Haueis and Colaço argue for incorporating metabolic constraints into cognitive modelling, suggesting that this may improve the biological plausibility of models of cognitive function. The authors propose evaluative and generative metabolic constraints that can refine existing cognitive models and guide the development of new ones. In this commentary, we build on the proposed framework by taking a developmental perspective, where metabolic constraints may play an even more critical role. At birth, the brain is reported to consume approximately 50–60% of total body metabolism (Kuzawa, Reference Kuzawa1998), decreasing over the first six months before rising alongside rapid synaptic and network growth (Kuzawa et al., Reference Kuzawa, Chugani, Grossman, Lipovich, Muzik, Hof, Wildman, Sherwood, Leonard and Lange2014). By age 5, brain energy use peaks at levels 2–3 times higher than in adults, utilising approximately 66% of resting metabolic rate and 43% of daily energy expenditure (Kuzawa & Blair, Reference Kuzawa and Blair2019). These shifts coincide with rapid structural development, network organisation, and the emergence of functional specialisation. A metabolism-informed developmental perspective, therefore, can extend the authors’ framework by demonstrating how emerging energetic constraints may shape the trajectory of cognitive development in both typical and neurodivergent developmental pathways. Within the proposed framework, evaluative approaches may determine whether a given model is metabolically feasible at a particular age, whereas generative approaches may use changes in energy availability and expenditure to explain how variation in developmental trajectories occurs.

In adult models, metabolic constraints are treated as relatively stable. However, in development, energy availability/expenditure changes over time. This makes evaluative metabolic considerations particularly important for assessing the plausibility of developmental cognitive models and helps rule out models that assume capacities, such as sustained neural activity or complex long-range connectivity, that may not be feasible given an infant brain’s limited energy budget. A clear example comes from early neural activity patterns observed in premature infants. Electroencephalography studies show characteristic “delta brush” activity, consisting of brief bursts of oscillatory activity followed by periods of relative quiescence (Whitehead, Pressler, & Fabrizi, Reference Whitehead, Pressler and Fabrizi2017). One interpretation is that these discontinuous patterns may reflect underlying metabolic constraints; sustained activity may be energetically too costly for the immature brain, necessitating intermittent bursts separated by recovery periods. From an evaluative metabolic perspective, this pattern suggests that models assuming continuous or sustained neural firing in early development may be implausible, as they do not account for the limited and fluctuating energy availability at this stage.

Using a generative approach, metabolism can also help explain developmental changes. Increasing energetic capacity may support the emergence of more complex processing and the gradual specialisation of brain networks, for example, in the development of executive functions or social cognition. Over the first few years of life, the brain becomes increasingly specialised and efficient, and this shift may be partly driven by changes in metabolic capacity. Thereby suggesting that metabolism may play an active role in shaping developmental changes and simultaneously act as a constraint that ultimately determines which neural pathways are strengthened or pruned and how cognitive functions emerge over time. Computational modelling work supports this view by demonstrating that brain network organisation emerges from trade-offs between energy cost, communication efficiency, and functional demands (Fakhar et al., Reference Fakhar, Akarca, Luppi, Oldham, Hadaeghi, Vértes, Bullmore, Hilgetag and Astle2025).

