about memory

5/20/2018 About Memory

1/44

5Sensory Memory,

Short-Term Memory,and Working Memory

C H A P T E R

135

Some Questions We Will Consider

What Is Memory?

What Are the Purposes of Memory?

The Plan of This Chapter

Introduction to the Modal Model of Memory

What Is Sensory Memory?

The Sparklers Trail and the Projectors Shutter

Sperlings Experiment: Measuring the Visual IconHow Can We Distinguish Between Short-Term

and Long-Term Memory?

The Serial-Position Curve

Demonstration: Remembering a List

Differences in Coding

Demonstration: Reading a Passage

Neuropsychological Evidence for Differences Between

STM and LTM Demonstration: Digit Span

Test Yourself 5.1

What Are the Properties of Short-Term Memory?

What Is the Capacity of Short-Term Memory?

What Is the Duration of Short-Term Memory?

Demonstration: Remembering Three Letters

Problems With the Modal Model

Demonstration: Reading Text and Remembering

Numbers

Working Memory: The Modern Approach to

Short-Term Memory

The Three Components of Working Memory

Operation of the Phonological Loop

Demonstration: Phonological Similarity Effect

Demonstration: Word-Length Effect

Demonstration: Repeating TheThe Visuospatial Sketch Pad

Status of Research on Working Memory

Working Memory and the Brain

The Delayed-Response Task in Monkeys

Neurons That Hold Information

Brain Imaging in Humans

Test Yourself 5.2

Think About It

Key Terms

CogLab: Apparent Movement; Partial Report; Serial Position;

Memory Span; Brown-Peterson; Phonological Similarity;

Irrelevant Speech Effect; Modality Effect; Operation

Span; Position Error; Sternberg Search; Suffix Effect


2/44

Some Questions We Will Consider

Why can we remember a telephone number long enough to place a call, but then we

forget it almost immediately? Is there a way to increase the ability to remember things that have just happened?

Do we use the same memory system to remember things we have seen and things wehave heard?

n n n

Everything in life is memory, save for the thin edge of the present.

(Gazzaniga, 2000)

The thin edge of the present is what is happening right at this moment, but a momentfrom now the present will become the past, and some of the past will become stored inmemory. What you will read in this chapter and the two that follow supports the idea thateverything in life is memory and shows how our memory of the past not only providesa record of a lifetime of events we have experienced and knowledge we have learned, butcan also affect our experience of what is happening right at this moment.

WHAT IS MEMORY?The definition of memoryprovides the first indication of its importance in our lives:Memory is the processes involved in retaining, retrieving, and using information about stimuli,images, events, ideas, and skills after the original information is no longer present.The fact thatmemory retains information that is no longer present means that memory can serve as aform of mental time travel, enabling us to bring back many different things that havehappened in the past. We can use our memory time machine to go back just a mo-mentto the words you read at the beginning of this sentenceor many yearsto eventsas early as a birthday party in early childhood.

What Are the Purposes of Memory?

Memory is important not only for recalling events from the distant past, but also for deal-ing with day-to-day activities. When I asked students in my cognitive psychology class tomake a Top 10 list of what they use memory for, they came up with over 30 differentuses, most of them related to day-to-day activities. The top five items on their list, in-

volved remembering the following things.

1. material for exams2. their daily schedule

3. names

Chapter 5136


3/44

Sensory Memory, Short-Term Memory, and Working Memory137

4. phone numbers5. directions to places

What would your list include? As a student, remembering material for exams is prob-ably on your list, but it is likely that people from different walks of life, such as businessexecutives, construction workers, homemakers, nurses, or football players, would createlists that would differ from the ones created by college students in ways that reflect the de-mands of their particular lives. Remembering the material that will be on the next cogni-tive psychology exam might not make a football players list, but remembering the play-book might.

One reason I ask students to create their memory list is to get them to think about

how important memory is in their day-to-day lives. But the main reason is to make themaware of how many important functions they dontinclude on their lists, because they takethem for granted. A few of these things include labeling familiar objects (you know youare reading a book because of your past experience with books), having conversations(you need memory to keep track of the flow of a conversation), knowing what to do in arestaurant (you need to remember a sequence of events, starting with being seated andending with paying the check and leaving a tip), and finding your way to class.

The list of things that depend on memory is an extremely long one because just abouteverything we do depends on remembering what we have experienced in the past. Butperhaps the most powerful way to demonstrate the importance of memory is to consider

what happens to peoples lives when they lose their memory. Consider, for example, thecase of Clive Wearing (Annenberg, 2000).

Wearing was a highly respected musician and choral director in England who, in his40s, contracted viral encephalitis, which destroyed parts of his temporal lobe that are im-portant for forming new memories. Because of his brain damage, Wearing lives totally

within the most recent one or two minutesof his life. He remembers what just hap-pened and forgets everything else. Whenhe meets someone, and the person leavesthe room and returns three minutes later,

Wearing reacts as if he hadnt met the per-son earlier. Because of his inability to formnew memories, he constantly feels he has

just become conscious for the first time.This feeling is made poignantly clear

by Wearings diary, which contains hun-dreds of entries like I have woken up forthe first time and I am alive (Figure5.1). But Wearing has no memory of ever

writing anything except for the sentence hehas just written. When questioned about

Figure 5.1 Clive Wearings diary looked like this. Sometimes he wouldcross out previous entries, because he could only remember writing the

most recent entry.


4/44

previous entries, Wearing acknowledges that they are in his handwriting, but because hehas no memory for writing them, he denies that they are his. It is no wonder that he isconfused, and not surprising that he describes his life as being like death. His loss of

memory has robbed him of his ability to participate in life in any meaningful way, and heneeds to be constantly cared for by others.

The Plan of This ChapterThe goal of this chapter is to begin describing the basic principles of memory so we canunderstand both cases like Clive Wearings and also the basic principles behind normal

memory processes (Figure 5.2). We begin by describing a model of memory that is calledthe modal model. This is an information-processing model, which contains a number of

Chapter 5138

Working memory

and the brain

What are the stages of the modal

model?

Do STM and LTM have different

mechanisms?

What are the properties of STM?

Compare STM

and LTM mechanisms

Short-term

memory

Working

memory

The modal

model

How does working memory explain

short-term memory processes?

How does the brain hold information

in working memory?

Figure 5.2 Flow diagram for this chapter.


5/44

stages, beginning with ones that hold information for only a short time (sensory memoryand short-term memory) and ends with one that can hold information for extremely longperiods of time (long-term memory).

After introducing the modal model, we will consider evidence for the idea that theshort-term and long-term components of the modal model involve different mechanisms.

We then focus on the short-term component, first looking at properties of short-termmemory and then at research that has led to a more modern way of looking at the short-term stage of memory, which is called working memory. Finally, we will describe researchon where some of the mechanisms of working memory are located in the brain. Thischapter therefore focuses mainly on the short-term component of the memory system. In

Chapters 6 and 7 we will consider the longer-term components.

Introduction to the Modal Model of MemoryIn 1968 Richard Atkinson and Richard Shiffrin proposed a model of memory that includedstages with different durations (Figure 5.3). The model has been so influential that it iscalled the modal model of memory.The various stages of the model are called the struc-tural features of the model. There are three major structural features: (1) sensory mem-

ory, an initial stage that holds information for seconds or fractions of a second; (2) short-term memory (STM),which holds information for only 1530 seconds; and (3) long-termmemory (LTM),which can hold information for years or even decades.

Atkinson and Shiffrin also describe the memory system as including controlprocesses,which are active processes that can be controlled by the person and may differfrom one task to another. An example of a control process is rehearsalrepeating a stim-ulus over and over, as you might repeat a telephone number in order to hold it in yourmind after looking it up in the phone book. Other examples of control processes are (1)strategies you might use to help make a stimulus more memorable, such as relating thenumbers in a phone number to a familiar date in history, and (2) strategies of attentionthat help you selectively focus on other information you want to remember.


Figure 5.3 Flow diagram for Atkinson and Shiffrins (1968) model of memory. This model, which is described

in the text, is called the modal modelbecause of the huge influence it has had on memory research.

InputSensorymemory

Short-termmemory

Long-termmemory

Output

Rehearsal: A control process


6/44

To illustrate how the structural features and control processes operate, lets considerwhat happens as Rachel looks up the number for Mineos Pizza in the phone book (Figure5.4). When she first looks at the book, all of the information that enters her eyes is regis-

tered in sensory memory (Figure 5.4a). But Rachel focuses on the number for Mineosusing the control process of selective attention, so the number enters STM (Figure 5.4b)and Rachel uses the control process of rehearsal to keep it there (Figure 5.4c).

