Distractor filtering and task load in working memory training in healthy older adults

Abstract

Filtering efficiency (FE) has been suggested as a task-related process of working memory (WM) in older adults, but the interaction of distractors with WM when implemented as a training to improve WM capacity is unclear. To investigate the effect of manipulations of WM and/or FE load in a multicomponent model-based WM training in improving WM capacity. 205 healthy older adults (129 women; aged 64.0 ± 8.3 years) were randomised into three WM training groups with manipulations of WM and FE load (high WM; low FE load, high WM; high FE load and low WM; high FE load). All groups underwent three online cognitive testing sessions and twelve 40-min training sessions over three weeks using the SmartBrain smartphone application. WM capacity was measured with complex span tasks and FE using the Change Detection Task. Linear mixed-effect models adjusted for age, sex, education and pre-baseline performance showed increased WM capacity post-training, but no difference between groups. While both distractors and WM are needed to improve WM in the absence of distraction, manipulations of the FE and WM load do not influence improvements following training.

Keywords: Filtering efficiency; Healthy aging; Training; Working memory.

Fig. 1

Flowchart of study procedures.

Fig. 2

Visual and spatial sketchpad tasks of the MB⁺⁺ training (high WM and high distractor load). Memory array contains the to-be-remembered items (in panel A, fully filled squares; panel B, sequence of planet objects) and distracting items (panel A, shaded squares; panel B, satellite object which lights up in sequence but has to be ignored). The irrelevant information tab contains the visually loaded (panel A) and haptic (panel B) distractor. In the recall array, the patten of the filled squares without distractor square (panel A) or correct sequence of to-be-remembered items (panel B) has to be filled in. The MB⁺⁻ has the same set up, without the shaded square (panel A) or distracting item (satellite) in panel B. The MB⁻⁺ training has the same set up as the MB⁺⁺, but WM load (increasing sequence length or level) is kept below a threshold.

Fig. 3

Flow chart of recruitment and study inclusion.

Fig. 4

Mean and standard error of the complex span composite score for each training group (MB⁺⁺, WM high, FE high; MB⁺⁻, WM high, FE low; MB⁻⁺, WM low, FE high) over the three sessions. S1: pre-baseline, S2: baseline, S3: post-training.