BADDELEY’S WORKING MEMORY MODEL

BADDELEY’S WORKING MEMORY MODEL
Baddeley’s (1986, 1997) model of working memory involves three main components: a central executive, and two socalled ‘slave’ systems – the phonological loop and the visuo-spatial sketch pad. To these Baddeley (2001) has added an episodic buffer. The central executive controls attention and coordinates the slave systems; the phonological loop contains a phonological store and an articulatory control process and is responsible for inner speech; the visuo-spatial sketch pad is responsible for setting up and manipulating mental images; the episodic buffer integrates and manipulates material in working memory.

Developing the models for Information processing systems

Developing the models for Information processing systems
While models like Atkinson and Shiffrin’s (1968) are useful ways to simplify and represent aspects of complex systems, this very complexity requires ongoing adjustment to enable these models to account for additional observations.


For example, the information-processing model made two assumptions:
1. that information could only reach long-term memory by passing through the short-term store; and
2. that rehearsing the information in the short-term store would both retain it in this store, and increase its chance of being selected for transfer to the long-term store.

The first of these assumptions was challenged by the identification of patients who had grossly impaired short-term memory spans and therefore (in terms of the model) severely damaged short-term memory stores, but who appeared to have no impairment in their long-term learning ability (Shallice & Warrington, 1970). The second assumption was called into question by studies where participants rehearsed the last few words of free recall lists for a longer time without showing improvement in the long-term recall of those words (Craik & Watkins, 1973). Under some circumstances, it became clear that encountering the same information on many occasions (which may also be assumed to lead to increased rehearsal) was not sufficient to lead to its retention. For example, people do not remember the details on the faces of the coins that they handle daily (Morris, 1988; Nickerson & Adams, 1979), as you discovered when you tried to sketch a penny at the beginning of the chapter. Other evidence that previously formed the basis for distinguishing between short-term and long-term memory stores has also come into question. For example, the recency effect in free recall had been attributed to the operation of a short-term store because it disappeared when the last few seconds before recall were filled with a task such as backward counting. But when recall was studied under different conditions, recency effects reappeared even without a contribution from short-term memory. When participants studied words and counted backwards after each word in the list, the last few items were better recalled than the middle of the list. This pattern was at odds with the model, because the short-term store should have been ‘filled’ with counting, and so no recency effect should have been observed (e.g. Baddeley & Hitch, 1977; Tzeng, 1973). Semantic encoding was also demonstrated in short-term learning under suitable conditions (Baddeley & Levy, 1971), showing that phonetic encoding was not the only form of coding relevant for the short-term store. Two major responses followed recognition of the problems with the Atkinson and Shiffrin (1968) information-processing model. One approach, especially associated with Baddeley et al. (e.g. Baddeley, 1986), was to enhance the short-term memory model in the light of its known limitations, along with more consideration of the functions that short-term remembering plays in cognition. This change in perspective led to Baddeley’s (1986, 1997, 2001) working memory model. The other response was to question the emphasis on memory stores and their capacity limitations, and to focus instead on an alternative approach based on the nature of the processing that takes place, and its consequences for remembering.

INFORMATION-PROCESSING METAPHOR

INFORMATION-PROCESSING METAPHOR
In the 1960s subdivisions of memory based upon informationprocessing models became popular. Following postwar developments in information technology, there had been a substantial growth in understanding the requirements of information storage during computer processing. A three-stage model of memory processing developed, reaching its fullest elaboration in the version proposed by Atkinson and Shiffrin (1968). In these stage models, information was considered to be first held very briefly in sensory memories before a selection of this information was transferred to a short-term store. From here, a yet smaller amount made its way into a long-term memory store.

Sensory memories
Evidence for sensory memory stores came from experiments such as Sperling’s (1960). He presented displays of 12 letters very briefly (e.g. 50 ms) to participants. Although they could report only about four letters, Sperling suspected that they might actually be able to remember more letters, but they could not hold them in mind long enough to report them. To test this hypothesis, Sperling briefly presented the letters as a matrix containing three rows, and then sounded a tone. Participants had been instructed to report only part of the array – which part depended on the pitch of the tone. Using this partial report procedure, Sperling found that people could recall about three letters from any row of four, which meant that they could actually potentially recall (for just a very brief period) about nine out of the twelve letters. Psychologists inferred from research such as this that there is a sensory memory store which holds a large amount of incoming perceptual information very briefly while selected elemens are processed. This sensory memory for visual information was termed iconic memory by Neisser (1967). Sensory memory for auditory information was referred to as echoic memory. Sensory memories are generally characterized as being rich in content, but very brief in duration.


Short- and long-term memories
Beyond the sensory memories, the information-processing models hypothesized one or more short-term stores that held information for a few seconds. The verbal short-term store has received the most research attention, its existence being inferred in part from the recency effect in free recall. For example, Postman and Phillips (1965) asked their participants to free recall lists of 10, 20 or 30 words. With immediate recall, the participants tended to be much better at recalling the last few words that had been presented than words from the middle of the list. But this recency effect disappeared if testing was delayed by as little as 15 seconds, so long as the delay involved verbal activity by the participant. The interpretation of the recency effect was that the last few items were being retrieved from a short-term store of rather limited capacity. The short-term store was believed to retain information primarily in an acoustic or phonological form (Baddeley, 1966) – a view that received additional support fro the errors that appear during short-term retention, even when the material to be retained is presented visually. Conrad and Hull (1964), for example, showed that visually presented sequences of letters that are similar in sound (e.g. P, D, B, V, C, T) were harder to recall correctly than were sequences of dissimilar-sounding letters (e.g. W, K, L, Y, R, Z).