After Rachel has dialed the phone number, she may forget it because it has not beentransferred into long-term memory. However, she decides to memorize the number sonext time she wont have to look it up in the phone book. The process she uses to memo-rize the number, a control process we will discuss in Chapter 6, transfers the number into

long-term memory, where it is stored (Figure 5.4d). A few days later, when Rachels urgefor pizza returns, she remembers the number. This process of remembering informationthat is stored in long-term memory is called retrieval because the information must be re-trieved from LTM so it can reenter STM to be used (Figure 5.4e). Retrieval is anothercontrol process that we will describe in Chapter 6.

One thing that becomes apparent from our example is that the components of mem-ory do not act in isolation. Long-term memory is essential for storing information but be-fore we can become aware of this stored information, it must be moved into STM. STM

is where information resides as we are working with it, as Rachel was doing when she firstlooked up the phone number and when she later retrieved it from LTM.

We have described the broad outline of the model, but what about the details? As wedelve deeper into the modal model, we will see that each stage handles information dif-ferently, and that our ability to remember depends on how these stages work together. Webegin by describing sensory memory.

WHAT IS SENSORY MEMORY?Sensory memoryis the retention, for brief periods of time, of the effects of sensory stim-ulation. We can demonstrate this brief retention for the effects of visual stimulation withtwo familiar examples: the trail left by a moving sparkler and the experience of seeing afilm.

The Sparklers Trail and the Projectors ShutterIt is dark, sometime around the Fourth of July, and you place a match to the tip of asparkler. As sparks begin radiating from the hot spot at the tip, you sweep the sparklerthrough the air, and create a trail of light (Figure 5.5). Although it appears that this trail iscreated by light left by the sparkler as you wave it through the air, there is, in fact, no lightalong this trail. The lighted trail is a creation of your mind because everywhere thesparkler goes you retain a perception of its light for a fraction of a second. This retention

of the perception of light in your mind is called the persistence of vision.

Chapter 5140


7/44


Figure 5.4 What happens in different parts of Rachels memory as she is (a and b) looking up the phone number,(c) calling the pizza shop, and (d) memorizing the number.A few days later, (e) she retrieves the number from long-

term memory to order pizza again. Darkened parts of the modal model indicate which processes are activated foreach action that Rachel takes.

Sensory STM LTM

All info on page

enters sensorymemory.

Sensory STM LTM

Focus on 555-5100.

It enters STM.

Sensory LTM

Rehearse the number

to keep it in STM whilemaking the phone call.

Sensory STM LTM

Storage

Store number in LTM.

Sensory STM LTM

Retrieve number from LTM.

It goes back to STM and isremembered.

(a)

(b)

(c)

(d)

(e)

STM

Remember numberto make call

Remember numberto make call again

Retrieval

Rehearsal555-5100555-5100555-5100

555-5100

Rehearsing

Memorizing

Retrieval


8/44

Something similar happens while you are watching a film in a darkened movie theater.You may see actions moving smoothly across the screen, but what is actually projected isquite different. We can appreciate what is happening on the screen by considering the se-quence of events that occur as a film is projected. First, a single film frame is positioned infront of the projector lens, and when the projectors shutter opens, the image on the filmframe is projected onto the screen. The shutter then closes, so the film can move to thenext frame without causing a blurred image, and during that time, the screen is dark.

When the next frame has arrived in front of the lens, the shutter opens again, flashing thenext image onto the screen. This process is repeated rapidly, 24 times per second, so 24still images are flashed on the screen every second, with each image separated by a briefperiod of darkness (see Table 5.1).

ApparentMovement

Chapter 5142

Figure 5.5 (a) A sparkler can cause a trail of light when it is moved rapidly. (b) This trail occurs because theperception of the light is briefly held in the mind.

Perceptual

trail

B

azukiMuhamm

ed/ReutersNewsmediaInc./

CORBIS

Table 5.1 PERSISTENCE OFVISION INFILM

What Happens? What Is on the Screen? What Do You Perceive?

Film frame 1 is projected. Picture 1 Picture 1

Shutter closes and film moves Darkness Picture 1 (persistence of vision)to the next frame.

Shutter opens and film Picture 2 Picture 2*frame 2 is projected.

*Note that the images appear so rapidly (24 per second) that you dont see individual images, but see a moving image created by the

rapid sequence of images.


9/44

A person viewing the film sees the progression of still images as movement anddoesnt see the dark intervals between the images because the persistence of vision fills inthe darkness by retaining the image of the previous frame. If the period between the im-

ages is too long, the mind cant fill in the darkness completely, and you perceive a flicker-ing effect. This is what happened in the early movies when the projectors flashed imagesmore slowly, causing longer dark intervals. This is why these early films were called flick-ers, a term that remains today, when we talk about going to the flicks.

Sperlings Experiment: Measuring the Visual Icon

The persistence of vision effect that adds a trail to our perception of moving sparklers andfills in the dark spaces between frames in a film has been known since the early days ofpsychology (Boring, 1942). This lingering of the visual stimulus in our mind was studiedby Sperling (1960) in a famous experiment in which he flashed an array of letters, like theone in Figure 5.6a, on the screen for 50 milliseconds (50/1,000 sec) and asked his partici-pants to report as many of the letters as possible. This procedure was called the whole-report procedure because participants were told to base their report on the whole display.

Sperlings participants were able to report an average of only 4 or 5 of the 12 letters in

the display, but they often commented that they had seen all of the letters at first, but theletters had faded away as they were reporting them. What were the participants seeing be-fore the letters faded? To answer this question, Sperling used the procedure shown in Fig-ure 5.6b.

Sperling flashed an array of 12 letters as before, but immediately after they were ex-tinguished he sounded a tone that told the participant which row of letters to report. Ahigh-pitched tone indicated that the participant should report the letters in the top row, amedium-pitched tone signaled the middle row, and a low-pitched tone signaled the bot-tom row. Note that since the tones were presented afterthe letters were turned off, theparticipants attention was directed not to the actual letters, which were no longer pres-ent, but to whatever trace remained in the participants mind after the letters were turnedoff. This procedure was called the partial-report procedure because participants wereasked to direct their attention to just part of the display.

The results showed that no matter which row the participants were instructed to re-port, they remembered an average of about 3.3 of the 4 letters (82 percent) in that row.

Starting with the fact that his participants were able to remember 82 percent of he let-ters no matter which row they reported, Sperling was able to calculate how many letterswere available to participants from the entire 12-item display just after the display wasturned off. Assuming that 82 percent of the entire display was available, Sperlings cal-culation, 12 0.82 = 9.8, indicated that about 10 letters were available from the wholedisplay.

We have already noted that Sperlings participants could report only 4 or 5 letterswhen tested using the whole-report procedure because as they were reporting these 4 or

5 letters, the other letters had faded. To determine the time course of this fading, Sperling


Partial Report


10/44

Chapter 5144

(a) Whole report

X

A

C

M

F

D

X F

ZD

C

L

N

Z

T

B

P

X

A

C

M

F

D

L

N

Z

T

B

P

(b) Partial report

Tone immediate

Immediate tone

X

A

C

M

F

D

B

L

N

Z

T

B

P

(c) Partial reportTone delayed

DelayDelayed tone

Medium

Low

High

Medium

Low

High

X M

L T

Figure 5.6 Procedure for three of Sperlings (1960) experiments. (a) Whole report procedure: Person saw all12 letters at once for 50 msec. and reported as many as he or she could remember. (b) Partial report: Person

saw all 12 letters, as before, but immediately after they were turned off, a tone indicated with row the person

was to report; (c) Partial report, delayed: Same as (b), but with a short delay between extinguishing the letters

and presentation of the tone.


11/44

repeated the partial-report procedure but instead of presenting the cue tones immediatelyafter the letters were extinguished, he delayed presentation of the tones so he could de-termine what a person could report from each row at various times after the display hadbeen extinguished (Figure 5.6c).

The result of the partial-report experiments showed that the participants memorydropped rapidly, so that by 1 second after the flash, they could report just slightly more

than 1 letter in a row, or a total of about 4 letters for all three rowsthe same number ofletters they reported using the whole-report technique. Figure 5.7 plots this result interms of the number of letters available to the participants from the entire display, whichSperling determined by multiplying the average number of letters reported for 1 rowtimes 3.

