Lake et al. draw from research in both artificial intelligence (AI) and cognitive development to suggest a set of core abilities necessary for building machines that think and learn like humans. We share the authors' view that children have a set of core cognitive abilities for learning and that these abilities should guide development in AI research. We also agree with the authors' focus on findings from theory theory research and their characterization of its principles as “developmental start-up software” that is adapted later in ontogeny for social learning. What is missing from this discussion, however, is the recognition that children's developmental start-up software is shaped by their culture-specific social environment. Children's early and ontogenetically persistent experiences with their cultural environment affect what learning “programs” children develop and have access to, particularly in the case of social learning.

Research suggests that from early infancy, children display a core set of abilities that shape their reasoning about the world, including reasoning about both inanimate objects (intuitive physics [e.g., Spelke Reference Spelke1990]) and animate social beings (intuitive psychology [e.g., Dennett Reference Dennett1987; Meltzoff & Moore Reference Meltzoff, Moore, Bermúdez, Marcel and Eilan1995]). Although the early onset of these abilities provides evidence that they may be universal, little research has examined their development in non-WEIRD (Western educated industrialized rich democratic) (Henrich et al. Reference Henrich, Heine and Norenzayan2010) cultures (Legare & Harris, Reference Legare and Harris2016). Moreover, research that has examined children's intuitive theories in different cultural settings has suggested the potential for both cross-cultural continuity and variation in their development. Take, for example, the development of children's theory of mind, a component of intuitive psychology. A large collection of research comparing the development of children's understanding of false belief in the United States, China, and Iran indicates that although typically developing children in all cultures show an improvement in false belief understanding over the course of ontogeny, the timing of this improvement differs widely—and such variability is potentially related to different sociocultural inputs (Davoodi et al. Reference Davoodi, Corriveau and Harris2016; Liu et al. Reference Liu, Wellman, Tardif and Sabbagh2008; Shahaeian et al. Reference Shahaeian, Peterson, Slaughter and Wellman2011). Thus, children's social environments may be shaping the development of these core abilities, “reprogramming” and updating their developmental start-up software.

To illustrate why considering the principles derived from theory theory are important for guiding AI development, Lake et al. point to AI's lack of human-like intuitive psychology as a key reason for why humans outperform AI. In their discussion of humans' superior performance in the Frostbite challenge, the authors highlight humans' ability to build on skills gained through the observation of an expert player,which requires reasoning about the expert player's mental state. AI can also draw on observations of expert players, but requires substantially greater input to achieve similar levels of performance. Humans' intuitive psychology and their corresponding ability to reason about others' mental states is just one element of why humans may be outperforming computers in this task. This situation also draws on humans' ability to learn by observing others and, like the development of false-belief understanding, children's ability to learn through observation as well as through verbal testimony, which is heavily influenced by sociocultural inputs (Harris Reference Harris2012).

Culturally specific ethno-theories of how children learn (Clegg et al. Reference Clegg, Wen and Legare2017; Corriveau et al. Reference Corriveau, Kim, Song and Harris2013; Harkness et al. Reference Harkness, Blom, Oliva, Moscardino, Zylicz, Bermudez and Super2007; Super & Harkness Reference Super and Harkness2002) and the learning opportunities to which children have access (Kline Reference Kline2015; Rogoff Reference Rogoff2003) shape their ability to learn through observation. As early as late infancy, sociocultural inputs such as how parents direct children's attention, or the typical structure of parent-child interaction, may lead to differences in the way children attend to events for the purpose of observational learning (Chavajay & Rogoff Reference Chavajay and Rogoff1999). By pre-school, children from non-WEIRD cultures where observational learning is expected and socialized outperform children from WEIRD cultures in observational learning tasks (Correa-Chávez & Rogoff Reference Correa-Chávez and Rogoff2009; Mejía-Arauz et al. Reference Mejía-Arauz, Rogoff and Paradise2005). Recent research also suggests that children from different cultural backgrounds attend to different types of information when engaging in observational learning. For example, Chinese-American children are more sensitive to whether there is consensus about a behavior or information than Euro-American children (Corriveau & Harris Reference Corriveau and Harris2010; Corriveau et al. Reference Corriveau, Kim, Song and Harris2013; DiYanni et al. Reference DiYanni, Corriveau, Kurkul, Nasrini and Nini2015). Such cultural differences in attending to social information in observational learning situations persist into adulthood (Mesoudi et al. Reference Mesoudi, Chang, Murray and Lu2015). Therefore, although the developmental start-up software children begin with may be universal, early in development, children's “software updates” may be culturally dependent. Over time, these updates may even result in distinct operating systems.

The flexibility of children's core cognitive abilities to be shaped by sociocultural input is what makes human learning unique (Henrich Reference Henrich2015). The role of this input is largely missing from Lake et al.'s discussion of creating human-like AI, but its inclusion would help move research even closer to machines that can learn and think like humans.