John et al. provided an underlying theoretical framework explaining the widely recognized phenomenon of why a measure ceases to be good when it becomes the target for what it aims to measure, capturing this phenomenon, termed proxy failure, in a range of domains. Associated problems have been widely debated in education, often under the banner of Campbell's law, and popularized in the notion of “training for the test,” where standardized tests only measure a subset of the properties they aim to measure (Koretz, Reference Koretz2008), leading to a variety of problems such as a too-narrow focus on end outcomes (i.e., the measure of learning chosen) rather than evaluating the activities, interactions, and environments that lead to those outcomes.
Paradoxically, relying solely on educational goals can lead to worse learning. This is identified in John et al.'s proxy failure concept. When proxies are adopted to steer the direction of education, even well-intentioned educational efforts eventually miss their goal. In childhood education, the counterintuitive results that academically goal-oriented educational programs do not produce better academic outcomes (e.g., Durkin, Lipsey, Farran, & Wiesen, Reference Durkin, Lipsey, Farran and Wiesen2022) might be explained by how the focused training children receive provides a too-narrow proxy for the complex phenomenon of learning. Problems with such programs have led scholars to turn to play as a cornerstone of learning and a serious educational alternative for young children (Nesbitt, Blinkoff, Golinkoff, & Hirsh-Pasek, Reference Nesbitt, Blinkoff, Golinkoff and Hirsh-Pasek2023). It can also hold other important implications, as play and its relationship with learning can provide insights into proxy failure mechanisms and provide alternatives for education and beyond.
Play as a phenomenon has long been a puzzle for scientists as it superficially looks like a nongoal-oriented and perhaps purposeless activity (Pellegrini, Reference Pellegrini2009). It is more puzzling still that play is adaptive for young organisms across species (Burghardt, Reference Burghardt2005). Crucially, because the focus of play is not on any external goal, engaging in play and related exploratory behavior trains the organism for uncertain outcomes in unstable environments (Bjorklund, Reference Bjorklund2022). For humans, this takes a cultural turn as children play with and socialize around the tools and technologies in their environment (Samuelsson, Price, & Jewitt, Reference Samuelsson, Price and Jewitt2022), leading to culturally appropriate learning potential as a by-product of this engagement (Samuelsson, Reference Samuelsson2023). Thus, play provides a valuable perspective on how less-regulated activities can lead to learning outcomes.
Play provides a peculiar window into the learning process that differs radically from many formal educational environments, for in play, learning is not a predetermined goal but follows as a by-product of engagement in an activity. Additionally, play is often a joyful experience for children (Weisberg, Hirsh-Pasek, Golinkoff, Kittredge, & Klahr, Reference Weisberg, Hirsh-Pasek, Golinkoff, Kittredge and Klahr2016). We can learn from studies of play, as they show how a different type of activity not associated with external goals also leads to learning results that are on par with or even better (e.g., Toub et al., Reference Toub, Hassinger-Das, Nesbitt, Ilgaz, Weisberg, Hirsh-Pasek and Dickinson2018) than those of a goal-oriented learning process.
This is also telling regarding proxy failures related to education. If education is based on narrow learning goals, it risks the proxy treadmilling effect identified by John et al., where teachers can too narrowly train for the test, or children find ways to “hack” the regulatory system. By contrast, playful learning is less concerned with external learning goals. It is centrally characterized by focused engagement in joyful activities and, if used correctly, engages curious children at the height of their abilities (Chu & Schulz, Reference Chu and Schulz2020). In “hacking” or “gaming,” the regulatory system becomes pointless, as it is not the goal of the activity (e.g., there is no question whether something will be on the test). However, play may involve practice in iterative engagement, which is typical of play. From this engagement, play becomes “training for the unexpected” (Spinka, Newberry, & Bekoff, Reference Spinka, Newberry and Bekoff2001, p. 143), a widely observed function of play. There is something to be learned from play's relation to learning because of its nongoal-oriented nature precisely because there is no proxy associated with a learning goal that leads to learning.
Playful learning has been widely adopted in childhood educational programs such as guided play activities (Weisberg et al., Reference Weisberg, Hirsh-Pasek, Golinkoff, Kittredge and Klahr2016). Overall, these initiatives can tailor the positive aspects of play in educational environments that are well adapted to childhood education. When using play for learning outcomes, however, one must be careful. Paradoxically, educational programs based on play-based learning risk are becoming examples of proxy failures themselves if they are abundantly goal-driven, reducing the robustness of playful engagement for narrower and more easily measured learning outcomes (e.g., the legibility problem). Here, John et al. warn us about the mechanistic properties of proxy failure, meaning that even if actors in an educational system are well-intentioned, proxies move toward failure. Educational systems wishing to use play as part of their learning program should learn from play that the loosening of educational proxies can lead to learning and consider how the inherent looseness of playful activities yields robust learning.
Understanding the problems instilled by creating measurable proxies for complex learning processes entails a considerable rethinking of how we conduct education. It is, however, a key lesson for not only creating educational settings that avoid the narrowing associated with failed educational proxies, but also for other areas wishing to navigate proxy failures. As play shows, engaging in activities with internally driven goals can lead to robust outcomes, such as long-term, meaningful learning. This can hold implications beyond the early childhood classroom and make us more humble in creating proxies for goal-driven processes, considering the often-nested hierarchies imposed when creating a target measure. Potential regulatory agents can benefit from an awareness of proxy failure mechanisms, here meaning all actors from the education providers of play-based learning to the teachers who engage in guided play interactions with children. This is one example of how John et al.'s contribution to proxy failure constitutes fundamental work for understanding educational processes and can be a crucial instrument for creating educational futures of genuine quality.