Sperling concluded from these results that a short-lived sensory memory registers allor most of the information that hits our visual receptors but that this information decays

within less than a second. This brief sensory memory for visual stimuli is called iconic

memoryor thevisual icon(icon means image), and corresponds to the sensory memorystage of Atkinson and Shiffrins model. Other research, using auditory stimuli, has shownthat sounds also persist in the mind. This persistence of sound, which is called echoic mem-ory, lasts for a few seconds after presentation of the original stimulus (Darwin et al., 1972).

Thus, sensory memory can register huge amounts of information (perhaps all of theinformation that reaches the receptors), but it retains this information for only seconds orfractions of a second. There has been some debate regarding the purpose of this large butrapidly fading store (Haber, 1983), but many cognitive psychologists believe that the sen-

sory store is important for (1) collecting information to be processed; (2) holding the


Figure 5.7 Results of Sperlings (1960) partial-report experiments. The decrease in performance is due to therapid decay of iconic memory (called sensory memoryin the modal model). (Reprinted from The Serial Position

Effect in Free Recall, by B. B. Murdoch, Journal of Experimental Psychology, 64, pp. 482488. Copyright

1962 with permission from the American Psychological Association. )

1.00.80.60.40.20

Delay of tone (sec)

2

4

6

8

10

12

Calculatednumberof

lettersavailabletoparticip

ant

Partialreport Whole

report


12/44

information briefly while initial processing is going on; and (3) filling in the blanks whenstimulation is intermittent. We now turn to the next boxes in the modal model, short-term memory, and long-term memory.

HOW CAN WE DISTINGUISH BETWEEN SHORT-TERM AND LONG-TERM MEMORY?Short-term memory and long-term memory are the central parts of the modal model.

When we described what happened in Rachels memory system as she ordered a pizza, wesaw that there is a rich interaction between STM and LTM. However, the way they are

represented as two separate boxes in the model implies that they are two different typesof memory with different properties. Certainly, our everyday experience seems to indicatethat there is one type of memory for remembering telephone numbers you have justlooked up in the phone book and another type for remembering the phone number a fewdays later or remembering what you did last summer. But could there be just one type ofmemory in which some information decays rapidly and some remains for long periods?

Although this is a possibility, there is good evidence to support the idea that STM andLTM are two different types of memory. We begin describing this evidence by consider-

ing a classic experiment that measured the relationship between a words position in a listand the chances that the word will be remembered later.

The Serial-Position CurveHow well can you remember a list of words? The following demonstration is based on aclassic experiment in memory research (Murdoch, 1962).

Demonstration

Remembering a List

Get someone to read the stimulus list (see end of chapter, on p. 178) to you at a rate of about one

word every two seconds. Right after the last word, write down all of the words you can remember.

You can analyze your results by noting how many words you remembered from thefirst five entries on the list, the middle five, and the last five. Did you remember more

words from the first or last five than from the middle? Individual results vary widely, butwhen Murdoch did this experiment on a large number of participants and plotted the per-centage recall for each word versus the words position on the list, he obtained the serial-position curve shown in Figure 5.8, which indicates that memory is better for words atthe beginning or end of the list.

Chapter 5146

Serial

Position


13/44

To understand why the words position in the list should matter lets first consider thebetter memory for words at the end of the list. This is called the recency effectbecausethese are the words that were presented most recently. One possible explanation for thebetter memory for words at the end of the list is that the most recently presented wordsare still in STM. To test this idea, Murray Glanzer and Anita Cunitz (1966) repeated Mur-

dochs experiment but had their participants count backwards for 30 seconds right afterhearing the last word. This counting prevented rehearsal and allowed time for informa-tion to be lost from STM. The result was what we would predict: The delay caused by thecounting eliminated the effect (Figure 5.9a). Glanzer and Cunitz therefore concluded thatthe recency effect is due to storage of recently presented items in STM.

But what about the words at the beginningof the list? Superior memory for stimulipresented at the beginning of a sequence is called the primacy effect.A possible explana-tion of the primacy effect is that these words have been transferred to LTM. One piece ofevidence supporting this idea is that the participants in Glanzer and Cunitzs experimentcontinued to remember these words even after they had finished counting backwards for30 seconds.

One reason that the words presented earlier in the list could have been transferredinto LTM is that participants had more time to rehearse them. Glanzer and Cunitiz testedthis idea by presenting the list at a slower pace, so there was more time between each wordand participants had more time to rehearse. Just as we would expect if the primacy effect


Figure 5.8 Serial-position curve (Murdoch, 1962). Notice that memory is better for words presented at the be-ginning of the list (primacy effect) and at the end (recency effect).

5 10 15 20

Serial position

20

40

60

80

100

Percentrecalled

Serial-position curve

Recencyeffect

Primaryeffect


14/44

is due to rehearsal, increasing the time between each word increased memory for the earlywords (see dotted curve in Figure 5.9b). Table 5.2 summarizes the results of the serial po-

sition experiments we have just described.

Chapter 5148

1 5 10 15

Serial position

Evidence that recency effect is due to STM

Evidence that primacy effect is due to LTM

20

30

40

50

60

70

Percentcorrect

No delay

30-sec delay eliminates

recency effect

1 5 10 15 20

Serial position

0

25

50

75

100

Percentcorrect

Short timebetween words

Longer timebetween wordsallows morerehearsal

(b)

(a)

Figure 5.9 Result of Glanzer and Cunitzs (1966) experiment. (a) The serial-position curve has a normal re-

cency effect when the memory test is immediate (solid line), but no recency effect occurs if the memory test isdelayed for 30 seconds (dashed line). (b) Memory for earlier words is better when words are presented more

slowly (solid line). (Reprinted from Journal of Verbal Learning and Verbal Behavior, 5, M. Glanzer et al., Two

Storage Mechanisms in Free Recall, pp. 351360 (Figures 1 & 2), copyright 1966, with permission from

Elsevier.)


15/44

Differences in CodingAnother way to distinguish between STM and LTM is to consider the way information iscoded in STM and LTM. Codingrefers to the way information is represented. Remember,for example, our discussion in Chapter 2 of how a persons face can be represented by thepattern of firing of a number of neurons. Determining how a stimulus is represented by thefiring of neurons is a physiological approach to coding. We can also take a mental approachto coding by asking how a stimulus or an experience is represented in the mind. For exam-ple, imagine that you have just finished listening to your cognitive psychology professorgive a lecture. We can describe different kinds of mental coding that occur for this experi-ence by considering some of the ways you might remember what happened in class.

Imagining what your professor looks like, perhaps by conjuring up an image in your

mind, is an example ofvisual coding. Remembering the sound of your professors voice isan example of auditory coding, which is called phonological coding. Finally, remember-ing what your professor was talking about is an example of coding in terms of meaning,

which is called semantic coding(see Table 5.3).Research on how information is coded in memory has demonstrated all of these kinds

of coding for both STM and LTM, but has shown that the most common type of codingfor short-term memory is phonological coding and the most common type of coding inlong-term memory is semantic coding.


Table 5.2 PRIMACY AND RECENCYEFFECTS

Effect Why Does It Occur? How Can It Be Changed?

Recency EffectBetter memory for words at the end Words are still in STM. To decrease, test after waiting 30

of the serial-position curve. seconds after end of the list soinformation is lost from STM

(see Figure 5.9a).

Primacy Effect

Better memory for words at the Words are rehearsed during To increase, present the list morebeginning of the serial-position curve. presentation of the list so slowly so there is more time for

they get into LTM. rehearsal (see Figure 5.9b).

Table 5.3 TYPES OFCODING

Type of Coding Example

Visual Image of a person

Phonological Sound of the persons voice

Semantic Meaning of what the person is saying


16/44

Coding in Short-Term Memory One of the early experiments that investigated cod-ing in STM was done by R. Conrad in 1964. In Conrads experiment, participants saw anumber of target letters flashed briefly on a screen, and were told to write down the let-

ters in the order they were presented. Conrad found that when participants made errors,they were most likely to misidentify the target letter as another letter thatsounded like thetarget. For example, F was most often misidentified as S or X, two letters that soundsimilar to F. Thus, even though the participantssaw the letters, the mistakes they made

were based on the letterssound.From these results Conrad concluded that the code for STM is phonological (based

on thesoundof the stimulus), rather than visual (based on the appearance of the stimulus).This conclusion fits with our common experience with telephone numbers. Even thoughour contact with them in the phone book is visual, we usually remember them by repeat-ing their sound over and over rather than by visualizing what the numbers look like in thephone book (also see Wickelgren, 1965).

