I find Ned Block's distinction between phenomenal and access consciousness incredibly useful (see attached 1995 paper). We use the word "conscious" to indicate both felt experience (phenomenal) and what we have "conscious" access to for reasoning and reporting purposes. Machines (currently) mimic access consciousness, but we have zero evidence that they are phenomenally conscious (and unfortunately such evidence may be inherently unobtainable).
Well, I believe we need more clarity on what would count as evidence. We are the very early stages of AI, even though the discipline was born in 1956. There needs to be a reconciliation between modern machine learning and some of the earlier insights of AI. I think it's important to note that our understanding of human consciousness is limited. How can we say that a machine is or is not conscious if we are unable to model it? One doesn't really understand what one cannot model. But what can be modeled can be implemented in a computer (possibly one with analogue components). Here is an example of requirements for a computational theory of human visual consciousness: [Aaron Sloman 1989 - On designing a visual system](https://cogaffarchive.org/Aaron.Sloman_vision.design.html). A long way to go before machines can implement this, which means before we can understand how humans do it.
Oof that paper is way above my pay grade! But I think we share the concern that until we understand human (phenomenal) consciousness better we cannot know whether machines can be conscious. What worries me is that we still have no idea how to prove that even other humans are conscious--eg, to disprove the idea that I am living in a simulation where I am the only conscious being--so how will we do it for machines? But I certainly hope it proves feasible, since there are huge ethical questions riding on this!
I just found where I read the Hebb incidents: This book: The Structure of Emotions: Investigations in Cognitive Philosophy by Robert Morris Gordon. Apparently Daniel Dennett often quoted this. I don't have the primary source, so this could be apocryphal. I'll try to reread the chapter later today.
Yes. I think there are different stances we can take to answer the question: (a) empirical stance; (b) intentional stance; (c) design stance. (b) is was proposed by Dan Dennett in his book [Intentional stance ](https://en.wikipedia.org/wiki/Intentional_stance) (and earlier papers). It's basically a pragmatic / functionalist approach. Even a coke machine, he says, has beliefs and desires. It's about the predictive utility of intentional ascriptions. If speaking about a machine or animal as having specific beliefs and desires gains you a better understanding of its behaviour, then it counts. There are famous studies out of McGill (Montreal, Canada) where *behavioral psychologists* avoided referring to their monkey subjects as having beliefs and desires. This led to all kinds of problems such as the experimenters being injured by the animals. Same goes for consciousness. Sure we can withhold intentional ascriptions, but we do that at our peril-- or so Dennett would have said.
Interesting. But we don't have ethical duties towards coke machines, right? Should there be a (d) ethical stance, or if that somehow in (a)-(c) already?
I took a few courses in moral philosophy as an undergraduate but that was not informed by AI. I can't really comment on ethical issues. Beyond my pay grade, but my intuition is that one day we will have to treat artificial *autonomous agents* as persons. I think if we search for a binary cutoff we will oversimplify the problem. Thus my advocacy for an integrative design-oriented perspective. I plan to submit a much revised version of the [ Manifesto for an Integrative Design-oriented Approach to Understanding Humans as Autonomous Agents – CogZest](https://cogzest.com/projects/a-manifesto-for-integrative-design-oriented-cognitive-science-and-ai/) to a journal like AGI -- I'm seeking cosignatories and co-authors.
But firsgt I will publish a revamped version on Substack.
As a humanistic psychologist who had the good fortune to meet Abraham Maslow while he was alive, I've followed your work with great interest. I very much agree and appreciate the distinctions you make here. Let's do stay in touch. Jed (Jed Diamond, PhD, Founder/VHS (Visionary, Healer, Scholar in residence) MenAlive.com
Thank you for pulling this apart. Once you read it - intelligence v. consciousness - it makes complete sense, but I don't think it's something most people are thinking about. It's definitely not something I've heard in the perils of AI debate yet.
