This commentary will focus on Haueis and Colaço’s notion of a metabolic turn in cognitive neuroscience and its broader implications for modeling cognitive systems. I find the core thesis of the relevance of metabolic considerations for cognitive modeling convincing, and my intention here is to provide an opportunity for further elaborating on this central thesis. To begin with, I will briefly present some potential compatriots of the metabolic turn in the cognitive sciences and the philosophy of biology.

In their target article, Haueis and Colaço note a growing metabolic turn in philosophy aligned with their core thesis (TA 2.2.3). But while the authors provide several illustrative examples of such metabolic approaches from cognitive neuroscience, less attention is given to existing literature on metabolism in philosophy of biology and cognitive science, much of which might add further support for their case.

Research from the philosophy of biology highlights the importance of metabolism in accounting for the development and functional role of various phenotypes – including cognitive capacities. Multiple accounts here align with the target paper in regarding metabolic considerations as critical to cognitive systems, their organization (Baedke, Reference Baedke2025; Bich & Bechtel, Reference Bich, Bechtel, Barandiaran and Etxeberria2026; Fábregas-Tejeda et al., Reference Fábregas-Tejeda, Baedke, Prieto, Radick, Fábregas-Tejeda, Baedke, Prieto and Radick2024; Nicholson, Reference Nicholson2014; Walsh, Reference Walsh2015), development, and evolution of cognition (Ginsburg & Jablonka, Reference Ginsburg and Jablonka2021; Moreno & Peretó, Reference Moreno and Peretó2026; Oyama, Griffiths, & Gray, Reference Oyama, Griffiths, Gray, Oyama, Griffiths and Gray2001). Other accounts argue that metabolic systems may even be primary to substantial biological individuality (Dupré, Reference Dupré2024; Jefferey, Pollack, & Rovelli, Reference Jeffery, Pollack and Rovelli2019; Nicholson & Dupré, Reference Nicholson and Dupré2018). Additionally, biologically informed approaches in cognitive science likewise highlight interdependencies between cognition and broader organismic dynamics, such as metabolic processes (Barandiaran & Moreno, Reference Barandiaran and Moreno2008; Huebner & Schulkin, Reference Huebner and Schulkin2022; Lyon et al., Reference Lyon, Keijzer, Arendt and Levin2021). These various biologically informed approaches to cognitive systems strike me as conducive to the overall line of argument of the target article.

In addition to general support, these accounts also suggest a potential explanation for the neglect of metabolic considerations documented by the target article. Dynamical models based on biological and metabolic considerations, including ones listed above, are often seen as conflicting with standard computational models (Brette, Reference Brette2022; Beer, Reference Beer2023; Lyon, Reference Lyon2020). Indeed, some of the vocal supporters of medium dependence within the philosophy of neuroscientific modeling (TA 4.2) have recently argued that standard computationalist approaches encourage viewing biological features “as background support structures that are separable from the truly cognitive (i.e. computational) processes of the brain” (Chirimuuta, Reference Chirimuuta2026, p. 4), suggesting there may be a conflict between metabolic considerations and standard computational modeling.

This brings me to my first question. While the authors seem to support the indivisibility of metabolism from cognition, they also reject the conflict between resource models and computational models (TA 4.2). How does this rejection of conflict apply to wholesale critiques of computational approaches, such as claims that computationalism inappropriately divides metabolic hardware from cognitive software? For example, are such wholesale contrasts overstating the limitations of computational analogues in modeling or the ramifications of a metabolic turn?

On the topic of how deep a metabolic turn in cognitive modeling should go, my second question concerns the potential connection between modeling considerations and ontological commitments. Throughout the target article, Haueis and Colaço deliberately opt for neutrality about ontology, regarding the ontological status of metabolic considerations as secondary (TA 2.1.1, 2.2.3). They also note that, while metabolic considerations may contribute to cognitive modeling across the spectrum, determining the relevance of metabolic considerations for modeling is a subject to be explored (TA 4.1). While this choice of focus is quite reasonable, I want to raise the possibility that determining the relevance of various metabolic considerations may be inextricable from questions about our background ontological commitments about cognitive systems.

As an example, consider the conspicuous connection between computationalism and medium independence. As demonstrated by various accounts above, metabolically informed approaches often highlight the importance of accounting for a wider array of neurobiological factors as well as broader, organismic factors. In this sense, advocates of a metabolic turn tend to converge on an identification of cognitive systems with a wider array of substrate processes. By contrast, backgrounding of metabolic considerations often converges with more constrained conceptualizations of cognitive systems (Barrett, Reference Barrett2016; Hurley, Reference Hurley2001).

This raises the potential connection between the attribution of relevance in cognitive modeling, such as the backgrounding of metabolism (TA 2.1.5), and ontological commitments, such as a narrower conceptualization of the cognitive system. As some philosophical work of modeling argues, models may be importantly constrained by background assumptions, including ontological commitments about which processes and parts form part of the system in question (Longino, Reference Longino2021; Potochnik & Sanches de Oliveira, Reference Potochnik and Sanches de Oliveira2020). If this is the case, ontological considerations may be unavoidable in specifying the relevance of various evaluative and generative metabolic considerations for modeling cognitive systems.

My second question is, do the authors see any connection between ontological commitments and future work in determining the relevance of metabolic considerations for modeling? For example, might ontological assumptions have influenced the present neglect of metabolic considerations in modeling cognitive systems? And could a metabolic turn in cognitive modeling potentially involve reconsidering some of our ontological background assumptions about cognitive systems?