Programme

Evidence for cognitive training-induced far transfer is mixed at best. One notable exception appears to be training interventions targeting speed of processing. In this pre-registered, multi-site training study, we test the hypotheses that (a) training tasks with stronger attentional control demands will induce larger transfer effects, and that (b) gains in the rate of information accumulation (i.e., drift rate) will be positively associated with these effects. For this purpose, we are collecting data from 400 adults (18-85 years) from three sites who are randomly allocated to one of four groups practising tasks with increasing attentional control demands. Transfer to working memory, executive functions, reasoning, and everyday cognitive functioning is assessed before, immediately after, and 3 months after 10 training sessions. In this talk, we will present the first results from this study. 

Processing speed is a crucial ability that changes over the course of the lifespan. Training interventions on processing speed have shown promising effects and have been associated with improved cognitive functioning. While training-related changes in processing speed are often studied using reaction times (RTs) and error rates, these measures provide limited insight into the mechanisms underlying changes during training. The drift-diffusion model provides estimates of the cognitive processes underlying speeded decision tasks, such as the rate of evidence accumulation (drift rate), response strategies (boundary separation), as well as time for other processes such as stimulus encoding and motor response (non-decision time). In the current study, we analyzed existing data of an extensive multi-session training intervention (von Bastian & Oberauer, 2013) to disentangle changes in drift rate, boundary separation, and non-decision time during training of different speeded choice-RT tasks. During this training intervention, 30 participants performed 20 training sessions over the course of four weeks, completing three tasks each session: a face-matching, a pattern-matching, and a digit-matching task. Our results show that processing speed training increased drift rates throughout training. Boundary separation and non-decision time decreased mostly during the initial parts of training. This pattern of prolonged training-related changes in rate of evidence accumulation as well as early changes in response strategy and non-decision processes was observed across all three tasks. Future research should investigate how these training-related changes relate to improvements in cognitive functioning more broadly.

Age-related cognitive decline can present challenges to the quality of life of older adults. Cognitive training can be used to maintain cognitive abilities and to potentially reinforce processes of cognitive reserve and neuroplasticity. In this presentation, I will share findings from my research, which investigated the efficacy of strategic cognitive training targeting memory and attentional control in older adults. Furthermore, I will assess the underlying brain mechanisms and compare its effects with those of early life education using functional magnetic resonance imaging. Additionally, the presentation will expound upon practical applications, bridging the gap between efficacy, mechanistic research, and real-world implementation. Various approaches will be explored, including the integration of technology such as computerized programs, online platforms, and virtual reality. I will also discuss studies that leverage participatory and implementation science tools and methods, with the goal of enhancing relevance and maximizing the practical impact of cognitive training programs. 

Processing speed is considered central in the operation of higher cognitive functions, including memory, language, and decision-making. It is hypothesized that the decline in processing speed may contribute to the deterioration of higher-level cognitive functions commonly associated with aging. If this holds true, training in processing speed has the potential to mitigate cognitive decline in aging populations and reduce the associated functional impacts. Notably, research indicates that the benefits of such interventions vary among individuals. Consequently, our focus is on identifying populations that derive optimal benefits from this type of intervention. To achieve this, we employ mixed linear models to predict progress based on individual differences. Today, we present the overall study and preliminary data on the individual variables incorporated in our models. 

When two tasks are presented simultaneously or in close succession, dual-task performance on those tasks is usually impaired compared to separate single-task performance. Numerous theories explain these emerging dual-task costs in terms of the existence of capacity limitations in the constituent component tasks. In addition, the framework of dual-task coordination adjustment proposes active dual-task coordination processes that work on the scheduling of these capacity-limited processes. Further, there are recent findings that point to a meta-cognitive control level in addition to these active coordination processes. This additional level’s responsibility is to adjust the dual-task coordination of capacity-limited stages (i.e., coordination adjustment). I review evidence focusing on the existence of dual-task coordination processes and processes of coordination adjustment. The talk elaborates on preliminary findings, pointing to the separability of these sets of processes, which is a key assumption of the framework of dual-task coordination adjustment. In particular, this talk focuses on how training regimes can be used as a tool to test the assumptions of this framework. 

