Cognitive Psychology 2341 Chapter 5: Memory Structures

Sensory Memory: The Echo

There is also a sensory memory for auditory material, which Neisser (1967) called the *echo*.

Daneman and Carpenter (1980): Working-Memory Span Task

A person is given a set of sentences to read (usually aloud), but at the same time is asked to remember the last word in each sentence for later recall. After reading the sentences aloud, the person is cued to recall the last word in each sentence. The number or sentences a participant can processes and reliably recall words from is said to be a measure of their span. This measure has been shown to correlate significantly with other cognitive measures, such as reading comprehension and other complex cognitive tasks, and vocabulary in children. Working memory capacity has also been linked to general fluid intelligence, as well as cognitive decline associated with normal aging.

Encoding

Acquiring information; it occurs when information is first translated into a form that other cognitive processes can use. It is held in *storage* in one form or another for later retrieval. We say that *forgetting* occurs when we cannot retrieve information.

Findings from Frontal-Lobe Brain-Damaged People

Findings from other brain damaged people have implicated areas in the frontal lobe as having to do with working memory. Damage to this brain region is often reported in disrupt attention, planning, and problem solving. These problems may arise not because attention and planning are located in the frontal lobe but rather because the frontal lobe inhibit activity in the posterior part of the brain. People with frontal-lobe damage seem to be more distractible and less able to ignore irrelevant stimuli.

Episodic Memory

Holds memories of specific events in which you yourself somehow participated.

LTM: Retention Duration

How long can information be stored in LTM? Although most laboratory experiments test recall after several hours or days, evidence is abundant that at least some information can last for decades or even a lifetime.

Short-Term Memory: Capacity

If you are going to store information for only a short period of time, how much room do you have in which to do so? In other words, how much information can you remember for only a brief period of time?

Keppel and Underwood (1962): Proactive Interference

Other evidence also supports the view that interference, not decay, accounts for forgetting in STM. Keppel and Underwood (1962), for instance, found that forgetting in the Brown-Peterson task doesn't happen until after a few trials. They suggested that, over time, *proactive interference* builds up. This term refers to the fact that material learned first can disrupt retention of subsequently learned material. They showed that even one trial's worth of practice recalling a three-letter trigram was enough to hurt subsequent memory for other trigrams.

State-Dependent Learning

Other studies have demonstrated similar effects, called *state-dependent learning*, with pharmacological states: Material learned while someone is chemically intoxicated is usually recalled better when the person recreates the state. The finding of interest was the participants who learned material while in chemically altered state showed significantly better recall if they were again chemically intoxicated at the time of the recall. Later studies suggest that *state-dependent memory* effect, like context effect, is found only with recall and not with recognition tasks.

STM vs. LTM Distinguishing Characteristics

Psychologists who make the distinction believe there are a number of such characteristics, including how much information can be stored (capacity), the from in which the information is stored (coding), the ways in which information is retained or forgotten, and the ways in which information is retrieved.

Paired Associates Learning to study Proactive Interference (PI)

Researchers have used this task to study interference in two ways. The first is through *proactive interference (PI)*. The term PI refers to the fact that *previous learning can make retention of subsequent learning more difficult*. Thus if a group of participants learns a list of varied associates and then learns a second list with the same set of first terms but new second ones, recalling information from the second list is much harder. Earlier learned information/material interferes with subsequent material.

Roger and Guynn (1966): Encoding Specificity Hypothesis

Roger and Guynn (1966) summarized the encoding specific hypothesis slightly differently: "A retrieval cue will be effective if and only if it reinstates the original encoding of the to-be-remembered event. The weak associate can serve as an excellent retrieval cue, but then the strong associate is completely ineffective."

Tulving: Semantic Memory

Semantic memory, in contrast, is thought to store general information about language and world knowledge. You can recall arithmetic facts, historical dates, or the past tense forms of various verbs, you are calling on semantic memory.

Wickens, Born, and Allen (1963): Release From Proactive Interference

Sickens, Born, and Allen (1963) extended the idea one more step. They reasoned as follows: If STM, like LTM, is subjected to proactive interference, then STM (like LTM) should also be subject to a related phenomenon, *release from proactive interference*. In other words, if you learn a number of pieces of similar information, after a while any new learning becomes more difficult because the old learning interferes with the retention of new (because of proactive interference). The greater the similarity among the pieces of information, the greater the interference. This implies that if a new and very distinct piece of information were presented, the degree of interference would be sharply reduced.

LTM: Retrieval of Information; Mnemonics

Techniques to improve memory, many of them having to do with visual imagery.

Sensory Memory: The Icon

The "after-image" from an experience, is a mental experience believed to persist in our sensory memory.

