Three-Stage Model of Memory

The three-stage model of our memory system has guided much of the research on memory since the late 1960s (Atkinson & Shiffrin, 1968). Diagrams depicting this memory model typically use a series of boxes to indicate the stages of information processing, and arrows connecting the boxes to show the flow of information within the system. Figure 5.1 is an informational flow chart for the three-stage memory model. In general, information enters from the physical environment through our senses into sensory memory and flows from sensory memory to short-term memory to long-term memory and then back to short-term memory when we need to use it. In this section, we will discuss each of these three stages of information processing and how the stages interact to provide us with memory. The initial stage of processing, sensory memory, is roughly comparable to what we called sensation in Chapter 3. Information in this stage is bottom-up sensory input that hasn’t been recognized yet. So let’s begin by discussing how memory researchers have demonstrated that a sensory-memory stage exists and how information is processed in this stage.

Figure 5.1 Three-Stage Model of Memory Information from the physical environment enters the sensory registers through each of our senses (vision, hearing, taste, smell, and touch). This set of registers is referred to collectively as sensory memory. These registers are temporary storage places for sensory information until it can be attended to, recognized, and moved further along in the memory system. Sensory information that is not attended to is quickly forgotten. The information in each register that we attend to goes on to be recognized and enters the second stage of memory, short-term memory, which is comparable to our present awareness. Top-down processing (using information stored in long-term memory) guides this encoding of sensory input from sensory memory into short-term memory. If attended to and studied, information in short-term memory will be encoded into long-term memory where it is stored for later use. If not attended to, the information will be forgotten. To use the information stored in long-term memory, we bring it back into short-term memory (a process called retrieval). If we cannot retrieve such information, it is said to be forgotten. Later in the chapter we will consider explanations for such forgetting.

Sensory Memory

Sensory memory (SM) consists of a set of registers, where we temporarily store incoming sensory information from the physical environment until we can attend to it, interpret it, and move it to the next stage of memory processing (short-term memory). We have a register for each of our senses—vision, hearing, taste, smell, and touch. The sensory information stored temporarily in these registers has not yet been recognized. Once it is recognized and we are consciously aware of it, it has moved into the next memory stage, short-term memory. Vision is our dominant sense, so we’ll focus on the visual sensory register, commonly referred to as iconic memory, to help you understand how these registers work.

A good way to think about iconic memory is that it is photographic memory but for less than 1 second. An exact copy of the visual information exists in iconic memory, but only for a very brief period of time. We cannot attend to everything we see; therefore, the visual information in the register that we attend to gets recognized and goes on to short-term memory, and the unattended information in the register fades away and is quickly forgotten. We said that iconic memory lasts less than a second. This means that its duration (how long information can remain in a memory stage if not attended to) is less than a second. How do we know this? We will examine how research psychologists have answered this duration question and a similar capacity question (how much information can be held in a memory stage at one time) for each of the three memory stages. We’ll start with a discussion of how the capacity and duration answers for iconic memory were found. Two different experimental procedures, one direct and one inferential, were used. We’ll consider the direct one, the temporal integration procedure, first, because it is easier to understand.

The temporal integration procedure.

In the temporal integration procedure, two random meaningless dot patterns are presented sequentially at the same visual location with a brief time delay between them. When these two meaningless patterns are integrated, a meaningful pattern is produced. So, if the meaningful pattern is seen, this means that the two patterns must have been integrated in our memory system (since the two patterns were not presented simultaneously).

To see how this works, let’s consider an example. Look at the first two dot patterns (a and b) in Figure 5.2. Neither has any meaning by itself. Each is just a random dot pattern. However, if you integrate the two patterns (as shown in c), you see a meaningful pattern—the letters V O H. If the first two patterns were shown simultaneously in the same visual location on a screen so that they were integrated, you would see V O H; but if they were presented sequentially, the only way you could see the three letters is if the two meaningless dot patterns are integrated somewhere within your memory system. This is exactly what happens. The two patterns are integrated in iconic memory. However, this is only the case if the interval between the two patterns is very brief, less than a second (Eriksen & Collins, 1967). Participants in experiments such as this do not see the meaningful pattern when the two dot patterns are separated by more than a second. The first pattern has already faded from the visual register, and the two patterns cannot be integrated in iconic memory.

