Mnemonic-trained brain tuning to a regular odd-even pattern subserves digit memory in children

Human memory benefits significantly from mnemonic techniques. Two hypotheses explain mnemonic training's effectiveness: the processing-efficiency hypothesis, suggesting less cognitive resource consumption after training, and the encoding-pattern hypothesis, proposing altered encoding patterns. The processing-efficiency hypothesis posits that extensive training leads to less effortful encoding and decreased neural activity, with lower neural dynamics correlating with better performance. The encoding-pattern hypothesis, supported by studies on world-class mnemonists using the digit-image method, suggests that mnemonists spend more encoding time on even-position digits due to their image-mapping strategy, resulting in stronger associations. This study proposes four theoretical models to depict neural activity concerning encoding patterns (odd vs. even positions) before and after training, considering both hypotheses. The study recruited children, as their low memory capacity often hinders educational progress, to investigate whether mnemonic training improves processing efficiency and/or alters encoding patterns to enhance memory. The study aimed to determine if 22 days of digit-image mnemonic training would significantly improve memory, induce changes in P200 (related to attention and mental imagery), alter theta power (linked to memory and mental imagery), and assess generalization to other cognitive abilities.

Prior research extensively demonstrates the effectiveness of mnemonic training in enhancing memory performance. Studies have explored the use of various mnemonic techniques, such as the Method of Loci and the digit-image method, showing improvement across different age groups and cognitive domains. The processing-efficiency hypothesis is supported by findings indicating reduced neural activity in experts compared to novices. Conversely, research on world-class mnemonists employing the digit-image method reveals altered encoding patterns, with more time spent on even-position digits, signifying stronger image associations. These studies highlight that mnemonic techniques can improve memorization either generally or specifically within a task, leading to either reduced neural activity or specific encoding pattern changes. Previous research using EEG has also implicated P200 and theta rhythms in mental imagery and memory processes, suggesting their potential involvement in mnemonic training.

This longitudinal study involved 41 children (22 boys), with no prior memory training experience, divided into mnemonic training (MT) and no-contact control (NC) groups based on interest. All participants underwent four sessions: pre-training tests (EEG task and cognitive test battery), a 22-day training period (MT group only), post-training tests (same as pre-training), and follow-up tests (four months post-training, subset of cognitive tests). The EEG task involved encoding and recalling 20 digit sequences; discriminability (d') measured performance. The cognitive test battery assessed various cognitive abilities (perceptual speed, inhibitory control, working memory, short-term memory, episodic memory, reasoning, spatial imagination, and divided attention). The MT group received intensive digit-image mnemonic training at a World Memory Championships Training Camp (6+ hours daily). EEG data (64 channels) were recorded during the EEG task, pre-processed (downsampling, artifact removal, re-referencing), and analyzed for ERPs (P200) and oscillatory power (theta). Statistical analyses involved ANOVAs (with Group and Session factors) for behavioral data and repeated-measures ANOVAs (with Position, Session, and Area factors) for EEG data. Correlation analyses examined the relationship between neural activity (P200 and theta power effects) and memory increment. Validation analyses using cluster-based methods were performed on ERPs and oscillatory power to investigate any general decreases in neural activity post-training.

Post-training, the MT group showed significantly higher memory performance (d') in the EEG task compared to the pre-training phase, while the NC group showed no significant difference. The MT group also showed a significant interaction effect on the number-noun pairs test in the cognitive battery, demonstrating better performance post-training, persisting in the follow-up. However, other cognitive abilities showed limited enhancement. EEG analysis revealed a significant Position x Session interaction in the MT group for P200 amplitude: post-training, even-position digits elicited greater P200 amplitude than odd-position digits, which was not observed in the NC group. Similarly, a significant three-way interaction (Position x Session x Area) was found for theta power in the MT group, with increased theta power for even-position digits in the left anterior area post-training, absent in the NC group. Crucially, the P200 effect and left anterior frontal theta power effect significantly correlated with memory improvement in the MT group only. Validation analyses showed no significant overall reduction in neural activity post-training except for a significant increase in alpha power (9-12 Hz) in the 1000-1400 ms window in the MT group, which did not correlate with memory improvement.

This study provides strong support for the encoding-pattern hypothesis, demonstrating that mnemonic training alters encoding patterns, leading to improved digit memory in children. The observed regular odd-even neural patterns (enhanced P200 and theta power for even-position digits) reflect the trained strategy of pairing digits and associating them with images, and this pattern directly predicted memory gains. The results suggest that both early (P200) and sustained (theta power) neural processing contribute to the mnemonic training effect. The limited generalization to other cognitive tasks aligns with previous findings suggesting domain-specificity of mnemonic training effects. While the processing-efficiency hypothesis was not strongly supported by this study's analyses, it is possible that more sophisticated network analyses could reveal such effects. The observed increase in alpha power warrants further investigation.

This study provides direct electrophysiological evidence of how mnemonic training alters encoding patterns in children's brains to improve digit memory. The observed regular odd-even neural activity (P200 and theta power) is functionally relevant to memory enhancement. Although limited generalization to other cognitive domains was found, the study highlights the efficacy of digit-image mnemonics in memory improvement. Future research should explore other mnemonics and utilize network analyses to further elucidate the neural mechanisms underlying mnemonic training.

The study's limitations include a potentially limited sample size that may have affected statistical power, especially for examining interactions. Group assignment based on interest might have introduced pre-existing differences in learning abilities. The lack of a main Group effect in some analyses could be due to insufficient power to detect such effects. Future studies should increase sample size and use more rigorous group matching techniques to address these limitations. The primary focus on planned contrasts in EEG analysis, rather than group comparisons, also limits generalizability and calls for replications with more comprehensive analyses.