Formulación del problema

There has been much debate regarding the structure of memory. Some argue that memory is a unitary system made up of one store (Melton, 1963). Others argue that memory comprises functionally separable systems: one termed short-term memory (STM), which is resource-limited and required for temporary storage and processing of information, and another, termed long-term memory (LTM), which is potentially unlimited in size and required for permanent storage and knowledge (e.g., Waugh & Norman, 1965). The debate regarding the structure of memory continues today. However, there is now substantial neuropsychological and behavioural data to support a dissociation between STM and LTM (e.g., Milner, Gorkin & Teuber, 1968; Shallice & Warrington, 1970).

Evidence supporting a dissociation between STM and LTM led to the proposal of a number of information processing models of memory. Atkinson and Shiffrin (1968)

proposed the modal model, which was a development of Broadbent’s (1958) model of

selective attention. According to the modal model, information enters the processing system via modality dependent sensory stores. This information is then passed into a short-term store and finally into a long-term store. The short-term store was assumed to be an active part of the system involved in processing information en route to the long-term store and incorporating a number of control processes, such as rehearsal, coding and retrieval. Atkinson and Shiffrin also proposed that the short-term store was involved in many aspects of human cognition and should therefore be regarded as a ‘working memory’.

However, Atkinson and Shiffrin’s model could not readily account for some aspects of the neuropsychological data. Shallice and Warrington (1970) identified a patient K.F. whose STM skills were impaired, yet his long-term learning was normal. Furthermore, K.F.’s STM impairment did not affect his performance of everyday activities. Until this time it had been assumed that any STM system was a unitary storage system of limited capacity. It was also assumed that STM played a fundamental role in a number of cognitive processes such as reasoning and comprehension. That is, all sensory information must be processed via STM. However, KF had impaired digit span yet his general cognitive processing was intact. Baddeley and Hitch (1974) explored this discrepancy by attempting to create, an

analogue of KF’s performance in normal subjects by ‘filling STM’ with a demanding task (dual task paradigm).

The digit span task had traditionally been regarded as an index of STM ability. This task requires immediate recall of digits in the order presented. If performance on a cognitive task, such as reasoning, relies entirely on STM processes, then requiring subjects to retain digits and perform a reasoning task simultaneously should impair reasoning performance. When subjects were required to recall a small digit load (three

digits) there was little or no effect on reasoning. When the load was increased to six digits performance on the reasoning task, although impaired, was still functional. This suggested that there was a “ ..considerable component of working memory which is not taken up by the digit span task” (Baddeley & Hitch, 1974, pg. 75). That is, STM may comprise of multiple systems, each related to a specific domain. Support for this interpretation came from STM patients such as PV whose memory impairment differed across different aspects of STM. PV’s recall o f auditory stimuli was substantially poorer than her recall of visually presented stimuli (Basso, Spinnler, Vallar & Zanobio, 1982).

To account for these data Baddeley and Hitch (1974) proposed the ‘Working Memory’ model. In this model STM was fractionated into a number of subsystems: a control system, termed the central executive, and a number of modality-specific slave systems (verbal and spatial). According to this model, patients such as PV and KF were able to perform everyday activities because they had only suffered damage to the part of the STM system that dealt with verbal material, yet the rest of their STM system was intact. The working memory model was also able to account for the differential effects of memory loads of three and six digits on reasoning ability. When verbal memory load exceeds the capacity of the verbal slave system, central executive processes are recruited to help retain the sequence. These processes were thought to be employed in cognitive tasks such as reasoning. Therefore when a large memory

load was imposed performance on the cognitive task was impaired. ,

In the UK, the term ‘working memory’ has become synonymous with the model of STM proposed by Baddeley and Hitch (1974), later revised to account for new findings (Baddeley, 1986). However, the term is used elsewhere to refer to rather different concepts (for a review of the use of the term see Richardson, 1996). In North America ‘working memory’ is often used to refer to a general processing resource that co-ordinates the functions of STM and LTM in relation to a specific task such as text comprehension. Although it is necessary to be aware of these different uses of the

term working memory, in this thesis the term working memory is only used to refer to the model proposed by Baddeley and Hitch (1974; Baddeley, 1986).

The working memory model has remained particularly influential, especially in the UK. It has generated much research, which has led to refinements in the model itself (Baddeley, 1986). It has also inspired the development of new models, particularly of phonological STM (see Gathercole, 1996; 1997; Chapter 7 in this thesis), which have their roots in the working memory model (e.g., Burgess & Hitch, 1992).

2.3. The working memory model (Baddeley & Hitch, 1974; Baddeley,