Figure 5.2 An Example of the Temporal Integration Procedure In this experimental procedure, two meaningless patterns (such as a and b) are shown sequentially at the same visual location. If the time interval between the two patterns is less than a second, a meaningful pattern (in this example, the letters V O H) is seen. The meaningful pattern can only be perceived when the two other patterns are integrated, so this integration must be taking place within our memory system, in what we call the visual sensory register or iconic memory.

This pattern of results using the temporal integration procedure is also observed for larger, more complex dot patterns (Hogben & Di Lollo, 1974). People see the meaningful integrated pattern if the interval between these larger patterns is less than a second. Hence the capacity of the visual sensory register must be fairly large or these more complex patterns couldn’t be integrated in iconic memory. Thus, the capacity of iconic memory is large, but its duration is very brief, less than a second.

Sperling’s full- and partial-report procedures.

We find the same results when we use a very different experimental procedure devised by George Sperling (1960). On each trial in Sperling’s research, participants were presented with a different matrix of unrelated consonants for 50 milliseconds, a very brief interval but long enough to process visual information. Different size matrices were used, but in our explanation we will consider 3 × 3, nine-letter matrices (as illustrated in Figure 5.3). The task was to report the letters in the matrix briefly flashed on each trial, but Sperling used two different report procedures. In Sperling's full-report procedure, the participant had to try to report the entire matrix of letters. Over trials, participants recalled 4.5 letters on average, usually those letters in the top row and the left section of the second row. However, participants also reported (subjectively) that they sensed the entire matrix, but that it had faded from memory by the time they reported the 4 or 5 letters. This sounds like iconic memory, doesn’t it? Let’s see how Sperling indirectly demonstrated through inference that the remaining unreported letters were indeed in iconic memory by using his partial-report procedure.

Figure 5.3 A 3 × 3 Letter Matrix Like Those Used in Sperling’s Iconic Memory Research On each trial, a different letter matrix is shown for 50 milliseconds. In the full-report procedure, participants attempt to recall all of the letters in the matrix. In the partial-report procedure, participants get an auditory cue following the matrix that tells them which row to report—high-pitched tone, recall top row; medium-pitched tone, recall middle row; and low-pitched tone, recall bottom row. The row that is cued is varied randomly across trials, so the participant has no way of knowing which row will be cued on any particular trial. In addition, the time between the letter matrix presentation and the auditory cue is varied.

In Sperling's partial-report procedure, the participant only had to report a small part of the presented letter matrix, a row indicated by an auditory cue on each trial. A high-pitched tone indicated that the top row was to be recalled, a medium-pitched tone the middle row, and a low-pitched tone the bottom row. These tones were easy to discriminate, so the participant had no difficulty in determining which row was indicated for recall. The letter matrix was different on every trial; therefore, the participant could not learn the matrix. In addition, the cued row was varied randomly across trials; therefore, the participant had no way of knowing which row would be cued on any particular trial. Regardless, when the auditory cue was given immediately after the brief presentation of the letter matrix, participants recalled the indicated row 100 percent of the time. From this result, we can infer that all of the rows must have been present in iconic memory; that is, an exact copy of the letter matrix must have been in iconic memory.

Based on the experimental results for the temporal integration procedure, what do you think would happen when Sperling inserted a time delay between the letter matrix and the auditory cue? Remember, the duration of iconic memory was estimated to be less than a second. Sperling found that as this time delay increased (up to 1 second), participants’ recall of the cued row worsened. This meant that the matrix was fading quickly from memory. We can conclude, then, based on two very different experimental procedures, that there is a visual sensory register and that it seems to hold an exact copy of the visual stimulus (indicating a large capacity), but only for a very brief time, less than a second (a very brief duration).

