Over the past few decades, researchers from across a variety of disciplines, including linguistics, machine learning, education, psychology, cognitive science, and others, have made tremendous progress in studying how learning occurs. Together, these insights have formed a new, interdisciplinary field coined the science of learning and this term was recently reviewed and operationalized in the pages of this journal [1]. This approach has led to a more comprehensive understanding of learning across the lifespan, but particularly during childhood [[2], [3], [4], [5], [6]].

Despite these insights, the translation and application of these findings to support children in everyday contexts remains lacking. However, while scientific insights have continued to accumulate, the challenges for education across the globe have remained. For example, according to the Sustainable Development Goal (SDG) 4 Scorecard, at least one in three counties showed backsliding, both in terms of teacher training at the pre-primary and primary levels as well as in learning proficiency [7]. Further, learning loss due to COVID-19 has resulted in children being behind, by eight months on average, of where they would have been without the pandemic and the impacts are only increasing inequalities [8].

Recent work from the science of learning has coalesced around a number of principles and pedagogical approaches that are effective in supporting learning, but much work remains to be done in supporting teachers in implementing those approaches in the classroom. This paper explores these approaches and then investigates how teachers of children ages 3-12 years across three county contexts (Bangladesh, Uganda, and Colombia) used a formative assessment tool designed to support their implementation of playful learning approaches in the classroom. This study helps us better understand how to support teachers in facilitation in their classrooms, taking a multi-context approach to look for similarities and differences across ages, learning goal differences, and country contexts. Finally, we explore how lessons from how teachers used the tool can be used to inspire research in the science of learning, arguing for a cyclical, rather than unidirectional, relationship between the science of learning and teacher professional development.

For decades, a false dichotomy between play and learning has done a disservice to the field of education and the implementation of pedagogy in the classroom [[9], [10], [11], [12]]. This dichotomy is likely rooted in a narrow conceptualization of play. However, evidence from the science of learning [13,14] and teacher practice within the field of education [15] have suggested that play and playful practices exist along a spectrum, or continuum, that captures the differing roles of the teacher and the student. When thinking about the spectrum of play, children's agency, marked by choice and the ability to direct, is central and required. What varies are the levels of adult facilitation and child agency and whether or not there is a learning goal [16,17].

Free play is sometimes heralded as the “gold standard” of play, and in free play, there is no specified learning goal, nor adult scaffolding or control [18]. In free play, children maintain agency, decision-making, and direction. Children are free to play, or not play, with whatever materials are available. Guided play and games [11,[19], [20], [21]] still maintain children's agency, but adults scaffold and support the play and there is an intended learning outcome. In teacher-directed play [15], children maintain limited agency, but adults have a heavier hand in both directing and supporting the play context. Direct instruction is outside the scope of the play spectrum. It eliminates child agency and aligns with the idea that children must be taught new information directly.

In a meta-analysis, Alfieri [22] analyzed the literature to compare learning across free play, assisted discovery methods (equivalent to guided play/games), and direct instruction. Alfieri found, perhaps unsurprisingly, free play was least likely to help support learning goals. However, Alfieri also found that guided play methods outperformed direct instruction methods for obtaining a variety of outcomes. Since then, additional work has extolled the benefits of guided play/games in supporting learning outcomes across domains - including mathematics [23], spatial thinking [24], literacy [[25], [26], [27], [28], [29]] and creativity [30]. But, research also suggests that there are some contexts in which direct instruction is better suited to support learning, especially of scientific thinking content [31].

Importantly, much of the research exploring the impacts of playful pedagogical practices are based in Western contexts, but there is growing interest in and program implementation supporting playful learning globally with some suggesting that playful practices have the potential to help leapfrog education [32]. A recent report highlights this potential and suggests that there are three “levers” that are needed to align educational policy around playful learning with implementation: namely shifting societal attitudes, through educational policy and implementation, and city design with a focus on intergenerational spaces centering children's learning [33]. However, we would argue that a fourth lever is also critical, direct partnership, training, and support of playful learning practices through teacher professional development.

Hirsh-Pasek, Zosh, and colleagues [3] reviewed the science of learning literature and argued that the literature supports the idea that there are pillars of learning - or characteristics that maximize learning. They suggested that when humans are active (minds-on), engaged (not distracted), learning meaningful content (connects to the larger world, their previous understanding, and potentially their passion), and socially interactive, learning is maximized. A few years later, Zosh and colleagues [14,16] expanded this model to specifically examine how play naturally leverages these characteristics and also supports learning via iterative (e.g., testing hypothesis and updating understanding) thinking in a joyful (including sustained or momentary positive affect and/or surprise) context.