Memory Structures

The ability of humans to revisit and reflect on events in the past is often thought of as one of the most basic, and important cognitive process humans possess.

Can we explain the Brown-Peterson Task Results in Terms of Interference?

Think once again about the counting task. Notice that it supposedly has very little purpose other than to distract the participant from rehearsing the trigram. Yet, maybe the counting task does more than prevent participants from rehearsing; it may actually interfere with their STM storage of the trigram. As participants count aloud, they compute and announce the values. As they compute and announce the values, they put them into STM. Thus, the counted values may actually be displacing the original information.

George Sperling (1960) Partial-Report Technique Experimental Results

Using this methods, Sperling showed that regardless of which tone sounded, participants reports indicated that they had roughly 9 of the 12 letters available in sensory memory. This was true as long as the tone was presented immediately after the display. If the tone cure was delayed, however, by 1 second, their level of recall dropped to 4 of the 12 letters, as in the whole report method. The data suggests that the visual store can hold about nine items, but that it held this information only briefly, as the advantage of the partial-over whole-report method was reduced when the delay stimuli and the cue tone was 0.3 seconds. Neisser (1967a) called this visual brief memory the *icon*.

Bousfield (1953): Categorization Experiment

Early work documents this principle: Bousfield (1953) presented participants with a list of 60 words. The words came from four categories (animals, names, professions, and vegetables), but were presented in scrambled order. Nevertheless, participants tended to recall the words in clusters. Memory improved depending on the extent to which organization, or semantic clustering, of the to-be-remembered information was used. It turns out that even if the material doesn't have apparent organization, asking people to organize it into their own subjective categories improves recall. Some people, senior citizens and young adults who've suffered a head injury to the front part of the brain, fail to make use of this principle and do not group similar items together during recall, and this partly explains their poor memory.

LTM: Capacity

A complete list of all information you have at one time or another put into LTM would be very long. This intuition has led psychologists to estimate that the capacity of LTM is virtually unlimited.

Saul Sternberg's (1966, 1969) Experimental Review by Hunt (1978)

A review by Hunt (1978) found that people of all sorts showed results consistent with the idea that retrieval from STM uses serial, exhaustive search, although search rate changes with the group, being faster for people with exceptional memories and slower for senior citizens.

LTM: Retrieval of Information; Principles of Retrieval Used to Aid Recall; *Encoding Specificity*

A second principle of retrieval, discovered by Thomas and Tulving (1970), is called *encoding specificity*. The idea is that memory is improved when information available at encoding is also available at retrieval. The principle provides a general theoretical framework for understanding how contextual, or background, information affects memory. At the same time as recall, it is a great advantage to have the same context information available, as aspects of context can function as cues and retrieval aids.

Sensory Memory

Sensory memory is closely connected to what we call perception because it refers to the initial brief storage of sensory information. Many cognitive psychologists hypothesize that separate sensory memories exist for each sensory modality. In other words, they believe there is a visual sensory memory, an auditory sensory memory, and olfactory (smell) sensory memory, a gustatory (taste) sensory memory, and a tactile (touch) sensory memory. The overwhelming bulk of research on sensory memories to date have focused on the first two types of sensory memory called the *icon* and the *echo*, respectively.

Is it decay or interference?

It may be that the question "is it decay or interference" is badly posed, because it rules out the possibility that both may be involved. That is, maybe STM loses information by more than one mechanism. Baddeley (1990) argues that some (although very little) trace decay does occur in STM along with interference. Altman and Gray (2002) propose that decay does occur and in fact is essential to avoid catastrophic proactive interference. These authors believed that when information must be updated frequently in memory, its current value decays to prevent interference with later values.

Baddeley (1996): STM Coding

Later, work by Baddeley confirmed this effect even when the stimuli were words rather than letters: Similar sounding words make for poor immediate recall, although similar-meaning words don't, and the reverse is true for delayed recall. Although an acoustic code is not only used in STM, researchers have regarded it as the dominant code used, at least by hearing adults and older children.

Short-Term Memory (STM): Free Recall Experiment

Many empirical findings seem to support the idea of different memory systems. One well-known finding comes from free-recall experiments, in which people are given a list of words to remember, and are asked to recall the words in any order. Next, the experimenter, using data from all the participants, computes the portability of recall of each word as a function of the words serial position in the original list. In the serial position curve graphical representation, the two ends of the curve are higher than the middle, indicating that more people recall more words at either the beginning or the end of the list than they do words in the middle. This is known as the *serial position effect*. The improved recall of the words and the beginning of the list is called the *primacy effect*; that at the end of the list, the *recency effect*.

