Even Indo-European could perhaps have a related first person singular pronoun. English "I" descends from Proto-Germanic "ik", "ek" (source of German "ich"), which descends from Proto-Indo-European "eg-" (source of Latin "ego", Greek "εγω"). It's not unreasonable that "eg-" could be a descendent of an earlier form "ŋ-". That would be closer to such a form than Mandarin 我 wǒ /wɔ/, which looks so dissimilar to ŋa that it would be difficult to see that the Mandarin form descends from the Sino-Tibetan form without the evidence from related Chinese languages (e.g., Cantonese /ŋɔ/).
Interesting! I had no idea about the Chinese case.
I actually do buy the line that most linguistic relationships are gone by something like 10k years. If pronouns were invented 20k years ago, there will be some branches from the original that are nigh impossible to trace.
Seeing you know something about linguistics, what did you think about the general argument? Does the distribution of pronouns imply their spread (and perhaps invention) after leaving Africa?
"This group would have spoken the same language, from which the extant languages descend"
Just to give a contrary linguistic opinion on this, we don't actually know that the group that crossed the Bering Strait spoke a single language - that's entirely speculative. It's entirely possible two or more linguistic groups crossed the strait at the same time (or perhaps more likely in short succession to one another - instead of a single migration, it could be two or more smaller migrations potentially hundreds of years apart).
We also don't know that all families in the Americas date from the initial migration(s) to the Americas. The Na-Dené family at the very least has been suspected to be a much more recent migration, even by Greenberg himself (and the Dené-Yeniseian hypothesis may be evidence of this, although it has also been suggested that Yeniseian, if it is indeed related to Na-Dené, was actually the result of a back migration across Beringia). We also know the Eskaleut family (aka Eskimo-Aleut or Inuit-Yupik-Unangam) is a much more recent arrival, probably around 4000 years or less.
It's important to note that while gene flow can be used as an assistive tool to help track the movement of prehistoric languages, it is not perfect, as groups may end up adopting the speech of unrelated groups even with only a small amount of genetic flow between the two.
*《The hypothetical language family has long been rejected by most comparative linguists, although it continues to be supported by a small but stable scholarly minority.》
I was idly thinking given French on and British English one used to mean I, can we bring Proto Indo European into the fold via *Hoi-no, one. They don't call it the FIRST person singular for nothing.
Uneasy about the low probability argument. It's a trite observation that 100 heads from 100 coin tosses is no less likely than any other given specified sequence, it's just that it is self-specifying. Or something - it's harder than it looks to explain when it is or isn't legitimate to say OMG what are the odds of that? I can think of a couple of 7 or 8 member genuinely coincidental letter sequences - the continents except Europe (still a vowel) and the leading British horse trials at Blair, Badminton, Burghley, Blenheim, Burgham, Bicton and Barbury (and, OK, chatsworth)
From my own n=2 experience the first wordlike sound babies make is da which I imagine fathers have coopted as Ooh look he recognises his daddy. An innate tendency to associate the na sound with I is quite plausible.
I've thought about this a lot, actually, because it effects how much time I invest in this particular problem. Funny enough, some of it hinges on whether I take a first-person view of the degrees of freedom.
You'll have to trust me that after considering "what if self-awareness was recent?" one of the first things I thought to check was the 1sg. There were really almost no degrees of freedom, and I assumed it would falsify the idea right there.
Now, for other people, the statistical calculation is different. What are the odds that someone on the internet finds such a coincidence? Now that is a different story!
Other than my word I didn't use degrees of freedom, there are other reasons to believe chance is unlikely. The linguists that say things like
“[It is] incorrect to claim that “chance resemblance” can play an important part in pronominal comparison between languages of different families. There are absolutely no coincidences in paradigm patterns between the languages which are not thought to be genetically related by modern long-range comparativists.”
are looking at much more than just the consonant in the 1sg. They compare grammar and whole baskets of words.