To get a feel for how iconic memory works in nonlaboratory situations, let’s think about seeing a bolt of lightning. It’s not really a singular, continuous bolt. It is actually three or more bolts that overlap in our iconic memory and lead to the perception of the single flash of lightning. If you turn off the lights and have a friend quickly move a flashlight in a circular motion, you see a circle of light. Why? Again, iconic memory is at work; it isn’t a continuous stream of light, but that’s what you see. This has larger implications. It is iconic memory that allows us to see the world as continuous and not as a series of unconnected snapshots.

Three or four bolts of lightning overlap in our iconic memory, leading to the perception of one continuous bolt.

NOAA.

All of our senses have sensory registers that have large capacities with very brief durations. For example, the auditory sensory register, called echoic memory, that processes sounds has a duration of 4–5 seconds, slightly longer than the duration of iconic memory (Darwin, Turvey, & Crowder, 1972; Glucksberg & Cowen, 1970). These sensory registers are what enable continuous perception of the physical environment. Collectively, these registers make up sensory memory, the first step of information processing in the three-stage memory model. However, when we think about memory, we aren’t usually thinking about sensory memory. We’re thinking about memory with a much longer duration. So let’s take a look at the next major stage of processing in the memory system—short-term memory, which has a little greater duration.

Short-Term Memory

Short-term memory (STM) is the memory stage in which the recognized information from sensory memory enters consciousness. We rehearse the information in short-term memory so we can transfer it into more permanent storage (long-term memory) and remember it at some later time. We also bring information from long-term memory back into short-term memory to use it to facilitate rehearsal, solve problems, reason, and make decisions; thus short-term memory is thought of as working memory (Baddeley, 1992, 2001, 2003; Engle, 2002). It is the workbench of the memory system. It is where you are doing your present conscious cognitive processing. What you’re thinking about right now as you read this sentence is in your short-term memory. As you work to understand and remember what you are reading, you are using your short-term working memory. The capacity of this type of memory is rather small. Humans just can’t process that many pieces of information simultaneously in consciousness. In addition, the information in this stage is in a rather fragile state and will be quickly lost from memory (in less than 30 seconds) if we do not concentrate on it. This is why it is called short-term memory. Now that we have a general understanding of the nature of this stage of memory, let’s see how researchers have learned about its capacity and duration.

The capacity of short-term memory.

To assess the capacity of short-term memory, researchers have used the memory span task. In this task, the participant is presented a series of items one at a time and has to remember the items in the order that they were presented. The list items could be any of several types of stimuli such as unrelated letters or words. On each trial, the specific list items change. For example, if words are used, then it is a different list of words on each trial. What have researchers found? George Miller provided the answer in his classic 1956 paper, “The Magical Number Seven, Plus or Minus Two: Some Limits on Our Ability to Process Information.” Your memory span is defined as the average number of items you can remember across a series of memory span trials. Humans remember 7 ± 2 (5 to 9) chunks of information on memory span tasks. To see what Miller meant by the term “chunk,” let’s consider the memory span task.

In the memory span task, different types of items can be used. If the items were unrelated letters, most participants would remember 5 to 9 unrelated letters. But if the items were three-letter acronyms (meaningful abbreviations like ABC or USA) or words (like dog or boy), participants would remember 5 to 9 three-letter acronyms or words (15 to 27 letters). In this latter case, participants remember more letters than in the first case, but they remember the same number of meaningful units. This is what is meant by the term “chunk.” A chunk is a meaningful unit in memory. The capacity limit in short-term memory is in terms of chunks, 7 ± 2 chunks. So if the chunks are larger for a particular type of material (words vs. letters), we remember more information but not more chunks. Experts in a particular area, such as chess masters, have larger chunks for information in their area of expertise (Chase & Simon, 1973). In the case of a chess master, for example, several chess pieces on the board form a chunk whereas for chess novices, each piece is a separate chunk.

The duration of short-term memory.

