Naturalizing relevance realization: why agency and cognition are fundamentally not computational

A theoretical paper arguing that how living things find what matters is not computation.

Author
Johannes Jaeger, Anna Riedl, Alex Djedovic, John Vervaeke and Denis Walsh, in Frontiers in Psychology
Published
25 June 2024
Kind
Peer-reviewed theoretical paper (Hypothesis and Theory), with no experiments
Original
www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2024.1362658/full
Subject
Mind, attention and practice

What it is

Five authors from universities in Vienna and Toronto wrote this for Frontiers in Psychology. It sets out to ground relevance realization in biology. Relevance realization is how an agent picks what matters. The paper then draws conclusions about artificial intelligence.

What it says

What evidence it gives

The paper offers conceptual argument only. It draws on Rosen's mathematics of living systems and Hofmeyr's models. Its main example is a bacterium swimming up a nutrient gradient. It runs no experiments and gives no data. It calls its claims about consciousness speculative. It also cites a footnote on a published simulation of Rosen's system.

Interests and limits

One author originated the relevance realization framework that the paper extends. The first author's funding came from the John Templeton Foundation. A footnote notes a published simulation of Rosen's system in lambda calculus. It says this does not change the argument. No working AI system is tested. The reference list was not read in detail.