Interesting discussion Eric. My first impression is that this seems like another reason to regard consciousness as an amorphous concept that our intuitions are inconsistent about.
But it seems like many of these features aren't exclusive to each other. Self representations seem to imply me-ness, some degree of at least momentary unity, access, and intentionality. I would think many are important prerequisites for flexibility in thought and behavior. And the limitations of these features might lead to the impression of determinacy, irreducibility, and privacy.
So we might have animals or AI systems with some of these features but not others. Per Henry Shevlin, we would then have to decide whether to be conservative and require all of the features to say they're conscious, liberal and only require one or two, or incrementalist and accept that there are varying degrees of consciousness. Or just reject the whole worry and focus on the system's capabilities.
I do think listing these out is productive. It seems to make clear that what we're talking about isn't simple and ineffable.
I agree our intuitions are inconsistent. I'm not sure the concept is amorphous, though. I do think there's an obvious property had by the positive examples of consciousness (e.g. pain, visual experience, inner speech) that negative examples lack (e.g., nonconscious processing, unretrieved memory stores). My view is that we can't just decide to attribute that property to AI systems with X, Y, or Z features. We could definitionally define "quasi-consciousness" in terms of some objective feature list, but then it will be an open question whether quasi-consciousness refers to the same thing as the property we pick out when defining consciousness by example.
I've come to suspect the idea that there's something obvious in the positive examples but missing in the negative ones, that isn't any combination of the features you listed (or others like episodic memory). An obvious property that we can't delineate, analyze, or objectively observe doesn't seem that obvious. (I don't think "illusion" is the right word, unless we mean a theoretical illusion.)
What does seem obvious to me is that for us to ponder and discuss the various positive examples, requires some of the features (self representation, access, etc). Of course, we could argue that the episodes we're recalling didn't have those features, but now we're at the methodological issues you discuss. How do we know a consciously remembered event was conscious at the time of the event? Or even if that's a meaningful question?
I certainly agree about the methodological worries in your second paragraph! On whether the property is not obvious, I'm not sure why analysis would be necessary. "Objective" observation, if that means that others can also observe not just the same type but the same token, seems inappropriate to the case. I'm not sure what "delineation" amounts to, but if it's drawing a distinction between positive and negative cases, I already do that, don't I?
Less fussily, my guess is that people who don't find the property obvious are trying too hard and thinking too much about theory. Isn't it clear that there's something in common between sensory experiences, felt pains, and inner speech that's absent from unrecalled memories and early visual processing? Don't theorize it yet; just point at it. That's the property I mean.
When I say "delineation", I'm looking for something like your description of the features in this post. I don't think you do the same thing for the definition by example cases. You do put them in distinct sets, but stay away from the actual characteristics that make them distinct.
I should clarify that this isn't criticism of you in particular. No one else who takes the pointing strategy does either as far as I can tell.
On not thinking about the theory, I think when you say the obvious property is not any of the features you list, it seems like you're actually making a theoretical assertion. (A negative one, but still a theoretical one I think.)
It seems like if we want to be theory free, we can't make or *deny* any linkages between our introspective impressions and features like self representation, access, etc. We have to be completely agnostic. (Which seems extremely difficult in practice.)
I suppose on a liberal view of what a "theory" is, every judgment is theory laden in some way. What I'm hoping to avoid is commitments on particular disputed theories about the nature of consciousness. People of a wide range of views can agree that imagery experiences, emotional experiences, inner speech, and pain are conscious and early visual processing is not. My assertion is that there's an obvious property they have in common, which we label with the word "conscious". Kind of like if I put a bunch of bright red things in a pile and asked what property they all have in common that these other things (a pile of blue, green, yellow, etc. things) lack. Ordinary people will know I mean redness, even without a commitment to, say, whether redness is a secondary quality, a reflectance property of surfaces, etc.
Hello - I've been thinking about your point that awareness is not a good way to capture the distinction between the positive and negative examples. You're right of course. I might be aware of (in the sense of thinking about or having knowledge of) the planet Pluto (still a planet to me, dammit!), or the iditarod race. That doesn't mean I'm having an experience of Pluto, or of the race (either from the perspective of spectator, musher, or dog.) I might be having an experience of thinking about these things, but that's another matter and, as you say, best to steer clear of "awareness" to avoid conflation.
