Yes. We can get into the ‘Turtles all the way down’ (as per Sapolsky) situation and it seems very convincing.

If I claim we are manufacturing information you could ask, ‘what determined that process?’

I could answer ‘it is a fluid process that adapts to circumstances in the pursuit of self-interest’.

You might then ask ‘ what determines the process of self-interest?’

To which I say evolution. It was evolutionarily economic to create consciousness and with sentience, navigate self-interest. What is the determining factor? Evolution to ensure survivability where survivability is the ability to safely navigate the environment and be around long enough to produce the next generation.

All the things you could mention like neural architecture, genetics, embryology, environmental circumstances past and present are all contributory. I don’t deny that.

If you look at one part of the last piece I shared which I think is a section called ‘Tim Palmers’ Pendulum’ I talk about self-determinism superimposing a directional bias on Chaos. Briefly to save you revisiting that section, it concerns one of those executive desk toys for ‘decision making’, consisting of a metal pendulum-bob and four magnetic bases. The motion is truly chaotic and the bob will spin around one magnetic base before shooting off elsewhere in an unpredictable way. Palmer’s point (which is actually about climate) is that despite the fact the motion is chaotic, over sufficient trials it will land on each of the four bases 25% of the time. However, if you superimpose a bias by putting a wedge under the plinth of the toy (in TP’s example, global warming), what happens? Well it is still chaotic and unpredictable but over many trials the lowest base will get more hits. The bigger the wedge the bigger the bias. This is exactly what I mean by superimposing free-will (self-determinism) over all the other influences. We don’t always get what we want but our choices give us a sense of self-direction. I say that this is not an illusion it is free-will.

You say:

I might be annoying and ask: what criteria are they using to measure most information? My initial thought was that I’m not sure there is a single level of abstraction that contains "the most" information - rather, each level contains different data (or input) that may or may not be relevant for our current needs and goals.

That is not annoying because that is precisely the question you needed to ask.

I would argue the informational content does not change at all. Why? Because the number of pixels on your screen is unchanged. The only thing that changed was abstraction level and that determines how relevant or useful the information is. Big data analytics can find patterns in what might look a lot like noise to us but what it is really doing is using information and energy to mine other information for useful abstractions. That is what we are doing too. We are mining information to determine how to align our actions to self-interest and experiential preferences. What do you think?