Do Conrads results mean that STM is always coded phonologically? Not necessarily.Some tasks, such as remembering the details of a diagram or an architectural floor plan, re-quire visual codes (Kroll, 1970; Posner & Keele, 1967; Shepard & Metzler, 1971a). This useof visual codes in STM was demonstrated in an experiment by Guojun Zhang and Herbert

Simon (1985), who presented Chinese language symbols to native-speaking Chinese par-ticipants. The stimuli for this experiment were radicals and characters (Figure 5.10a).

Radicalsare symbols that are part of the Chinese language and that are not associated withany sound. Charactersconsist of a radical plus another symbol, and do have a sound.

When participants were asked to reproduce a series of radicals presented one after an-other, or a series of characters, they were able to reproduce a string of 2.7 radicals, on the

Chapter 5150

Figure 5.10 (a) Examples of radicaland character stimuli for Zhang and Simons (1985) coding experiment.

(b) Results showing evidence for visual coding (left bar) and phonological coding (right bar).

(a) (b)

Radicals Characters

6

4

2Num

berrecalled

0

8

Greater recall whenphonological codingpossible

Recall based onvisual coding

Radical(no sound)

Character(has sound)


17/44

average, and a string of 6.4 characters, on average (Figure 5.10b). The participants abil-ity to remember the radicals must be due to visualcoding, since the radicals have no soundor meaning. The participants superior memory for the characters is most likely due to the

addition of phonological coding, since each character is associated with a sound.Thus, information can be coded in STM both visually and phonologically. In addi-

tion, there is also evidence for semantic coding in STM. This is illustrated by an experi-ment in which Delos Wickens and coworkers (1976) had three different groups of partic-ipants (the fruit group, the meat group, and the professions group) listen to three

words, count backwards for 15 seconds, and then remember the three words. They didthis for a total of four trials, with different words presented on each trial.

Table 5.4 shows the experimental design of Wickenss experiment for the three groupsof participants. Looking down the fruit column indicates that the fruit group was askedto remember names of fruits for all four trials. The meat group was asked to remembermeats for the first three trials and then was switched to fruit on trial 4, and the professionsgroup was asked to remember professions for the first three trials, and then was switchedto fruit on trial 4.

The results for these three groups are indicated in Figure 5.11. There are two impor-tant things to notice about these results. First, participants in three all groups remembered

about 87 percent of the words on trial 1 (Figure 5.11a), and the performance for all threegroups dropped on trials 2 and 3 (Figures 5.11b and c), so by trial 3 they remembered onlyabout 30 percent of the words. The decrease in performance on the second and third trialsis caused by an effect called proactive interference (PI)information learned previouslyinterferes with learning new information.

The effect of proactive interference is illustrated by what might happen when a fre-quently used phone number is changed. Consider, for example, what might happen whenRachel calls the number 521-5100, she had memorized for Mineos Pizza, only to get a

recording saying that the phone number has been changed to 522-4100. Although Racheltries to remember the new number, she makes mistakes at first because proactive interfer-ence is causing her memory for the old number to interfere with her memory for the new


Table 5.4 WICKENSS EXPERIMENT DEMONSTRATING SEMANTIC CODING INSTM

Groups

Fruit Meat Profession

Trial 1 banana, peach, apple salami, pork, chicken lawyer, firefighter, teacher

Trial 2 plum, apricot, lime bacon, hot dog, beef dancer, minister, executive

Trial 3 melon, lemon, grape hamburger, turkey, veal accountant, doctor, editor

Trial 4 orange, cherry, pineapple orange, cherry, pineapple orange, cherry, pineapple

(same category) (switch category) (switch category)


18/44

number. The fact that the new number is similar to the old one adds to the interferenceand makes it harder to remember the new number.

The decrease in performance on trials 2 and 3 of Wickenss experiment was caused byproactive interference because the new items on each trial were from the same categoryas those on the previous trials. Participants in the fruit group heard only names of fruitsfor the first 3 trials, those in the meat group heard only names of meats, and those in theprofessions group heard only names of professions. But on the fourth trial, all of thegroups heard names of fruits. Performance remained low for the participants who hadbeen hearing the names of fruits because proactive interference continued for that group,but performance increased for the professions group because shifting to fruits eliminatedthe proactive interference that had built up on trials 13 for the names of professions (Fig-ure 5.11d). The resulting increase in performance is called release from proactive inter-ference, or release from PI.

Figure 5.11d also indicates that the release from PI is not as pronounced for theswitch from meats to fruits because meats and fruits are more similar to each other thanare professions and fruits. Can you relate these results to Rachels problem in remember-ing the new number for the pizza shop? Would the switch have been easier or harder forRachel if the new phone number had been very different from the old one?

What does release from PI tell us about coding in STM? The key to answering thisquestion is to realize that the release from PI that occurs in the Wickens experiment de-pends on the words categories (fruits, meats, professions). Because placing words into cat-

Chapter 5152

Figure 5.11 Results of Wickens et al.s (1976) proactive inhibition experiment. (See Table 5.4 for design).(a) Initial performance on trial 1. (b and c) On trials 2 and 3 performance for all groups (professions, mean,

and fruit) drops due to proactive interference. (d) On trial 4, performance recovers for the professions and

meat group due to release from proactive interference.

(a) Trial 1.

P MGroup

F

Percentrecalled 80

100

60

40

20

0

(b) Trial 2. More words in same categories.

P MGroup

F

Percentrecalled 80

100

60

40

20

0

Decrease due tobuildup of PI

(c) Trial 3. Words still in same category.

P MGroup

F

Percentrecalled 80

100

60

40

20

0

(d) Trial 4. Shift to fruit category for professions and meat. No shift for fruit.

P MGroup

F

Percentrecalled 80

100

60

40

20

0

Release from PI

More PI

Meat

Fruit

Professions


19/44

egories involves the meaningsof the words, the results of the Wickens experiment demon-strate the operation ofsemantic codingin STM.

Coding in Long-Term Memory Althoughsome semantic coding does occur in STM, se-mantic coding is thepredominanttype of coding in LTM. Semantic encoding is illustratedby the kinds of errors that people make in tasks that involve LTM. For example, remem-bering the word tree as bush would indicate that the meaning of the word tree (ratherthan its visual appearance or the sound of saying tree) is what was registered in LTM.

Jacqueline Sachs (1967) demonstrated the importance of meaning in LTM by havingparticipants listen to a tape recording of a passage like the one in the following demon-stration. Try this yourself.

Demonstration

Reading a Passage

Read the following passage.

There is an interesting story about the telescope. In Holland, a man named Lippershey was

an eyeglass maker. One day his children were playing with some lenses. They discoveredthat things seemed very close if two lenses were held about a foot apart. Lippershey began

experimenting and his spyglass attracted much attention. He sent a letter about it to

Galileo, the great Italian scientist. Galileo at once realized the importance of the discovery

and set about to build an instrument of his own.

Now cover up the passage and indicate which of the following sentences is identical to a sentence

in the passage and which sentences are changed.

1. He sent a letter about it to Galileo, the great Italian scientist.

2. Galileo, the great Italian scientist, sent him a letter about it.

3. A letter about it was sent to Galileo, the great Italian scientist.

4. He sent Galileo, the great Italian scientist, a letter about it.

Which sentence did you pick? Sentence 1 is the only one that is identical to one in thepassage. Many of Sachss participants (who heard a passage about two times as long as the

one you read) correctly identified (1) as being identical and knew that (2) was changed, buta number identified (3) and (4) as matching one in the passage, even though the wording

was different. The participants apparently remembered the sentences meaning and not itsexact wording.

But just as STM can be encoded in a few different ways, so can LTM. For example,you use a visual code when you recognize someone based on his or her appearance, andyou are using a phonological code when you recognize them based on the sound of his orher voice.



20/44

Neuropsychological Evidence for Differences Between STM and LTMIn Chapter 2 we saw that a technique used in neuropsychology is identifying dissocia-

tionssituations in which one function is absent but others are present. This techniquehas been used in memory research to differentiate between STM and LTM by studyingpeople with brain damage that has affected one of these functions while sparing the other.

People With Functioning STM but Poor LTM Clive Wearing, the musician who losthis memory due to viral encephalitis, is an example of a person who has a functioningSTM but is unable to form new LTMs. Another case of functioning STM but absentLTM is the case of H.M., who became one of the most famous cases in neuropsychology

when surgeons removed his hippocampus (see Figure 5.12) in an attempt to eliminateepileptic seizures that had not responded to other treatments (Scoville & Milner, 1957).

