Just a week after I wrote my last blog about unsupervised learning, cognitive science and Andy Clark’s “predictive processing” an article appears in the New Scientist magazine entitled “Reality? It’s what you make it” [i].
This piece, by Philip Ball, discusses some very new ideas in quantum theory that go under the general title of participatory realism [ii]. This new concept, championed by Christopher Fuchs, Markus Müller and others, argues that the world, as we experience it from a first person perspective, is the emergent property arising from something much deeper, more complex and mysterious than we can imagine. On his web site Müller asks “Can we have a fully information-theoretic approach to physics in which not a notion of “external world”, but of “observation” is the fundamental starting point?” [iii]. It seems we can.
We are but social apes, limited beings of a certain scale, that experience some aspects of the world quite reliably – our unsupervised learning converges on constructs that look and behave like laws even though they might prove to be metaphysical constructs as we learn more about the underlying reality. (Remember, as Nancy Cartwright has argued [iv], even our most robust physical laws are ceteris paribus laws: founded upon “all other things being equal…”)
At other scales, and in other disciplines, we face complementarity – apparently incommensurate statements about reality. Müller, for example, argues that in areas like quantum physics, cosmology and philosophy of mind (and I would add some aspects of ecology and environmental science) we face fundamental problems with the usual scientific approach.
Now the standard view of science is the ontic view: that any incommensurability arises from the real world itself. The new – and to many the “mad” view – is that these problems arise, not from Nature, but from what we as observers construct. The physics we see is an emergent property and the conclusions we draw from our observations can be explained by algorithmic information theory and a form of induction in which the observer merely conditions his or her theories about future observations on past experience.
This is exactly what Clark proposes as cognitive science based on unsupervised learning, As we observe the world – as our embodied heuristics and predictive processing construct the world for us – there need be no ground and we end up with merely useful generalisations that might occasionally be complementary to others, or undecidable and incommensurate.
So much is entirely consistent with my last blog. There is nothing to say that our embodied mental heuristics that merely “keep us in touch with the world” are in any way designed to make fully detailed and accurate representations of the world. Evolution has merely ensured that our cognitive information processing has given us heuristics with survival value at relevant scales. Prehistoric apes were never selected for their knowledge of quantum physics or the finer points of cosmology!
There was a small meeting that was held last June at the Stellenbosch Institute for Advanced Studies (a meeting that Philip Ball attended) to discuss aspects of this participatory realism. Some sub-editor in New Scientist who drafted the title of Ball’s article based around that meeting added a subordinate clause that stated “an audacious new take on quantum theory suggests alternative facts are a fact of nature”.
In a world of flat Earth believers, climate change deniers and those in high places who peddle alternative facts, how long is it before we hear distorted accounts of participatory realism with the cries of “we told you so”?
(Although I am intrigued that as we develop new tools and ways of thinking about artificial intelligence, information technologies and machine learning it is simultaneously leading both to fundamental conceptual advances as well as to false reasoning and backward looking “echo chambers” on social media.)
I guess we can only keep on repeating that participatory realism is a form of realism: there is a real world out there, the Earth is round, satellites stay up, and the planet is warming. It’s not always clear where our understanding leads and some aspects of what our senses reveal are downright strange – we should not be surprised about that – but evidence and reason remain our best approach.
Despite our limited and partial understanding of the world there are constraints, there are consistencies, the macroscopic laws of physics and engineering are repeatable and, generally speaking, our empiricism and cognitive powers are reliable and useful. (We would not be here if they weren’t!)
[i] Philip Ball (2017) Reality? It’s what you make it. New Scientist, 11 November 2017, No. 3151, p18-32
[ii] See, for example, Christopher Fuchs (2016) On participatory realism, arXiv:1601.04360v3 [quant-ph] 28 Jun 2016
[iv] Nancy Cartwright (1983) How the laws of physics lie. Oxford.