Another property I've been mulling, that perhaps is more on point, is that what distinguishes experiences from non-experiences is that experiences seem to be present to us, in some (broad) sense of happening "here and now." When I look out my window and see a Douglas fir, it seems to be present. I assume that is also true of a mental image of the tree (though as an aphantasic I can't personally attest.) Same with an occurrent feeling of anxiety (e.g. that as I type this I may be missing something obvious that will later make me cringe.) The feeling is maybe not "here" in a specifically ostensible spatial sense, but still I think the broader sense of the terms captures something important about the experience. When I'm thinking about the iditarod race (rare as that may be), the race doesn't seem to me to be present, but the act of thinking about it arguably does.
This notion maybe has some connection to the idea (as emphasized by David Chalmers in his metaproblem paper - maybe elsewhere that I'm not thinking of) of a sense of presentation as being a key aspect of our purported acquaintance with our experiences, though I'm not sure if it's an exact match. There might also be some overlap with your principle of luminosity, though I'm also not sure about that. It could also be further analyzed - the "now" part, that is - on the lines of the specious or thick present, but that's not necessary, I think.
Anyway, this strikes me as potentially a good way to zero in on the core concept that we're sorting on when we distinguish the positive from the negative examples, but I wonder what you think?
Interesting thought. I wonder if this can be folded into the "specious present" idea or whether it's separable. Experiences occur at a time. But then so also do lots of things. Experience can represent things that occur at times (a tree is here now). But then maybe not always and not always at the present time (a tree was there before). There's something attractive in saying that there's some here-and-now-ness that isn't captured by either of those ways of describing things.
Thank you for the reply, and for the very thought-provoking post! There is interesting overlap I think between some of the purported features. For example on the specious present, though of course it goes back quite a ways at least to William James (and I guess he took the idea from E.R. Clay), I'm sure you're familiar with the emphasis Nicholas Humphrey puts on the idea of the "thick present" in his work. For Humphrey, as I understand him, the brain representing experienced time in this paradoxical way triggers an introspective sense of wonder and mystery. I'm not totally sure I buy in to his ideas about this sense of wonder being adaptive, but a fascinating thought anyway.
Interesting discussion Eric. My first impression is that this seems like another reason to regard consciousness as an amorphous concept that our intuitions are inconsistent about.
But it seems like many of these features aren't exclusive to each other. Self representations seem to imply me-ness, some degree of at least momentary unity, access, and intentionality. I would think many are important prerequisites for flexibility in thought and behavior. And the limitations of these features might lead to the impression of determinacy, irreducibility, and privacy.
This list also reminds me of Kevin Schilback's discussion of the concept of religion, and how polythetic (anti-essentialist) approaches to defining it could be useful. https://plato.stanford.edu/entries/concept-religion/#PolyAppr
So we might have animals or AI systems with some of these features but not others. Per Henry Shevlin, we would then have to decide whether to be conservative and require all of the features to say they're conscious, liberal and only require one or two, or incrementalist and accept that there are varying degrees of consciousness. Or just reject the whole worry and focus on the system's capabilities.
I do think listing these out is productive. It seems to make clear that what we're talking about isn't simple and ineffable.
I agree our intuitions are inconsistent. I'm not sure the concept is amorphous, though. I do think there's an obvious property had by the positive examples of consciousness (e.g. pain, visual experience, inner speech) that negative examples lack (e.g., nonconscious processing, unretrieved memory stores). My view is that we can't just decide to attribute that property to AI systems with X, Y, or Z features. We could definitionally define "quasi-consciousness" in terms of some objective feature list, but then it will be an open question whether quasi-consciousness refers to the same thing as the property we pick out when defining consciousness by example.
I've come to suspect the idea that there's something obvious in the positive examples but missing in the negative ones, that isn't any combination of the features you listed (or others like episodic memory). An obvious property that we can't delineate, analyze, or objectively observe doesn't seem that obvious. (I don't think "illusion" is the right word, unless we mean a theoretical illusion.)
What does seem obvious to me is that for us to ponder and discuss the various positive examples, requires some of the features (self representation, access, etc). Of course, we could argue that the episodes we're recalling didn't have those features, but now we're at the methodological issues you discuss. How do we know a consciously remembered event was conscious at the time of the event? Or even if that's a meaningful question?
I certainly agree about the methodological worries in your second paragraph! On whether the property is not obvious, I'm not sure why analysis would be necessary. "Objective" observation, if that means that others can also observe not just the same type but the same token, seems inappropriate to the case. I'm not sure what "delineation" amounts to, but if it's drawing a distinction between positive and negative cases, I already do that, don't I?