(Expanding the argument a bit, feel free to skip this point.) From the consciousness side as well, it turns out that members of the "consciousness basket" that I had posited (eg. snakes and creation, primordial matriarchy, male initiation into sacred knowledge) are some of the best evidence of global diffusion: https://www.vectorsofmind.com/p/evidence-for-global-cultural-diffusion
>An innate tendency to associate the na sound with I is quite plausible.
It is weird that pronouns have tended to _diverge_ from na in the last 10,000 years. That seems to be at least soft evidence against this.
But, in general, yes. There is a plausible explanation for the sound "na" and the 1sg. Have to look to grammar and other potentially-diffused ideas to build the case that it can't _all_ be chance.
It's amazing how often I think I have discovered a new and imponderable problem, and discover 1. that there's an incredibly lucid explanation of it on the internet, and 2. that the explanation is by Scott.
Expert pronouncements that" There are absolutely no coincidences..." are hard to accept. There can never not be the possibility of coincidence unless there's a wholly unknown cosmic principle ruling them out. A coin tossing expert could tell you that there's no evidence in his personal experience or in the literature of a 100 heads sequence.
Jul 3, 2023·edited Jul 3, 2023Liked by Andrew Cutler
Interesting. My mother tongue is Hungarian, where the 1sg is 'én', which does contain the 'n' sound. But the 2sg is 'te', which is quite far from the sound 'm'.
Futhermore, Hungarian has no linguistic gender whatsoever: so much so that the 3sg is 'ő', which doesn't convey the sex of the referent: the listener has to know that from some other source. But my understanding is that this is the case with Turkic languages (with which Hungarian apparently has lots of commonality).
Also regarding animate-inanimate distinction: the only case where something similar features in Hungarian (at least which comes to mind to me) is the 3sg: as mentioned before, when the referent is a human person, it is 'ő', but for any other referent (animate or inanimate) it is 'az' (for entities further away; for ones closer to the speaker, it is 'ez').
I'm telling you all this only to throw further confusion into the discussion. :-P
The linked paper "Where Do Personal Pronouns Come From?" lists t as one of the most common consonants for 1sg. I was just interested in the 1sg, obviously, but there are several streams of evidence for these long-range connections.
> where instead of mindvector (a name) the actor-word started out as a title, such as “father” or “mother”. Therefore “Nana go this way”, eventually became our worldwide friend na.
Isn't that, together with nana/mama having been converged on for voice-mechanical reasons, a perfectly sufficient explanation for independent 1sg convergence?
Yeah, but if it happened significantly after OoA, the mechanism would have to really work like clockwork. Again and again all over the world. It would be surprising if that's sort of a rule. The paper that put forth the model also discusses it happening ~100 kya. But then you run into a lot of problems with cognates lasting that long. (The author has already bit this bullet and elsewhere has suggested dozens of world-wide cognates. Pronouns, but also words like fire, hand, or think.)
So sufficient yes. But it does seem unlikely. Could say the same of self-awareness spreading. One of my main reasons for engaging is there is no good date for when recursion evolved (which is related to self-awareness). Recursion evolving recently solves a bunch of problems matching evolution to archeology. Worth exploring what that would look like, even if each individual observation (such as the pronoun distribution) has other sufficient explanations.
I would push back some. You consistently state that self-consciousness and a first person pronoun would have evolved simultaneously. This is not necessarily obvious. Necessarily, self consciousness would have needed to exist *before* terms referring to it emerged. Secondly, the pronouns may have been an innovation, and most languages may have used a system similar to modern Japanese, where people often use their own name to refer to themselves, and their partner's name instead of a second person pronoun. Replacing this name system with pronouns may have instead been the innovation which spread like wildfire.
I can go for pronouns originally referring to the first person, but not the first person identifying with their inner life. But you can't get around that identification causing many linguistic changes to the 1sg. Feel Jaynes (and crew) really dropped the ball here.
This stuff is deliciously insane. I'm not sure how much credence I should be giving this thesis, but this sure is entertaining and thought-provoking! I can't get it out of my head now!
One thing that's occurred to me: have you considered how this stuff may relate to hypnosis? Presumably, if consciousness is more about software than hardware and triggerable, maybe it is still reversible? Maybe hypnosis is people tricking other people's brains into thinking that their voice is the brain's own auditory hallucination telling it what to do?
I'd love to talk to a hypnotist and ask them how they think it works. How fast they can tell if someone is open to it, etc. Elsewhere I talk more about what the first inner voice would have said. In that post (Consequences of Conscience) I argue that it could have been related to moral behavior. Now I'm actually more partial to it being something like "run!". Herd animals are in a sense hypnotized when they hear one of their group give a call that indicates a predator; they don't think before they stop moving. I wonder if there was a pretty established circuit between life-and-death commands and action that our first inner voice (and now hypnosis) tapped into.
All that said, I do think there have been hardware changes in the last 10k years, but obviously emphasize the plasticity of how we experience/construct "self".
Wow! What a fascinating read. I rapidly consumed this and all your other posts on recursion and the snake cult of consciousness. Just such fun stuff and wonderful to have my mind bend a little bit as the threads came together. I shared your Substack with a friend of mine who has a phd in philosophy and he was intrigued as well. He suggested I read ‘the first person’ by Elizabeth Anscombe. I find her arguments about "self" contained therein to be quite reinforcing, from a philosophical perspective, of what you are getting at from a linguistic perspective in this post. I would guess you have read it already, but if not, you can access it here: https://ifac.univ-nantes.fr/IMG/pdf/Anscombe-The_First_Person.pdf
I don't know much about the linguistics here, but I want to point out that you skip over a possible explanation as to how unrelated languages might converge on 'n' that's different from how they converge on 'mama'/'papa': perhaps there is some other sense, intrinsic to either the universe or to how the human body is constructed, which makes "n" sounds appropriate for 1sg. I don't particularly believe this as I can't imagine how it would work in detail, but it's still worth keeping around as a possible explanation.
A sketch of how it might work is: there is some 'shape' in neuronal activations in the brain that corresponds to "I" in an entirely pre-language sense (maybe it could also be detected in other animals? perhaps in the other primates that understand what they're seeing in mirrors?), and this shape, like, is somehow mechanically related to the neuronal activation required to make an 'n' sound. Or perhaps the correlation is even more abstraction, such as existing entirely in mathematical space: the "I" concept as a program in thought-space is topologically similar to the "n" sound in vocal-chord-activation-space, or something.
This is very much a "just saying" comment, I have no idea how you would even tell if it's plausible or insane... but it seems to me that it should stick around as a possible theory unless there's really a way to discard it.
Yeah, that is one possibility, but the force of convergence would have to be quite strong. There are a bunch of examples of divergence from "na", which goes against the convergence force being strong. There's also the expert that says,
“[It is] incorrect to claim that “chance resemblance” can play an important part in pronominal comparison between languages of different families. There are absolutely no coincidences in paradigm patterns between the languages which are not thought to be genetically related by modern long-range comparativists.”
Could be wrong of course. Most linguists probably find "long-range comparativists" to be off track (maybe in the same way physicists think about string theorists?). But my argument is fairly soft; that the pronouns are admissible evidence in the debate about the evolution of self-awareness.
As I've gotten more curious about id/ego/superego after reading Snake Cult, I came across this bit from Wikipedia which seems very relevant to the connection between pronouns and recursive sapience.
> Freud himself used the German terms das Es, Ich, and Über-Ich, which literally translate as "the it", "I", and "over-I". The Latin terms id, ego and super-ego were chosen by his original translators and have remained in use.
There's a lot of room for studying convergence through the lens of biology. Assuming that the tongue, diaphragm and vocal cords of humans for 50,000 years were functionally identical over time it could be that the similarities in language were due to convergence from optimization constraints. In other words peoples over time separately arrived at the same linguistic norms because physical biology only allows so many ways to optimize.
ChatGPT responds to questions with "I" answers, though, often like "Sorry, I can't answer that as I am just an AI, a tool for humans..." How do we differentiate?
Well, the AI didn't invent the word. I'm concerned about the first person who communicated to another about this internal experience we call "I". Parrots using pronouns also doesn't imply they have internal experiences.
