Bonn 2025 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

AGPhil: Arbeitsgruppe Philosophie der Physik

AGPhil 2: Foundations of Physics II

AGPhil 2.3: Vortrag

Montag, 10. März 2025, 18:00–18:30, HS XVII

The Probabilistic Turn across Physics: From Classical to Quantum Physics and from Psychophysics to AI — •Ken Archer — Linkoping University, Linkoping, Sweden

The meaning and interpretation of probability within quantum physics is illuminated in this paper by identifying parallels in the probabilistic turn across multiple areas of physics. The probabilistic turn from classical physics to statistical mechanics has important parallels with the probabilistic turn from classical physics to quantum physics. Critically, this paper shows these same parallels within another probabilistic turn in a field whose association with physics is controversial - the probabilistic turn from psychophysics to artificial neural networks (ANNs) that are the basis for AI.

In all three fields, probability enables physical models to account for stability. Just as statistical mechanics accounts for the stability of fields and quantum mechanics accounts for the stability of matter, ANNs enable cognitive models to account for the stability of cognitive capacities across heterogenous and even damaged neural networks. Furthermore, this role of probability across physics points to another common feature - the absence of pre-given distributions (Gaussian, binomial, Bayesian, etc) such that softmax in ANNs plays an analogous role as Born's Rule in quantum mechanics. In both cases, the particular mathematization of the phenomena is the theory - there's no deeper human intuition about the phenomena to leverage in a pre-given distribution, as probabilities emerge naturally from the mathematical formalism.