A review article by Borjan Milinkovic and Jaan Aru argues that treating the mind as software running on interchangeable hardware is a poor fit for how brains actually compute. The authors propose “biological computationalism,” a framework that ties cognition and (potentially) consciousness to computation that is hybrid, multi-scale, and shaped by energy constraints.
The scientific and philosophical debate over consciousness is often framed as a standoff between two camps: computational functionalism, which treats thinking as abstract information processing that could in principle be implemented in many physical systems, and biological naturalism, which holds that conscious experience depends on the concrete physical processes of living organisms.
In a new review article, researchers Borjan Milinkovic and Jaan Aru argue for what they call biological computationalism, describing it as a way to move beyond the “software versus biology” framing. Their core claim is that standard, computer-inspired notions of computation are mismatched to the way brains operate, and that this mismatch matters for debates about consciousness and artificial minds.
According to the authors, biological computation has three central features:
First, it is hybrid, combining discrete events (such as neuronal spikes and neurotransmitter release) with continuous, time-evolving physical dynamics (including changing voltage fields and chemical gradients) that continually influence one another.
Second, it is scale-inseparable. The authors argue that brain activity cannot be cleanly divided into an abstract “algorithm” on one side and a separate “implementation” on the other, because causal influences run across levels—from ion channels to circuits to whole-brain dynamics—and changing the physical organization changes what the system computes.
Third, it is metabolically (or energetically) grounded. In their account, strict energy constraints shape what the brain can represent and how it learns and maintains stable activity, treating this coupling as part of how biological intelligence is organized.
Taken together, the framework emphasizes the idea that, in brains, “the algorithm is the substrate”—computation is not merely symbol manipulation layered onto hardware, but a physical process unfolding in real time.
The authors also argue that this perspective exposes limits in how modern artificial intelligence is often described. While AI systems can learn powerful input-output mappings as digital procedures, the paper contends that biological computation relies on real-time physical dynamics and multiscale coupling that today’s digital architectures generally do not instantiate.
The article does not claim consciousness is restricted to carbon-based life. Instead, it suggests that if consciousness depends on this kind of computation, then building synthetic minds may require physical systems that reproduce key features of biological-style computing—hybrid dynamics, multiscale coupling without clean interfaces, and strong energetic constraints—rather than only better software.
The review article, titled “On biological and artificial consciousness: A case for biological computationalism,” appears in Neuroscience & Biobehavioral Reviews (Volume 181, February 2026) as article number 106524. Materials for the ScienceDaily summary were provided by the Estonian Research Council.
