The title is from a blog post by Brian Leiter. Leiter links to an article in the LA Review of books: Imitation and Extinction: The Case Against Reality. The article is written by Donald Hoffman.
We have discussed the general topic before, in several threads. So maybe this is a good time to revisit the topic.
Hoffman asks: “I see a green pear. Does the shape and color that I experience match the true shape and color of the real pear?”
My take is that there is no such thing as the “true shape and color of the pear.”
It is a common presumption, that there is an external standard of truth. Here, I mean “external to humans”. Truth is presumed to come from somewhere else. And our perceptual systems evolved to present us with what is true.
As I see it, this is backwards. Yes, our perceptions are mostly true. But this is not because perception is based on truth. Rather, it is because our human ideas of truth are based on what we perceive.
Open for discussion.
Regarding The Turing test, and whether p-zombies can be conceived of.
I would say that a useful conception of p-zombie would require an operational definition, a method of distinguishing soul possessing from soulless. We might call it the Turing test.
I don’t think consciousness ever ceases. It is either hidden or manifest. And I would say it is manifest in any individual organism which possess a nervous system from the most rudimentary up to the human nervous system which is the most complex physical feature of living nature.
Look for a nervous system and you will find consciousness in some form. As in Alan’s reply above, it is not either present or absent, it is on a scale of manifestation.
As best I can tell, what we consider to be a pattern is entirely dependent on human conventions. What can be cycles and epicycles with geocentric conventions becomes ellipses with heliocentric conventions. What can be a donut shape with one set of conventions becomes a coffee cup shape with another set of measuring conventions.
Our perception gives us information. The senses, by themselves, do not.
Perception is engaged in constructing information.
Neil Rickert,
So sayeth Gibson.
So sayeth Steiner.
I’m not so sure that’s right. Gibson talks a lot about picking up information from the optical array.
petrushka,
Remember that p-zombies are defined as behaving exactly like humans. If a human passed the Turing test, then so would their p-zombie counterpart.
The Turing test is a test of intelligence, not of conscious experience.
keiths:
Bruce:
Yes. I was responding to KN’s blanket statements regarding the conceivability of the impossible, such as this one:
Corneel,
The problem is with the phrase “true states of the world”. Does a state include the position and momentum of every particle in the system being perceived? Obviously not, because then there would be far too many states for a 1:1 mapping between states and distinct perceptions.
So there is already some coarse-graining going on, but Hoffman never offers a rationale for coarse-graining on one variable (resource quantity) versus another (expected payoff). It appears that the former just seems more objective to him than the latter, but I can’t see why.
CharlieM,
It’s the product of perception, not the process itself, that can be veridical or non-veridical. Think of perceptual illusions.
But he is a giant with shoulders to stand on, metaphorically speaking.
Bruce,
I’m thinking of information generically as something that reduces uncertainty in the receiver.
Eyes closed, you’re uncertain of what’s in your visual field. Eyes open, your uncertainty is reduced. You’ve received information from your visual sense.
If I want to pour myself a glass of milk, it doesn’t suffice that we, human beings, agree that HERE is a glass. There actually has to be a container capable of holding the liquid.
I ask again: How do I succeed in accomplishing any task when there is no correspondence between perceived objects and true structures in the world, save that established by human convention? I simply cannot see how that would work.
In fact, I believe that is exactly what Hoffman and colleagues mean. They call that the omniscient realist strategy.
All realist strategies preserve the one-to-one mapping for a subset of W. Only the interface strategy allows mapping of multiple world states onto a single perception event. I admit that I don’t really understand what that means. They could have done a better job linking up the definitions to some real life examples.
Firstly, the claim that we can’t conceive of impossibilities is not my claim — it’s Chalmers’s claim, and he uses it to argue from zombies to the hard problem. And to be clear, we are talking about logical possibilities, not physical (‘nomological’) possibilities.
But I think there’s something to be said in defense of the idea that we cannot conceive of things that are logically impossible, ergo whatever we can conceive of is logically possible.
Suppose I say, “I”m conceiving of a square circle”. You press me a further on what I mean. I tell you that I’m conceiving of a plane geometric figure with equal sides and four 90 angles that has all points on its perimeter equidistant from the center. You say that this is impossible, but I insist that I’m conceiving of it anyway. Of course this means I’m committed to affirming that this object has contradictory properties, e.g. that the perimeter and area of the figure are and also are not calculated using an irrational number.
At this point one might quite plausibly think that if I’m committed to affirming contradictions, there’s no point in having a conversation with me.
The point here being that I’m not sure it makes any sense to say that we can conceive of logical impossibilities — that we can even entertain, as a candidate for endorsement, a claim that entails a contradiction.