Now let’s consider the duration of short-term memory, less than 30 seconds. Why is the duration of short-term memory said to be less than 30 seconds if this type of memory is equivalent to our conscious workspace? If we choose to do so, we could keep information in our consciousness for as long as we want, clearly longer than 30 seconds. The duration estimate refers to how long information can stay in short-term memory if we cannot attend to it. To measure this duration, researchers developed the distractor task (Brown, 1958; Peterson & Peterson, 1959). In the distractor task, a small amount of information (three unrelated consonants such as CWZ) is presented, the participant is immediately distracted from concentrating on the information for a brief interval of time, and then the information must be recalled. How is the participant distracted? A number is immediately presented, and the participant has to count rapidly aloud backward by 3s (or by some other interval). Counting backward rapidly occupies the short-term work space and prevents the participant from attending to the three letters. The experimenter varies the length of the distraction period. When the distraction period is over, the participant must try to recall the letters. What happens? Some typical data are presented in Figure 5.4.

Figure 5.4 Results for the Short-Term Memory Distractor Task This figure shows how forgetting in short-term memory occurs over time. As the length of the distractor interval increases, forgetting increases very rapidly. In less than 30 seconds, recall is essentially zero.

(From Peterson & Peterson, 1959.)

As you can see in Figure 5.4 the estimated duration for information in short-term memory is rather brief, less than 30 seconds. To relate this to everyday life, think about looking up a new phone number. You find the number in the phone book. It goes into your iconic memory, and you attend to it and recognize it. It enters your conscious short-term memory. You start to concentrate on it so you can dial the number. Now what would happen to that number if you heard someone screaming outside and you ran to see what had happened? Chances are that you would forget the phone number (just like participants forget the three consonants in studies using the distractor task), and then have to look it up again. Information in short-term memory is in a temporary storage state, and we need to concentrate on it to prevent it from being lost. Usually we use maintenance rehearsal to accomplish this. Maintenance rehearsal is repeating the information in short-term memory over and over again in order to maintain it. For example, in the case of the phone number, we rehearse it over and over again to ourselves until we dial it.

Maintaining information in short-term memory is usually not our goal, certainly not when we are trying to learn. Our goal is to put that information into long-term storage so that we can retrieve and use it in the future. In the last two major sections of this chapter, we will look at the process of encoding information from short-term memory into long-term memory and of retrieving that information from long-term memory at some later time. But first we need to get an overview of long-term memory, the last memory stage in the three-stage model.

Long-Term Memory

When we use the word “memory,” we normally mean what psychologists call long-term memory. Long-term memory (LTM) allows storage of information for a long period of time (perhaps permanently), and its capacity is essentially unlimited. Remember the trillions of possible synaptic connections in the brain that we discussed in Chapter 2? They represent the capacity of long-term memory. The brain’s memory storage capacity has been estimated to be around 2.5 petabytes (a million gigabytes), which would be enough to hold three million hours of television shows if your brain worked like a video recorder (Reber, 2010). You would have to leave the television running continuously for more than 300 years to use up all that storage. We will consider the duration or permanence of information in long-term memory in more detail later in this chapter when we consider theories of forgetting.

The durations and capacities for all three stages of memory are summarized in Table 5.1. Review these so that you have a better understanding of how these three stages differ. Next, let’s consider different types of long-term memories.

Types of long-term memories.

Memory researchers make many distinctions between various types of long-term memories (Squire, 2004). The first distinction is between memories that we have to recall consciously and make declarative statements about and those that don’t require conscious recall or declarative statements (see Figure 5.5). What if someone asked you, “Who was the first president of the United States?” You would retrieve the answer from your long-term memory and consciously declare, “George Washington.” This is an example of what is called explicit (declarative) memory—long-term memory for factual knowledge and personal experiences. Explicit memory requires a conscious explicit effort to remember.

Figure 5.5 Types of Long-Term Memory

A further distinction is made in explicit memory between semantic memory, memory for factual knowledge, and episodic memory, memory of personal life experiences (Tulving, 1972). Remembering that George Washington was the first president of the United States (semantic memory) is very different than remembering your first kiss (episodic memory). Semantic and episodic memories blend together in autobiographical memory (Williams, Conway, & Cohen, 2008). Autobiographical memories obviously include episodic memories of your past personal experiences but they can also be about factual, semantic aspects of your personal history, such as remembering your birth date or what high school you attended.

