1

The hard problem of consciousness is stated as- 'why objective, mechanical processing can give rise to subjective experiences.'

The reason I ask this question is that if we do not even know what a possible solution to this problem should look like, then perhaps this is an unsolvable problem.

So, my question is, what would a possible solution look like? What issues should it address so as to qualify as a solution?

For example, if I have a problem to produce a theory that explains the behavior of a physical system, then a possible theory should constitute a way to mathematically represent the state of the physical system and a law as to how this state evolves over time, etc. This can be formalized using differential equations, etc.

What would an analogue theory to solve the Hard problem of consciousness look like?

13
  • 1
    The reason the hard problem is hard is that we do not know at present what a possible solution might look like. Does it mean it is unsolvable? No, it does not. Ancients did not know what a possible solution to squaring the circle looks like, yet it was solved 2 millenia later. In a way they could hardly imagine. Some answers require conceptual apparatus we did not develop yet, trying to time skip and anticipate them in terms of available concepts won't help.
    – Conifold
    Jun 25, 2023 at 9:06
  • @Conifold Is my answer to this question also addressing only the easy problem? I would be interested to know your opinion. In anticipation of your answer being yes, I want to emphasize that my answer proposes the need for a new law the axiomatically proposes the existence of an entity, yet not understood, called 'conscious observer', associated with some physical systems in the universe. Deciding which physical systems have a 'conscious observer' associated with them and which do not is a decision problem also to be solved by the theory.
    – Prem
    Jun 25, 2023 at 9:18
  • Postulation of entities hardly explains something, we can postulate qualia, correlate them to something measurable, dissect and map them, etc., none of that is what hard problem asks for. As formulated by those who believe in it, is not about "explaining that", which underlies all current forms of scientific explanation, because it is not asking to explain a "that", something propositional. Their presuppositions might be mistaken and the problem might be dissolved as misguided, but it cannot be solved by postulating, detecting, correlating... We'll need a new mode of explanation.
    – Conifold
    Jun 25, 2023 at 9:30
  • 1
    @Conifold, who squared the circle? Seems like that should have been huge news, especially since it was proven impossible. Jun 25, 2023 at 14:21
  • 1
    @DavidGudeman I am saying that what was done is proving its impossibility. That was the solution to what "squaring the circle problem" meant, since already Greeks guessed that literally doing it (with straightedge and compass) was impossible.
    – Conifold
    Jun 26, 2023 at 4:29

3 Answers 3

2

A solution to the hard problem of consciousness would be a mathematical formula to determine what consciousness is present, if any, in a formally described physical system, and what that consciousness is experiencing.

We can think of the solution as a "black box."

  • The black box takes as input a mathematical description of a physical system. Precisely what form this description takes is open to discussion. For example, it might be a set of differential equations giving the time evolution of the physical system, depending on its inputs.
  • The black box yields as output a mathematical description of any consciousness present in the physical system. The precise form of this description is also open to discussion, but perhaps it would involve variables for the different elements of qualia, similar to the pixels on a computer screen, and describe their evolution over time, perhaps using differential equations.
  • The output yielded by the black box should substantially agree with an individual's direct observations of their own consciousness.
  • The black box should be a fairly simple mathematical theory, in accordance with Occam's razor.

Anticipating some objections:

  1. There are philosophers who argue that this is not a solution at all, that there is an insurmountable gap between the mathematical description of qualia akin to pixels on a computer screen, and the actual experienced qualia. Perhaps there is such a gap. It doesn't seem that gap can be crossed with anything we could ever write down; there will always be a difference between the words written and the actual qualia. However, the black box would at least be a substantial stride forward in our understanding, and would model the mathematical structure of the qualia.
  2. There are philosophers who think that this is inherently impossible for some reason. Non-functionalists, for example.
8
  • 3
    What you describe is solution to the "easy problem", finding and mapping physical correlates. There are many philosophers who consider "hard problem" to be a pseudo-problem, but I can't think of any who take it as a genuine problem and consider something like this a solution to it.
    – Conifold
    Jun 25, 2023 at 8:53
  • @Conifold No, the easy problem is the domain of neuroscience in application to specific complex biological organisms we're familiar with. What I've described is instead about fundamental mathematical principles of consciousness, with the key characteristic that it could generalize to alien organisms or artificial minds. That's not part of the easy problem.
    – causative
    Jun 25, 2023 at 8:59
  • @Conifold Integrated Information Theory is an example of a (partial) attempt to provide something similar to the "black box" I described, and IIT is considered (by its proponents, not its detractors) an attempt to solve the hard problem. You're only repeating objection (1).
    – causative
    Jun 25, 2023 at 9:03
  • 3
    Chalmers and others do not tie hard/easy problems to neuroscience. "Explaining why consciousness occurs at all can be contrasted with so-called “easy problems” of consciousness: the problems of explaining the function, dynamics, and structure of consciousness", IEP. IIT is a prime example of a theory that would say nothing about the hard problem even if information integration perfectly correlated with consciousness (which it does not). Indeed, it is a theory of access consciousness only, just like HOT theories.
    – Conifold
    Jun 25, 2023 at 9:17
  • 1
    The "usual methods of science" are hardly restricted to neuroscience, and the arguments of Chalmers et al. appeal to apparent non-susceptibility of what they see as the hard problem to any "objective scientific description" broadly construed. Your abstraction from specifics of neuroscience, does not bring in anything new by their lights. And what individuals can formulate and report is not qualia, so the black box will not even be "validating" what's intended, only detachable access consciousness they accompany.
    – Conifold
    Jun 25, 2023 at 9:43
1

I am old now, but I remember in my teens being introduced to the idea of dimensional analysis in physics, and it struck me then, as it still does now, that it would be surprising to find a physical explanation for consciousness, since that would require reducing consciousness into some kind of unit based on length, mass, time etc, which seems inherently unpromising.

Perhaps information theory might offer a more useful perspective. Imagine a computer screen upon which is printed what seems to be a random stream of numbers. Physics can only go so far in explaining the phenomenon of the numbers on the screen. Physics can account for why certain pixels are more luminous than others, by reference to electrical potentials in circuits and so on, but it cannot account for the numbers being meaningful symbols, and it cannot impart meaning to them. Are they just a random stream? Are they the face of Beyonce encoded in some way? Are they my bank details encrypted with some hash? Those questions can't be addressed by physics, although you do need physics to count for the way in which the numbers are displayed by the underlying substrate of the computer screen. I suspect some kind of analogous explanation will be required for consciousness. There will be an enabling workspace, as it were, for consciousness, which is accounted for by the electrical states of neurons in the brain (in the same way that pixel intensities can be accounted for by electrical states in the computer display), but consciousness per se may be a level of abstraction above that, incapable of being described by physics alone.

0

Assume the Universe to be well modeled by a Turing Machine (TM), with the data on the tape encoding the current state of the Universe. Then, to solve the hard problem of consciousness, we need a theory that can do the following things-

  1. Give a method to (approximately) recognize which bit string on the tape of this TM are 'conscious'.
  2. Propose an extra law that formally associates these 'conscious' bit strings with an entity called 'conscious observer'.
  3. Study the different types of subjective experience and their relationship with each other.
  4. Study if the conscious observer is just a mute observer or can also have influence on the TM

Some comments-

  • Given the state of the universe, it may not be easy to recognize which subsystem is conscious and which is not. For example, I am tempted to say that every time there is a functional brain, then it is conscious, but even AI may become conscious in future, so it is tricky.
  • The law that formally associates these 'conscious' bits with entities called 'conscious observers' basically explicitly acknowledges that we live in a Universe which, when modelled by a TM, has extra entities called 'conscious observers' associated with it.
  • The theory should define different types of conscious experiences, like sensation of 'redness', 'heat' or 'fear', and study the relationship between these entities.
  • If the conscious observer has influence on the TM, is the influence algorithmic? If yes, then it would be equivalent to it having no influence, because any such algorithm can just be incorporated in the original TM. On the other hand, if the influence itself is not algorithmic, then this raises the question- Is the universe well modelled by a TM?

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .