Rationale

Researchers, practitioners, and the general public have continued to exhibit growing interest in the Science of Learning, “the scientific study of the underlying bases of learning with the goal of describing, understanding, or improving learning across developmental stages and diverse contexts” [1]. Recent advances in the cognitive and neural sciences have illuminated our understanding of the principles, processes, and mechanisms of learning across a diverse range of species, including our own [e.g., 2]. Findings from these studies support the conclusion that learning is a complex lifelong process involving the interplay of cognitive, neural, and other factors [e.g., 3]. Furthermore, the experience of learning modifies underlying cognitive processes and brain structure and function in a manner that impacts on future cognition and behavior [4]. Given the relevance of this work outside the research laboratory, there has been growing interest in the application of findings from the Science of Learning to a number of other fields, including education [e.g., 5–7].

While there is tremendous interest in understanding human learning, previous investigations have generally not been developmentally comprehensive in their focus. Existing studies on the Science of Learning have predominantly investigated samples drawn from developmental stages during early and late life [8], although some notable exceptions do exist [e.g., 9]. This same pattern can be observed across published reviews with most focusing on research in samples of younger children [e.g., 10], or older adults, including clinical populations diagnosed with neurodegenerative diseases [11]. Collectively, this has created a gap in our understanding of learning during healthy adulthood as well as effective ways in which that learning can be improved. Given that the human brain retains the capacity for plasticity across the adult lifespan [12], and that adult learning trajectories can be altered based on the influence of both internal and external factors [13], there is significant value in better understanding learning during this period of development in the interests of maximizing human potential.

Developing a more comprehensive understanding of learning during adulthood, and effective ways of improving that learning, are crucial goals given current global demographic trends. Many nations around the world are continuing to struggle in addressing the significant consequences of a rapidly aging population [14]. This issue is further complicated by the dynamic nature of the modern workforce where regular career changes are becoming the norm [15]. Consequently, governments have recently shifted attention and resources toward extending and translating findings from the Science of Learning in order to inform educational policy and intervention strategies. One of the overarching goals of this shift is to support the development of a citizenry that is both productive and agile into late adulthood. Presently, the absence of a comprehensive understanding of learning across adulthood in healthy populations places a significant hurdle on the path to accomplishing this goal. The conduct of an exhaustive review synthesizing existing research on the cognitive and neural mechanisms of learning and interventions to improve learning in healthy adults would contribute significantly to the accomplishment of this goal, with the additional benefit of identifying areas in need of further research.

Objectives

The main objective of the proposed systematic review is to identify and synthesize all recent cognitive and brain research investigating learning across the adult lifespan. This review will provide an overview of the principles, processes, and mechanisms of learning in healthy adults. Additionally, this review aims to synthesize all recent findings on interventions to improve learning in adults. Findings from the proposed review will have implications for our understanding of learning in adulthood, the scoping of funding calls for future Science of Learning research, and for public education and outreach efforts to support lifelong learning.

Research questions

The primary research question (RQ) of the proposed review is, “How does learning change across a healthy adult’s lifespan?” The research sub-questions are:

1. What learning-related cognitive changes occur in healthy adults as they age?
   1. Which specific dimensions of cognitive function have been investigated?
   2. How do changes in these dimensions influence learning?
   3. What factors influence these changes?
2. What learning-related brain changes occur in healthy adults as they age?
   1. Which specific dimensions of brain structure/function have been investigated?
   2. How do changes in these dimensions influence learning?
   3. What factors influence these changes?
3. What interventions based on evidence from 1 and 2 are most effective in the improvement of learning in healthy adults?
   1. What are the reported effects of interventions on cognitive function?
   2. What are the reported effects of interventions on brain structure/function?
   3. What are the mediating and moderating factors that influence the efficacy of interventions?
4. What are the current knowledge gaps identified in 1–3?