Explicit memory is contrasted with implicit (nondeclarative) memory. Implicit memory is long-term memory that influences our behavior, but does not require conscious awareness or declarative statements. Implicit memory happens automatically, without deliberate conscious effort. For example, you remember how to drive a car and you do so without consciously recalling and describing what you are doing as you drive. Some implicit memories (like driving a car, typing, or hitting a tennis ball) are referred to as procedural memories because they have a physical procedural aspect (the execution of an ordered set of movements) to them. In contrast with declarative explicit memories, procedural implicit memories are like knowing “how” versus knowing “what.” Not all implicit memories, however, are procedural memories (see Figure 5.5). We learned about another type of implicit memory, classical conditioning, in Chapter 4. Classically conditioned responses elicited automatically by conditioned stimuli are also implicit memories.

Another type of nonprocedural implicit memory is priming. In priming, an earlier stimulus influences the response to a later stimulus. Priming is classed as implicit memory because it occurs independent of a person’s conscious memory for the priming stimulus. There are several experimental priming procedures, but let’s consider one called repetition priming so that you can gain a better understanding of how priming works. In repetition priming, a person first studies a list of words and then at some later time is asked to complete a list of word fragments with the first word that comes to mind for each fragment. For example, the fragment s _ c _ _ _ might be presented. The likelihood that the person answers s o c i a l (the primed word because the word social was on the earlier word list) is much higher than for unprimed words, such as s o c c e r or s o c k e t, that fit the fragment but were not on the list. Such priming occurs even when people had not recognized the list word on an earlier recognition test (Tulving, Schacter, & Stark, 1982). Thus, priming occurs when explicit memory for the word does not, which means that priming is an involuntary, nonconscious implicit process. Graf, Squire, and Mandler (1984) provided further evidence that priming is an implicit, nonconscious type of memory. They found that amnesics who had no explicit memory for new information could perform as well as normal adults on a repetition priming word fragment task even though the amnesics had no conscious memory of having seen the words before.

Next we’ll see how other memory research with amnesics has allowed us to further differentiate explicit and implicit memory and even discover what parts of the brain seem to be involved in each type of memory.

Amnesia, the loss of long-term memories.

There is some evidence from the studies of amnesics, people with severe memory deficits following brain surgery or injury, that explicit and implicit memories are processed in different parts of the brain. We will focus our discussion of amnesics on the most studied amnesic in psychological research, H. M., whom we discussed briefly in Chapter 1. When H. M. was 7 years old, he was hit by a bicyclist and suffered a brain injury that later led to severe epileptic seizures in his teens (Hilts, 1995). In 1953 at the age of 27, H. M. had his hippocampus and surrounding temporal lobe areas removed with the hope of reducing his epileptic seizures (Corkin, Amaral, Gonzalez, Johnson, & Hyman, 1997). The seizures were reduced, but the operation drastically affected his long-term memory. Before the operation, he had normal short-term memory and long-term memory. After the operation, he had normal short-term memory, above-average intelligence, and no perceptual or language deficits, but he didn’t seem able to store any new information in long-term memory (Scoville & Milner, 1957). For example, H. M. could read the same magazine over and over again and think it was a new magazine each time. He could no longer follow the plot of a television show because the commercials would interfere with his memory of the story line. If he did not know someone before his operation and that person introduced herself to him and then left the room for a few minutes, he would not know the person when she returned. Even Brenda Milner and Suzanne Corkin, the two neuroscientists who studied H. M. for decades, had to introduce themselves anew each time they met with him.

H. M. had anterograde amnesia—the inability to form new explicit long-term memories following surgery or trauma to the brain. Anterograde amnesia is contrasted with retrograde amnesia—the disruption of memory for the past, especially episodic information for events before, especially just before, brain surgery or trauma. Such amnesia is typical in cases of brain concussions. H. M. had some retrograde amnesia, especially for the several days preceding the surgery, but this was mild compared to his severe, pervasive anterograde amnesia.