Even Indo-European could perhaps have a related first person singular pronoun. English "I" descends from Proto-Germanic "ik", "ek" (source of German "ich"), which descends from Proto-Indo-European "eg-" (source of Latin "ego", Greek "εγω"). It's not unreasonable that "eg-" could be a descendent of an earlier form "ŋ-". That would be closer to such a form than Mandarin 我 wǒ /wɔ/, which looks so dissimilar to ŋa that it would be difficult to see that the Mandarin form descends from the Sino-Tibetan form without the evidence from related Chinese languages (e.g., Cantonese /ŋɔ/).
Interesting! I had no idea about the Chinese case.
I actually do buy the line that most linguistic relationships are gone by something like 10k years. If pronouns were invented 20k years ago, there will be some branches from the original that are nigh impossible to trace.
Seeing you know something about linguistics, what did you think about the general argument? Does the distribution of pronouns imply their spread (and perhaps invention) after leaving Africa?
"This group would have spoken the same language, from which the extant languages descend"
Just to give a contrary linguistic opinion on this, we don't actually know that the group that crossed the Bering Strait spoke a single language - that's entirely speculative. It's entirely possible two or more linguistic groups crossed the strait at the same time (or perhaps more likely in short succession to one another - instead of a single migration, it could be two or more smaller migrations potentially hundreds of years apart).
We also don't know that all families in the Americas date from the initial migration(s) to the Americas. The Na-Dené family at the very least has been suspected to be a much more recent migration, even by Greenberg himself (and the Dené-Yeniseian hypothesis may be evidence of this, although it has also been suggested that Yeniseian, if it is indeed related to Na-Dené, was actually the result of a back migration across Beringia). We also know the Eskaleut family (aka Eskimo-Aleut or Inuit-Yupik-Unangam) is a much more recent arrival, probably around 4000 years or less.
It's important to note that while gene flow can be used as an assistive tool to help track the movement of prehistoric languages, it is not perfect, as groups may end up adopting the speech of unrelated groups even with only a small amount of genetic flow between the two.
Altaic: *mi
*《The hypothetical language family has long been rejected by most comparative linguists, although it continues to be supported by a small but stable scholarly minority.》
https://en.m.wikipedia.org/wiki/Altaic_languages
*But 1sg pronouns have 'n' consonant in almost every turkic languages.
Turkish= ben
Kazakh = men
Sakha = min
https://en.m.wikipedia.org/wiki/Proto-Turkic_language
* Korean language
First person = 저 (jeo), 나 (na)
https://en.m.wikipedia.org/wiki/Korean_pronouns
Another wild adventure. Begs for follow up reading. Excellent work!
I was idly thinking given French on and British English one used to mean I, can we bring Proto Indo European into the fold via *Hoi-no, one. They don't call it the FIRST person singular for nothing.
Uneasy about the low probability argument. It's a trite observation that 100 heads from 100 coin tosses is no less likely than any other given specified sequence, it's just that it is self-specifying. Or something - it's harder than it looks to explain when it is or isn't legitimate to say OMG what are the odds of that? I can think of a couple of 7 or 8 member genuinely coincidental letter sequences - the continents except Europe (still a vowel) and the leading British horse trials at Blair, Badminton, Burghley, Blenheim, Burgham, Bicton and Barbury (and, OK, chatsworth)
From my own n=2 experience the first wordlike sound babies make is da which I imagine fathers have coopted as Ooh look he recognises his daddy. An innate tendency to associate the na sound with I is quite plausible.
>Uneasy about the low probability argument.
The real statistical question is how many degrees of freedom I had to find this particular outcome. (Great treatment of the problem by Scott himself: https://slatestarcodex.com/2016/11/05/the-pyramid-and-the-garden/)
I've thought about this a lot, actually, because it effects how much time I invest in this particular problem. Funny enough, some of it hinges on whether I take a first-person view of the degrees of freedom.
You'll have to trust me that after considering "what if self-awareness was recent?" one of the first things I thought to check was the 1sg. There were really almost no degrees of freedom, and I assumed it would falsify the idea right there.
