The Development of Children's Early Memory Skills: Longitudinal Multi-Task Study Notes

Authors and Document Information

• Authors: Catherine A. Haden (Loyola University Chicago), Peter A. Ornstein (University of North Carolina at Chapel Hill), Barbara S. O’Brien (Loyola University Chicago), Holger B. Elischberger (Albion College), Caroline S. Tyler (University of North Carolina at Chapel Hill), and Margaret J. Burchinal (University of North Carolina at Chapel Hill).
• Title of Study: The Development of Children’s Early Memory Skills.
• Publication Context: Journal of Experimental Child Psychology ($2011$), Volume $108(1)$, pages $44-60$.
• Key Subjects: Cognitive development, Event memory, Deliberate memory, Nonverbal memory, Language skills, Longitudinal study, Working memory.

Theoretical Background and Study Motivation

• Mnemonic Competence in Infants: Young children demonstrate significant memory abilities long before they possess the verbal skills to describe past experiences. Nonverbal tasks like conditioning and elicited/deferred imitation have documented this competence.
• Scientific Contributions of Memory Literature:
  • Contributed to a reappraisal of infant cognitive competencies.
  • Suggested developmental continuities in basic memory systems.
  • Explored critical constructs like reinstatement.
  • Provided a foundation for addressing "infantile amnesia."
• Identified Gaps in Current Literature:
  • Lack of Longitudinal Data: Most studies are cross-sectional, making it difficult to describe the developmental course within individual children or relate early skills to later abilities.
  • Single-Paradigm Focus: Memory assessment typically emphasizes one paradigm. Comprehensive characterization requires multi-task assessments that vary in information processing and effort requirements.
• Study Uniqueness: This study couples a longitudinal design with a multi-task assessment strategy, tracking memory and linguistic skills across the first 6 years of life, focusing here on ages $18$, $24$, and $30$ months.

Principal Hypotheses and Goals

• Main Objective: To determine how children’s abilities to reproduce modeled event sequences (Elicited Imitation) are associated at each age with language, deliberate memory, and working memory.
• Developmental Linkages: To examine if the patterns of association among these skills change over the course of development from $18$ to $30$ months.
• Rationale for Elicited Imitation: This task is considered a nonverbal analogue to explicit memory. Toddlers' imitation is based on processes comparable to older children's verbal recall and deliberate memory preparation.
• Potential Predictors:
  • Working Memory: Capacity may set fundamental limits on representing and reproducing action sequences.
  • Deliberate Memory: Processes involved in strategies (naming, pointing, seeking) may underlie imitation performance.
  • Language: Narrative descriptions during modeling might help children attend to salient aspects, make sense of events, and form more robust representations.

Participants and General Procedure

• Initial Sample: $60$ children ($30$ in North Carolina, $30$ in Illinois; $33$ female, $27$ male).
• Final Sample: $57$ children (loss of $3$ due to moving or loss of contact).
• Demographics: Middle to upper-middle class; $50$ European Americans, $3$ Asians, $2$ African Americans, $2$ American Indians.
• Maternal Education: $12\%$ some college; $50\%$ college degree; $38\%$ advanced degree.
• Assessment Intervals:
  • Visit 1 to Visit 2: Delay of approximately $14.03$ days at $18$ months; $21.12$ days at $24$ months; $22.33$ days at $30$ months.
  • Visit 2 to Visit 3: Delay of approximately $1.1$ to $2.3$ days.
• Age at Sessions:
  • $18$ months: $18$ months $10$ days.
  • $24$ months: $24$ months $18$ days.
  • $30$ months: $30$ months $21$ days.

The Elicited Imitation Task

• Task Description: Children use props to reproduce a sequence of actions modeled by an examiner.
• Sequence Difficulty:
  • $18$ months: $3$ steps (E.g., making a gong, bunny, party hat).
  • $24$ months: $5$ steps (E.g., picnic, train, tugboat, garden).
  • $30$ months: $7$ steps (E.g., trucks, snow, birthday art, school).
• Types of Sequences:
  • Arbitrary: No inherent constraints on the temporal order of actions.
  • Enabling/Mixed: Actions must be performed in a specific order to reach an end state.
• Procedure:
  1. Baseline: Child manipulates props before modeling.
  2. Modeling: Researcher demonstrates and narrates the sequence twice.
  3. Immediate Test: Child is encouraged to reproduce the actions immediately.
  4. Delayed Test: Child is prompted to reproduce actions after a $2-3$ week delay without further modeling.
• Scoring Variables:
  1. Total individual target actions produced (max = $3$, $5$, or $7$).
  2. Number of pairs of actions produced in the target order (max = $2$, $4$, or $6$).
• Reliability: Average agreement between coders was $95\%$, $95\%$, and $94\%$ at $18$, $24$, and $30$ months.

Deliberate Memory: The Hide-and-Seek Task

• Procedure: Modeled after DeLoache et al. ($1985$).
  • Researcher hides objects in various room locations while the child watches.
  • Objects hidden: $2$ objects ($18$ months), $3$ objects ($24$ months), $4$ objects ($30$ months).
  • Delay: $2$ minutes ($18$ months), $3$ minutes ($24$ and $30$ months).
  • Retrieval: Child searches for objects after the delay while performing an interim activity (coloring/stacking).
• Scoring: Total toys correctly retrieved across two trials. Maximum possible scores: $4$ ($18$ months), $6$ ($24$ months), $8$ ($30$ months).
• Reliability: $100\%$ agreement between coders.