What Chalmers wants to claim is that zombies are conceivable, hence logically possible — since if they were logically impossible we could not conceive of them.
Where the argument seems to get wonky, to me, is the blurring of the distinction between logical modality and physical modality. The neuroscience of consciousness could, speculatively, tell us why zombies are physically impossible in this universe — and so the existence of zombies in some possible world is perfectly compatible with a solution to the hard problem.
I know there’s a discussion of logical and physical modality in Chalmers but I don’t recall it right now. I’ll look it up if this continues to bother me.
He does take it to be metaphysical, but the argument for that claim is grounded in epistemic and semantic considerations, including (most importantly, for Chalmers) two-dimensional semantics. (Needless to say, as a Brandomian inferentialist about meaning I’m not on board with this aspect of Chalmers’s entire way of doing philosophy.) Chalmers became a big deal not because he invented the idea of the explanatory gap but because of the sophistication of his arguments for why the explanatory gap cannot be bridged.
Just saying, your p-zombie is an empty construct, without an operational definition.
There are, of course, people who fail the usual consciousness tests.
glass of milk: an abstraction established by a human naming convention
pour: an abstraction established by a human naming convention.
actually be: an abstraction established by human convention.
holding: an abstraction established by human convention.
liquid: an abstraction established by human convention.
And I ask: “How can there be a correspondence except by virtue of human conventions?” How could the word “correspondence” have a meaning, except by virtue of human meaning conventions?
Human conventions are ubiquitous. A newborn child, not yet in touch with social conventions, will have to start making his/her own private “conventions” as a way of getting started. I used quotes around that last “conventions”, because perhaps if it is private, it shouldn’t really be called a convention.
petrushka,
Note the following sentence from the quote you provided:
That’s precisely the point I was making here:
You’re thinking of something distinct from the Turing test.
Neil,
Whether you name it or not, the glass has to be there in order to hold the milk. Reality calls the shots, not human naming conventions.
petrushka,
It can’t be investigated scientifically without testable consequences, but that doesn’t make it an “empty construct”. Hence the richness of the philosophical literature devoted to the topic.
KN,
I think there’s a difference between directly conceiving of a contradiction and conceiving of something that entails a contradiction.
The square circle is an example of the former. Here’s an example of the latter:
Suppose someone specifies two large numbers x and y and asks us to conceive of a prime number lying between them. I can conceive of a number that satisfies those criteria even if it turns out that none actually exists. To use your phrasing, I can entertain, as a candidate for endorsement, the claim that such a prime exists. The contradiction isn’t obvious to me at first glance, the way it is with the square circle. It takes a lot of work to figure out that there is no such prime.
I have not been dismissing reality.
Corneel,
Is that really true? Suppose that W1, W2, … Wn are distinct world states that are perceptually indistinguishable. That means that all of them map to a single perception, Xm. It’s a many-to-one mapping, not one-to-one.
I question that for the reason given above.
Yeah, the shortage of concrete examples is a real problem with the paper. Perhaps Hoffman provides more of them in his book.
Neil,
Sure you have, by denying that there is structure to it. Reality is highly structured, and it calls the shots.
Structure is conventional.
If you want to say that reality is calling the shots, then you have to leave the structure out of it.
That hardly disqualifies it as an empty construct. Free will and determinism are empty constructs, in the absence of operational definitions.
If it were an empty construct then there would be nothing to discuss. There’s plenty to discuss, and we’re discussing it in this very thread.
Neil,
Why?
Already answered, in the sentence that you did not quote.
Then you would have no trouble listing the well supported conclusions reached by consensus.
What I see is a bunch of assertions and empty definitions.
I haven’t seen much added to the concept of objects, beyond the common sense understanding. I stub my toe when I kick it.
Is ‘truth is based on perception’ true? It’s not been my perception. The two seem quite often at odds.
petrushka,
That doesn’t follow. “Non-empty constructs” can be controversial.
Neil,
“Structure is conventional” doesn’t answer my question. I’m asking why you believe that
You’ve already acknowledged that reality is “lumpy”. Lumpiness is structure. Why ignore that or try to “leave the structure out of it”?
I am studying cognition. A large part of that is related to how we manage to connect to reality. This is roughly what KN calls “cognitive friction”.
It turns out that it depends on arbitrary pragmatic behaviors. These are roughly the private equivalent of conventions. To do this in a way that can be shared by a society requires social conventions.
You miss the point.
To say that it is lumpy is just to say that it is not homogeneous, and that falls short of being structure. In any case, whether or not reality is homogeneous is already conventional.