Research regarding the relationship between first-person shooter (FPS) gameplay and cognitive performance yields mixed results, perhaps due to methodological limitations. This online cross-sectional study examined the association between FPS expertise and processing speed, task mixing and switching performance. Our large sample of N = 235 Counter-Strike (CS) players completed a CS expertise questionnaire and were allocated into four cluster groups of expertise using k-means cluster analysis: Casual (N = 78), Experienced (N = 101), Aspiring (N = 22) and Semi/Professional (N = 34) players. A colour/shape switching task was also administered to measure processing speed, task mixing and switching performance. Significant main effects of the expertise cluster group were observed for processing speed however, no significant effects were found for mixing or switching costs. Drift-diffusion modelling of reaction times and accuracy scores revealed processing speed advantages to be due to faster non-decision times (encoding and response execution). Findings demonstrate that processing speed, but not task mixing or switching performance, was enhanced in CS players with greater expertise. 

Four popular visual working memory (VWM) models - the Standard Mixture Model (SMM, Zhang & Luck, 2008), Swap Model (Bays et al., 2009), Target Confusability Competition Model (TCC, Schurgin et al., 2020), and Signal Discrimination Model (Oberauer et al., 2021) – were previously compared in a VWM training study (N = 64) among young adults (aged 18–35 years). Our previous findings suggested that SMM best fitted when there were substantial changes that were induced by training or test-retest effects. In contrast, TCC was the best fitted model when substantial changes were absent. Using the same VWM paradigm but as a transfer task, the present study aims to replicate the previous patterns by comparing these models with a large sample (N = 408) completing a processing speed training intervention across the adult lifespan (aged 18–85 years). Preliminary results from N = 290 replicated the latter pattern. Furthermore, the young adulthood group (aged 18–39 years) had smaller recall errors compared to the middle adulthood group (aged 40–59 years) while the older adulthood group (aged 60–85 years) had the bigger recall errors. Although TCC were largely preferred by the middle adulthood group and the older adulthood group, TCC and SMM were closely preferred by the young adulthood group after training. The overall pattern of results cautioned the use of one single model when considering the presence or absence of training-induced changes as well as age-related changes. 

Working memory (WM) is highly correlated with fluid intelligence (Gf), but it is yet unclear why. Previous research has suggested that common attentional processes explain this relation without, however, distinguishing between attentional control and sustained attention. Therefore, the present study investigated whether sustained attention – attention fluctuation and lapses of attention – mediates the relationship between WM and Gf above and beyond attention control. N = 146 participants completed a battery of 12 tasks, with 3 tasks each assessing WM, Gf, attention control and sustained attention. Contrary to our expectations, preliminary analyses suggest that neither attention control nor sustained attention mediated this relationship; further analyses with latent-variable modelling techniques are currently ongoing. These preliminary findings challenge the prevailing theory that attention control is the primary driver of the WM-Gf association, highlighting the need for further research to better understand the underlying mechanisms at play.

The Three-Embedded Components (3Cs) model of working memory (WM) propositions the Focus of Attention (FA), representing  a single attended memorandum, the Region of Direct Access (RDA), representing memoranda bound to active context, and the Activated Long-Term Memory (aLTM), representing activated memoranda not bound to active context. We used structural equation modelling to investigate individual differences in the 3Cs and mediation analysis to see whether the 3Cs can relate WM capacity (WMC) and fluid intelligence (Gf). Additionally, to address reliability concerns regarding individual-differences investigations of experimentally derived variables, bi-factor analysis and homogenous measures were used to isolate variables of interest. Participants (N = 198) completed tasks to assess WMC, each of the 3Cs, and Gf across verbal-numerical and visual-spatial domains. Individual differences in each of the 3Cs were identified, but model fit was slightly improved when FA and aLTM were combined. Neither any of the 3Cs nor the combined FA – aLTM variable mediated the WMC – Gf relationship. All these candidates regressed significantly on WMC but not Gf. These results suggest that individual differences specific to the individual components of the 3Cs model of WM cannot account for the WMC-Gf relationship. Also, these components represent constructs of WM that do not relate to Gf.