Drawing on the interactive specialisation framework (Johnson, Reference Johnson2011), the developing brain can be seen as a self-organising network in which cortical regions cooperate and compete to acquire functional specialisation. Metabolic efficiency and energetic costs may shape these competitive and cooperative interactions. That is, regions that require lower energy expenditure to perform computations are more likely to be repeatedly recruited, thereby strengthening connections. Similarly, cooperative interactions may emerge when regions coordinate activity in ways that minimise total energy expenditure, stabilising networks that consistently activate together. Changes in energetic capacity across development, owing to factors such as vascular development (Catalina Camacho et al., Reference Catalina Camacho, King, Ojha, Garcia, Sisk, Cichocki, Humphreys and Gotlib2020; Chiron et al., Reference Chiron, Raynaud, Mazière, Zilbovicius, Laflamme, Masure, Dulac, Bourguignon and Syrota1992; Takahashi et al., Reference Takahashi, Shirane, Sato and Yoshimoto1999; Yu et al., Reference Yu, Ouyang, Detre, Kang, Hu, Hong, Fang, Peng and Huang2023), neural circuitry development (Bianchi et al., Reference Bianchi, Stimpson, Duka, Larsen, Janssen, Collins, Bauernfeind, Schapiro, Baze, McArthur, Hopkins, Wildman, Lipovich, Kuzawa, Jacobs, Hof and Sherwood2013; Huttenlocher, Reference Huttenlocher1990; Huttenlocher & Dabholkar, Reference Huttenlocher and Dabholkar1997; Le Magueresse & Monyer, Reference Le Magueresse and Monyer2013), and the metabolic environment (Chugani, Phelps, & Mazziotta, Reference Chugani, Phelps and Mazziotta1987; Takahashi et al., Reference Takahashi, Shirane, Sato and Yoshimoto1999), may constrain which neural computations are feasible at each developmental stage, biasing the formation of short-range neural connections before more costly long-range connections can be supported. Resting-state functional MRI studies (Zhang, Shen, & Lin, Reference Zhang, Shen and Lin2019) illustrate the developmental emergence of long-range functional connectivity over the first year. This provides evidence for the brain favouring local, low-cost connections, with integration across distant regions becoming possible as energetic capacity increases.

Lastly, a metabolism-informed account of development can potentially provide insight into neurodivergent development more broadly, where early differences in neural structure, circuitry, and functional networks may interact with metabolic constraints to shape alternative trajectories of cortical specialisation. Evidence suggests that mitochondrial dysfunction, region-specific differences in metabolic activity, and atypical oxygen and glucose metabolism may limit the energy available to support neural computations across brain regions (Anitha et al., Reference Anitha, Nakamura, Thanseem, Yamada, Iwayama, Toyota, Matsuzaki, Miyachi, Yamada, Tsujii, Tsuchiya, Matsumoto, Iwata, Suzuki, Ichikawa, Sugiyama, Yoshikawa and Mori2012; Ginsberg et al., Reference Ginsberg, Rubin, Falcone, Ting and Natowicz2012). These alterations in energy metabolism have been implicated across a range of neurodevelopmental and neurological conditions, including autism (Chauhan & others, Reference Chauhan, Gu, Essa, Wegiel, Kaur, Brown and Chauhan2011; Gu et al., Reference Gu, Chauhan, Kaur, Brown, LaFauci, Wegiel and Chauhan2013), attention deficit hyperactivity disorder (Pinto Payares et al., Reference Pinto Payares, Spooner, Vosters, Dominguez, Patrick, Harris and Kanungo2024), Alzheimer’s disease and Parkinson’s disease (Zong et al., Reference Zong, Li, Liao, Chen, Pan, Zheng, Zhang, Liu, Zheng and Gao2024), and are likely to influence which regions can sustain neural activity and support computations at different developmental stages, possibly guiding competitive and collaborative interactions. Such interactions between metabolic constraints and neural circuitry may contribute to the diverse cognitive, sensory, and behavioural profiles observed across neurodevelopmental and neurological conditions, and conversely, the fact that pronounced cognitive impairments are reliably observed in mitochondrial disease suggests that energy metabolism plays a fundamental role in supporting normal brain function.

To conclude, a metabolism-informed developmental perspective extends Haueis and Colaço’s framework. Across typical development, changes in metabolic capacity both constrain and guide the emergence of functional specialisation, shaping which neural pathways are strengthened or pruned and supporting the transition from specialised and integrated networks. By including metabolic considerations into developmental models, we improve our understanding of the emergence of typical mental functions and of the patterns of cognitive, sensory, and behavioural profiles observed in neurodevelopmental conditions.