Latent Cognitive Effects from Cognitive Training in the Elderly: An Experimental Psychometric Approach to the Fadeout Effect

When an intervention raises intelligence in children, the effects are temporary, lasting around two to five years as an upper limit. Two explanations are put forward to explain this fadeout effect: one being that the... [ view full abstract ]

When an intervention raises intelligence in children, the effects are temporary, lasting around two to five years as an upper limit. Two explanations are put forward to explain this fadeout effect: one being that the intervention group loses their gains, the other being that the control group catches up. These two explanations diverge in their predictions of what happens to intervention effects in the elderly. The fading explanation predicts that intervention effects will be temporary due to the experimental group losing their gains, the catching up explanation predicts that intervention effects will be permanent due to the control group no longer experiencing an increase in cognitive abilities. We test these competing explanations in a large randomized controlled trial that randomly assigned 2382 elderly adults (age=73.6) to either memory, reasoning, speed of processing training or a control group. Due to similarities in the training and posttests (including instances of exact teaching to the test) we investigate latent effects using measurement invariance testing within a multi-group confirmatory factor analysis framework. Results show local effects of memory training on latent memory ability (β=.149, 95%CI=.089 to .208) and reasoning training on latent reasoning abilities (β=.051, 95%CI=.007 to .095). There was no transfer across abilities or upwards to g. One year after the intervention ended, there were no more effects in any of the groups at the latent level. Thus, the fadeout effect occurs because intervention groups lose the gains they made. Booster training was able to preserve latent memory abilities for longer, but again, not permanently. We discuss implications for the fadeout effect, the structure of intelligence, and future training studies.