Brilliant and articulate, thank you. Two comments. First, do Freudians and behaviorists approach this issue differently but come to the same conclusions? Secondly, how will humanists respond to an amoral machine? How will justice be enforced? How will unchecked, amoral goals be dealt with? Exciting, amazing, risky and uncertain times ahead! Tristan Harris at Your Individed Attention is Addressing these issues l8ke you .
I like the two-axis point you make, and the anaesthesia example. But I think that example does more than separate intelligence from consciousness. Under anaesthesia the brain still makes 'intelligent', life-preserving decisions while the felt experience vanishes completely. What drops away isn't the machinery of processing but the subjective feel, the "what it is like." These two can come apart even inside one's skull, which is a deeper cut than smart-versus-sentient alone.
Tessa's Ned Block reference upthread names this precisely: the difference between what a system can report and reason with, and whether anything is felt on the inside. And this matters most when the stakes are high. A machine can get better at the reporting-and-reasoning kind which is measurable, and it's improving rapidly. Whether anything is felt is a different question entirely. But "genuinely suffering systems" would require the felt kind. You can't suffer in processing alone. So the welfare question rides entirely on an axis we can't measure which is either terrifying or clarifying, depending on the day.
the jury is still out. There are different ways of conceiving of consciousness. The most famous distinction is that between phenomenal consciousness and access consciousness.
I propose instead the distinction between phenomenonal and information processing architecture-based consciousness:
I think the more one takes an integrative design-oriented approach to experience, the more one can ascribe it to different kinds of agents: other animals, machines and of course humans. That should not be surprising to those of us who take the _design-stance_ because the measure of how well psychological constructs can be understood is the extent to which they can be modeled. See:
(He used "design-based" but design-oriented is less confusing. "Design-based" has come to mean something else.)
> intelligence and consciousness pull apart.
yes and no. The more one takes an integrative design-oriented approach to autonomous agency (which is a more general category than human or machine), the more one sees a variety of interconnected mechanisms that give rise to that which requires explanation.
This does a fantastic job of characterizing consciousness across species and through human evolution. It's no stretch to see that he is characterizing different forms of consciousness, different types of competence, and thus different types of intelligence.
If one is to talk about consciousness, I think they need to respond to that book — it's the best book on consciousness that I have read. (And the author is Canadian 🇨🇦 😉 ).
> Humanistic psychologists believe that the sacred thing about a human is the inner life, including the experiences of awe, love, creativity, giving, meaning, presence, and peak experiences.
I disagree with saying that humanism is [some simple statement]. Who is the authority who gets to say that?
I believe psychology must "embrace complexity" as neuroscientist Luiz Pessoa put it on ["The Cognitive-Emotional Brain" (BSP 106) — Brain Science Podcast](https://brainsciencepodcast.com/bsp/2014/the-cognitive-emotional-brain-bsp-106) in Feb 2014. (I highly recommend the episode, it does a great job of communicating what emotion is about and what neuroscience should be about.)
> I do think with sufficient complexity they very well could be in the very near future
3. multidimensional (has more or less of various dimensions of it)
4. the architectural view: about the mechanisms and their organization in an information processing architecture.
"sufficient complexity" sounds like a scalar view of intelligence and consciousness. Each of 1-4 is useful in some circumstances. Sometimes we need to know whether something is sufficiently intelligent to have legal responsibilities for instance. But ultimately as scientists we need to gravitate toward the most complex view, which is #4.
> I have a real issue with tying worth to intelligence. We have an ugly history of doing just that.
this does get to important moral intuitions that can't be adjudicated in short form. However, I would recommend:
> The space of possible architectures for intelligent systems is very large. This essay takes steps towards a survey of the space, by examining some environmental and functional constraints, and discussing mechanisms capable of fulfilling them. In particular, we examine a subspace close to the human mind, by illustrating the variety of motives to be expected in a human-like system, and types of processes they can produce in meeting some of the constraints.
there we see that modeling intelligence and consciousness amount to the same thing, as this deals with emotions. One would be hard pressed to deny consciousness to machines that can experience emotions, and yet this article shows that machines will be able to have some types of emotions.