The operation eliminated H.M.s seizures, but unfortunately also eliminated his abil-ity to form new LTMs. Thus, the outcome of H.M.s case is similar to that of Clive Wear-ings, except Clive Wearings brain damage was caused by disease and H.M.s was causedby surgery. H.M.s unfortunate situation occurred because in 1953 the surgeons did notrealize that the hippocampus is crucial for the formation of LTMs. Once they realized thedevastating effects of removing the hippocampus, H.M.s operation was never repeated.However, H.M. has been studied for over 50 years and has taught us a great deal aboutmemory. The property of H.M.s memory that is important for distinguishing betweenSTM and LTM is the demonstration that it is possible to lose the ability to form newLTMs while still retaining STM. This loss of one ability while the other remains intact isa single dissociation. To determine a double dissociation we need to find another person

who has the opposite problem (i.e., good LTM, poor STM).

Chapter 5154

Frontalcortex

Prefrontalcortex

Amygdala

Hippocampus

Figure 5.12 Cross-section of the brain showing some of the key structures that are involved in memory.


21/44

People With Functioning LTM but Poor STM T. Shallice and Elizabeth Warrington(1970) describe K.F., a patient with normal LTM but poor STM. One indication of K.Fsproblems with STM is her reduced digit spanthe number of digits a person can re-

member. You can determine your digit span by doing the following demonstration.

Demonstration

Digit Span

Using an index card or piece of paper, cover all of the numbers below. Move the card down to un-

cover the first string of numbers. Read the numbers, cover them up, and then write them down.

Then move the card to the next string and repeat this procedure until you begin making errors.The longest string you are able to reproduce without error is your digit span.

2149

39678

649784

7382015

84261432

4823928075852981637

If you succeeded in remembering the longest string of digits, you have a digit span of10. The typical span is between 5 and 8 digits. Patient K.F. had a digit span of 2 and, inaddition, the recency effect in her serial-position curve, which is associated with STM,

was reduced.Table 5.5, which indicates which aspects of memory are impaired and which are intact

for Clive Wearing, H.M., and K.F., demonstrates that a double dissociation exists forSTM and LTM. In Chapter 2 we saw that this means that the two functions are caused bydifferent mechanisms, which act independently.

The evidence we have described involving coding, the serial-position curve, andneuropsychological case studies provides good reasons to differentiate between STM andLTM. We will see that this distinction has proven to be an extremely valuable way to ap-proach the study of memory. In the remainder of this chapter we will describe the prop-

erties of STM and then focus on working memory, which is an updated way of looking atshort-term memory.


Memory

Span

Table 5.5 A DOUBLE DISSOCIATION FORSTM AND LTM

STM LTM

Clive Wearing and H.M. OK Impaired

K.F. Impaired OK


22/44

Test Yourself 5.1

1. Why can we say that memory is life? Answer this question by considering whatmemory does for people with the ability to remember, and what happens when thisability is lost, as in cases like Clive Wearing (p. 137).

2. Describe Atkinson and Shiffrins modal model of memory both in terms of its struc-ture (the boxes connected by arrows) and the control processes. Then describe howeach part of the model comes into play when you decide you want to order pizza butcant remember the phone number of the pizzeria.

3. What do seeing the trail left by a moving sparkler and watching a film have in com-

mon? How are both related to the modal model of memory?4. Sperling showed that we can see 9 or 10 out of 12 letters right after they are brieflyflashed, but we can report only about 4 of these letters just a short time later. How didhe show this? How is this result related to the modal model?

5. The idea that there are two different types of memoryone short-term and the otherlong-termis supported by experiments that measured the serial-position curve andby experiments that investigated the form of the memory code. Describe these exper-iments and the reasoning behind them.

6. The proactive interference experiment described in the chapter is presented as evi-dence for a particular type of coding in short-term memory. What type of coding wasdemonstrated? Why does the conclusion hold for STM, even though the experimenttook much longer than the 3060 seconds that information is usually held in STM?

7. Studying the behavior of people with brain damage has provided further evidence forthe idea of two types of memory. Describe the cases of Clive Wearing, H.M., and K.F.in terms of their symptoms and how we can draw general conclusions about memoryfrom these symptoms. Why can we draw stronger conclusions about memory by con-sidering more than one case?

WHAT ARE THE PROPERTIES OF SHORT-TERM MEMORY?As we think about memory, considering both what psychologists have learned about it and

how we use it every day, it is easy to downplay the importance of STM compared to LTM.In my class survey of the uses of memory, my students focused almost entirely on howmemory enables us to hold information for long periods, such as remembering directions,peoples names, or material that might appear on an exam. Certainly, our ability to storeinformation for long periods is important, as attested by cases such as H.M. and Clive

Wearing, whose inability to form LTMs makes it impossible for them to function inde-pendently. But, as we will see, STM (and working memory) is also crucial for normalfunctioning. Consider, for example, the following sentence.

Chapter 5156


23/44

The human brain is involved in everything we know about the important things in life,like football.

How do we understand this sentence? First, the beginning of the sentence is stored inSTM. Then we read the rest of the sentence and determine the overall meaning by com-paring the information at the end of the sentence to the information at the beginning.

But what if we couldnt hold the beginning of the sentence in STM? If the informationin the first phrase faded before you completed the sentence, you might think that the topicof the sentence is football and wouldnt realize that the sentence is really about the brain.

Holding small amounts of information for brief periods is the basis of a great deal ofour mental life. Everything we think about or know at a particular moment in time in-

volves STM because short-term memory is our window on the present. (Remember fromFigure 5.4e that Rachel remembered the phone number of the pizzeria by transferring itfrom LTM to STM.) Early research on STM focused on answering the following twoquestions: (1) How much information can STM hold? and (2) How long can STM holdthis information?

What Is the Capacity of Short-Term Memory?

One measure of the capacity of STM is the digit span of 58 items (or more for some peo-ple). A famous paper by George Miller (1956), who was one of the pioneers in the devel-opment of modern cognitive psychology, begins with the title The Magical NumberSeven, Plus or Minus Two, and goes on to present evidence that we can hold 59 itemsin our short-term memory.

One of the things Miller grapples with in his discussion of STM is the problem ofdefining exactly what an item is. Although the answer to this question might seem obvi-ous if we simply consider the memory span for digits, it becomes more complicated when

we consider trying to remember the following words: team, noise, room, crowd, training,screening, football, film.

How many units are there in this list? There are 8 words, but if we group them differ-ently, they can form the following 4 pairs: football team, training film, crowd noise, screen-ing room. We can take this one step further by arranging these groups of words into onesentence: Thefootball teamviewed the training film about crowd noise, in thescreening room.

Is this stimulus 8 items, 4 items, or 1 item? Miller introduced the concept of chunking

to describe the fact that small units (like words) can be combined into larger meaningfulunits, like phrases or sentences. A chunkhas been defined as a collection of elements thatare strongly associated with one another but are weakly associated with elements in otherchunks (Gobet et al., 2001). Thus the word noise is strongly associated with the word crowdbut is not as strongly associated with the other words, such asfilm or room.

Research has shown that chunking in terms of meaning can increase our ability tohold information in STM. Thus, we can recall a sequence of 58 unrelated words, but ar-ranging the words to form a meaningful sentence so that the words become more strongly



24/44

associated with one another increases the memory span to 20 words or more (Butterworthet al., 1990).

K. Anders Ericcson and coworkers (1980) demonstrated an effect of chunking by

showing how a college student with average memory ability was able to achieve amazingfeats of memory. Their participant, S.F., was asked to repeat strings of random digits that

were read to him. Although S.F. had a typical memory span of 7 digits, after 230 one-hoursessions, he was able to repeat sequences of up to 79 digits without error. How did he doit? S.F. used chunking to recode the digits into larger units that formed meaningful se-quences. For example, 3492 became 3 minutes and 49 point 2 seconds, near world-record mile time, and 893 became 89 point 3, very old man. This example illustrateshow STM and LTM can interact with each other, because S.F., who was a runner, created

his chunks based on his knowledge of running times that were stored in LTM.Another example of chunking that is based on an interaction between STM and LTM

is an experiment by William Chase and Herbert Simon (1973a, 1973b) in which theyshowed chess players pictures of chess pieces on a chess board for 5 seconds. The chessplayers were then asked to reproduce the positions they had seen. Chase and Simon com-pared the performance of a chess master who had played or studied chess for over 10,000hours to the performance of a beginner who had less than 100 hours of experience. The

results, shown in Figure 5.13a, show that the chess master placed 16 pieces out of 24 cor-rectly on his first try, compared to just 4 out of 24 for the beginner. Moreover, the masterrequired only four trials to reproduce all of the positions exactly, whereas even after seventrials the beginner was still making errors.