Remember, as we just learned, amnesics have shown implicit repetition priming effects. So what about H. M.? Although he didn’t form any new explicit long-term memories (but see O’Kane, Kensinger, & Corkin, 2004, and Skotko et al., 2004), did he too form new implicit memories? The answer is a resounding “yes” (Corkin, 2002). Let’s briefly consider some of the experiments demonstrating this. H. M. demonstrated implicit procedural memory on a mirror-tracing task. In this task, you have to trace a pattern that can be seen only in its mirror image, which also shows your tracing hand moving in the direction opposite to its actual movement. This is a difficult motor task, but there is a practice effect in that the number of errors decreases across practice sessions. H. M.’s performance on this task showed a normal practice effect even when months elapsed between the sessions (Gabrieli, Corkin, Mickel, & Growdon, 1993). However, he did not remember ever having done this task and had to have the instructions repeated for each session. Corkin (1968) also found that H. M. improved with practice on another manual skill learning task, one in which he had to keep a stylus on a small dot that was spinning around on a turntable. H. M. got better at this task with practice, but he had no conscious memory of his earlier experiences that led to his improved performance.

The Mirror-Tracing Task| The task is to trace the outline of a star (or some other shape) with a metal stylus when the star and your hand can be seen only in the mirror. Thus, the tracing movements have to be made in the direction opposite from the way in which they appear in the mirror. When the stylus moves off of the star outline (each red section in the illustrated tracing), it makes electrical contact with the underlying aluminum plate and a tracing error is recorded. There is nonconducting tape on the star outline so as long as the stylus stays on the outline, no electrical contact is made. Just like we would, H. M. improved from session to session (the number of errors he made decreased) as he gained more experience in this task. However, unlike us, he could not remember ever having performed the task before and had to have the task explained to him each session. As explained in the text, this means that he formed new implicit procedural memories for how to do the task, but he did not form new explicit episodic memories of having performed the task.

H. M. also demonstrated implicit repetition priming effects on a word fragment completion task without conscious awareness of the earlier presented priming words (Corkin, 2002) and implicit memory for a classically conditioned eyeblink response (Woodruff-Pak, 1993). In the latter case, after he was classically conditioned to give an eyeblink response to a tone, he could not consciously remember the conditioning episodes; however, he stored an implicit memory of the conditioned association between the tone and eyeblink response. Thus, he blinked when he was exposed to the tone but had no idea why he did so. LeDoux (1996) tells the story of a similar finding with a female amnesic with memory deficits like those of H. M. She was unable to recognize her doctor (Edouard Claparède), so each day Dr. Claparède shook her hand and introduced himself. One day, however, the doctor concealed a tack in his hand so that when shaking hands, the tack pricked her. The next time the doctor tried to introduce himself and shake hands, she refused to do so but didn’t know why. She had been conditioned without any explicit awareness of it.

How is it possible that H. M. and other amnesics like him can form new implicit but not explicit memories? Research indicates that other parts of the brain, such as the cerebellum and basal ganglia, and not the hippocampus are important for implicit memory formation and storage (Green & Woodruff-Pak, 2000; Knowlton et al., 1996; Krebs, Hogan, Hening, Adamovich, & Poizner, 2001; Krupa, Thompson, & Thompson, 1993; Squire, 2004). Implicit memory formation is functional in these amnesics because the cerebellum, basal ganglia, and the other parts of the brain necessary for such memories are intact, but explicit memory formation isn’t because the hippocampus has been removed. This explanation is also one of those proposed for infantile/child amnesia, our inability as adults to remember events that occurred in our lives before about 3 years of age. According to this explanation, we cannot remember our experiences during this period because the hippocampus, which is crucial to the formation of episodic explicit long-term memories, is not yet fully developed. It is also important to realize that although the hippocampus is critical to the formation of such memories, it is not the final repository for these memories, but more like a holding zone for them. As explicit memories age, the hippocampus’s participation wanes (Smith & Squire, 2009). These memories are distributed throughout many areas in the cortex, but how this happens and how these memories are represented remain questions to be answered (Miller, 2005).

