I would think a large part of the problem is the difficulty in observing a working brain.
There are very few non-destructive ways we can get a look into a brain while it is living. And even research into ways we can look into a living brain is hindered by the fact we don't want to harm the person being observed.
The even bigger challenge is determining _what_ you need to observe in the first place.
As a simplistic analogy, evolutionary designs of FPGA boards can end up relying upon idiosyncratic properties of the board(s) and create circuits that "shouldn't work" based on an idealized electrical circuit model. And they may not be transferable to other boards. In other words, to "understand" some evolutionary FPGA circuits, you need to "observe" more than just the gate configurations and idealized schematic.
Brains are not FPGAs or even circuits, but I think the analogy holds. They're not _just_ idealized representations of spiking neural networks.
I remember reading an article about this years ago - a non-transferable FPGA configuration which had some logic that should be unreachable but didn't work without it. Very faschinating but I've not been able to find it since.
I agree and think the way forward is computer simulation. Figure out a simulation equivalent of a neuron and put a lot together and see how well the simulated brain portion matches the real thing. That hasn't really been possible due to lack capability of computer power but we are getting to where it could be. They have of course done stuff with "artificial neural networks" of the ChatGPT type but those are very far from biology.
Yeah and you can do a simulation using a rough approximation and then it is interesting to see in what ways it matches real brains and in what it does not.
It seems the DeepMind rat brain didn't really try to get close to biological neurons but still produced outputs close to a real brain. (vid from here for about 2 mins https://youtu.be/hIuXXvWY6-4?t=125)
Look up Wada Test, where we shut down each hemisphere (one at a time) of the brain, while the patient is awake, to see how their personality changes without 1/2 their brain. Also stereo EEG. 200+ sensors placed in/on/around the brain, patient stays in bed for a week measuring brain activity. Here's the kicker... neurologists can plug into the brain and send teeny bits of electricity to each of the 200+ sensors individually to see what happens (typically this is for identifying seizure spots). Wish I didn't have to know this, but my partner is very epileptic, and has an RNS device implanted in her head. We upload her brain activity daily!
I did mention people specifically. Mammalian brains are fairly more complex than piscine brains.
People are important because we can't tell a fish to perform an action or to think of an elephant, etc. So while we can image, we can't know what correlates with a high degree of certainty.
But even in other cases, a working brain is an entirely different object than a non-working one. Finding those foundational ideas and basic operating principles because we can't test a living brain. It's almost chicken/egg. Being able to experiment on a living human brain would probably give us much better ideas of the basic operating principles. And once we know those principles, we could probably move from experimenting on living human brains.
Like, we can dick around with muscle. We can force it to perform as it were living and receiving signals. And that's partly because you can cut open a living person and watch those muscles operate. You can manipulate those muscles in a living person. You can measure things, etc. It's much harder to do that with a brain. Because it's incredibly easy to make irredeemable mistakes.
>So while we can image, we can't know what correlates with a high degree of certainty.
I don't think I agree just on this piece, though you are making important points here. We can know, for instance, of visual or spacial awareness, reactions to certain stimuli or objects introduced in their environment, and so on.
You're not wrong on the importance of learning from active brain functions but I think we can at least make meaningful inroads with animals.
> Finding those foundational ideas and basic operating principles
I know a lot of people talk this way about brains and consciousness, that we need to first have our definitional elements before we proceed with research. But I think we often have, as the object of research, closing in on those elements, while gathering all kinds of functional and theoretical data from the mass of complicated noise that is brain structure and activity. We can build out correlations, mechanisms, and so on and close in on foundational elements.
It may even be the case that foundational organizing principles turn our understanding inside out in important ways, but I have never understood the idea that we are stuck if we can't start with such things in hand.
I'm talking about very specific correlation. Like when I type these words, exactly how and what fires to achieve that. Which neurons and pathways are the thought "hit these keys", what's the signal to the fingers etc. We know "When this area is active, this group of actions can happen".
It's hard to get good resolution.
And I believe it would be easier to replicate consciousness if we knew how it worked, I also believe it's possible we eventually figure it out. It will just be much harder. And we may not know when we do it either. I believe that is where we will need common, agreed upon definitions.
But defining intelligence and consciousness is difficult because you have to define it in a way that includes everything we agree is conscious and excludes everything we agree is not. And we'll also have to agree what is and is not conscious. It's a way more involved problem than a lot of people give credit.
>But defining intelligence and consciousness is difficult because you have to define it in a way that includes everything we agree is conscious and excludes everything we agree is not.
I think this is extremely important, but I would say that methodologically, messy definitions can nevertheless be stepping stones that bring us closer to full comprehension of definitions. I suppose I just have a specific hobby horse in this context and I don't know that I'm responding to you so much as a general idea I've seen elsewhere, that you're rhyming with, which is this idea that research has to start from agreed upon definitions.
I would say instead that there are such things as characteristic instances that embody lots of the stuff we are interested in, such that we can say "whatever consciousness is, it involves that" and we can just point to something that's a wild, underdefined phenomenon, and make it subject of study. Without a definition, I can still understand that states of mind that involve, say, wakeful awareness and intention are going to be about consciousness. We can know necessary conditions even if we don't know "sufficient" conditions, and the former are a stepping stone to comprehending the latter.
Animals are quite easy to get to do all sorts of experiments, and per very similar anatomy it seems very likely that the basic fundamental operating principles are same across species. If we can't figure out how a fruit fly's nervous system works, we have no hope at figuring out the human brain.
There are very few non-destructive ways we can get a look into a brain while it is living. And even research into ways we can look into a living brain is hindered by the fact we don't want to harm the person being observed.
I could see that making progress a lot slower