Chapter 5158

Figure 5.13 Results of Chase and Simons (1973a, 1973b) chess memory experiment. (a) Master is better atreproducing actual game positions. (b) Masters performance drops to level of beginner when pieces are

arranged randomly.

(a) Actual game positions (b) Random placement

Master Beginner

12

8

4Correctplacements

0

16

Master Beginner

12

8

16

4Correctplacements

0

No advantage formaster if can't chunk

Master does betterbecause can chunkbased on game positions


25/44

We know that the masters superior performance was caused by chunking because itoccurred only when the chess pieces were arranged in positions from a real chess game.

When the pieces were arranged randomly, however, the chess master performed as poorly

as the beginner (Figure 5.13b). Chase and Simon concluded that the chess masters ad-vantage was due not to a more highly developed short-term memory, but to his ability togroup the chess pieces into meaningful chunks. Because the chess master had stored manyof the patterns that occur in real chess games in LTM, he saw the layout of chess piecesnot in terms of individual pieces but in terms of 4 to 6 chunks, each made up of a group ofpieces that formed familiar, meaningful patterns. When the pieces were arranged ran-domly, the familiar patterns were destroyed and the chess masters advantage vanished(also see DeGroot, 1965; Gobet et al., 2001).

Chunking is an essential feature of STM because it enables this limited capacity sys-tem to deal with the large amount of information involved in many of the tasks we per-form every day, such as chunking letters into words as you read this, remembering the firstthree numbers of familiar telephone exchanges as a unit, and transforming long conver-sations into smaller units of meaning.

What Is the Duration of Short-Term Memory?How long does information stay in short-term memory if we are prevented from rehears-ing it? In a famous experiment carried out independently by John Brown (1958) in Eng-land and Lloyd Peterson and Margaret Peterson (1959) in the United States, participants

were given a task similar to the one in the following demonstration.

Demonstration

Remembering Three Letters

You will need another person to serve as a participant in this experiment. Tell the person that you

are going to read them three letters followed by a number. Once they hear the number they should

start counting backwards by 3s from that number, and then when you say Recall they should

write down the three letters that they heard at the beginning. Once they start counting, time 20

seconds and say Recall. Note their accuracy and repeat this procedure for a few more trials,

using a new set of letters and a new three-digit number on each trial.

Trial 1: B F T 100

Trial 2: Q S D 96

Trial 3: K H J 104

Trial 4: L W G 50

Trial 5: C M Y 52

Trial 6: Z F H 75

Trial 7: F N C 120


Brown-Peterson


26/44

Peterson and Peterson found that their participants were able to remember about 80percent of the letters after a 3-second delay (left bar in Figure 5.14a), but could rememberan average of only 10 percent of the three-letter groups after an 18-second delay (right barin Figure 5.14a). Peterson and Peterson initially interpreted this result as demonstratingthat STM decays within 18 seconds. According to this interpretation, participants forgotthe letters because of the passage of time. However, when Keppel and Underwood (1962)looked closely at Peterson and Petersons results, they found that if they just considered

the participants performance on the first trial, there is little falloff between the 3-secondand the 18-second delay (Figure 5.14b). However, when they analyzed the results for thethird trial, they began seeing a drop-off in performance between the 3-second and the 18-second delay (Figure 5.14c).

The fact that memory for the letters becomes worse after a number of trials (also seeWaugh & Norman, 1965) suggests that the difficulty may have been caused by proactiveinterference. [Remember from our description of Wickenss experiment (see Figure 5.11and Table 5.4) that proactive interference occurs when material that was presented earlier

interferes with memory for new information.] The rapid fading of memory observed inBrowns and Peterson and Petersons experiments was therefore actually caused not bydecay, but by interference.

What does it mean that the reason for the decrease in memory is proactive interfer-ence, rather than decay? Because there is a great deal of interference in our everyday ex-perience, it doesnt really matter that PI causes the memory decrease, and we can still con-clude that the effective duration of STM, when rehearsal is prevented, is about 1520

seconds.

Chapter 5160

Figure 5.14 Results of Peterson and Petersons (1959) duration of STM experiment. (a) The result originallypresented by Peterson and Peterson, showing a large drop in memory for letters for a delay of 18 seconds be-

tween presentation and test; (b) analysis of Peterson and Petersons results by Keppel and Underwood, showing

little decrease in performance on trial 1, and (c) more decrease by trial 3.

3 18

50

Percentcorrect

0

100

Delay

3 18

50

100

Percentcorrect

0

Delay

3 18

50

100

0

Percentcorrect

Delay

(a) Average performanceover many trials

(b) First-trial performance (c) Third-trial performance


27/44

Problems With the Modal ModelAtkinson and Shiffrins model of memory has served the study of memory well because by

breaking the process of memory into stages, it generated a huge amount of research oneach stage and on how the stages interact with one another. There are, however, some re-sults that pose problems for the model. The model does not account for how a personsuch as K.F. who had a very small digit span could have poor STM but normal LTM. Ac-cording to the model, all information must pass through STM to reach LTM. But howcould this work for K.F., if her STM is defective? Shallice and Warrington (1970), whostudied K.F., suggested that the Atkinson and Shiffrin model is oversimplified and thatperhaps there may be another pathway that enables information to enter LTM without

passing through STM.In addition, other research has shown that STM is not just a single process but con-

sists of a number of different processes, perhaps corresponding to the way information iscoded into STM (remember that there is evidence for visual, phonological, and semanticencoding in STM; Freedman & Martin, 2001). Some of the most influential evidence forthis idea has been provided by Alan Baddeley, who has shown that the short-term processconsists of a number of specialized components.

One of the first experiments that suggested that the short-term process consists of a

number of components was the observation by Baddeley and Hitch (1974) that undersome conditions participants could do two tasks at once. You can demonstrate this to

yourself with the following demonstration.

Demonstration

Reading Text and Remembering Numbers

Keep these numbers in your mind (7, 1, 4, 9) as you read the following passage.

Baddeley reasoned that if STM had a limited storage capacity of about the length of a tele-

phone number, filling up the storage capacity should make it difficult to do other tasks that

depend on STM. But he found that participants could hold a short string of numbers in their

memory while carrying out another task, such as reading or even solving a simple word

problem. How are you doing with this task? What are the numbers? What is the gist of what

you have just read?

Because Baddeleys participants were able to read while simultaneously rememberingnumbers, he concluded that the short-term process must consist of a number of compo-nents that can function separately. In the foregoing example, the digit span task in which

you held numbers in your memory is handled by one component while comprehendingthe paragraph is handled by another component. Based on results such as this, Baddeleydecided that the name of the short-term process should be changed from short-term

memory to working memory.



28/44

WORKING MEMORY: THE MODERN APPROACH TO SHORT-TERM MEMORY

Baddeley (2000) definesworking memoryas follows: Working memory is a limited capacitysystem for temporary storage and manipulation of information for complex tasks such as compre-hension, learning, and reasoning. From this definition we can see that working memory dif-fers from STM in two ways:

1. Working memory consists of a number of parts.2. Its function is not just to brieflystore information but to manipulate information

to help us carry out complex cognitive tasks.

Thus, the emphasis of the working memory concept is not on how information is brieflyretained but on how information is manipulated to achieve complex cognitions such asthinking and comprehension (Baddeley, 2000).

The Three Components of Working MemoryWorking memory accomplishes the manipulation of information through the action ofthree components: thephonological loop, the visuospatial sketch pad, and the central executive(Figure 5.15).

The Phonological Loop The phonological loop holds verbal and auditory informa-tion. Thus, when you are trying to remember a telephone number or a persons name, orto understand what your cognitive psychology professor is talking about, you are using

your phonological loop (Figure 5.16a). This loop is divided into two parts:

1. Storagea place that holds the memory trace. The trace fades in about 2 sec-

onds unless it is refreshed by rehearsal. This is the passive part of the phonologi-cal loop.

2. Rehearsalthe part of the phonological loop responsible for the repetition thatrefreshes the memory trace. This is the active part of the phonological loop.

Chapter 5162

Figure 5.15 Diagram of the three main components of Baddeley and Hitchs (1974; Baddleley, 2000) model of

working memory: the phonological loop, the visuospatial sketch pad, and the central executive.

Storage(passive)

Rehearsal(active)

Phonologicalloop

Baddeley's working memory model

Centralexecutive

Visualinformation

Spatialinformation

Visuospatialsketch pad


29/44

The Visuospatial Sketch Pad Thevisuospatial sketch pad holds visual and spatialinformation. When you form a picture in your mind or do tasks like solving a puzzle orfinding your way around campus, you are using your visuospatial sketch pad (Figure5.16b). As you can see from the diagram, the phonological loop and the visuospatial sketchpad are attached to the central executive.

