your analysis of causal emergence is of great importance for understanding of complex cognitive systems. I have just realized that a similar phenomenon is known in quantum mechanics: the best possible knowledge of the states of the whole system is not sufficient to have the best possible knowledge of all parts of the system, and vice versa, perfect knowledge of all parts does not imply perfect knowledge of the whole system.
My paper is now 33 years old, and written from quite different perspective. I have almost forgotten about my work in quantum mechanics, but you may find some analogies and conclusions interesting. Causal emergence seems to be common in quantum as well as classical complex systems.
Duch, W. (1989). Schrödinger’s thoughts on perfect knowledge. In: The concept of probability. Ed. Bitsakis EI and Nicolaides CA (pp. 5–14). Kluwer Academic Publishers.
Wish I had more time to get a better grasp of your theory! If my questions totally miss the point, feel free to tell me. Here goes:
Does your theory of causal emergence have any connection to the concept of downward causation? Or is downward causation a "spooky" idea--i.e. does it require strong emergence?
Secondly, does your theory have any conceptual connection with the "integrated information" of IIT? That is, if the higher-level causal relationships are stronger than the lower-level ones, is that at all similar to the idea information being more integrated for a whole than for any of its parts?
Thanks Ryan! Causal emergence is, in my opinion, basically the "correct" form of downward causation, wherein downward causation as it's usually framed doesn't make sense. Causal emergence is neither strong nor weak emergence, it seems to slip through the cracks as it really is the case causal relationships are stronger at macroscales, but this doesn't violate supervenience, likely because causation is a type of information and macroscales convert information into being causally-relevant https://arxiv.org/abs/2104.13368. Integrated Information is merely another measure of causation, and we've shown it too shows causal emergence https://academic.oup.com/nc/article/2016/1/niw012/2757132
Forgive the naive question, but as I read this Quanta article, I couldn't help wondering, could Fernando Rosas's "computational approach to hierarchical emergence" framework support / complement / dovetail with your "macro can beat micro" take on causal emergence ?
Oops, I had not finished reading the article and just now see that you were actually cited and interviewed in it !
Tononi's quote seems to address my question :
'... Tononi says that, while his approach and that of Rosas and colleagues address the same kinds of systems, they have somewhat different criteria for causal emergence. “They define emergence as being when the macro system can predict itself as much as it can be predicted from the micro level,” he said. “But we require more causal information at the macro level than at the micro level.” ...'
I'm not sure if i agree... Sounds wierd. Maybe not. How would you even measure that? hehe... not easy. You are writing about abstractions. Kinda... I dont wanna go into how i think now i would have to lay down a bit... here and now.
Dear Eric,
your analysis of causal emergence is of great importance for understanding of complex cognitive systems. I have just realized that a similar phenomenon is known in quantum mechanics: the best possible knowledge of the states of the whole system is not sufficient to have the best possible knowledge of all parts of the system, and vice versa, perfect knowledge of all parts does not imply perfect knowledge of the whole system.
My paper is now 33 years old, and written from quite different perspective. I have almost forgotten about my work in quantum mechanics, but you may find some analogies and conclusions interesting. Causal emergence seems to be common in quantum as well as classical complex systems.
Duch, W. (1989). Schrödinger’s thoughts on perfect knowledge. In: The concept of probability. Ed. Bitsakis EI and Nicolaides CA (pp. 5–14). Kluwer Academic Publishers.
https://fizyka.umk.pl/publications/kmk/89-Schrodinger.html
Regards, Wlodek Duch
Google: Wlodzislaw Duch
Wish I had more time to get a better grasp of your theory! If my questions totally miss the point, feel free to tell me. Here goes:
Does your theory of causal emergence have any connection to the concept of downward causation? Or is downward causation a "spooky" idea--i.e. does it require strong emergence?
Secondly, does your theory have any conceptual connection with the "integrated information" of IIT? That is, if the higher-level causal relationships are stronger than the lower-level ones, is that at all similar to the idea information being more integrated for a whole than for any of its parts?
Thanks Ryan! Causal emergence is, in my opinion, basically the "correct" form of downward causation, wherein downward causation as it's usually framed doesn't make sense. Causal emergence is neither strong nor weak emergence, it seems to slip through the cracks as it really is the case causal relationships are stronger at macroscales, but this doesn't violate supervenience, likely because causation is a type of information and macroscales convert information into being causally-relevant https://arxiv.org/abs/2104.13368. Integrated Information is merely another measure of causation, and we've shown it too shows causal emergence https://academic.oup.com/nc/article/2016/1/niw012/2757132
Forgive the naive question, but as I read this Quanta article, I couldn't help wondering, could Fernando Rosas's "computational approach to hierarchical emergence" framework support / complement / dovetail with your "macro can beat micro" take on causal emergence ?
"The New Math of How Large-Scale Order Emerges"
https://www.quantamagazine.org/the-new-math-of-how-large-scale-order-emerges-20240610/
"Software in the natural world: A computational approach to hierarchical emergence"
https://arxiv.org/abs/2402.09090
Oops, I had not finished reading the article and just now see that you were actually cited and interviewed in it !
Tononi's quote seems to address my question :
'... Tononi says that, while his approach and that of Rosas and colleagues address the same kinds of systems, they have somewhat different criteria for causal emergence. “They define emergence as being when the macro system can predict itself as much as it can be predicted from the micro level,” he said. “But we require more causal information at the macro level than at the micro level.” ...'
I'm not sure if i agree... Sounds wierd. Maybe not. How would you even measure that? hehe... not easy. You are writing about abstractions. Kinda... I dont wanna go into how i think now i would have to lay down a bit... here and now.