Now, for other people, the statistical calculation is different. What are the odds that someone on the internet finds such a coincidence? Now that is a different story!
Other than my word I didn't use degrees of freedom, there are other reasons to believe chance is unlikely. The linguists that say things like
“[It is] incorrect to claim that “chance resemblance” can play an important part in pronominal comparison between languages of different families. There are absolutely no coincidences in paradigm patterns between the languages which are not thought to be genetically related by modern long-range comparativists.”
are looking at much more than just the consonant in the 1sg. They compare grammar and whole baskets of words.
(Expanding the argument a bit, feel free to skip this point.) From the consciousness side as well, it turns out that members of the "consciousness basket" that I had posited (eg. snakes and creation, primordial matriarchy, male initiation into sacred knowledge) are some of the best evidence of global diffusion: https://www.vectorsofmind.com/p/evidence-for-global-cultural-diffusion
>An innate tendency to associate the na sound with I is quite plausible.
It is weird that pronouns have tended to _diverge_ from na in the last 10,000 years. That seems to be at least soft evidence against this.
But, in general, yes. There is a plausible explanation for the sound "na" and the 1sg. Have to look to grammar and other potentially-diffused ideas to build the case that it can't _all_ be chance.
It's amazing how often I think I have discovered a new and imponderable problem, and discover 1. that there's an incredibly lucid explanation of it on the internet, and 2. that the explanation is by Scott.
Expert pronouncements that" There are absolutely no coincidences..." are hard to accept. There can never not be the possibility of coincidence unless there's a wholly unknown cosmic principle ruling them out. A coin tossing expert could tell you that there's no evidence in his personal experience or in the literature of a 100 heads sequence.
Interesting. My mother tongue is Hungarian, where the 1sg is 'én', which does contain the 'n' sound. But the 2sg is 'te', which is quite far from the sound 'm'.
Futhermore, Hungarian has no linguistic gender whatsoever: so much so that the 3sg is 'ő', which doesn't convey the sex of the referent: the listener has to know that from some other source. But my understanding is that this is the case with Turkic languages (with which Hungarian apparently has lots of commonality).
Also regarding animate-inanimate distinction: the only case where something similar features in Hungarian (at least which comes to mind to me) is the 3sg: as mentioned before, when the referent is a human person, it is 'ő', but for any other referent (animate or inanimate) it is 'az' (for entities further away; for ones closer to the speaker, it is 'ez').
I'm telling you all this only to throw further confusion into the discussion. :-P
Uralic? Cool! Hyperborean confirmed :p
The linked paper "Where Do Personal Pronouns Come From?" lists t as one of the most common consonants for 1sg. I was just interested in the 1sg, obviously, but there are several streams of evidence for these long-range connections.
I think your first 1sg here wants to be 2sg?
right!
> where instead of mindvector (a name) the actor-word started out as a title, such as “father” or “mother”. Therefore “Nana go this way”, eventually became our worldwide friend na.
Isn't that, together with nana/mama having been converged on for voice-mechanical reasons, a perfectly sufficient explanation for independent 1sg convergence?
Yeah, but if it happened significantly after OoA, the mechanism would have to really work like clockwork. Again and again all over the world. It would be surprising if that's sort of a rule. The paper that put forth the model also discusses it happening ~100 kya. But then you run into a lot of problems with cognates lasting that long. (The author has already bit this bullet and elsewhere has suggested dozens of world-wide cognates. Pronouns, but also words like fire, hand, or think.)
So sufficient yes. But it does seem unlikely. Could say the same of self-awareness spreading. One of my main reasons for engaging is there is no good date for when recursion evolved (which is related to self-awareness). Recursion evolving recently solves a bunch of problems matching evolution to archeology. Worth exploring what that would look like, even if each individual observation (such as the pronoun distribution) has other sufficient explanations.
I would push back some. You consistently state that self-consciousness and a first person pronoun would have evolved simultaneously. This is not necessarily obvious. Necessarily, self consciousness would have needed to exist *before* terms referring to it emerged. Secondly, the pronouns may have been an innovation, and most languages may have used a system similar to modern Japanese, where people often use their own name to refer to themselves, and their partner's name instead of a second person pronoun. Replacing this name system with pronouns may have instead been the innovation which spread like wildfire.