Great text, but I, on other hand, wonder sometimes if we do not put too much attention to feelings. In part of the world I am from there is a concept of intellectual friendship for example. Same: can you prefer truth over lie irregardless of your attitude to particular statements? So many things do not require subjective perspective at all.
The post separates intelligence from subjective experience and notes that outward competence alone does not establish an inner life. This video develops the related question of why naturalistic evolution would give humans reliable knowledge that subjective experience exists.
New Argument from Consciousness
This video explores the meta problem of consciousness proposed by David Chalmers and argues that naturalistic processes couldn’t have given us knowledge of consciousness
Thanks for pointing out Dr. Scott. I just finished reading "The Transcendent Brain" by Alan Lightman, who conflates intelligence with consciousness over and over... which is, in my view, typical of "nuthin' but stuff" materialists.
“Smart is not enough. How fast a mind can compute does not tell us much about what the mind is worth, or what a life is for.” A thought is measured up to 8m outside of the body, where an emotion is measured up to 16m away. The body has more than one brain - a second in the heart, a third in the gut and a fourth in the pelvis. All are developed before the one in the head. All four brains need to be in harmony for one to wake up and transcend, connecting consciousness of Self with universal consciousness.
I find Ned Block's distinction between phenomenal and access consciousness incredibly useful (see attached 1995 paper). We use the word "conscious" to indicate both felt experience (phenomenal) and what we have "conscious" access to for reasoning and reporting purposes. Machines (currently) mimic access consciousness, but we have zero evidence that they are phenomenally conscious (and unfortunately such evidence may be inherently unobtainable).
https://tinyurl.com/4b2edwhe
Well, I believe we need more clarity on what would count as evidence. We are the very early stages of AI, even though the discipline was born in 1956. There needs to be a reconciliation between modern machine learning and some of the earlier insights of AI. I think it's important to note that our understanding of human consciousness is limited. How can we say that a machine is or is not conscious if we are unable to model it? One doesn't really understand what one cannot model. But what can be modeled can be implemented in a computer (possibly one with analogue components). Here is an example of requirements for a computational theory of human visual consciousness: [Aaron Sloman 1989 - On designing a visual system](https://cogaffarchive.org/Aaron.Sloman_vision.design.html). A long way to go before machines can implement this, which means before we can understand how humans do it.
Oof that paper is way above my pay grade! But I think we share the concern that until we understand human (phenomenal) consciousness better we cannot know whether machines can be conscious. What worries me is that we still have no idea how to prove that even other humans are conscious--eg, to disprove the idea that I am living in a simulation where I am the only conscious being--so how will we do it for machines? But I certainly hope it proves feasible, since there are huge ethical questions riding on this!
I just found where I read the Hebb incidents: This book: The Structure of Emotions: Investigations in Cognitive Philosophy by Robert Morris Gordon. Apparently Daniel Dennett often quoted this. I don't have the primary source, so this could be apocryphal. I'll try to reread the chapter later today.
Yes. I think there are different stances we can take to answer the question: (a) empirical stance; (b) intentional stance; (c) design stance. (b) is was proposed by Dan Dennett in his book [Intentional stance ](https://en.wikipedia.org/wiki/Intentional_stance) (and earlier papers). It's basically a pragmatic / functionalist approach. Even a coke machine, he says, has beliefs and desires. It's about the predictive utility of intentional ascriptions. If speaking about a machine or animal as having specific beliefs and desires gains you a better understanding of its behaviour, then it counts. There are famous studies out of McGill (Montreal, Canada) where *behavioral psychologists* avoided referring to their monkey subjects as having beliefs and desires. This led to all kinds of problems such as the experimenters being injured by the animals. Same goes for consciousness. Sure we can withhold intentional ascriptions, but we do that at our peril-- or so Dennett would have said.
Interesting. But we don't have ethical duties towards coke machines, right? Should there be a (d) ethical stance, or if that somehow in (a)-(c) already?
