Introduction: Efficient learning is critical to adapting to different environments. There are well-known learning principles in cognitive rehabilitation, including errorless (EL) and trial-and-error (T&E) learning; however, little is known about their underlying neural mechanisms. In the current study, to understand the age-related changes in learning benefits and neural mechanisms applying EL and T&E learning methods in healthy middle-aged adults, we conducted a graph theoretical analysis using functional magnetic resonance imaging data and analyzed the relationship between learning benefits and age, as well as functional network connectivity and age, with both learning principles.
Method: A total of 43 participants performed a color-name association task through EL and T&E learning methods. We focused on the functional connectivity patterns of the default mode network (DMN) since previous studies demonstrated this network to be more distinctive and important for the T&E learning method than EL. Within-network functional connectivity was used as the graph metric.
Results: Age showed significant moderate negative correlations with T&E scores and within-DMN functional connectivity in the test state following T&E learning. Conversely, age was not significantly correlated with EL scores or within-DMN functional connectivity in either the EL learning or test states.
Conclusions: Our findings demonstrate the age-related learning decline associated with decreased DMN integration with aging, when applying the T&E method but not the EL method, even in healthy middle-aged adults. Relationships between the underlying neural network and age are different depending on the learning method. This suggests the need to take into consideration the remaining learning ability through the T&E learning method compared to normal aging and to utilize residual DMN functioning, in addition to the comparison between score differences between EL and T&E methods, when tailoring an individual learning approach.
Keywords: Aging; cognitive rehabilitation; default mode network; errorless learning; functional magnetic resonance imaging; trial-and-error learning.