Evidence for short-term versus long-term memory distinction.

The memory findings for amnesics like H. M. also indicate that short-term memory and long-term memory are different stages of memory. H. M.’s short-term memory did not suffer any substantial deficits after the operation. For example, his memory span was within the normal range. He could repeat a telephone number with no difficulty. Researchers examining the free recall task have found additional evidence that long-term memory and short-term memory are different stages. In the free recall task, participants are given a list of words one at a time and then asked to recall them in any order they wish. Kirkpatrick (1894) introduced the free recall task, noting that some word positions are recalled much better than others. If recall performance for the words is plotted in terms of the order the words were presented (their position in the list—first, second,…, last), the figure has a very distinctive shape (Figure 5.6). The ends of the list are recalled much more often than the middle of the list. The superior recall of the early portion of the list is called the primacy effect. The superior recall of the latter portion of the list is called the recency effect.

Figure 5.6 Serial Position Effects for a One-Trial Free Recall Task The superior recall of the first few items presented relative to those in the middle of the list is called the primacy effect. This effect is due to the fact that the primary items in the list are studied more and so have a higher probability of being in long-term memory for later recall. The recency effect refers to the superior recall of the last few items presented versus those in the middle of the list. This effect is due to the easy immediate recall of the items presently in short-term memory (those recently presented).

How do the primacy effect and recency effect relate to the distinction between short-term and long-term memory? The recency effect is caused by recall from short-term memory. Items at the end of the list, the most recent items, have a high probability of still being in short-term memory, so they can be recalled immediately and very well. The primacy effect, however, is the result of superior recall from long-term memory of the first few words in the list versus those in the middle (Rundus & Atkinson, 1970). Let’s think about the task. The words are presented one at a time. The first word comes into your empty short-term memory to be rehearsed. It gets 100 percent of your attention. Then the second word appears and is rehearsed with the first word (each gets 50 percent of your attention). This goes on until short-term memory capacity is reached and each new word causes a word already in short-term memory to be bumped out. This results in the first few items on the list being rehearsed more than the later words in the middle and thus having a higher probability of being stored in long-term memory and recalled better. The items in the middle of the list come into a filled short-term memory, get little rehearsal, and thus have a low probability of being stored in long-term memory and recalled later. This is similar to what sometimes happens to students on exams. They study one topic more than another, and this translates to better test performance on the more-studied topic.

How do we know that this is so? Researchers have demonstrated that the primacy effect and the recency effect can be manipulated independently, indicating that different memory stages are involved. For example, if recall is delayed by having the participants count rapidly backward by 3s for 30 seconds, the recency effect is eliminated, but the primacy effect remains (Glanzer & Cunitz, 1966). Thus, the distractor period only disturbed recall from short-term memory. Similarly, we can eliminate the primacy effect but still observe a recency effect if we force participants to rehearse each of the list items equally by having them only rehearse each word aloud after it is presented. With equal rehearsal, the primary and middle items are recalled equally well so there is no primacy effect; and because amount of rehearsal is not critical to recall from short-term memory, a recency effect (although smaller) remains. In addition, recent fMRI neuroimaging data indicate that both short-term memory and long-term memory are involved in these serial position effects (Talmi, Grady, Goshen-Gottstein, & Moscovitch, 2005). Recognition of early items in the word list uniquely activated brain areas traditionally associated with long-term memory, but none of these areas were activated for retrieval of late items in the list.

We learned that more rehearsal leads to better long-term memory, but is better memory just a matter of more rehearsal (more study)? This question is addressed in the next section where we discuss moving information from short-term to long-term memory (what you try to do when you study). You will learn that the type of rehearsal (study method) you use is important. Thus, the information in the next section can help you to develop better study strategies and better memory.