Context: The current state of extended-mind research involves different frameworks, predictive processing and enactivism, among others. It is unclear to what degree these two frameworks converge toward a unified conception of the extended mind. Problem: The third wave of extended-mind research expands the scope of what has been acknowledged as a legitimate case of extended mind under the parity principle and complementarity principle of the first two waves. The two central commitments of the third wave are: (a) That extended cognitive agents exhibit plasticity (b) that extended cognitive systems may not be organism-centered. I explore a general notion of boundary that might accommodate those two claims and provide a general criterion of what constitutes a case of extended mind. Method: I employ the method of conceptual analysis. I explore several conceptions of mental boundary and plasticity with regard to the context of the wider conceptual frameworks within which they are embedded, namely predictive processing and enactivism. I confront the two frameworks with regard to how the notions they provide fare against the issues of third-wave extended-mind research. Results: a) I confront the notion of boundary of cognitive agents based on the Markov-blanket formalism with the enactivist notion of mental boundary based on operational closure and argue for the latter approach. (b) I argue that plasticity of mental boundaries exhibits two fundamental facets that can be distinguished and accounted for by recourse to Ashby’s conception of ultrastability and the notion of perceptual inference as employed by the free-energy principle, if the two notions are integrated into the enactivist framework. Implications: Predictive processing and enactivism cannot be reconciled regarding their respective notions of boundary of cognitive agents. In this regard, enactivism provides a better point of departure for the third wave of extended-mind research. Notions of active and perceptual inference based on the free-energy principle might nevertheless provide insights that enrich the enactivist position and lead to a more nuanced perspective on the extended mind. Constructivist content: Constructivist epistemology forms the theoretical background of some key notions, utilized throughout the article, namely the conception of mind as autonomous and self-organizing.

Key words: Enactivism, extended mind, free-energy principle, Markov blanket, predictive processing.

Abstract: The commentaries on my target article can be clustered into three topics. First, I discuss the broader theoretical background underlying the notion of cognitive boundary, and argue that this notion should be taken to refer to the boundary by which cognitive autonomous agents actively distinguish themselves from their environment. Subsequently, I address the difference between the notions of Markov blanket and operational boundary, pointing out that the two notions of boundary are compatible in some, but not all cases. In the last section I deal with some concrete examples of extended mind and argue that the proposed account does justice to instances of distributed assembly.

Upshot: Self and Other takes on an ambitious task of articulating an account of selfhood that would do justice both to the experiential and the intersubjectively constructed aspect of the self. We claim that Zahavi argues convincingly against the possibility of reducing the former to the latter, but that he needs to develop his position further in order to offer a convincing account of their interconnectedness.