On the other hand, long-term memory was believed to be stored primarily in terms of the meaning of the information. So, when asked to remember meaningful sentences, people usually cannot reproduce the exact wording, but they can generally report the meaning of what has been encountered (e.g. Sachs, 1967).

EXPLICIT AND IMPLICIT MEMORY

EXPLICIT AND IMPLICIT MEMORY
Another common distinction is between explicit and implicit memory. Explicit memory (explicit memory memory with conscious awareness of the original information or the situation in which the learning occurred) involves conscious awareness of the original information or the situation in which the learning occurred, and recollection of the original information or experience that is subsequently recalled. As these experiences involve a recollective experience, Baddeley (1997), among others, prefers to refer to ‘recollective’, rather than explicit, memory. Implicit memory refers to an influence on behaviour, feelings or thoughts as a result of prior experience, but without conscious recollection of the original events. (implicit memory influence on behaviour, affect or thought as a result of prior experience but without conscious recollection of the original events) For example, if you pass the fish counter in a supermarket, you might later think of having fish for dinner without being aware that ‘fish’ had been primed by the supermarket experience. Baddeley (1997) argues that, rather than a single implicit memory system, there is probably an array of learning mechanisms that are similar in that they influence subsequent behaviour but they do not generate recollective memories.

Demonstrating the distinction
Distinctions between implicit and explicit memory are sometimes demonstrated by studies that measure priming. One task used in many priming studies is completion of word fragments (described previously for the word ‘computer’). Solutions are generally faster or more certain for recently encountered words than for new ones, even when the words are not consciously recognized. One source of evidence for the implicit/explicit distinction comes from studies involving patients with amnesia. Their amnesia means that they cannot consciously recognize words or pictures that have been previously presented, but they are nevertheless better at completing the corresponding word fragments later on (Warrington & Weiskrantz, 1968). Tulving, Schacter and Stark (1982) found a similar difference between priming and recognition test results for healthy participants not suffering from amnesia. The effect of studying a list of words on later recognition of those words declined considerably over a seven-day period, but there was no imilar decline in the effects of priming of the presented words. These studies suggest that there is a fundamental difference in the functional nature of memory, depending upon whether the test requires conscious awareness of the previous event. Jacoby (1983) provided further evidence for this view. As in the previous studies, there were two types of test: recognition (involving conscious remembering) and unconscious remembering (in this case tested via perceptual identification, i.e. identifying a word that appeared in a brief flash). Jacoby also manipulated how the words were studied. Each target word was shown with no context (e.g. ‘woman’), or shown with its opposite as a context (e.g. ‘man – woman’), or generated by the participant when shown its opposite (e.g. ‘man’ shown and ‘woman’ generated by the participant). Subsequently, the explicit memory test involved showing a mixture of target words and new words to participants and asking them to identify which words they had studied (‘Studied’ words included both read and generated words, as described above). The implicit memory test was a perceptual identification test: a mixture of targets and new words were shown very briefly (40 ms) one at a time, and the participants attempted to identify the word. on the implicit memory measure of identification and the explicit memory task of recognition. Explicit recognition improved from the ‘no context’ condition to the ‘generate’ condition, but the reverse was the case for the implicit perceptual identification task. Because the pattern of results is reversed for the two tests, it suggests that the underlying processes (i.e. implicit and explicit memories) are distinct and involve possibly independent memory mechanisms.


The nature of the task
The implicit/explicit memory distinction is often tangled up (and therefore potentially confused) with two different types of task. Some tasks require people to think about meanings and concepts; these are called concept-driven tasks. Others require people to focus on the materials in front of them; these are called data-driven tasks (Roediger, 1990). For example, if you are asked to remember words from a list that you studied, you would be explicitly recalling words and you would be likely to recall the meanings of the words as well. On the other hand, if your task was to complete word fragments, without reference to the studied list, then the influence of the study session would be implicit rather than explicit, and you would be working with the visual patterns of letters, but less so (if at all) with the meanings. It is challenging to separate the nature of the task (i.e. conceptor data-driven) and the nature of the memory being tested (i.e. explicit or implicit) (see Roediger & Blaxton, 1987). Roediger, Buckner and McDermott (1999) review the debate between explanations based upon memory systems and memory processes. The experience of remembering (or knowing) Related to the explicit/implicit memory distinction is the experience that accompanies performance on the memory task. A participant may remember having seen the item under test in a recognition experiment at the original learning trial, or they may simply ‘know’ that the word was in the original list without specifically recalling it. This ‘remember/know’ distinction was first used by Tulving (1985). He required each response in the memory test to be judged as being accompanied by an experience of remembering having studied the item, or, alternatively, of knowing that the item had been presented without specifically remembering the event. Gardiner and associates have since carried out extensive investigations of ‘remember/know’ judgements under a range of different conditions (reviewed by Gardiner & Java, 1993). A number of conditions have been shown to influence ‘remember’ and ‘know’ judgements differently. For example, semantic processing (where the meaning of the items is foremost) leads to more ‘remember’ responses than does acoustic processing (which emphasizes the sound of the words studied). In contrast, ‘know’ responses do not differ between the semantic and acoustic conditions.