The Central Executive The central executive is where the major work of workingmemory occurs. The central executive pulls information from long-term memory and co-ordinates the activity of the phonological loop and visuospatial sketch pad by focusing onspecific parts of a task and switching attention from one part to another. For example,imagine that you are driving in a strange city, and a friend in the passenger seat is reading

you directions to a restaurant. As your phonological loop takes in the verbal directions,your sketch pad is helping you visualize a map of the streets leading to the restaurant (Fig-ure 5.17). These two kinds of information are coordinated and combined by the centralexecutive to create a coherent episode that we might title Finding the Restaurant.

As we describe working memory, keep in mind that it is a hypothesis about how themind works that needs to be tested by doing experiments. A number of experiments havebeen conducted to test the idea that the phonological loop and visuospatial sketch pad deal

with different types of information.


Figure 5.16 Tasks handled by components of working memory. (a) The phonological loop handles language.Reading is shown here, but the phonological loop also handles information that is received verbally, as when lis-

tening to someone speak. (b) The visuospatial sketch pad handles visual and spatial information.

(a) (b)

Reading


Phonologicalloop

Puzzle


30/44

Operation of the Phonological LoopThree phenomena support the idea of a system specialized for language: the phonologicalsimilarity effect, the word-length effect, and articulatory suppression.

Phonological Similarity Effect The phonological similarity effectoccurs when let-ters or words that sound similar are confused. Remember Conrads experiment in whichhe showed that people often confuse similar-sounding letters, such as T and P. Con-rad interpreted this result to support the idea of phonological coding in STM. In pres-ent-day terminology Conrads result would be described as a demonstration of thephonological similarity effect that occurs as words are processed in the phonologicalloop of working memory. Here is another demonstration of the phonological similarityeffect:

Chapter 5164

Figure 5.17 Tasks processed by the phonological loop (hearing directions) and visuospatial sketch pad(visualizing the route) being coordinated by the central executive.

verbal and visual information

Phonologicalloop


Go leftat the

secondcorner

PhonologicalSimilarity

Effect


31/44

Demonstration

Phonological Similarity Effect

Task 1: Slowly read the following words. Look away and count to 15. Then write them down.

mac, can, cap, man, map

Task 2: Now do the same thing for these words.

pen, pay, cow, bar, rig

Which of the two tasks was more difficult? Many people find that they confuse thesimilar-sounding words in Task 1 and that it is easier to remember the different-soundingwords in Task 2. This confusion of words in Task 1 is an example of the phonological sim-ilarity effect.

Word-Length Effect Theword-length effectrefers to the finding that memory forlists of words is better for short words than for long words.

Demonstration

Word-Length Effect

Task 1: Read the following words, look away, and then write down the words you remember.

beast, bronze, wife, golf, inn, limp, dirt, star

Task 2: Now do the same thing for the following list.

alcohol, property, amplifier, officer, gallery, mosquito, orchestra, bricklayer

Each list contains eight words but according to the word-length effectthe second list will be more difficult to remember because the words arelonger. Results of an experiment by Baddeley and coworkers (1984) thatillustrate this advantage for short words is shown in Figure 5.18. The

word-length effect occurs because the larger words fill up the capacityof the phonological loop, and so rehearsal is less effective for the longer

words because of the extra time needed to rehearse them.The limited capacity of the loop explains the initially surprising find-

ing that American children have a larger digit span than Welsh children.Before you conclude that American children are smarter than Welshchildren, consider that the names of numbers in Welsh (un, dau, tri, ped-

war, pump, chwech . . .) are longer than the names of the numbers in


Shortwords

Longwords

50

Percentcorrectrecall

0

100

Normal word-length effect

Figure 5.18 How word length af-fects memory, showing that recall is

better for short words (Baddeley et

al., 1984).


32/44

English (one, two, three, four, five, six . . .). Since it takes longer to pronounce Welsh num-bers, fewer can be held in the phonological loop, and the memory span for these numbersis therefore less (Ellis & Hennelly, 1980).

In another study of memory for verbal material, Baddeley and coworkers (1975)found that people are able to remember the number of items that they can pronounce inabout 1.52.0 seconds (also see Schweickert & Boruff, 1986). Try counting out loud, asfast as you can, for two seconds. According to Baddeley, the number you reach should beclose to your digit span.

Ar ti cula tory Suppressi on A phenomenon called articulatory suppressionoccurswhen a person repeats an irrelevant sound such as the while hearing words to remem-

ber. Saying the, the, the, the . . . impairs memory for the words by interfering with op-eration of the phonological loop (Baddeley et al., 1984; Murray, 1968). The followingdemonstration, which is based on an experiment by Baddeley and coworkers (1984), illus-trates an effect of articulatory suppression:

Demonstration

Repeating TheTask 1: Repeat the word the out loud (i.e., the, the, the . . .) as you read the following

list. Then turn away and recall as many words as you can.

automobile mathematics

apartment syllogism

basketball Catholicism

Task 2: Now do the same thing for the following list.

story ant towel

car coffee swing

According to the word-length effect, the second list should be easier to recall than thefirst because the words are shorter (see Figure 5.18). Remember that one of the advan-tages of shorter words is that they leave more space in the phonological loop for rehearsal.

However, Baddeley et al. (1984) showed that articulatory suppression caused by sayingthe, the, the . . . reduces performance on both lists and reduces the advantage for shortwords (Figure 5.19a). This occurs because repeating the, the, the . . . prevents rehearsalin the phonological loop (Figure 5.19b).

What about the effect of saying the, the, the . . . on the phonological similarity ef-fect, in which similar sounding words are confused? If participants heara list of words

while saying the, the, the . . ., the phonological similarity effect still occurs because eventhough the phonological loop is dealing with the, the, the . . ., asking participants to

Chapter 5166


33/44

listen to the words causes the words to enter the phonological loop directly (Figure 5.20a).Once in the phonological loop these words can be confused based on their sound, so thephonological similarity effect still occurs.


Figure 5.19 (a) Saying the, the, the . . . abolishes the word-length effect, so there is little difference in per-formance for short words and long words (Baddeley et al., 1984). (b) Diagram of how working memory explains

this result by proposing that saying the, the, the . . . reduces rehearsal in the phonological loop.

(a) (b)

Short

words

Long

words

50

Percentcorrect

recall

0

100

Articulatorysuppression

the, the,the . . .

Phonologicalloop


Reduces rehearsal

advantage forshort words

Figure 5.20 Working memory and the effect of saying the, the, the . . . on the phonological similarity effect.(a) Phonological similarity effect still occurs for spoken words because listening causes words to enter phono-

logical loop directly. (b) The effect does not occur for visually presented words because they are prevented from

being recoded verbally in the phonological loop.

the,the,the . . .

Phonological

loopVisuospatialsketch pad

Blocked from changing tophonological code by the, the, the . . .

Pear,Bear,Tear

Spoken words enterphonological loop directly

Effect of saying the, the, the . . .

the,the,the . . .

Phonological

loop

Visuospatial

sketch pad

Pear,Bear,

Tear

(a) Phonological similarity effect occurs if words are heard (b) No phonological similarity effect if words are read


34/44

However, if participants reada visually presented list of words while saying the, the,the . . ., the phonological similarity effect does notoccur, because the words are initiallyregistered in the visuospatial sketch pad and saying the, the, the . . . prevents them frombeing changed to a phonological code by the phonological loop (Figure 5.20b). Withouta phonological code, similar-sounding words cant be confused based on their sound sothe phonological similarity effect is eliminated. Table 5.6 summarizes the effects of sayingthe, the, the . . . on the phonological similarity effect.

The Visuospatial Sketch PadWhen we discussed divided attention in Chapter 4 (see page 113), we described some ex-periments by Lee Brooks (1968) in which participants were asked to do two tasks at once.

In one experiment, participants were asked to indicate whether each word in a sentencethey had memorized was a noun by either sayingyesor no (task 1), or by pointing to Yor

Nin a visual display (task 2) like the one in Figure 4.18 (see Figure 5.21a). In another ex-periment, participants were asked to indicate whether each corner of a block letter Fthat they were imagining was an outside corner by either sayingyesor no (task 3) or point-ing to YorN(task 4) (see Figure 5.21b).