I took a few courses in moral philosophy as an undergraduate but that was not informed by AI. I can't really comment on ethical issues. Beyond my pay grade, but my intuition is that one day we will have to treat artificial *autonomous agents* as persons. I think if we search for a binary cutoff we will oversimplify the problem. Thus my advocacy for an integrative design-oriented perspective. I plan to submit a much revised version of the [ Manifesto for an Integrative Design-oriented Approach to Understanding Humans as Autonomous Agents – CogZest](https://cogzest.com/projects/a-manifesto-for-integrative-design-oriented-cognitive-science-and-ai/) to a journal like AGI -- I'm seeking cosignatories and co-authors.
But firsgt I will publish a revamped version on Substack.
“Yes, they may surpass our intelligence, but it won’t touch the things that make being alive matter in the first place.” Love this!
Yes, I agree! If/ when we realize this distinction, I believe the cruelty of factory farming will become a thing of the past. Hopefully 🤞
As a humanistic psychologist who had the good fortune to meet Abraham Maslow while he was alive, I've followed your work with great interest. I very much agree and appreciate the distinctions you make here. Let's do stay in touch. Jed (Jed Diamond, PhD, Founder/VHS (Visionary, Healer, Scholar in residence) MenAlive.com
Thank you for pulling this apart. Once you read it - intelligence v. consciousness - it makes complete sense, but I don't think it's something most people are thinking about. It's definitely not something I've heard in the perils of AI debate yet.
Brilliant and articulate, thank you. Two comments. First, do Freudians and behaviorists approach this issue differently but come to the same conclusions? Secondly, how will humanists respond to an amoral machine? How will justice be enforced? How will unchecked, amoral goals be dealt with? Exciting, amazing, risky and uncertain times ahead! Tristan Harris at Your Individed Attention is Addressing these issues l8ke you .
As always succinct, inspiring , and comforting
Hear hear
I like the two-axis point you make, and the anaesthesia example. But I think that example does more than separate intelligence from consciousness. Under anaesthesia the brain still makes 'intelligent', life-preserving decisions while the felt experience vanishes completely. What drops away isn't the machinery of processing but the subjective feel, the "what it is like." These two can come apart even inside one's skull, which is a deeper cut than smart-versus-sentient alone.
Tessa's Ned Block reference upthread names this precisely: the difference between what a system can report and reason with, and whether anything is felt on the inside. And this matters most when the stakes are high. A machine can get better at the reporting-and-reasoning kind which is measurable, and it's improving rapidly. Whether anything is felt is a different question entirely. But "genuinely suffering systems" would require the felt kind. You can't suffer in processing alone. So the welfare question rides entirely on an axis we can't measure which is either terrifying or clarifying, depending on the day.
I worked through this gap in a recent essay [https://rajeshachanta.substack.com/p/a-miracle-occurs?utm_source=publication-search]. The point where every mechanistic account, however complete, still has a missing step.
Thanks, Scott, that's a very timely article.
> consciousness is subjective experience
the jury is still out. There are different ways of conceiving of consciousness. The most famous distinction is that between phenomenal consciousness and access consciousness.
I propose instead the distinction between phenomenonal and information processing architecture-based consciousness:
https://luccogzest.substack.com/p/if-intelligence-isnt-binary-what
> they experience anything from the inside.
I think the more one takes an integrative design-oriented approach to experience, the more one can ascribe it to different kinds of agents: other animals, machines and of course humans. That should not be surprising to those of us who take the _design-stance_ because the measure of how well psychological constructs can be understood is the extent to which they can be modeled. See:
[Sloman-1993-Prospects for AI is the General Science of Intelligence](https://cogaffarchive.org/Aaron.Sloman_prospects.pdf)
(He used "design-based" but design-oriented is less confusing. "Design-based" has come to mean something else.)
> intelligence and consciousness pull apart.
yes and no. The more one takes an integrative design-oriented approach to autonomous agency (which is a more general category than human or machine), the more one sees a variety of interconnected mechanisms that give rise to that which requires explanation.
Take for instance [_A Mind So Rare_](http://books.wwnorton.com/books/A-Mind-So-Rare/),
This does a fantastic job of characterizing consciousness across species and through human evolution. It's no stretch to see that he is characterizing different forms of consciousness, different types of competence, and thus different types of intelligence.