Researchers and the Suffix Effect

Researchers think the recall cue, called the *suffix*, functions as an auditory "mask" of sorts, because when the suffix is simply a beep or tone, or a visual stimulus, there is usually not much effect. Nor is there any effect if the items on the list are presented visually-say, on a computer screen. Finally, the more auditory stimulus there is between the suffix and the items on the list, the greater the suffix effect.

Saul Sternberg (1966, 1969) Experimental Task

Saul Sternberg (1966, 1969) experimental task was the following: First, participants were presented with a set of seven or fewer letters. These were to be encoded and held in STM and hence could be called the *memory set*. After the participant had the set in memory, he indicated readiness for an upcoming trial. A single letter, called the *probe*, was presented, and the participants task was to decide, as quickly as possible, whether the probe was in the memory set.

Short-Term Memory: Retrieval of Information; Saul Sternberg (1966, 1969)

Saul Sternberg (1966, 1969), in a series of experiments found some surprising things about how we retrieve information from STM. His first question was whether or not we search for information held in STM in a parallel or serial manner.

Neurological Studies of Memory

Schacter (1996) offered a number of case studies of people suffering from different kinds of amnesia that support the episodic/semantic distinction. Schacter also described neuropsychological cause studies of people with deficits. These studies provided clinical neuropsychological evidence supporting the idea that episodic memory and semantic memory operate independently. That is, the existence of people in whom one type of memory seems seriously impaired while the other appears spared gives concrete evidence for the existence of two separate systems of memory. Studies with patients with amnesia most often show a deficit in episodic but preserved semantic memory.

Retrograde Amnesia

Amnesia for for old events.

Working Memory: Baddeley and Hitch (1974)

Baddeley and Hitch (1974) performed a series of experiments to test this model. The general design was to have participants temporarily store a number of digits while simultaneously performing another task, such a reasoning or language comprehension, that was thought to require resources from STS-specifically, the control processes. Because researchers think STM has the capacity of about 7 items, plus or minus two, the six-digit memory load should have essentially stopped any other cognitive activity. Baddeley and Hitch (1974) therefore argued for the existence of what they call *working memory (WM)*. They see WM as consisting of a limited-capacity workspace that can be divided between storage and control processing.

Central Executive

Baddeley conceived of WM as consisting as three components. The first was the *central executive*. This component directs the flow of information, choosing which information will be operated on, when, and how. Researchers assume it has a limited amount of resources and capacity to carry out its tasks. The central executive is thought to function more as an attentional system than a memory store, meaning that rather than dealing with the storage and retrieval of information, it deals with parcelling out resources needed for cognitive tasks. It is thought to coordinate incoming information from the current environment with the retrieval of information from the past, enabling people to use this information to select options or strategies. Baddeley equated this coordination with conscious awareness.

LTM: Retention Duration; Harry Bahrick (1983, 1984) Experimental Findings

Bahrick (1984) interpreted the findings as follows: "Large portions of the originally acquired information remain accessible for over 50 years in spite of the fact that the information is not used or rehearsed. This portion of the information in a permastore state is a function of the level of original training, the grades received in Spanish courses, and the method of testing (recall vs. recognition), but it appears to be unaffected by ordinary conditions of interference.

R. Conrad (1964): STM Coding Experiment

A study by R. Conrad (1964) addressed this question. He presented participants with lists of consonants for later recall. Although the letters were presented visually, participants were likely to make errors that were similar in *sound* to the original stimuli. So, if P had been presented, and participants later miscalled this stimulus, they were much more likely to report a letter that sounded like P (for example, G or C) than to report a letter that looked like P (such as F). Remember, the original presentation was visual, but participants apparently were confused by the sound. Participants were apparently forming a mental representation of the stimuli that involved the acoustic rather than visual properties.

Sensory Memory: Modality Specific

Although research continues to refine our understanding of both the icon and the echo, sensory memory can currently be best described by a number of properties. First, sensory memories are *modality specific*: the visual sensory memory contains visual information; the auditory sensory memory, auditory information; and so forth. Second, sensory memory capacities appear relatively larger for visual than auditory sensory memory, but the length of time information can be stored is longer in the auditory than the visual store. Third, the information that can be stored appears relatively unprocessed, meaning that most of it has to do with physical aspects of the stimuli rather than the meaningful ones. Sensory memory is useful in the real world, as it guarantees a minimum of time during which information presented to us (that we pay attention to) is available for processing. In other words, by this argument sensory memory plays an important role in everyday workings of normal memory: It ensures that we will be able to "re-inspect" incoming data, if not with our actual eyes and ears, then with the minds eye and minds ear.

Anterograde Amnesia

Amnesia for new events.

Anderson and Neely (1996)

Anderson and Neely (1996) speculated that forgetting may not be so much a shortcoming of memory as a side effect of our ability to direct memory. In particular, they wonder whether it is sometimes beneficial to be able to forget voluntarily. Laboratory work that they reviewed suggests that when people lose information through directed (voluntary or intentional) forgetting, they experience such less proactive interference. Forgetting then, can be useful.

Saul Sternberg (1966, 1969): Exhaustive Search

Another kind of serial search is an *exhaustive search*, meaning that even if a match is found, you continue looking through every other item in the set. With this kind of search, it takes just as long for successful as unsuccessful searches.

Golden and Baddeley (1975): Context Effect

Apparently, even information unrelated to the material, such as environmental stimuli present at the time of encoding, can be a good retrieval cue. Results from Golden and Baddeley's (1975) experiment showed that recall was best when the environment was the same as the learning environment. This finding, that recall is best performed in the original environment, is called a *context effect*. Interestingly, researchers found later that recognition and recall work differently. In particular, this finding suggests that recognition and recall work differently. In particular, this finding suggests that physical context affects recall but not recognition. Presumably, in the former task, the participant must do more work to generate his or her own retrieval cues, which may include certain features of the learning environment, whereas on recognition tasks the test itself supplies some retrieval cues (in the form of the question and the possible answers)

George Sperling (1960) Partial-Report Technique

Armed with this information, Sperling (1960) devised a method of more accurately measuring the available content in his participants sensory store. He used the logic of a multiple-choice test in which only a subset of the total material was tested, and the percentage correct is taken as a measure of the students knowledge of the entire material. He created what has become known as the *partial-report technique*. Sperling informed participants that they would have to recall only a single row of the display, but they wouldn't know which row until after the display was shown. An auditory tone was cued to participants as to the row they were to report. After seeing the display, participants were presented with either a low, medium or high pitched tone. A low pitch indicated that they were to report only the letters in the bottom of the display; high pitch for top row; middle pitch for middle row.

Saul Sternberg (1966, 1969) Experimental Task Results

As counterintuitive as it sounds, Sternberg's results argue for serial exhaustive search as the way we retrieve information from STM. His explanation is that the search process itself may be so rapid and have such momentum it is hard to stop once it starts. From a processing point of view, it may be more efficient to just let the search process finish and then make one decision at the end, instead of making several decisions, one after each item in the memory set.

LTM: Forgetting; Paired Associates Learning

As with STM, many psychologists believe that interference, not decay, accounts for forgetting from LTM. They believe that material of which can't be retrieved successfully from LTM is there but buried or in some other way unavailable. Much of the literature on interference has used a task called *paired-associates learning*. Participants hear lists of pairs of words such as flag-spoon and drawer-switch. After one or more presentations of list, the experimenter then presents participants with the first word in each pair and the participant is asked to recall the word original paired with it.

Modal Model of Memory

Assumes that information is received, processed, and stored differently for each kind of memory. Unattended information presented very quickly is stored only briefly in *sensory memory*. Attended information is held in *short-term memory (STM)* for periods of up to 20 or 30 seconds. (Synonyms for STM include *primary memory* and *short term storage (STS)*.) Information needed for longer periods of time is transferred to *long-term memory (LTM)*, sometimes called *secondary memory* or *long-term storage (LTS)*.

Tulving and Thomson (1973) and Encoding Specificity Experimental Results

At recall, memory performance for participants in the control condition (no cues) was aided if highly related cues were presented, even if these cues hadn't been seen in the learning phase of the experiment. In contrast, as you might expect, the not very related cues were not very effective in promoting recall in the control group. However, the results were very different for participants who had seen the cues during the learning phase. For these participants, the not very related cues were in fact very effective in aiding recall, even better than highly related cues that had not been presented during the learning phase. They interpreted these results to mean that even a weakly related word can act as a potent retrieval cue if it is presented at the time of encoding.

George Miller (1956): Chunking

Chunking involves taking individual units in a list of random digits and chunking them into larger units. If you notice that the 12 letters are really four organized sets, you'll be more likely to remember the entire string. In recognizing that the three sets really "go together" and in forming them into a single unit, you are said to be chunking them. Chunking depends on knowledge. Miller regarded the process of forming chunks (he called it *recoding*) as a fundamental process of memory-a very powerful means of increasing the amount of information we can process at any given time, and one we use constantly in our daily lives. The process of chunking can be seen as an important strategy in overcoming the severe limitation of having only seven or so slots to temporarily store information.

Crowder (1976): The suffix effect

Crowder (1976), reviewing the literature on echoic memory, proposed that echoic memory has a larger capacity than iconic memory. A demonstration called the *suffix effect* also reveals something about the nature of echoic memory. Imagine you are a research participant in a memory experiment, and a list of random digits, letters, to the like is being presented to you. If the list is presented to you auditorily (as opposed to visually), and if there is an auditory recall cue such as a spoken word or specific item, recall of the last few items on the list is seriously hindered.

Types of Memory: History and Analogies

Fascination with what memory is and how it works has a long tradition is philosophy, predating any psychological investigation. Neath and Surprenant (2003) noted that the Greek Philosopher Plato wrote about memory, comparing it both with an aviary and to a wax tablet on which impressions are made. Throughout the Middle Ages and Renaissance, other analogies were made between memory and a cave, an empty cabinet, and a body in need of exercise. In the 1950s, memory was compared to a telephone system, and later it was compared to a computer. One theoretical approach to studying memory, which dominated cognitive psychology through the 1960s and 1970s, distinguishes among kinds of memory according to the length of time information is stored. More recently, cognitive neuroscience is attempting to establish the brain basis of memory.

George Sperling (1960) Experiment

George Sperling first conducted experiments, now considered classic, to investigate the properties of a visual, iconic, sensory memory. He presented participants with displays containing sets of letters and asked them to recall all the letters they saw. The displays were presented briefly, for only 50 milliseconds. Sperling found that, on average, people could report only 4 or 5 of the 12 letters presented. Extending the display time, even to 500 milliseconds, did not improve performance. The problem wasn't perceptual; 500 milliseconds, or half a second, is plenty of time to perceive something about all the letters. Sperling spoke to his participants afterward and mainly said they had seen all the stimuli quite clearly, but once they started reporting what they saw, forgot the rest. Put another way, even as participants were recalling the display, the information was fading from wherever it was being stored. This implies that information lasts only briefly in this memory system.

LTM: Retention Duration; Harry Bahrick (1983, 1984)

Harry Barrack (1983, 1984) has studied peoples memory for material learned to varying degrees at varying times, including memory for the faces of college classmates 20 or 30 or even 50 years after graduation. In one study, Bahrick (1984) tested 733 adults who had taken or were taking a high school or university course in Spanish. The participants who were not currently enrolled in Spanish had not studied Spanish for periods ranging from 1 to 50 years. They also varied in their original degree of learning of Spanish. Bahrick plotted "forgetting curves" for the different aspects of knowledge of Spanish. Although forgetting differed slightly as a function of the measure, the pattern of results was remarkably consistent. For the first three years after completing Spanish study, participants recall declined. But for the next three decades or so, the forgetting curve was flat, suggesting no further loss of information. Retention showed a final decline after about 30-35 years.

LTM: Forgetting; Hermann Ebbinghaus Experimental Results

He assumed that the more forgetting, the more effort it would take to relearn a list; conversely, the less forgetting, the less effort to relearn. The curve suggests that forgetting is not a simple linear function of time. Instead *forgetting is rapid at first and then levels off*. Notice how well his laboratory finding anticipates the real-world memory studies of Bahrick reported earlier.

LTM: Forgetting; Hermann Ebbinghaus

He pioneered the empirical study of memory under controlled conditions. His master work reported on 19 of his studies using himself as a subject. He created stimuli he thought were carefully controlled and free from any contamination from prior learning; he called the *nonsense syllables*. He carefully and precisely presented, at a controlled rate, hundreds of lists of these syllables to himself. Day after day, he memorized, tested himself, recorded the results, and prepared new stimuli. Altogether, he spent about 830 hours memorizing 85,000 syllables in 6,600 lists The primary question he asked had to do with the number of repetitions needed for perfect recall, the nature of forgetting, the effects of fatigue on learning, and the effects of widely spaced versus closely spaced practice.

Semantic Memory

Holds information that has entered your general knowledge base: you can recall on parts of that base, but the information recall is generic-it doesn't have much to do wit your personal experience.

Anderson and Neely (1996): Retrieval Cues

How exactly does interference work? Anderson and Neely (1996) presented several possibilities. They started with the assumption that *retrieval cues* point to, and leads to the recovery of, a target memory. However, when that retrieval cue becomes associated with another target, during retrieval that second target competes with the first. The more possible targets associated with the cue, the less the chances of finding any particular one of them. A retrieval cue can become associated with different targets as well, or other cues, leading to even more complexity, and making it that much harder to traverse a path for the cue to get to the correct target.

Interference

However, other cognitive psychologists soon began to challenge this decay explanation of forgetting. The proposed a different mechanism, called *interference*, that worked as follows: Some information can "displace" other information, making the former hard to retrieve. You can think of the interference explanation as being akin to finding a piece of paper on your professors desk. At the start of each academic term, her desk is relatively free of clutter. Any price of paper placed on the desktop is trivially easy to find. However, as the term goes on and time grows short, your prof tends to allow all kinds of memos, papers, journals, and the like to accumulate. Papers placed on the desk at th beginning of the term become buried; they're there, all right, but can be very difficult to find at any given moment. The late-arriving papers have "displaced" the early papers.

