The episodic buffer

for an array of findings that could not be accommodated easily by the original three-component model (Baddeley & Hitch, 1974). For instance, there was strong evidence to suggest that long-term memory can play a role in supporting

short-term and working memory (e.g., Chase & Ericsson, 1981), and that under some circumstances temporary storage of materials in quantities that exceed the capacities of the short-term stores can be achieved (e.g., Baddeley, et al., 1987).

The episodic buffer was proposed as a limited capacity system capable of binding information from both working and long-term memory in a unitary episodic representation. One key suggestion was that the buffer stores

information in a multimodal code making it capable of integrating information from various resources.

1.2.3: Measures of short-term and working memory

Short-term and working memory both involve temporary storage, but are distinguished by whether or not significant processing activity is required

concurrently. The tasks assumed to tap these resources differ along similar lines:

short-term memory tasks impose storage but minimal processing demands, whereas working memory tasks engage the participant in significant processing activity in addition to storage. Short-term memory tasks typically involve the immediate recall of information, and may employ either serial or free recall, serial recognition, or recreation of a pattern. The most widely employed measures of working memory are complex memory span paradigms, in which participants engage in some form of processing activity such as reading sentences or performing mental rotation, and simultaneously maintain

information for subsequent recall. Typically, these tasks are administered in a span procedure aimed at measuring capacity of the resource by increasing the sequence length of trials until recall errors are made. Verbal and visuospatial measures have been developed for both short-term and working memory in line

with the research reviewed above indicating domain-specific resources in short-term memory (see section 1.2.1), and raising the possibility of domain-specific resources in working memory (see section 1.2.2).

Conventional measures of verbal short-term memory include serial recall of words, letters or digits (e.g., Conrad & Hull, 1964). It must be noted that long-term memory may play a role in such tasks (i.e., the word frequency effect and the imageability effect; see section 1.2.1.1 above). Nonword repetition is an additional measure in common use that requires the repetition of novel

phonological forms such as /!"#$%&'()*/. It has been suggested that nonword repetition provides a relatively pure index of verbal short-term memory because of the reduced availability of long-term lexical knowledge to support the

unfamiliar phonological forms (e.g., Gathercole & Baddeley, 1989, 1993). It has proved a simple and effective task with valid measurements reported for children as young as two years of age (Roy & Chiat, 2004). Even nonword recall, however, may tap long-term knowledge: nonword repetition performance has been found to be influenced by the wordlikeness effect (e.g., Gathercole et al., 1991; Nimmo & Roodenrys, 2002), the phonotactic frequency effect (e.g., Munson, 2001; see also section 1.2.1.1 above), and by the prosodic pattern of a nonword (e.g., Dollaghan, Biber, & Campbell, 1995; Roy & Chiat, 2004).

It should be noted that this interpretation of nonword repetition is not universally held. Alternative accounts suggest that nonword repetition taps other cognitive processes including lexical knowledge (Snowling, Chiat, &

Hulme, 1991), phonological sensitivity (e.g., Bowey, 1996; Metsala, 1999;

Reuterskiold-Wagner, Sahlen, & Nymen, 2005), and output phonology (e.g., Sahlen, Reuterskiold-Wagner, Nettelbladt, & Radeborg, 1999; Wells, 1995).

Understanding of the cognitive processes that support nonword repetition is an important issue in this thesis and will be explored in some detail, predominantly in chapters 5 and 6.

Visuospatial versions of short-term memory tasks involve the retention of either visual patterns or sequences of movements (e.g., Smyth & Scholey, 1996;

Wilson, Scott & Power, 1987). One challenge in designing visuospatial tasks generally is that many individuals have a tendency to recode visuospatial

information verbally (i.e., describing a shape to themselves with the verbal label

‘circle’), and once the information has been entered into verbal short-term memory the task is no longer a visuospatial measure. In this area, then, it is particularly important to make use of well-designed, validated measures, as was the case in this thesis.

Domain-specific complex memory measures have been developed to assess working memory as well. An example of a verbal complex memory task is reading span, in which the participant is asked to make a meaning-based judgment about each of a series of sentences and then remember the last word of each sentence in sequence (e.g., Daneman & Carpenter, 1980). A corresponding visuospatial task is spatial span, in which the participant is asked to judge the orientation of a set of letters, and then remember the sequence of degrees of rotation of the letters (Shah & Miyake, 1996). Within the working memory model, the storage demands of complex memory tasks are suggested to depend on appropriate short-term subsystems, with processing supported principally by the central executive (Baddeley & Logie, 1999; Cocchini, Logie, Della Sala, MacPherson, & Baddeley, 2002).

Several proposals exist concerning the cognitive processes that may be engaged in complex memory tasks. One account consistent with the working memory model (Baddeley & Hitch, 1974) is that of task-switching (Towse &

Hitch, 1995), according to which individuals alternate between processing and storage aspects of the task. Increased processing demands have the effect of extending the time over which items may be forgotten. A more explicit view has been advanced by Barrouillet and Camos and colleagues (Barrouillet, Bernadin, & Camos, 2004; Barrouillet & Camos, 2001; Gavens & Barrouillet, 2004), who suggest that performance on complex span tasks is constrained by a limited attentional resource that is required to support both processing activities involving memory retrievals and item storage. Barrouillet et al. introduced the notion of cognitive load – the extent to which attention is switched away from maintenance to retrieval during a particular period – as the crucial determinant of complex memory span. Still another proposal comes from resource sharing accounts of working memory, according to which the limited capacity resource pool is employed for both processing and storage (e.g., Daneman & Carpenter, 1980; Just & Carpenter, 1992), and individual differences in task performance arise due to differences in processing speed.