If one is to talk about consciousness, I think they need to respond to that book — it's the best book on consciousness that I have read. (And the author is Canadian 🇨🇦 😉 ).
> Humanistic psychologists believe that the sacred thing about a human is the inner life, including the experiences of awe, love, creativity, giving, meaning, presence, and peak experiences.
I disagree with saying that humanism is [some simple statement]. Who is the authority who gets to say that?
If we look at the [The Amsterdam Declaration of humanism](https://humanists.international/what-is-humanism/the-amsterdam-declaration/), we see something much richer than this.
I believe psychology must "embrace complexity" as neuroscientist Luiz Pessoa put it on ["The Cognitive-Emotional Brain" (BSP 106) — Brain Science Podcast](https://brainsciencepodcast.com/bsp/2014/the-cognitive-emotional-brain-bsp-106) in Feb 2014. (I highly recommend the episode, it does a great job of communicating what emotion is about and what neuroscience should be about.)
> I do think with sufficient complexity they very well could be in the very near future
Well, there are different types of consciousness for machines as there are for animals. In [an article on intelligence](https://luccogzest.substack.com/p/beyond-is-ai-intelligent-intelligence) I argue that there are different conceptions of intelligence.
1. binary (has it or not)
2. scalar (has more or less of it)
3. multidimensional (has more or less of various dimensions of it)
4. the architectural view: about the mechanisms and their organization in an information processing architecture.
"sufficient complexity" sounds like a scalar view of intelligence and consciousness. Each of 1-4 is useful in some circumstances. Sometimes we need to know whether something is sufficiently intelligent to have legal responsibilities for instance. But ultimately as scientists we need to gravitate toward the most complex view, which is #4.
> I have a real issue with tying worth to intelligence. We have an ugly history of doing just that.
this does get to important moral intuitions that can't be adjudicated in short form. However, I would recommend:
[Sloman-1981-You Don't Need a Soft Skin to Have a Warm Heart Towards a Computational Analysis of Motives and Emotions](https://cogaffarchive.org/sloman-croucher-warm-heart.html) which elaborates on:
> The space of possible architectures for intelligent systems is very large. This essay takes steps towards a survey of the space, by examining some environmental and functional constraints, and discussing mechanisms capable of fulfilling them. In particular, we examine a subspace close to the human mind, by illustrating the variety of motives to be expected in a human-like system, and types of processes they can produce in meeting some of the constraints.
there we see that modeling intelligence and consciousness amount to the same thing, as this deals with emotions. One would be hard pressed to deny consciousness to machines that can experience emotions, and yet this article shows that machines will be able to have some types of emotions.
Great text, but I, on other hand, wonder sometimes if we do not put too much attention to feelings. In part of the world I am from there is a concept of intellectual friendship for example. Same: can you prefer truth over lie irregardless of your attitude to particular statements? So many things do not require subjective perspective at all.
The post separates intelligence from subjective experience and notes that outward competence alone does not establish an inner life. This video develops the related question of why naturalistic evolution would give humans reliable knowledge that subjective experience exists.
New Argument from Consciousness
This video explores the meta problem of consciousness proposed by David Chalmers and argues that naturalistic processes couldn’t have given us knowledge of consciousness
This is my video: https://www.youtube.com/watch?v=WJVvZNi0Fi8
Thanks for pointing out Dr. Scott. I just finished reading "The Transcendent Brain" by Alan Lightman, who conflates intelligence with consciousness over and over... which is, in my view, typical of "nuthin' but stuff" materialists.
“Smart is not enough. How fast a mind can compute does not tell us much about what the mind is worth, or what a life is for.” A thought is measured up to 8m outside of the body, where an emotion is measured up to 16m away. The body has more than one brain - a second in the heart, a third in the gut and a fourth in the pelvis. All are developed before the one in the head. All four brains need to be in harmony for one to wake up and transcend, connecting consciousness of Self with universal consciousness.