Saul Sternberg (1966, 1969): Parallel Search

If you compare Titanic simultaneously to all the titles on the list of movies, you are performing a *parallel search*. Essentially, no matter what the number of titles is, you examine them at the same time, and it takes you no more time to compare Titanic to 1 title than to 10 titles.

Long-Term Memory

In the modal model, LTM is thought to differ from STM in many ways. LTM is described as a place for storing large amounts of information for indefinite periods of time. Note the contrast here with the modal description of STM as holding a very limited amount of information ( seven plus or minus 2, pieces of unrelated information) for a very short period of time (seconds, or at most a few minutes). In other words, LTM is commonly thought to be a sort of mental scrap book: the material you have cognitively collected in your lifetime is stored there in some form.

LTM: Coding

Many studies of recall from LTM report a common finding: Errors made while recalling information from LTM are likely to be *semantic confusions*. That is, words or phrases that mean things similar to the words or phrases actually presented are likely to be recalled in error, if errors are made. Baddeley demonstrated this phenomenon experimentally. He presented participants with lists of words that sounded similar or that were matched to the first list but did not sound alike. Others also saw a list of words with similar meanings (semantically similar words) and another list of control words that were matched to the third list but did not share meaning. Recall was tested after a 20 minute interval, during which participants worked on another task, to prevent rehearsal and to ensure the material would be drawn from LTM rather than STM. The results showed that acoustic similarity produced little effect on performance but that the list of semantically similar words was harder to learn. Baddeley (1976) reviewing this and other work, concluded that the following generalization, although not absolute, is roughly true: *Acoustic similarity affect STM; Semantically similar information affects LTM*.

Tulving: Memory

Memory is a biological abstraction. There is no place in the brain that one could point and say, Here is memory. There is no single activity, or class of activities, of the organism that could be identified with the concept that the term denotes. There is no known molecular change that corresponds to memory, no behavioural response of a living organism that is memory. Yet, the term memory encompasses all these changes and activities.

Moray, Bates, and Barnett (1965) Demonstration of Echo Experiment

Moray, Bates, and Barnett (1965) offered a clever demonstration of the echo. Participants were given a "four eared" listening task, similar to a dichotic listening task. They heard, simultaneously over headphones, four channels of incoming information, each apparently coming from a different location, consisting of a string of random letters. The four channels were created by stereophonic mixing. In one condition, similar to Sperling's (1960) whole-report condition, participants were asked to report all the letters they had heard. In another condition, each participant held a board with four lights on it, each light corresponding to one of the channels, cueing the participant to report only the letters from a particular channel. As did Sperling, Moray and colleagues found that participants giving partial reports could report proportionately more letters, This suggests that the echo, like the icon, stores information only briefly.

Neath and Surprenant (2003): Sensory Memory Research

Neath and Surprenant (2003) argued that sensory memory research has a very practical use outside of the laboratory: Having directory assistance operators say "Have a nice day" after giving a phone number should (and apparently does) disrupt recall for the phone number because their pleasant sign-off acts as a suffix!

Long-Term Potentiation

Neil Carlson described some basic physiological mechanisms for learning new information. One mechanism is the *Hebb Rule*, which states that if a synapse between two neurons is repeatedly activated at about the same time the postsynaptic neuron fires, the structure or chemistry of the synapse changes. A more general and more complex mechanism calle *long-term potentiation*. In this process, neural circuits in the hippocampus that are subjected to repeated and intense electrical stimulation develop hippocampal cells become more sensitive to stimuli. This effect of enhanced response can last for weeks or even longer, suggesting that this could be a mechanism for long-term learning and retention. As you may suspect, disrupting the process of long-term potentiation also disrupts learning and remembering.

Averbach and Coriell (1961): Masking Phenomenon

Other researchers built on this work to define the properties of this type of memory. Averbach and Coriell (1961) showed that the icon can be erased by other stimuli presented immediately after the icon, a phenomenon known as *masking*. For instance, if the display with letters was followed by a display with circles, and if the participant was told to report which letters had been in the locations of the circles, the circles appeared to "erase" the memory of the letter originally shown. Other work investigated how many ways participants could be cued to give partial reports. Different investigators showed that such things as the colour or brightness of the letters could be used to cue partial reports. Interestingly, cueing partial reports by category or phonological sound is all but impossible. This suggests that information available in the icon is only visual, not auditory or related to type of stimulus.

What Account for these two (primacy and recency) effects?