Structure is abstract. That should already tell you that it is not real (unless you are a platonist). Structure is geometric, and geometry is conventional. Traditional Euclidean geometry is the study of the consequences of using our standard measuring conventions (such as a yard stick).
To be very clear — when I refer to conventions, I am talking about pragmatic choices that are shared with the community. I am not talking about purely random choices. It is important that these are pragmatic choices. And part of what makes them conventions, is that they are underdetermined by reality. That is to say, we could have made different pragmatic choices.
We are not talking if you think our perceptions are mostly accurate or only somewhat inaccurate at times. I was talking about what Hoffman believes, which is that our perceptions are completely wrong about reality. As I said, he goes so far as to say space-time does not exist.
So I don’t know what one means when one says science is not so “fragile” that it can’t contend with some aspects being wrong. If one says space-time is a construct of our perception myopia, claiming other aspects of science are unaffected (such as our perceptions on natural selection) is just silly rationalizing.
I understand your point that we use human conventions to make sense of the world, and that those conventions are somewhat arbitrary and pragmatic. Fine. What I don’t agree with is your suggestion that patterns are exclusively decided by human convention.
Suppose we agreed that I should pour my milk 10 centimeters* left of the spot where we observed to be a glass. Then we end up with the abstraction that human naming convention has labeled “an enormous mess”. Hence, those two locations are not equal in terms of usefulness for holding liquids. We tune our conventions so they enable us to successfully interact with the outside world. To establish useful conventions, we use feedback from the world outside us.
*Or inches, or whatever measuring convention suits you
I think you are right.
The way I read “preserves all structures on W” is that multiple mappings were not allowed in a realist strategy, but in figure 2 it seems it is OK to bin large sets of world states onto a single perception, as long as the ordering is preserved.
ETA: correction
I have a really hard time getting my head around Hoffman’s argument. Needless to say, as long as I don’t understand his arguments, I am not really convinced by them and I don’t feel the need to defend them.
Every party to the argument agrees that if zombies can exist in some conceptually (metaphysically) possible world, then physicalism is false.
The argument is under study is, roughly: zombies are conceivable; conceivable means metaphysically possible; hence zombies are metaphysically possible; hence physicalism is false.
Chalmers defines two types of physicalists: type A and type B. Both Chalmers and Dennett think type A physicalism is correct. Ned Block, for example, thinks type B.
Physicalism is defined in terms of the language of physics only. Further work of some kind is needed to go from the language of physics to all everyday and non-physics-sciences concepts.
Type A physicalists think reflection in an armchair is enough for that further work. That is, upon reflection only of all the physical facts zombies will not even be conceivable. It’s important to qualify that conceivability to include reflection because it may not be obvious without reflection upon all the physical facts. (That’s the reason you and KeithS disagree, I think). If zombies are not even conceivable, the argument fails.
Type B physicalists think that even after we know all the physical facts, we will still need empirical work to understand how all our concepts relate to the physical facts. So before we do that work for zombies, they would be conceivable. Type B physicalists rely on Kripke rigid designator ideas to claim that we once we discovered the right identities (or possibly physical realizations), then we would could conclude that zombies are conceivable but not metaphysically possible and so the argument fails at a later step.
A standard example is water and H2O. Type A physicalists think that once we understand the physical facts and the concept water (that wet stuff in lakes and rivers), then we will be able to deduce H2O=water. Type B think we would still need empirical work to do this. The same idea applies to your neuroscience example: type A would say once we know all the facts expressed in language of physics, we would know from our everyday concepts and the concepts of neuroscience that zombies are not conceivable. Type B would say they would remain conceivable until further empirical work is needed to relate the language of physics to all of our concepts.
I did spend some time with 2D semantics and how Chalmers uses it at one time. Looking at my notes, he uses it against the type B physicalist to show why in detail Kripke’s work fails for the zombie case (Kripke said this himself, but not as formally).
Chalmers starting point paper for type A/B physicalism is Consciousness and its Place in Nature
http://www.consc.net/papers/nature.html
There is an intro to the discussion here:
https://reducing-suffering.org/hard-problem-consciousness/
Since I understand this to be your field of study as a mathematician, I am not sure in what sense your are using the word “geometric”. But when I refer to structure, I am assuming only some general mathematical structure such as
– the topological spaces used in GR
– group theory for the standard model of particle physics
– some other kind of abstract algebra for QFT
– Bayesian probabilistic causal networks and the corresponding joint pdfs for PP perception
It is true that there is excess mathematical structure in scientific theories so there is underdetermination in that sense. For example, gauge invariance is a fundamental principle used in building scientific theories, and it can be interpreted as implying excess mathematical structure. Another example is the convention used in SR that we assume the light travels at the same speed in the to/from directions when synchronizing clocks in a reference frame.