John et al. provided an underlying theoretical framework explaining the widely recognized phenomenon of why a measure ceases to be good when it becomes the target for what it aims to measure, capturing this phenomenon, termed proxy failure, in a range of domains. Associated problems have been widely debated in education, often under the banner of Campbell's law, and popularized in the notion of “training for the test,” where standardized tests only measure a subset of the properties they aim to measure (Koretz, Reference Koretz2008), leading to a variety of problems such as a too-narrow focus on end outcomes (i.e., the measure of learning chosen) rather than evaluating the activities, interactions, and environments that lead to those outcomes.
Paradoxically, relying solely on educational goals can lead to worse learning. This is identified in John et al.'s proxy failure concept. When proxies are adopted to steer the direction of education, even well-intentioned educational efforts eventually miss their goal. In childhood education, the counterintuitive results that academically goal-oriented educational programs do not produce better academic outcomes (e.g., Durkin, Lipsey, Farran, & Wiesen, Reference Durkin, Lipsey, Farran and Wiesen2022) might be explained by how the focused training children receive provides a too-narrow proxy for the complex phenomenon of learning. Problems with such programs have led scholars to turn to play as a cornerstone of learning and a serious educational alternative for young children (Nesbitt, Blinkoff, Golinkoff, & Hirsh-Pasek, Reference Nesbitt, Blinkoff, Golinkoff and Hirsh-Pasek2023). It can also hold other important implications, as play and its relationship with learning can provide insights into proxy failure mechanisms and provide alternatives for education and beyond.
Play as a phenomenon has long been a puzzle for scientists as it superficially looks like a nongoal-oriented and perhaps purposeless activity (Pellegrini, Reference Pellegrini2009). It is more puzzling still that play is adaptive for young organisms across species (Burghardt, Reference Burghardt2005). Crucially, because the focus of play is not on any external goal, engaging in play and related exploratory behavior trains the organism for uncertain outcomes in unstable environments (Bjorklund, Reference Bjorklund2022). For humans, this takes a cultural turn as children play with and socialize around the tools and technologies in their environment (Samuelsson, Price, & Jewitt, Reference Samuelsson, Price and Jewitt2022), leading to culturally appropriate learning potential as a by-product of this engagement (Samuelsson, Reference Samuelsson2023). Thus, play provides a valuable perspective on how less-regulated activities can lead to learning outcomes.
Play provides a peculiar window into the learning process that differs radically from many formal educational environments, for in play, learning is not a predetermined goal but follows as a by-product of engagement in an activity. Additionally, play is often a joyful experience for children (Weisberg, Hirsh-Pasek, Golinkoff, Kittredge, & Klahr, Reference Weisberg, Hirsh-Pasek, Golinkoff, Kittredge and Klahr2016). We can learn from studies of play, as they show how a different type of activity not associated with external goals also leads to learning results that are on par with or even better (e.g., Toub et al., Reference Toub, Hassinger-Das, Nesbitt, Ilgaz, Weisberg, Hirsh-Pasek and Dickinson2018) than those of a goal-oriented learning process.
This is also telling regarding proxy failures related to education. If education is based on narrow learning goals, it risks the proxy treadmilling effect identified by John et al., where teachers can too narrowly train for the test, or children find ways to “hack” the regulatory system. By contrast, playful learning is less concerned with external learning goals. It is centrally characterized by focused engagement in joyful activities and, if used correctly, engages curious children at the height of their abilities (Chu & Schulz, Reference Chu and Schulz2020). In “hacking” or “gaming,” the regulatory system becomes pointless, as it is not the goal of the activity (e.g., there is no question whether something will be on the test). However, play may involve practice in iterative engagement, which is typical of play. From this engagement, play becomes “training for the unexpected” (Spinka, Newberry, & Bekoff, Reference Spinka, Newberry and Bekoff2001, p. 143), a widely observed function of play. There is something to be learned from play's relation to learning because of its nongoal-oriented nature precisely because there is no proxy associated with a learning goal that leads to learning.
Playful learning has been widely adopted in childhood educational programs such as guided play activities (Weisberg et al., Reference Weisberg, Hirsh-Pasek, Golinkoff, Kittredge and Klahr2016). Overall, these initiatives can tailor the positive aspects of play in educational environments that are well adapted to childhood education. When using play for learning outcomes, however, one must be careful. Paradoxically, educational programs based on play-based learning risk are becoming examples of proxy failures themselves if they are abundantly goal-driven, reducing the robustness of playful engagement for narrower and more easily measured learning outcomes (e.g., the legibility problem). Here, John et al. warn us about the mechanistic properties of proxy failure, meaning that even if actors in an educational system are well-intentioned, proxies move toward failure. Educational systems wishing to use play as part of their learning program should learn from play that the loosening of educational proxies can lead to learning and consider how the inherent looseness of playful activities yields robust learning.
Understanding the problems instilled by creating measurable proxies for complex learning processes entails a considerable rethinking of how we conduct education. It is, however, a key lesson for not only creating educational settings that avoid the narrowing associated with failed educational proxies, but also for other areas wishing to navigate proxy failures. As play shows, engaging in activities with internally driven goals can lead to robust outcomes, such as long-term, meaningful learning. This can hold implications beyond the early childhood classroom and make us more humble in creating proxies for goal-driven processes, considering the often-nested hierarchies imposed when creating a target measure. Potential regulatory agents can benefit from an awareness of proxy failure mechanisms, here meaning all actors from the education providers of play-based learning to the teachers who engage in guided play interactions with children. This is one example of how John et al.'s contribution to proxy failure constitutes fundamental work for understanding educational processes and can be a crucial instrument for creating educational futures of genuine quality.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interests
None.