Working memory training improves episodic memory in older people: transfer based on controlled retrieval processes.

Abstract

The results of working memory (WM) training to improve episodic memory in older people are inconsistent. This inconsistency could be due to the fact that the episodic memory tests used do not share the same cognitive resources as the trained WM task. The aim of this study was to assess whether performance on an episodic memory test will improve only when this test requires controlled processes of retrieval of information from secondary memory or recollection, similar to the processes exercised during WM training. Fifty-five people over 60 years of age participated in the study: 27 were randomly assigned to the experimental group (EG) and the rest to the control group (CG). The EG was trained in complex span tasks. Before and after training, both groups were tested on episodic memory tests (a verbal and a visuospatial recognition test) and WM span tasks (reading, digit and spatial location). ANOVAs revealed a greater improvement of recollection estimates in the EG than in the CG for both verbal recognition (p = 0.023) and visuospatial recognition (p = 0.014). Our results provide support for a cognitive mechanism whose shared presence favored transfer from training on a WM task to a test of episodic memory. Consistent with our predictions, training on complex span tasks improved performance on recognition tests only when recall required a controlled search process in secondary memory, or recollection. We therefore stress the importance of identifying other cognitive resources that are susceptible to transfer from a training task to other untrained tasks. A better understanding of the phenomenon of transfer is crucial for the design of increasingly effective intervention programs for older people.