In their target article, Phillips and colleagues focus on the important problem of how people understand others' minds, and they argue that children's ability to represent knowledge develops before their ability to represent beliefs. They recognize that in focusing on knowledge they are obliged to be clear about their assumptions. And in their “call to arms,” they state that their goal is to reach a “deeper understanding of the nature of knowledge.” We applaud this goal because if we are concerned with how children learn about the world, including other people, then we must be aware of our assumptions regarding knowledge (Chapman, Reference Chapman1999). Yet, in Phillips et al.'s following sentence, and indeed throughout, it is clear that they fail to explicate or assess their own assumptions and simply assume a representational view, according to which knowledge is based on representations that match the world. This view is also labeled a correspondence or copy theory (Piaget, Reference Piaget1970), or spectator notion of knowledge (Dewey, Reference Dewey1960), and has been extensively critiqued (e.g., Bickhard, Reference Bickhard2009; Carpendale & Lewis, Reference Carpendale and Lewis2004, Reference Carpendale and Lewis2006, Reference Carpendale and Lewis2021; Dewey, Reference Dewey1960; Piaget, Reference Piaget1970). Russell (Reference Russell1992) refers to representation as a humpty dumpty term because it can be used in any way desired to mean absolutely nothing.

This is a passive view of knowledge which, it has often been argued, presupposes “what it is meant to explain” (Chapman, Reference Chapman1999, p. 31). That is, from this perspective knowledge consists of representing or forming a copy of the world. But, in that case, the only way to check our knowledge is by forming another copy to compare it to because we do not have direct access to the world. This does not solve the problem, and thus we cannot tell if the copy is accurate. Yet becoming aware of errors and learning from them is essential in understanding development (Bickhard, Reference Bickhard2009). Thus, this view already presupposes knowledge when that is what it is meant to explain, and thus the account is inadequate. Instead of taking for granted the ability to re-present and think about what is not immediately there, we view this as an important challenge to explain. We follow others in not taking our biggest problem and making it our premise (Bickhard, Reference Bickhard2009; Piaget, Reference Piaget1970).

Understanding how children learn about knowledge and beliefs depends on an adequate conception of the more general issue of how children develop knowledge. As an option that deals with the criticism of the representational view, we endorse an action-based or constructivist approach, according to which knowledge is conceptualized in active terms of learning what can be done with the world, that is, of its interactive potential. Thus, knowledge is viewed as skill rather than “representation” (Bibok, Carpendale, & Lewis, Reference Bibok, Carpendale, Lewis, Müller, Carpendale, Budwig and Sokol2008). Instead of assuming that the child possesses a pre-existing mind and faces “the problem of other minds” when trying to understand others, human infants develop within a social system. They gain mastery of the world through coming to anticipate what happens when they do things. For example, they learn how to reach for and grasp objects of interest. Because this action typically occurs in a social context, infants' interests are manifest for other people. Others may respond if the infant is not initially successful in reaching the desired object. Thus, their goal may be reached not because they attain it but because it prompts facilitation from a caregiver. Infants can then gradually become aware that their actions are of significance for others, that is, they learn that they are communicating (Mead, Reference Mead1934). Infants can then begin to communicate intentionally, first with gestures and later with the addition of words (Carpendale, Reference Carpendale, Dick and Müller2018).

From this perspective, infants first form of knowledge of other people in terms of an anticipation of their action based on their prior activity. This is achieved first at the practical sensorimotor level. It is this form of knowledge that can be evident in infant false-belief tasks (Carpendale & Lewis, Reference Carpendale, Lewis, Liben and Müller2015). This means of understanding others is also evident in research with chimpanzees, which shows that in competitive situations they can anticipate dominant co-specifics' actions such as moving toward food depending on whether they have seen it (e.g., Tomasello, Call, & Hare, Reference Tomasello, Call and Hare2003).

We believe that the distinction between knowledge and belief that Phillips et al. discuss refers to much later distinctions involving language. Children come to use terms referring to others' anticipated action in terms of their knowledge or beliefs, as in “she thought that…” or “she knows that…” To be able to talk about and think about others in terms of knowledge and belief depends on learning how to use such words. These words refer to patterns of human activity not to inner mental entities that mental-state words are mapped onto (Canfield, Reference Canfield2007; Wittgenstein, Reference Wittgenstein2009). Children gradually learn some of the many ways in which words such as think and know are used. To consider just a few early uses of such terms, the word “know” can be used in various ways – in relation to whether someone has seen something, to refer to someone's ability or lack of ability, to deny responsibility as in “I didn't know,” or to avoid answering as in “I don't know.” The word “think” can be used to modulate certainty or to refer to a process of decision-making.

We have focused on a more fundamental problem with Phillips et al.'s position than the concern with whether children “compute” belief representations before or after knowledge representations. This is because their way of conceptualizing both belief and knowledge is problematic because of being based on a problematic conception of knowledge and mind.

What we are suggesting as an alternative to Phillips et al. is not empiricism, which neglects the activity of the subject (Piaget, Reference Piaget1970, Reference Piaget1972), but instead is an action-based, process-relational approach (Carpendale, Hammond, & Atwood, Reference Carpendale, Hammond, Atwood, Lerner and Benson2013b). Our focus on activity should not be mistaken for behaviorism. Instead, we begin from action and interaction in order to explain psychological development (Carpendale, Atwood, & Kettner, Reference Carpendale, Atwood and Kettner2013a; Carpendale & Lewis, Reference Carpendale, Lewis, Liben and Müller2015).