Those viewing play as a spectrum argue that playful learning practices (including free play, guided play, games, and teacher-directed play) naturally leverage the characteristics that lead to learning [14,16]. Indeed, a 2020 policy report explores the potential of playful learning as an approach to promote 21st-century in schools (and beyond) using these characteristics as a foundation [34]. In other words, the authors use these characteristics of learning to ensure that “… we are teaching in ways most compatible with the ways human brains learn.” [p. 12]. Implementing these practices in the classrooms across varying contexts, ages, and in the service of different domains, does, however, create challenges.

There is ongoing work exploring how these characteristics and playful pedagogies can be a path for much-needed educational reform in the United States of America [34], but it is also important to better understand how these characteristics of learning, based in the science of learning and designed to be universal in benefits, are instantiated across contexts in other countries.

While concrete, content-based learning outcomes are typically the focus of academic testing. Especially in Western contexts, much work has been done recently to expand the conceptualization of what children need to learn. Children's education is typically responsible for supporting learning in areas such as literacy and mathematics, but studies grounded in the science of learning have focused on learning skills, such as executive function, as well as 21st-century skills (Fadel, 2008), for example, Golinkoff and Hirsh-Pasek's [35] 6 C's - content, collaboration, creative innovation, communication, critical thinking, and confidence.

This expansion of thinking about what children need to learn is certainly important, but it still tends to be focused on Western contexts and recent work suggests that researchers need to be more cautious about using a cultural lens when examining outcomes [36,37]. OECD [38] expands the conceptualization of learning outcomes to include knowledge (e.g., academic proficiencies such as literacy and numeracy and interdisciplinary knowledge such as science, health and nutrition, humanities, arts and culture, digital literacy), skills (e.g., 21st-century skills like communication, collaboration, creativity, critical thinking, leadership, problem-solving, and entrepreneurship), as well as attitudes and values (e.g., personal, local, societal, and global). Similarly, Fadel [39] proposed a 4-dimensional framework for education that includes character (e.g., mindfulness, curiosity, courage, resilience, ethics, and leadership), skills (creativity, critical thinking, communication, and collaboration), knowledge (interdisciplinary, traditional knowledge [e.g., math], modern knowledge [e.g., entrepreneurship], themes [e.g., global literacy]), and meta-learning (metacognition and growth mindset).

Beyond context-specific differences, there may also be differences in intended learning outcomes based on the ages of the children in a classroom context. Teachers must also determine how to create lessons that support learning amongst their students. To date, we are not aware of research comparing intended learning goals across ages and country contexts while also investigating the pedagogies used to support those goals.

While resources, training, and situations may differ across contexts, it is universal that teachers are in the position to select the learning goals that are important to them, determine how to teach the students in their classrooms, and evaluate how well they meet their curricular goals. The larger question becomes how can we, as a field, support teachers’ use of evidence provided by the science of learning in their classrooms? How can we create tools or conduct training in ways that are based in science but are not so prescriptive that they can only be applied in limited contexts?

There are a number of challenges in both implementation and assessment of programs designed to help support teachers. In terms of implementation, simply sharing information about the spectrum of playful practices and descriptions of the characteristics is likely not enough to change teacher practice. Indeed, there are a number of review articles exploring the various challenges facing professional development of teachers [[40], [41], [42]] with Hill and colleagues [40] stressing that more research is needed at earlier stages of professional development design rather than simply evaluating a program's efficacy at the end. Sancar and colleagues [42] highlight that professional development needs to be attentive to a variety of factors including context, comprehensiveness, support and control and others.

Danniels and colleagues [43] outline two main types of assessment that are commonly used in research and policy. Summative assessment/assessment of learning is based on outcomes (e.g., student test scores, outcomes at the end of a project) whereas formative assessment/assessment for learning/assessment focuses on ongoing assessment that occurs throughout the project that provides continuous feedback and opportunities for reflection and changes throughout the implementation of a project. Formative assessment has been used in teacher professional development across contexts and ages [[43], [44], [45]]. This suggests that one potential pathway for supporting teacher practice that is based in the science of learning literature is through the use of formative assessment tools.

Beyond determining the type of assessments that may support teacher practice, another consideration comes from the fundamental challenge of bridging theory and practice. In other words, the description of active, engaged, meaningful, iterative, joyful, and socially interactive states is just one piece of the puzzle. The larger challenge is determining how one could even assess the presence of these characteristics in their classrooms. Notably, these are internal characteristics, thus leading to additional issues of measurement. The literature supports the idea that leveraging these characteristics supports learning [14,16], but it is a challenge to determine how well each one of these characteristics is engaged for individual learners. Yet another challenge is that each of these characteristics is not a present/absent concrete state. Instead, their engagement is dynamic within a given context, changes over the course of a single lesson, and the characteristics interact with one another. This presents a challenge for both implementation and measurement.