Participants typically report subvocalizing to themselves when they first start the experiment. The participants repetition of items, or *rehearsal*, is thought to help items enter long-term storage. In fact, if the experimenter reads the list rapidly enough to prevent the participant from having enough time to rehearse, the primary effect disappears, although the recency effect stays intact. That the primacy and recency effects cab be independently affected suggests they reflect two kinds of memory. Those who endorse the idea of sensory memory believe that incoming information first passes through this rapidly decaying storage system. If attended to, the information next moves to the STM. To be held for longer than a minute or two, the information must be transferred again, this time to LTM.

Saul Sternberg (1966, 1969): Serial Search

Suppose, instead you use a *serial search*. In our movie titles example, this would mean comparing Titanic to the first movie title on the list, then to the second title on the list, and so on until you come to the last title. The comparisons are done one at a time. In this model, the longer the list is, the longer it should take to decide if Titanic matches a title on that list. Successful searches are indicated by the yes line; unsuccessful searches (where a target is not found) by the no line.

George Miller (1956): STM Capacity

Th integer plaguing Miller was 7 (plus or minus 2). Among other things, 7 (plus or minus 2, depending on the individual, the material, and other situational factors) seems to be the maximum number of independent units we can hold in STM. We call this the *capacity* of STM. Miller (1956) reviewed evidence demonstrating that if you are presented with a string of random digits, you will be able to recall them only if the string contains about seven or fewer digits. The same is true is you a presented with random strings of any kinds of units. Th only way the overcome this limitation is by *chunking*.

STM: Brown-Peterson Task

The Brown-Peterson task works as follows. Participants are presented with a three-consonant trigram, such as BKG. They are also given a number, such as 347, and asked to count backwards outlaid by threes, at the same rate of two counts per second, in time to a metronome. The purpose of the counting task is to prevent the participant from rehearsing the trigram. The length of time a participant must count varies. If asked to count backward for only 3 seconds, roughly 80% of participants can recall the trigram. If asked to count for another 18 seconds, this drops to about 7%. The authors of the study interpreted this finding as meaning that the *memory trace*-the encoded mental representation of the to-be-remembered information that is not rehearsed-*decays*, or breaks apart, within about 20 seconds.

Retrieval

The calling to mind of previously stored information

LTM: Retrieval of Information; Principles of Retrieval Used to Aid Recall; *Categorization*

The first is the principle of *categorization*. This states that material organized into categories or other units is more easily recalled than information with no apparent organization. This effect happens when organized material is initially presented in a random order.

Summary of STM Systems

The general picture that emerged in the 1096s and 1970s was the STM is a short-term, limited capacity storehouse where information is coded acoustically and maintained through rehearsal. Information can be retrieved from this storage using high speed, serial, exhaustive search. The nature of the information in STM, however, can help change the capacity and processing of stored information.

Episodic Buffer

The original model of working memory proposed by Baddelely and Hitch comprised three elements: the central executive and two "slave" systems, the phonological loop and the visuospatial pad. A fourth component was later added called the *episodic buffer*. This buffer is believed to be required when remembering specific events from the past. The buffer is needed to link information across domains (visual, spatial, verbal), and to allow one to sequence various events inc chronological order. Such a slave system was hypothesized because of the recent findings that she amnesic patients are perfectly able to remember entire stories that are told to them, as long as recall is assessed within a short period of hearing the story. Remembering details of a story, and the ordering of events, requires more storage capacity than can be held in the phonological loop; hence the proposal of the episodic buffer.

Paired Associates Learning to study Retroactive Interference

The other kind of interference is called *retroactive interference*. Some researchers have argued that interference plays a role in most, if not all, forgetting of material from the LTM storage system. Of course, it is impossible to rule out the idea that decay occurs, because it is impossible to design a task in which interference cannot occur.

The Recency Effect

The recency effect is thought to result from the participants using either sensory memory or STM. Participants often report they can still sort of hear the last few words, and they often report these first and quickly. If the experimenter prevent the participants from reporting the words right away, by having them first perform an unrelated task, the recency effect (but not the primacy effect) disappears.

Short-Term Memory: Coding

The term *coding* refers to the way in which information is mentally represented-that is, the form in which the information is held.

