Stone, James Michael and Towse, John (2015) How does practice affect working memory? : The efficacy of adaptive-difficulty working memory training programs. PhD thesis, Lancaster University.
2015stonephd.pdf - Published Version
Available under License Creative Commons Attribution-NoDerivs.
Download (6MB)
Abstract
Working memory refers to a mental ’workbench’ whereby new or goal relevant information is held in a readily accessible state in order to achieve success with cognitive problems. Working memory has been shown to be relevant to individual differences in many aspect of cognition including fluid intelligence, while also identified as a core deficit in cognitive developmental disorders. Therefore the possibility that working memory capacity is trainable, and that such interventions produce generalisable cognitive benefits is highly noteworthy. Following these initial claims from early studies using an adaptive working memory training intervention, commercial products have been developed and marketed, based on the premise that working memory can indeed be trained and lead to transfer to a wider range of cognitive abilities. A close examination of the literature suggests that these claims are based on a combination of mixed generalisable results and often a lack of evidence for genuine working memory improvement. In some examples the analyses testing for working memory improvement fail to show such effects while in others the chosen assessment tasks are too similar to the trained tasks to be evidence for general working memory improvement. The potential for improved fluid intelligence and amelioration of deficits leading to developmental issues due to a working memory training intervention is of clear practical importance. However, for such arguments to be made convincingly it is of critical importance that there is increased understanding of the effects of working memory training on the construct itself. Thus far the effects seen in the literature have not proved to be robust and therefore the mechanisms of any proposed effect need to be examined to illuminate the conceptual and empirical benefits of such training interventions, whilst also establishing their robustness and reliability. In addition, there is reason to believe that tighter methodological designs are needed to make the domain more credible with particular emphasis on suitable active control groups. Therefore this thesis pursued a series of studies assessing the efficacy of adaptive difficulty WM training interventions with an emphasis on the impact on the working memory construct (near-transfer). Thus, the assessment of potential improvements in working memory formed the core analyses within this thesis. Therefore, in addition to the three working memory training studies, this thesis will also addressed various methods of assessing working memory performance from typical behavioural assessments, so as to inform the methods utilised in the pre-post changes in the training studies. This was achieved by means of analysing alternative scoring methods, assessment protocols, and modelling the difficulty of different list length trials using a Rasch model in working memory tasks. Each training study utilised a randomised pre-post (experiment one also incorporated a follow-up phase) intervention design where the control group completed a demanding regime of tasks with minimal stress on working memory. In each experiment participants completed a battery of tasks prior to the onset of the training phase that lasted 5-6 weeks. During the training phase an intervention group would complete the specified training regime while an active control group would complete different tasks. Following the conclusion of the training phase the initial battery of tasks was re-administered (experiment one also incorporated a follow-up phase). Additionally, all software used was developed specifically for this project including control group tasks therefore maintaining consistency in the ’look and feel’ between the two groups. In experiment one (N = 55) an intervention using a variety of working memory based tasks (Working Memory Period, Memory Updating, Colour Corsi, and Stroop) was assessed and compared with an active control intervention in children aged 9-10. No improvements were found in the individual tasks measuring verbal and visuo-spatial working memory, nor were there training effects. In experiment two (N = 76) children aged 9-11 participated in three working memory interventions with each consisting of a single task (Working Memory Period, Colour Corsi, and N-Back). No improvements were seen in any training group beyond what was also seen in an active control group in composite measures of verbal and visuo-spatial working memory, or processing speed. A final working memory training experiment (N = 55) replicated the results from experiment two (Dual N-Back replacing standard N-Back) using a sample of healthy adults. These data suggest, in typically developing children and adults, that adaptive working memory training interventions may not improve working memory functions. These results cast doubt on the potential for such interventions to produce improved performance in a wider range of higher order cognitive abilities. Potential reasons for the existence of a number of positive results in the literature are considered including ineffective controls, publication bias, and potential false positives as a consequence of multiple comparisons - with regards the number of tasks used in a pre-post battery in addition to the number of measures one can extract from each task compounded by the possibility of analysing these data using multiple methods.