Similarly, play types exist along a continuum where there are stronger and weaker ways of facilitating free play, guided play, and teacher-directed play. While this creates a challenge for measurement, it also provides a benefit for implementation. By viewing these facilitation styles and characteristics as continuums of their own, teachers are afforded the opportunity to facilitate in a variety of ways– for example, ways that are suitable for their context, the lesson at hand, the children's age and educational needs, and their own strengths. Recently, Zosh and colleagues [46] suggest a framework that characterizes teacher practices that are likely to engage each of the characteristics to varying degrees within the classroom. For example, seating children in groups is a positive step towards promoting social interaction, but providing multiple opportunities for children to work together and with peers of varying abilities is even more likely to promote high quality social interaction. Relatedly, there are different ways of facilitating each type of play, which provides a number of opportunities for teachers to make small changes to maximize learning through play even within a single type of play (e.g., within guided play, increasing how children engage with the characteristics). Similarly, as reviewed by Parker and Thomsen [47], the term “play” is often not used when discussing pedagogical approaches after the age of about 8 years old . Instead, they review evidence suggesting that other more typical “pedagogical” terms such as active learning, cooperative and collaborative learning, experiential learning, guided discovery learning, inquiry-based learning, problem-based learning, and Montessori education are related to learning through play and, to varying degrees, align with learning through play approaches and with the five characteristics of playful learning discussed here.

Finally, research in the science of learning has primarily come from Western contexts and, while implementation of these practices is starting to have global reach [32], it is critical to understand how direct practitioners - the teachers - approach these core principles of pedagogy. In other words, how do educators approach selecting learning goals, supporting the characteristics of learning in their classroom experiences, and facilitating playful pedagogies across ages - and how do these factors interact with an eye towards uncovering universals and differences across contexts?

Taken together, the larger question is how can we support teacher practice in ways that are grounded in the science of learning literature but that are actionable and observable for teachers across ages and contexts? And, more broadly, how can we create a feedback loop such that these types of professional development efforts can inform researchers from the science of learning to ask better questions that are rooted in practice (see Figure 1)?

The purpose of the current study was to investigate teachers’ approaches to supporting playful practices in their classroom across three country contexts. To do so, we created a formative assessment tool that was designed to support their facilitation of playful practices in the classroom. Importantly, we aimed to create a flexible tool with teachers’ professional development and reflection at the fore (rather than assessing teachers or measuring child outcomes). To do so, the tool allowed teachers to self-select a learning goal as well as the facilitation style and the characteristic of learning (actively engaging, meaningful, socially interactive, joyful, iterative) that they would like to examine in their classroom. The tool provided behavioral indicators that could be used by teachers to examine whether individual students, small groups of students, or the entire class were exhibiting the characteristics of learning as they engaged in playful learning practices.

Teachers were able to control their own data (enter and access) and we purposefully moved away from evaluative models that have the potential to negatively impact teachers based on their performance as well as limit the potential for teachers to select responses that they thought were “correct” so that they would look better to anyone monitoring their results. While this purposeful decision to use authentic formative assessment rather than accountability approaches [48] limited our ability to monitor changes in teacher behavior over time, we decided that the ability of teachers to use the tool with no concerns about repercussions would provide a higher likelihood of authentic use of the tool to support student learning.

Importantly, this work sought to investigate the use of this tool across three diverse country contexts: Bangladesh, Colombia, and Uganda. The countries were purposefully selected to represent different geographic regions and education. Colombia has a more central focus on play as a pedagogy in the early childhood education system [49], but has less of a focus on play in primary and secondary school contexts [50]. While play has largely been traditionally viewed as a leisure activity in Bangladesh, there is a shift occurring with the 2021 National Curriculum Framework emphasizing learning through play for pre-primary and grades 1-3 [51]. However, class size, teacher professional development, availability of play materials, and ambiguous perception of play based pedagogies have been identified as barriers to implementation in pre-primary classes [52]. While there is some recognition of learning through play in Uganda, it is important to note that there is no official guidance or policy about its use in schools and rote learning is the norm [53]. Further, Uganda identifies as having a low quality of education (e.g., low rates of literacy and numeracy, high dropout rate) and improving education has been identified as a priority in their national development plan [54]. Indeed, a recent analysis suggests that even for early childhood, the discourse and approach focus on survival rather than thriving and carers view cognitive development as fixed and innate [55].

Thus, we also sought to investigate how teachers across these contexts engaged with the tool to determine whether there were meaningful differences in their selection of learning goals, facilitation types, and characteristics. Learning how teachers used this type of tool can help provide guidance to anyone seeking to support teacher practice by determining areas of alignment and areas of misalignment across age and context.

More specifically, this study examined the following research questions:

RQ 1: What facilitation styles were selected by teachers across student ages and teacher learning goals and are these choices impacted by context?

RQ 2: Which characteristics of learning were selected by teachers across student ages and teacher learning goals and are these choices impacted by context?