The central executive

responsible for the imaginability effect on recall. Evidence from neuropsychological studies suggests separate visual and spatial components of imagery (cf. Baddeley, 1997).

a capacity limited attentional controller, they conducted a series of studies to examine the executive function of normal subjects and patients with Alzheimer’s disease or dysexecutive syndromes. In the experiments, a dual-task paradigm was employed where subjects had to perform a pursuit tracking task, which relies on the visuo-spatial sketchpad, plus a digit span task, which depends on the phonological loop. It turned out that in comparison to normal subjects including elderly and young people, the performance of the patients with Alzheimer’s disease was significantly worse. However, when the two tasks were performed independently, there was no evidence that those patients’ performance was differently affected by increasing task difficulty. Since Alzheimer patients are characterized by an impairment of episodic memory and attentional deficits, the difference in single and dual task performance suggests a separable executive capacity to allocate attention to coordinate two tasks (see Baddeley et al., 1991, 2001; Logie et al., 2000). It is important to bear in mind that, with normal subjects, a performance decrement in dual tasks is quite normal; it indicates that the capacity to coordinate information from the slave systems is limited (Bourke et al., 1996; Baddeley, 1997).

In addition to the capacity to coordinate two tasks, Baddeley (1996, 2001) suggested that the central executive also has the capacity to: 1) switch retrieval strategies used for the random generation task; 2) to selectively focus attention on one stimulus and inhibit the disrupting effect of others, which is based on the assumption that “anything that limited attentional capacity would impair performance” (Baddeley, 2001: 856); and 3) to hold and manipulate information from long-term memory, which is reflected in the working memory span (Daneman and Carpenter, 1980). Yet it remains an open question whether the central executive should be regarded as “a single coordinated system that serves multiple functions, a true executive, or a cluster of largely autonomous control processesan executive committee” (Baddeley, 1996: 26). In a recent article, Baddeley (2001) has suggested that the central executive involves the capacity to focus attention as well as the capacity to divide and switch attention (Baddeley et al., 2001), but it has no capacity for storing information (Baddeley and Logie, 1999) and might not be much involved in retrieval from long-term memory (Baddeley et al., 1984; Craik et al., 1996). As to the last assumption, retrieval from LTM was tested by means of a dual-task experiment. Subjects were required to perform a demanding secondary task while learning or retrieving lists of words. The concurrent load from the secondary task did affect learning, but it had little effect on recall.

In Baddeley’s working memory model (1986), the central executive is supposed to be capable of combining the information from working memory with that from LTM. Yet the functions of the central executive described so far have not

yielded any concrete information about how the central executive and LTM interact.

There are some other phenomena that cannot be well explained by the current model provided that the central executive is purely an attention system without any storage capacity of its own (Baddeley, 1996; Baddeley and Logie, 1999). Firstly, some patients with dense amnesia have been able to use a chunking strategy to perform well in a task involving immediate recall of prose. Secondly, the model does not explain how and where the central executive combines the verbal and the visual information from the two subsystems. To overcome these problems, Baddeley (2000) has modified his model by adding a fourth component the episodic bufferto the working memory system (see Figure 6).

The episodic buffer is assumed to be the place where information from the subsystems of working memory and that from LTM are integrated. “It is assumed to be episodic in the sense that it holds integrated episodes or scenes and to be a buffer in providing a limited capacity interface between systems using different codes”

(Baddeley, 2001: 858). The integration of different codes from the two subsystems and from LTM is explained by assuming that the buffer uses a kind of common code.

It is capable of chunking information and storing it coherently in a multimodal fashion. Moreover, the episodic buffer is supposed to depend heavily on the central executive because there is no direct link between the buffer and the phonological loop as well as the visuo-spatial sketchpad. The integration of information from the two subsystems and that from LTM is still mainly controlled by the central executive. In my opinion, the buffer does exist because it provides a workspace where the integration of information as well as the influence of LTM in the process of information processing (e.g., chunking) may take place and also provides a temporary storage for the integrated information. With the assumption of an episodic buffer, the problems of the previous model can be solved.

It is further assumed that information is retrieved from the buffer through conscious awareness. “This allows multiple sources of information to be considered simultaneously, creating a model of the environment that may be manipulated to solve problems and plan future behavior” (Baddeley, 2001: 858). The episodic buffer, in my view, takes over some functions that previously had been implicitly ascribed to the central executive. Therefore, the buffer may actually be regarded as a fraction of the central executive which carries out the information processing but leaves other executive functions to the central executive. “The executive is now assumed to be a purely attentional system whose role extends beyond memory function, whereas the episodic buffer is assumed to be purely mnemonic in character” (Baddeley, 2001: 858).

Figure 6: The revised working memory model (Taken from Baddeley, 2000: 421)

Another noteworthy update to Baddeley’s working memory model is that there are direct links between the phonological loop and the long-term verbal memory as well as between the visuo-spatial sketchpad and the long-term visual memory. Evidence that supports this assumption is based on studies which show that knowledge or previous experience stored in LTM is also involved in the processing of information within the subsystems of working memory. For example, non-words that resemble English in their phonotactic structure are easier to remember than those differing from English (Baddeley, 1996; Adams and Gathercole, 1995; Baddeley et al., 1998), which suggests that the working of the phonological loop is sensible to LTM retrieval. Moreover, the links between the two subsystems and LTM indicate that the transmission of information between them can be carried out directly without a bypass through the episodic buffer or the central executive. That is, the interface function between working memory and LTM is not exclusively taken over by the central executive as suggested in the previous model, nor is it solely controlled by the episodic buffer.

As an addendum to my outline of Baddeley’s model of working memory, I would like to point out that there are several alternative accounts of working memory to be found in the literature (see Miyake and Shah, 1999). However, I have concentrated on Baddeley’s model because it is widely accepted, well supported both by psychological and by neurophysiological evidence, and is relevant for the purpose of my study. Its recent developments emphasize the role of LTM in information processing, a point that has largely been ignored before. In my opinion, the greatest challenge in memory research is to find the boundary between working memory and

LTM. This is an issue that needs to be solved in the future. In the following section, I shall address the role of LTM in learning.