For the purpose of the present review, learning will be broadly defined to include traditional measures related to change in achievement or attainment, as well as changes in underlying cognitive and neural processes such as executive function, intelligence, and brain network activation.

Registration

The protocol for the proposed review was designed following recommendations provided in the Preferred Reporting in Systematic Reviews and Meta-Analysis Protocols (PRISMA-P) checklist [16] and was registered with the Centre for Open Science (https://osf.io/d9yev/). Deviations from the registered protocol will be explicitly noted in the final version of the paper. A timeline for the conduct of the proposed review is presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Timeline for the proposed review.

https://doi.org/10.1371/journal.pone.0301935.t001

Eligibility criteria

Studies will be included if they meet the following criteria:

• Journal articles, theses, dissertations, conference papers, and preprints
• Cross-sectional, longitudinal, and sequential studies or research syntheses
• Published or posted between January 1st, 2012 and September 1st, 2023
• Reports one or more of the following:
  1. ○ RQ1: Relationship between measure(s) of learning-related cognition and age.
  2. ○ RQ2: Relationship between measure(s) of learning-related brain structure or function and age.
  3. ○ RQ3: Impact of any intervention aimed at improving learning.
• Focused on healthy adult humans
• Reports findings (quantitative or qualitative) from participants age ≥ 20 years either exclusively or as a separate analysis from a wider age group.
• Written in English
• Full text available

Studies will be excluded if any of the above criteria are not met or if the document is a purely theoretical paper, non-systematic narrative review, magazine article, abstract, protocol, preregistration, blog post, editorial, commentary, corrigendum, book, or book review.

Information sources

The search for relevant literature will occur using Scopus, Web of Science (Core Collection and Preprints), and ProQuest databases. These databases were selected due to their extensive indexing of relevant published work as well as conference papers, theses, dissertations, and preprints. Search dates will be limited from January 1st, 2012 to September 1st, 2023. Studies identified across all databases will be pooled and filtered for duplicate results prior to screening for inclusion. If the full text for an article cannot be located, corresponding authors will be contacted. If no reply is received after two weeks, the study in question will be excluded.

Search strategy

For each database used, search strings will be written following the database-specific requirements related to Boolean operators and search limiters. Strings for each database will contain a combination of terms aimed at identifying all potentially relevant articles including “learning” and “adult”, as well terms related to different measures of interest including “memory”, “intelligence”, “brain”, and “executive function”. Finally, additional terms will be included using the “NOT” operator to prevent the inclusion of large numbers of irrelevant articles including “dementia”, “impairment”, and “brain damage”. When appropriate, “wildcard” search terms will be included to allow for inclusion of highly similar terms (e.g., “adult*” for “adult”, “adulthood”, “adults”). An example draft search strategy for Web of Science (Core Collection) is presented in Table 2. Searches will be limited to title and abstract fields and will be run using the “exact search” option or equivalent for each database. Search string performance will be evaluated based on the recovery of known sentinel papers [9,17,18].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Example draft web of science (core collection) search strategy.

https://doi.org/10.1371/journal.pone.0301935.t002

Study records

Data items

The following data will be extracted from all selected studies: 1) publication year; 2) authors; 3) source format (e.g., journal article); 4) country the study was conducted in; 5) study setting; 6) study design; 7) study duration (if applicable); 8) population of interest 9) number of participants by sex per group; 10) average age of participants per group; 11) sociodemographic background of participants per group; 12) cognitive, brain, or learning outcome(s) reported; 13) additional variables, and; 14) main results. Additionally, 15) intervention format; 16) comparison group, 17) outcome of interest, and; 18) measure of impact of intervention on outcome of interest will be extracted from intervention studies. In the event necessary data were not reported in the original articles, corresponding authors will be contacted. Authors not responding to a request for data within two weeks will not be contacted again and the study in question will be included with incomplete details. Extracted data will be summarized in tabular format organized by research question. Data from single studies reporting findings relevant to multiple research questions (e.g., study that reports the relationship between both cognitive and brain measures with learning), will be summarized under a “combined studies” heading in the synthesis table with relevant findings contributing to their respective narrative synthesis or, in the case of intervention studies, quantitative synthesis. Extraction table finalization is expected to be completed by March 31st, 2024.