Working Memory for Locations Task

• Procedure: Modeled after Pelphrey and Reznick ($2003$).
  • Researcher hides a small object under one of several inverted cups.
  • View is obstructed for a short delay ($1$, $5$, or $10$ seconds).
  • Level of difficulty increase: First delay length, then number of cups ($2$ cups to $4$ cups).
  • Failure Criteria: Task ends after two consecutive errors at the same difficulty level.
• Scoring: Total correct responses (number of trials successfully cleared).
• Reliability: Average agreement ranged from $98\%$ to $99\%$.

Language Skill Assessment

• Tool: Preschool Language Scale - 3 (PLS-3).
• Components: Auditory Comprehension and Expressive Communication subscales.
• Assessment: Pointing to pictures, describing actions, defining words/actions.
• Scoring: Raw scores converted to Total Language standard scores.

Descriptive Findings and Age-Related Changes

• Elicited Imitation Performance:
  • Children learned the sequences: Immediate and Delayed performance significantly exceeded Baseline ($p < .001$).
  • Forgetting occurred: Immediate performance was significantly better than Delayed performance ($p < .001$).
  • Gender Effect: Girls consistently outperformed boys on all Elicited Imitation measures, primarily due to higher immediate production levels.
• Language Skills:
  • Mean scores: $100.45$ ($18$ mo), $112.88$ ($24$ mo), $118.17$ ($30$ mo).
  • Significant improvement between $18$ and $24$ months ($p < .001$).
• Memory Task Improvements:
  • Hide-and-Seek: Increased from $2.10$ ($18$ mo) to $5.18$ ($30$ mo) objects ($F = 70.12, p < .01$).
  • Working Memory: Increased from $3.05$ ($18$ mo) to $6.02$ ($30$ mo) trials ($F = 20.79, p < .01$).

Stability and Interrelations of Measures

• Stability Over Time:
  • Language skills showed high stability across all intervals (E.g., $r = .69$ between $24$ and $30$ months, $p < .001$).
  • Memory measures generally showed little stability across ages, with only modest correlations for Elicited Imitation actions between $18$ and other ages.
• Concurrent Relations Between Tasks:
  • Deliberate Memory (Hide-and-Seek) was not significantly correlated with Working Memory ($r = .11 - .20$) or Language ($r = .08 - .21$) at any age.
  • Working Memory was linked to Language only at $24$ months ($r = .29, p < .05$).

Concurrent Prediction of Elicited Imitation

• The Predictive Role of Language:
  • Immediate Performance: Language predicted immediate production of pairs by $24$ months and actions/pairs by $30$ months (Interactions $p < .05$). At $30$ months, regression coefficient $B = .042$ for actions.
  • Delayed Performance: Language was a significant positive predictor of delayed action and pair production at all ages ($p < .001$). Effect sizes were moderate ($d = .24$).
  • Unique Contribution: Even when immediate performance was used as a covariate, language uniquely contributed to delayed performance.
• The Predictive Role of Deliberate Memory:
  • Deliberate memory skill (Hide-and-Seek) predicted immediate production and ordering of actions at all ages ($p < .01$).
  • Encoding Effect: When immediate performance was controlled, the association between deliberate memory and delayed performance became non-significant, suggesting these skills primarily impact the learning/encoding of the modeled actions.
• The Predictive Role of Working Memory:
  • Working memory capacity positively predicted both immediate and delayed performance of actions and pairs across all ages ($p < .05$).
  • Delayed Sequencing at $30$ Months: A significant interaction showed that at $30$ months, working memory was a particularly strong predictor of sequencing action pairs after a $3$-week delay ($B = .091, p < .001$).

Discussion of Results and Implications

• Replication of Cross-Sectional Data: The mean performance levels found longitudinally were consistent with previous cross-sectional studies (E.g., Bauer et al., $2000$ for imitation; DeLoache et al., $1985$ for hide-and-seek).
• Why Language Matters: The study suggests that by age $2$, the language available during an event determines what is encoded into memory. Verbal narratives provided by examiners likely help children organize and represent the nonverbal actions.
• Differentiation of Memory Tasks: The independence of deliberate and working memory predictors suggests that they contribute to memory in different ways. Deliberate memory relates to attentional focus and strategy at encoding, whereas working memory capacity sets limits on information storage and retrieval quality.
• Future Directions: Future research should seek to replicate these findings in more linguistically diverse samples and specify underlying mechanisms for task interrelations. The data provides a foundation for predicting later verbal memory skills in preschool and school-age years.

Questions & Discussion

• Statistical Controls: Why was immediate performance used as a covariate in delayed models?
  • Response: Controlling for immediate performance allows researchers to determine if a predictor (like language) affects the retention or forgetting process specifically, rather than just being an artifact of how much was learned during the initial modeling.
• Gender Differences: Why were girls outperforming boys in Elicited Imitation?
  • Response: The study notes this is consistent with some previous literature (Bauer et al., $1998$) for children aged two and older, though the exact developmental driver for this nonverbal advantage was not the primary focus of analysis here.
• Handling "Rattle" Task: Why was the rattle event excluded from the $18$-month analysis?
  • Response: Preliminary analyses showed baseline performance was significantly higher for the rattle (children could do it before modeling), which would have contaminated the learning and memory scores.