Home – Home – Jornals – Philosophy of Sciences 2025 number 4

Classically, physicists strived to assign some rigorous explanatory models to what happens in our phenomenal scope, and they surely succeeded. Nowadays, cognitive science tells us that our phenomenal representations, even being presumably structurally veridical, are made out of our inborn “mental paints” (Harman), such as light, colours, warmth, density, etc. According to professionally trained physics, the wave-particle duality is not at all paradoxical mathematically, only is it perceptually. Thus, one may conjecture that the rest of items of our physical ontologies may have been induced by our perceptual constitution, and if bats or robots did physics, their science, the set of its problems and their representations of facts could have substantially differed. Our physics uses categories (particles, waves) that feel natural to us because they’re built from our “mental paints” and predictive habits. A robot’s correlation-based science would skip these, suggesting our ontology isn’t inevitable. If perception is an evolved interface (Hoffman) and knowledge is a data-trained predictive model (Friston), then bats or robots would indeed craft different sciences. Our “problems” (e.g., wave-particle duality) might not even register with them. Big Data science hints that theory itself - our need to explain with entities like “fields” - might be a human peculiarity. An AI could deal with the same reality without our conceptual baggage, echoing Hoffman’s idea that we don’t see the “real” code. In sum, this suggests that the ontology of physics isn’t a direct window into reality but a human-constructed map, shaped by how evolution painted our interface and how our brains model the world. The mathematics might be universal (e.g., quantum equations), but the entities we project beyond - like waves or particles - are our own. Moreover, match or mismatch of an ontology and a theory can affect the latter’s efficiency.,