Research from Bielefeld University offers groundbreaking insights into how brain metabolism influences cognitive processes such as memory, perception, and attention. Dr. Philipp Haueis, a philosopher of science at the university, co-authored the study with Dr. David J. Colaço from Ludwig-Maximilians-Universität Munich, recently published in the journal Behavioral and Brain Sciences. The findings challenge traditional cognitive models that often neglect the brain’s energy constraints.
Haueis emphasizes the significance of understanding the brain as a biological entity rather than an abstract computer. “We wanted to find out what happens when you take the brain’s energy demands seriously,” he stated. He noted that the human brain, which constitutes roughly 2% of total body mass, consumes about 20% of the body’s energy. This high energy requirement underscores the need for cognitive models to consider metabolic limitations.
Metabolism’s Role in Cognitive Models
The research identifies two key roles of metabolism in cognitive modeling. Firstly, it assesses whether existing cognitive models are biologically feasible. Any theoretical model that demands more energy than the brain can supply is deemed unrealistic. Secondly, insights into metabolism can pave the way for developing new models that may better reflect the relationship between brain structure and information processing. These models could illustrate how neural networks efficiently use energy during learning processes.
Haueis and Colaço’s publication stands out as the first comprehensive summary of how metabolic insights can inform cognitive modeling. The study aims to stimulate discussion across various disciplines, particularly through the journal’s “Open Peer Commentary” format, which invites input from a broad range of researchers.
Broader Implications for AI and Intelligence
This research has significant implications for understanding mental effort and the differences between computation in biological and artificial systems. Haueis pointed out that recognizing the energy costs of thinking can shed light on why attention is limited. It also highlights how machine learning, which operates without biological constraints, may follow different developmental paths.
The study is expected to contribute not only to academic research but also to public discourse on artificial intelligence, energy efficiency, and the broader nature of intelligence itself. The findings were produced within the Institute for Studies of Science (ISoS) at Bielefeld University, which was designated a central academic unit in May 2025. ISoS integrates interdisciplinary research focused on science, medicine, and technology, examining how these fields interact with societal contexts.
For further details, refer to the work by Haueis et al., titled “Metabolic considerations for cognitive modeling,” published in Behavioral and Brain Sciences. The article can be accessed via DOI: 10.1017/s0140525x25103956.
