the distorting pane. The author asserts that "many scholars" have observed that children seem to blend memory, perception, and imagination in describing reality.

Another experiment (Appel 1972) demonstrates that rehearsal has little benefit for children of certain ages. Tests were done on toddlers, first-graders, and fifth-graders that involved asking them to memorize pictures of fifteen objects and remember their names. The older children performed better, but almost none of the younger ones showed any benefit from rehearsal. While some of them seemed to know about memory strategies, most did not use them, and few of those who tried them showed any benefit.

Further experiments with mentally retarded children (Brown 1973) show the same effects regardless of the age of subjects - that is, they did not apply any metamemory techniques to remember, did not remember, and did not show the ability to transfer knowledge from one task or situation to another.

Additional studies (Kearins 1981 and Rogoff 1986) demonstrate that memory strategies are learned by experiments that contrasted the memory skills and related practices among children in western cultures (where formal education includes teaching metamemory techniques) against those raised in undeveloped and even primitive societies, naturally finding that those who had been taught how to remember things applied the techniques and, as a result, had more accurate memory skills.

Other experimentation suggests that cognitive monitoring, rather than memory in itself, may be the difference. That is, children trained to remember things have the ability to distinguish what is significant, particularly identifying things that are known from things that are not known, and give greater attention to the unknown data as a means of committing it to memory. In other words, it may not be that there are differences in the ability to make memories, merely in the difference in the ability to perceive things about which memories may be made.

It is reckoned that, in addition to practicing metamemory techniques, older children and adults also have the ability to remember more because they have developed contexts and schemas for information - as earlier mentioned, the more educated and experienced a person becomes, the greater their facility to absorb and retain information in increasingly complex schemas. In essence, a more sophisticated mind observers, recalls, and learns more.