LTM: Thomas Landauer (1968)

Thomas Landauer (1968) has tried to provide a more quantitative answer to this question. He begins with two previous estimates. The first is that the human memory is equal to the number of synapses in the cerebral cortex of the brain. Recall: A synapse is the gap between two neurons, basic cells of the body, across which neurotransmitters pass chemical messages. The cerebral cortex has 10^(13) synapses, so some believe that the human memory can hold 10^(13) distinct bits of information. Another estimate is 10^(20) bits of information, the estimated number of neural impulses, or electrical messages, transmitted in the brain during a persons lifetime. Landauer argued that both these estimates are probably too high: Not every neural impulse or synaptic connection results in a memory . Through various different analyses, in which he tried to estimate the rate at which new information is learned and the rate at which information is forgotten to lost, he came to an estimate of about 1 billion bits of information for an adult in midlife (say, about age 35). Whatever the actual number of bits of information stored in LTM, not all that information is retrievable at any given moment. Indeed, there are some everyday examples of failures to retrieve information. The information is probably in your LTM somewhere, but you somehow can't access it.

John Anderson (1974): Fan Effect

To account for some of these results, psychologist John Anderson (1974) described a phenomenon known as the *fan effect*. His idea is that research participants study more facts about a particular concept, the time they need to retrieve a particular fact about that concept increases.

Tulving and Thomson (1973) and Encoding Specificity Experiment

Tulving and Thomson (1973) demonstrated the encoding specificity as follows. Participants saw lists of words, with the to-be-remembered words printed in capital letters. Some participants saw these target words paired with other words printed in small letters. Both groups were told that the words in the small letters were cues or hints. Cues were either highly related to the target or not very related. Participants in the control condition were presented with the target words, but no cues.

Tulving: Memory Systems

Tulving described episodic and semantic memory as *memory systems* that operate on different principles and hold onto different kinds of information. He pointed out a number of differences in the ways episodic memory and semantic memory seem to work. Organization of episodic memory is temporal; that is, one event will be recognized as having occurred before, after, or at the same time as another. Organization of semantic memory is arranged more on the basis of meanings and meaning relationships among different pieces of information. Memory processes are ultimately instantiated in the brain.

Tulving: Episodic Memory

Tulving proposed a classification of long-term memories in two kinds: episodic and semantic. Episodic memory is memory for information about ones own personal experiences. As Tulving put it, episodic memory enables people to travel back in time, as it were, in their personal past, and to become consciously aware of having witnessed or participated in events and happenings at earlier times. Episodic memory has also been described as containing memories that are temporally dated; the information has some sort of marker for when it was originally encountered.

Phonological Loop

Used to carry out subvocal rehearsal to maintain verbal material. Plays important role in tasks such as learning to read comprehending language, and acquiring vocabulary. Consists of two structures: The *short-term phonological buffer* (which holds onto verbal information for short periods of time, assumes rehearsal is not prevented), and a *subvocal rehearsal loop* used to compensate for the rapid decay of information is the phonological buffer. The idea here is that when a person initially encounters information, particularly verbal information, they translate it into some sort of auditory code and processes it through the phonological loop. Because the information from the phonological buffer decays rapidly, the person must sub vocally rehearse the information, and the faster the rehearsal process, the more information can be maintained. If the phonological buffer is "filled up"-say, by having a person repeat a syllable or count aloud-then less capacity from this system would be available to devote to other tasks.

Visuospatial Sketch Pad

Used to maintain visual material through visualization. Involves the creation and use of mental images.

Watkins and Watkins (1980): Echoic memory capacity

Watkins and Watkins (1980) provide evidence that echoes can last longer than icons, perhaps even as long as 20 seconds.

Saul Sternberg (1966, 1969): Self-Terminating Search

We can also ask if the search is self-terminating or exhaustive. A *self-terminating search* stops when a match is found. On average, then, successful searches take less time (because you don't continue to search after you ahem found the match) than unsuccessful searches (where you have to search through everything).

STM: Retention Duration

We regard STM as the storage of information for short periods of time. But how short is short? John Brown (1958) and Peterson and Peterson (1959), working independently, came to the same conclusion: If not rehearsed, information is lost in the STM in as little as 20 seconds. That length of time is called the *retention duration* of the memory.

Sickens, Born, and Allen (1963): Release From Proactive Interference Experiment

Wickens and colleagues (1963) demonstrated release from proactive interference in a clever experiment. They gave participants a series of either three-digit strings (such as 179) or three-letter strings (such as DKQ). There were 10 trials in all. Some participants received 10 trials of the same type. Others saw a switch in the stimuli partway through the 10 trials. Participants getting a switch performed almost as well immediately after the switch as they did on the first trial. Their memory is said to have been released, or freed, from the clutches of proactive interference!

Short-Term Memory (STM)

You use this kind of memory system when you look up a phone number, walk across a room to a telephone, and dial the number. STM only lasts for a short while. Cognitive psychologists typically regard STM as lasting for a minute or two, if rehearsal is not prevented; however, neuropsychologists sometimes consider information in STM as lasting for up to a day, which can lead to some confusion. When we talk about STM, we are talking about material stored for up to about a minute.