Less fussily, my guess is that people who don't find the property obvious are trying too hard and thinking too much about theory. Isn't it clear that there's something in common between sensory experiences, felt pains, and inner speech that's absent from unrecalled memories and early visual processing? Don't theorize it yet; just point at it. That's the property I mean.
When I say "delineation", I'm looking for something like your description of the features in this post. I don't think you do the same thing for the definition by example cases. You do put them in distinct sets, but stay away from the actual characteristics that make them distinct.
I should clarify that this isn't criticism of you in particular. No one else who takes the pointing strategy does either as far as I can tell.
On not thinking about the theory, I think when you say the obvious property is not any of the features you list, it seems like you're actually making a theoretical assertion. (A negative one, but still a theoretical one I think.)
It seems like if we want to be theory free, we can't make or *deny* any linkages between our introspective impressions and features like self representation, access, etc. We have to be completely agnostic. (Which seems extremely difficult in practice.)
I suppose on a liberal view of what a "theory" is, every judgment is theory laden in some way. What I'm hoping to avoid is commitments on particular disputed theories about the nature of consciousness. People of a wide range of views can agree that imagery experiences, emotional experiences, inner speech, and pain are conscious and early visual processing is not. My assertion is that there's an obvious property they have in common, which we label with the word "conscious". Kind of like if I put a bunch of bright red things in a pile and asked what property they all have in common that these other things (a pile of blue, green, yellow, etc. things) lack. Ordinary people will know I mean redness, even without a commitment to, say, whether redness is a secondary quality, a reflectance property of surfaces, etc.
Hello - I've been thinking about your point that awareness is not a good way to capture the distinction between the positive and negative examples. You're right of course. I might be aware of (in the sense of thinking about or having knowledge of) the planet Pluto (still a planet to me, dammit!), or the iditarod race. That doesn't mean I'm having an experience of Pluto, or of the race (either from the perspective of spectator, musher, or dog.) I might be having an experience of thinking about these things, but that's another matter and, as you say, best to steer clear of "awareness" to avoid conflation.
Another property I've been mulling, that perhaps is more on point, is that what distinguishes experiences from non-experiences is that experiences seem to be present to us, in some (broad) sense of happening "here and now." When I look out my window and see a Douglas fir, it seems to be present. I assume that is also true of a mental image of the tree (though as an aphantasic I can't personally attest.) Same with an occurrent feeling of anxiety (e.g. that as I type this I may be missing something obvious that will later make me cringe.) The feeling is maybe not "here" in a specifically ostensible spatial sense, but still I think the broader sense of the terms captures something important about the experience. When I'm thinking about the iditarod race (rare as that may be), the race doesn't seem to me to be present, but the act of thinking about it arguably does.
This notion maybe has some connection to the idea (as emphasized by David Chalmers in his metaproblem paper - maybe elsewhere that I'm not thinking of) of a sense of presentation as being a key aspect of our purported acquaintance with our experiences, though I'm not sure if it's an exact match. There might also be some overlap with your principle of luminosity, though I'm also not sure about that. It could also be further analyzed - the "now" part, that is - on the lines of the specious or thick present, but that's not necessary, I think.
Anyway, this strikes me as potentially a good way to zero in on the core concept that we're sorting on when we distinguish the positive from the negative examples, but I wonder what you think?
Interesting thought. I wonder if this can be folded into the "specious present" idea or whether it's separable. Experiences occur at a time. But then so also do lots of things. Experience can represent things that occur at times (a tree is here now). But then maybe not always and not always at the present time (a tree was there before). There's something attractive in saying that there's some here-and-now-ness that isn't captured by either of those ways of describing things.
Thank you for the reply, and for the very thought-provoking post! There is interesting overlap I think between some of the purported features. For example on the specious present, though of course it goes back quite a ways at least to William James (and I guess he took the idea from E.R. Clay), I'm sure you're familiar with the emphasis Nicholas Humphrey puts on the idea of the "thick present" in his work. For Humphrey, as I understand him, the brain representing experienced time in this paradoxical way triggers an introspective sense of wonder and mystery. I'm not totally sure I buy in to his ideas about this sense of wonder being adaptive, but a fascinating thought anyway.
Yes, Humphreys is always interesting!
Consciousness.
Some of us have it to some extent some of the time.
This observation seems to make a non-trivial number of people uncomfortable.