So when we argue that the success of science implies that reality has the structure of the math theory, then we have to add the restriction that only the part of the structure that affects the empirical predictions matters.
Of course, that is still just an intuitive case and needs to be justified for each structure. And there is the deeper issue that since different sciences might use different math structures, in what sense is there a single structure to reality. Those are reasonable and open questions that various philosophers have different answers to.
On gauge invariance
https://en.wikipedia.org/wiki/Introduction_to_gauge_theory
https://physics.stackexchange.com/questions/266992/what-in-simplest-terms-is-gauge-invariance
On to/from in light speed, dga and PdotQ posts in
https://discourse.peacefulscience.org/t/one-and-two-way-speed-of-light/3919
If you are willing to share them, I’d like to get your views on a comparison of consciousness between, say, adult humans, embryos, plants, roundworms, and bacteria? Are they all equally conscious? Are only some of them conscious? Do they have different levels of consciousness, different types of consciousness?
But isn’t uncertainty dependent on the nature of the receiver?
If we are not talking about Shannon information, then perhaps you are thinking of semantic information in the sense that Dennett describes in Chapter 6 of the Bacteria to Bach book.
Here, information is a “difference that makes a difference” and, as he says, you need to include in that a discussion of the organism and how it makes a difference to that organism.
My basic point is that, as far as perception goes, any type of information is a property of the relation between the organism and the environment; it is not just a property of the environment.
If we move beyond perception to science, that may not be true, at least for quantum information. (Shannon info may still involve the observer for the probabilities, eg as in the discussion of TD entropy).
Bruce,
Sure, and it’s even a function of time. The same stimulus at two different times can reduce the receiver’s uncertainty by different amounts.
None of that changes the fact that information is flowing from the source (external reality) to the receiver (the organism). Interrupt the flow of photons (in Neil’s cat example) and you interrupt the flow of visual information.
If we see the rays of the sun spreading out towards the earth from behind a cloud, they make various angles the apex of which originates in the sun. Some would say that this is a perceptual illusion as the streams of light are in actual fact virtually parallel to each other. But, because the retina is a two dimensional surface, the patterns of light falling on it do in reality possess the angles as we see them.
It is our concepts which can be veridical or non-veridical. But our concepts are not the product of perception, we add them to our perceptions so as to gain the full reality. We gain a more complete knowledge of sunlight with the concept of perspective.
‘Seeing’ through understanding (conceptually), puts back together what we had initially separated out by our ‘seeing’ through our senses (perceptually). Our perceptions are not so much wrong as incomplete without the added concepts.
Animals do not have this level of understanding and nor do they need it to survive. An archer fish can compensate for light refraction and the parabolic path of projected water while knowing nothing of these concepts. All it need do is learn to orientate itself in such a way so as to successfully hit the target. It doesn’t give any thought to the veracity of its perceptions, all it cares about is the accuracy of its shot.
Through thinking we can come to an understanding of how our perceptions relate to each other, we perceive shapes and colours in relation to their surroundings and make adjustments accordingly.
I think that perception is not just bottom up; it is top down as well. The evidence starts with the neural connectivity of the brain. It also comes from higher level descriptions: many modern theories of perception, such as Bayeisan approaches like PP, rqeuire simultaneous bottom up and top down processing (along with action and not just static sensing).
Because of that, I don’t think it is correct to say information flows from the environment. Rather, information from perception is something that depends on the processing an organism does based on its internal state and using both data sensed in the environment and acting in that environment.
This is one where I think Neil is basically right, although no doubt we differ on details and implications.
That is likely alll I have to say on this subtopic.
Did you miss the word “pragmatic” that I have been using throughout this discussion?
Conventions are arbitrary in the sense that different choices could have been made. But they are not random. They are guided by pragmatic considerations.
That clears things up. I certainly agree that our dividing up the world is informed by pragmatic considerations. But doesn’t that severely restrict the conventions we can come up with? Wouldn’t some of those conventions necessarily converge on similar concepts?
To return to a previous example: your chair on the patio. Ants and birds don’t label your chair the way humans do (which we can agree on is an abstraction). But they still recognise it as a solid object distinct from its surroundings; Ants will climb over it. Birds will perch on it.
Do they recognize it as distinct from surroundings? How would you know that?
Yes, ants climb over it, just as they would if they did not see it as distinct from surroundings. And a similar comment applies to birds.
I have no way of knowing whether ants and birds see it as distinct from their surroundings. And I suspect the same limitations apply to you.
Well, yes. But it still comes down to decisions that we make.
Understanding the decisions we make about how we see the world is an important part of understanding cognition.