Outcomes and prioritization

The primary outcomes of the proposed review are measures of the relationship between cognition or brain structure/function and age (RQ1 & RQ2), and learning difference scores or qualitative descriptions from comparisons between control and experimental groups or pre- and post-intervention timepoints (RQ3). Cognitive measures can include reaction time or accuracy rates from behavioral tasks (e.g., Simon task, n-back task), scores from cognitive assessments (e.g., intelligence test score), or self-reported measures (e.g., Behavior Rating Inventory of Executive Function for Adults). Brain measures can include data from any neuroanatomical or neurophysiological instrument related to structure (e.g., magnetic resonance imaging [MRI]) or function (e.g., electroencephalogram [EEG]). Measures of age can include those that are quantified (e.g., age in years) as well as categorical labels (e.g., young adults and elders). Measures of learning can either be objectively measured (e.g., performance on a memory task, motor skill performance) or self-reported (e.g., GPA, educational attainment). For RQ3, the impact of an intervention can be represented as either a quantitative measurement or qualitative assessment of change in reported learning-related outcome.

Risk of bias in individual studies

We will evaluate the risk of bias in each individual study by employing the relevant checklists provided by the Joanna Briggs Institute [20]. Every included study will receive a rating of "Yes," "No," "Unclear," or "NA" for each criterion assessed. A team of three trained reviewers will conduct this assessment, and inter-reviewer reliability will be calculated across 20% of the included studies. No studies will be excluded from the planned review based on the results of the risk of bias assessment.

Data synthesis

Regarding RQ3, if there are at least two studies that report the impact of a specific intervention on learning, we will conduct a quantitative synthesis. Analyses will be performed in jamovi [21]. Each type of intervention will undergo separate analysis. Either Hedge’s g or Fisher’s r-to-z transformed correlation coefficient will be used as the outcome measure depending on the outcome reported in a given study. To model the data, we will utilize a random-effects model and will estimate heterogeneity (i.e., tau²) using the restricted maximum-likelihood estimator [22]. Additionally, heterogeneity will be reported as the I² statistic [23]. We will also assess the presence of outliers or influential data points using studentized residuals and Cook’s distances. For each model, we will generate forest plots, which will include the pooled effect size. However, if the data reported across the included intervention studies exhibit significant heterogeneity that prevents a quantitative synthesis, we will assess the effectiveness of interventions using effect direction analysis [24,25]. Finally, if we do not identify a sufficient number of studies for a particular intervention or outcome (i.e., fewer than 2), we will provide a narrative synthesis.

Meta-biases

If data from included studies supports the conduct of a quantitative synthesis for RQ3, publication bias will be assessed through generation of Egger’s statistics for each intervention type [26], and funnel plot visual analysis. Both Egger’s statistics and funnel plots will be reported in the final version of the proposed review.

Confidence in cumulative evidence

For RQ3, we will evaluate the quality of evidence following the methodology outlined by the Grading of Recommendations Assessment, Development and Evaluation (GRADE) working group [27]. This assessment will encompass several domains, including risk of bias, consistency, precision, and publication bias. The overall quality of evidence will be categorized using the GRADE certainty ratings [28,29], which include:

• High: indicating a high level of confidence that the true effect aligns closely with the estimated effect.
• Moderate: signifying a belief that the true effect is likely to be close to the estimated effect.
• Low: suggesting that the true effect might significantly differ from the estimated effect.
• Very low: indicating a high likelihood that the true effect significantly differs from the estimated effect.

Quality of evidence assessment will be modified accordingly in the event that we are unable to perform a quantitative synthesis.