Shayne Wissler
Imagine a world where we figured out the right direction to push, and then we pushed in that right direction…

Understanding understanding

January 02 2017

Is it possible to understand our own understanding, in the sense that we understand how the underlying physical factors can produce such a thing as understanding? We understand the question, but it’s unknown whether we are intelligent enough to answer it. We don’t even know whether the universe is such that the most intelligent being possible to this universe could.

All scientific questions start off this way: we proceed with a sort of trust (some might say faith) that we can find answers, and, lo and behold, we do find them much of the time (more precisely: the most brilliant among us create new theories, lesser minds learn and teach these, and then most of humanity lives their lives without ever learning more than the marketing brochure version and yet feeling proud and patriotic about “how far we’ve come” anyway).

A simpler sort of problem: Can we understand consciousness? Understanding, in the sense meant here, is unique to human beings; but consciousness is a far less ambitious thing to understand. Our pet dog is conscious. Can we understand how it is possible that physical factors create its subjective experience? Could we go one step further and engineer an artificial consciousness? And not that cheap parlor trick of making something that merely appears to be conscious, but a physical object that has what we have, namely a subjective first-person experience? (Such a trick is inherently relative, it depends on how sophisticated the audience is.) It’d be foolish to think we could do this accidentally, we have to know what we’re doing. Is it even possible to know what consciousness is in the requisite sense of something that we can engineer?

If understanding our understanding is possible, then this may be the ultimate kind of understanding. Indeed, it may even be the case that in order to know how it can be possible that conscious minds physically exist, we must also know how simple particles can give rise to such a thing as the universe, not just in a general sense but in the sense of understanding precisely how all particular qualities emerge from fundamental particles. We aren’t even close to this – consider the fact that we can’t even predict the properties of water from “first” (are they really first?) principles, let alone basic properties of our own biology (consider the incredible difficulty of just one of the elemental parts – protein folding). There are no guarantees that consciousness is some easy set of principles like Newtonian Mechanics or Special Relativity. And it may be the case that understanding how it arises from matter necessitates a complete understanding of matter and how it combines to create a life form. We just don’t know – we can’t, until we actually arrive at an understanding.

We don’t know what the units of an explanation would be. Some of us like to think that simplistic models of neurons, networked together, would be enough. But what if the specific molecular internals need to be understood too, including their quantum behavior? Or what if the sub-quantum behavior is relevant too? What if the requisite physical knowledge is only available through smashing particles an accelerator bigger than the Earth? As much as Nature permits us to discover, there is no particular reason why we should think that everything about it is knowable, and what is needed to understand consciousness just might be physically unknowable by us. That may be disturbing to those who put ambition before competence, but humility before Nature is just as important as the bold exercise of the capabilities she has given us.

If we could go back in time and tell Pythagoras about our modern computers, surely he’d want to know how they worked. If we wanted to help him, the first thing might want to tell him is how difficult such an understanding is to reach: our brightest spend the first part of their lives learning just one aspect of them, and even then they typically can’t contribute to further development. Generations of genius, and indeed starting with men like Pythagoras, were required. Given his metaphysical inclinations, we might want to sternly warn him about cargo cults – taking that kind of short cut would completely undermine his understanding. Instead, he should very patiently develop his knowledge of all the prerequisites (modern physics, chemistry, material science, transistors, lithography, digital logic, computer architecture, compilers, etc.), and only then would he see how our modern computers are possible. The urge to take shortcuts around so daunting a project might be ever-present, but such is only a Siren song that will dash the possibility of understanding on the shores of pseudoscience.

The tribalism of humans makes them like to believe that whatever institution they are part of is some sort of paragon of wisdom, but if we just look back at all the previous humans who thought that about their institutions, we’d have a hint to take: to think that we are even somewhat close to knowing everything there is to know, or even to be convinced that our institutions are on the right track, is a brutally unforgiving trap. For all of history, our institutions have been disastrously wrong about crucially important things, and have served more often to slow down our progress than to advance it. Usually it was the lone individual willing to suffer institutional persecution who has actually been the cause of our increasing institutional knowledge. Anointed leaders of our institutions will laud their alleged virtues, but when in history has any such pimp been a source of progress?

If an understanding of understanding is to come, it will come from a line of patient intelligence working on the prerequisite problems while assiduously avoiding cheap shortcuts. What are these prerequisite problems?

One problem is philosophical. We need to know how to think clearly about consciousness, intelligence, and understanding, and we need a proper philosophy of science.

Other problems concern science and engineering. I don’t intend to create a comprehensive list, but let me point out one area I think would bear particularly worthwhile fruit: man-machine neuronal interfaces. Instrumentation of our brains leads directly to all sorts of immediately useful things for the handicapped[1]; we can, with enough scientific and engineering insight, make the blind see, the deaf hear, the paralyzed walk. Furthermore, our best source of empirical knowledge concerning the mind is indeed our own minds; to be able to finely instrument them enables a new type of scientific work: the systematic gathering of knowledge of precisely how and where the physical brain produces various types of internal and external awareness and control.

  1. Consider the disadvantages of the multielectrode arrays and its inferiority relative to our native biology to get an idea of how much work remains to be done.