MEMORY MODELS

KINDS OF REMEMBERING
Psychologists have applied a number of techniques in their efforts to understand memory. One approach has been to subdivide the vast field of memory into areas that seem to function differently from one another. Cast your mind back to the last time you arrived home. How does that memory differ from remembering how to spell ‘table’, or that there are 11 players in a soccer team, or remembering how to ride a bicycle? Our intuition would suggest that there are different kinds of remembering. But what is the evidence? Episodic and semantic memory One distinction made by psychologists is between episodic and semantic memory (Tulving, 1983). Episodic memory (episodic memory memory for personally experienced events) can be defined as memory for the personally experienced events of your life. Such memories naturally tend to retain details of the time and situation in which they were acquired.

Semantic memory, (semantic memory abstract knowledge that is retained irrespective of the circumstances under which it was acquired (e.g. ‘the world’s largest ocean is the Pacific’)) by contrast, is knowledge that is retained irrespective of the circumstances under which it was acquired. For example, your memory of eating breakfast this morning will be an episodic one involving when, where and what you ate. On the other hand, remembering the meaning of the term ‘breakfast’ involves semantic memory. You can describe what ‘breakfast’ means but you probably have no recollection of when and how you learned the concept. Autobiographical memory – the recall of events from our earlier life – has become a particular aspect of episodic memory that has attracted considerable interest in recent years (Cohen, 1996; Conway, 1996). (autobiographical memory the recall of events from our earlier life – a type of episodic memory)

Declarative and procedural knowledge
Another sub-division of memory is between declarative and procedural knowledge (Anderson, 1976; 1995). Declarative knowledge is explicit knowledge that people are consciously aware of and can report. For example, you can probably remember eating breakfast this morning. Ryle (1949) described this type of memory as ‘Knowing That’. Procedural knowledge is a knowledge of how to do things, such as riding a bicycle or typing. Ryle referred to it as ‘Knowing How’. The skills of typing, driving and so forth may be well learned and highly developed, but it is generally not easy to describe in detail how to carry them out. So an accomplished typist might find it difficult to identify each finger movement required to type this sentence, while being quite capable of typing it quickly and correctly.

Comparison groups

Comparison groups
But it clearly is not enough to observe how often ‘gold’ appeared in the lists. Many people, when asked to think of metals, would include gold, even without having it read to them while they slept. Researchers can overcome this type of problem by examining the difference between the performance of a comparison group or condition and an experimental group or condition. So Wood et al. (1992) made two comparisons. One comparison was between groups. Some participants were awake while the words were read to them, and some were asleep. Because people were randomly assigned to the groups, comparing how often the target words appeared in each of the groups showed whether people were more influenced by presentations while they were awake or by presentations while they were asleep. People who were awake during the presentations were more than twice as likely to report the target words as people who slept. This comparison shows that learning while awake is better than learning while asleep, but it does not rule out the possibility that the sleepers’ performance was influenced by the presentations. Multiple observations were made for each participant and then compared. There were actually two different lists of words – one included ‘a metal: gold’ and the other included ‘a flower: pansy’. Each participant was read only one of the lists, but all participants were tested on both categories. This allowed the experimenters to measure how often people produced words that had been read to them compared to words that had not been read to them.

There was no real difference between individuals’ subsequent reports of key words when the words had been read to them and when the words had not been read to them. The pair of bars furthest to the left provides the same comparison for people who were awake during the word presentations. It is pretty clear that if people were awake during word presentation, then the presentations of the lists had a big effect on subsequent memory for those key words.

OVERCOMING THE PROBLEM USING MEMORY

To address this problem, memory is often studied by comparing two groups of participants or information, organized such that the ‘past event’ occurs for one group but not for the other. Because the only known difference between the groups is the presence or absence of the event, differences observed at the later time are assumed to reflect memory for that event. It is therefore essential to determine that there are no other differences between the groups. The sleep learning experiment Suppose you played tapes of information to yourself in your sleep. Would you remember the information later? (For a review of ‘sleep learning’, see Druckman & Bjork, 1994.) To answer the question, you might present some information to people while they sleep, wake them up, and then observe whether their subsequent behaviour reflects any memory for that information. Wood, Bootzin, Kihlstrom and Schacter (1992) did just this. While people slept, the researchers read out pairs of category names and member names (e.g. ‘a metal: gold ’), repeating each pair several times. After ten minutes, the sleepers were awakened and asked to list members of named categories – such as metals – as they came to mind. The assumption was that if participants had any memory for having ‘a metal: gold’ read to them while they slept, then they would be more likely to include gold in their list of metals.