The important result for our purposes is that for both of these experiments the taskwas easier if the stimulus that participants was holding in their mind and the operation

they were performing on that stimulus involved differentcapacities. Thus, in the first ex-periment, in which the stimulus was verbal(a memorized sentence), thespatial task (task 2)

was easier, but in the second experiment, in which the stimulus was spatial(an imaginedletter), the verbal task (task 3) was easier. We can explain these results in terms of thephonological loop and the visuospatial sketch pad by recognizing that verbal tasks dependon the phonological loop and spatial tasks depend on the visuospatial sketch pad. Thus,for task 1 in the first experiment, when the stimulus and task were both verbal (Figure

Chapter 5168

Table 5.6 EFFECT OFSAYING THE, THE, THE . . . ON THE PHONOLOGICALSIMILARITYEFFECT FORAUDITORY ANDVISUAL PRESENTATIONS

Presentation Effect?

Auditory No(hear word list) Phonological similarity

effect still occurs

(see Figure 5.20a)

Visual Yes

(read a word list) Phonological similarity effect no longer occurs

(see Figure 5.20b)


35/44

5.22a), the phonological loop was overloaded and the task became difficult. But for task 2in the first experiment, when the stimulus was verbal and the task was spatial (Figure5.22b), the loop and sketch pad shared responsibility and the task became easier. Figures5.22c and d show the spatial stimulus in the sketch pad for tasks 3 and 4 of experiment 2.Fill in the tasks and results for each diagram to match the information in Figure 5.21b.

All of these demonstrations show that working memory is set up to process phono-logical and visual-spatial information separately so it can handle differenttypes of infor-mation that are presented simultaneously, but that working memory has trouble handling

similartypes of information that are presented simultaneously.Before leaving the loop and the sketch pad, lets consider an experiment by M. A.

Brandimonte and coworkers (1992) that illustrates the principles we have been discussingin a different way. Brandimonte briefly presented a picture of an object like the one in Fig-ure 5.23a and then briefly presented a picture of part of the object, as in Figure 5.23b. Theparticipants task was to subtract the second picture from the first and indicate what the newpicture represented. Remember that the pictures were not present while the participants


Figure 5.21 Summary of the results of Brookss (1968) experiment described in Chapter 4. (a) The tasks in-volving the verbal stimulus.Task (2), which involves a spatial response, is easier. (b) The tasks involving the

spatial stimulus.Task (3), which involves a verbal response, is easier.

John ran to the store to buy some oranges

(a) Verbal stimulus

Task 1: Saying yes or no

Task 2: Pointing to Y or N in Figure 4.18

(b) Spatial stimulus

Visualize F and indicate whether each corner is outside by:

Memorize sentence and indicate whether each word is a noun by:

Easier

(spatial task)

Easier

(verbal task) Task 3: Saying yes or no

Task 4: Pointing to Y or N


36/44

Chapter 5170

Figure 5.22 The results of Brookss (1968) experiments from Figure 5.21, explained in terms of workingmemory. When the stimulus is verbal, (a) task 1, which requires a verbal response, is difficult because the

phonological loop has to process both the verbal stimulus and verbal response. (b) Task 2, which requires a

visual response, is easier because processing is divided between the phonological loop and the visuospatial

sketch pad. (c and d) Fill in these diagrams for tasks 3 and 4 in Figure 5.21b.

Remembersentence+

Sayyes

Phonologicalloop


(a) Task 1

Remembersentence

Pointto

yes

(b) Task 2: Easier

(c) Task 3: You fill in the task and result

(d) Task 4: You fill it in

Overload

d h b k h b h d b d h


37/44

carried out the subtraction task so the subtraction had to be done in theirminds. In the example in Figure 5.23, the initial picture is a piece of candy,and after subtraction, it becomes a fish. Brandimonte then had another

group of participants do the same task while saying la, la, la. . . .Based on what you know about operation of the phonological loopand visuospatial sketch pad, how do you think saying la, la, la . . .

would affect participants performance on the subtraction task? One wayto answer this question is to reason that saying la, la, la . . . would notaffect participants performance because the words la, la, la . . . areprocessed by the phonological loop and the visual images are processedby the visuospatial sketch pad, so they wouldnt interfere with one an-

other. However, as it turns out, saying the words improvedparticipantsperformance on the subtraction task.

Why did the la, la, la group do better? The key to answering thisquestion is to realize that in the first part of the experiment there are two

ways that the participants can memorize the objects. They can name them(verbal coding) so the object in Figure 5.23a would become the wordcandy, or they can create visual images of them (visual coding) so theobject would be a picture in the participants minds.

The working memory diagram in Figure 5.24 shows that saying la,la, la . . . fills up the phonological loop and so prevents the image in Fig-ure 5.23a from being coded phonologically. Consequently, the partici-pants must code the objects visually in the sketch pad (Figure 5.24). Be-cause the subtraction task in the second part of the experiment is easier

when starting with a visual image, being forced to use the sketch pad im-proves performance. Thus, the Brandimonte experiment shows how per-

formance can be enhanced when activity in one component of working memory forcesparticipants to use the system that is most well-suited to the task.


(a)

(b)

Figure 5.23 Stimuli used inBrandimonte et al.s (1992) exper-

iment. (a) The initial stimulus;

(b) part that the participant is

asked to mentally subtract from

the initial stimulus. (Reprinted

from Influence of Short-Term

Memory Codes on Visual Image

Processing: Evidence from Image

Transformation Tasks, by M.A.

Brandimonte, Journal of Experi-

mental Psychology: Learning,

Memory and Cognition, 18, pp.

157165. Copyright 1992 with

permission from the American

Psychological Association.)

Figure 5.24 Explanation of the Brandimonte result in terms of working memory. Saying la, la, la . . . , preventsthe initial stimulus from being recoded verbally by the phonological loop. Therefore, it must be coded visually by

the visuospatial sketch pad.

la, la,la, la,

la

Phonologicalloop


Phonological codinginto candy prevented

Must becodedvisually

St t f R h W ki M


38/44

Status of Research on Working MemoryOne of the characteristics of a good theory is that it suggests new experiments. Atkinsonand Shiffrins modal model did this, and so has Baddeleys working memory model. But

just as Baddeleys early observations led to a rethinking of the STM part of the modalmodel, other observations (including some made by Baddeley) have raised questions aboutthe working memory model.

For example, there are patients with brain damage whose memory spans have been re-duced to just one or two digits, indicating that the phonological loop is severely impaired.However, these same patients can repeat back sequences of 5 or 6 words if the words forma sentence (Vallar & Baddeley, 1984; Wilson & Baddeley, 1988). How can this memory

for sentences occur if the phonological loop is impaired? One possibility is that perhapslong-term memory (which is not impaired in these patients) helps out for meaningful ma-terial like sentences. Results such as these have led Baddeley (2000) to propose that thereare additional components in the working memory system that communicate closely withlong-term memory. It is likely that future research on working memory will be concerned

with the connection between working memory and long-term memory (see Ericsson &Kintsch, 1995).

More research is also needed to better determine how the central executive works.

You may have noticed that we said little about the central executive. One reason for this isbecause the central executives task of coordinating and controlling the two subsystems is

very complex, and only recently have researchers begun studying the ways that the cen-tral executive operates (see Andrade, 2001; Baddeley, 1996).

WORKING MEMORY AND THE BRAINEarly physiological studies of the short-term component of memory demonstrated thatbehaviors that depended on working memory can be disrupted by damage to specific areasof the brain, especially the prefrontal cortex (PF cortex; see Figure 5.12). The PF cortexreceives inputs from the sensory areas, which are involved in processing incoming visualand auditory information. It also receives signals from areas involved in carrying out ac-tions and is connected to areas in the temporal cortex that are important for forming long-term memories (see Figure 2.3 for location of the temporal lobe). Thus the wiring dia-

gram of the PF cortex is exactly what we would expect for the operation of a memorysystem like working memory that has to take incoming information from the environmentand pass some of this information on to longer-term storage.

The Delayed-Response Task in MonkeysEarly work on the physiology of working memory used a task called the delayed-responsetask,which required a monkey to hold information in working memory during a delay pe-

riod. Figure 5.25 shows the setup for this task. The monkey sees a food reward in one of

Chapter 5172


39/44

two food wells. Both wells are then covered, a screen is lowered during a delay, and then thescreen is raised and the monkey obtains the food if it reaches for the correct food well.

Monkeys can be trained to accomplish this task but if their PF cortex is removed, theirperformance drops to chance level so they pick the correct food well only about half of thetime. This result supports the idea that the PF cortex is important for holding informa-tion for brief periods of time. In fact, it has been suggested that one reason we can de-scribe the memory behavior of very young infants (younger than about 8 months of age)as out of sight, out of mind (so when an object that the infant can see is then hiddenfrom view, the infant behaves as if the object no longer exists) is that their frontal an

about memory

Documents