I can go for pronouns originally referring to the first person, but not the first person identifying with their inner life. But you can't get around that identification causing many linguistic changes to the 1sg. Feel Jaynes (and crew) really dropped the ball here.
This stuff is deliciously insane. I'm not sure how much credence I should be giving this thesis, but this sure is entertaining and thought-provoking! I can't get it out of my head now!
One thing that's occurred to me: have you considered how this stuff may relate to hypnosis? Presumably, if consciousness is more about software than hardware and triggerable, maybe it is still reversible? Maybe hypnosis is people tricking other people's brains into thinking that their voice is the brain's own auditory hallucination telling it what to do?
Thanks!
I'd love to talk to a hypnotist and ask them how they think it works. How fast they can tell if someone is open to it, etc. Elsewhere I talk more about what the first inner voice would have said. In that post (Consequences of Conscience) I argue that it could have been related to moral behavior. Now I'm actually more partial to it being something like "run!". Herd animals are in a sense hypnotized when they hear one of their group give a call that indicates a predator; they don't think before they stop moving. I wonder if there was a pretty established circuit between life-and-death commands and action that our first inner voice (and now hypnosis) tapped into.
All that said, I do think there have been hardware changes in the last 10k years, but obviously emphasize the plasticity of how we experience/construct "self".
Well, if you happen to find a correlation between susceptibility to hypnosis and resistance to hallucinogenics you'll make me update hard on this!
Andrew,
Wow! What a fascinating read. I rapidly consumed this and all your other posts on recursion and the snake cult of consciousness. Just such fun stuff and wonderful to have my mind bend a little bit as the threads came together. I shared your Substack with a friend of mine who has a phd in philosophy and he was intrigued as well. He suggested I read ‘the first person’ by Elizabeth Anscombe. I find her arguments about "self" contained therein to be quite reinforcing, from a philosophical perspective, of what you are getting at from a linguistic perspective in this post. I would guess you have read it already, but if not, you can access it here: https://ifac.univ-nantes.fr/IMG/pdf/Anscombe-The_First_Person.pdf
I haven't read the essay, thanks for the share! Very much learning the philosophy as I go
I don't know much about the linguistics here, but I want to point out that you skip over a possible explanation as to how unrelated languages might converge on 'n' that's different from how they converge on 'mama'/'papa': perhaps there is some other sense, intrinsic to either the universe or to how the human body is constructed, which makes "n" sounds appropriate for 1sg. I don't particularly believe this as I can't imagine how it would work in detail, but it's still worth keeping around as a possible explanation.
A sketch of how it might work is: there is some 'shape' in neuronal activations in the brain that corresponds to "I" in an entirely pre-language sense (maybe it could also be detected in other animals? perhaps in the other primates that understand what they're seeing in mirrors?), and this shape, like, is somehow mechanically related to the neuronal activation required to make an 'n' sound. Or perhaps the correlation is even more abstraction, such as existing entirely in mathematical space: the "I" concept as a program in thought-space is topologically similar to the "n" sound in vocal-chord-activation-space, or something.
This is very much a "just saying" comment, I have no idea how you would even tell if it's plausible or insane... but it seems to me that it should stick around as a possible theory unless there's really a way to discard it.
Yeah, that is one possibility, but the force of convergence would have to be quite strong. There are a bunch of examples of divergence from "na", which goes against the convergence force being strong. There's also the expert that says,
“[It is] incorrect to claim that “chance resemblance” can play an important part in pronominal comparison between languages of different families. There are absolutely no coincidences in paradigm patterns between the languages which are not thought to be genetically related by modern long-range comparativists.”
Could be wrong of course. Most linguists probably find "long-range comparativists" to be off track (maybe in the same way physicists think about string theorists?). But my argument is fairly soft; that the pronouns are admissible evidence in the debate about the evolution of self-awareness.
"nor does it correspond to any Paradox or Revolution that perplexes other fields."
Wrong. Fall of bicameral mind corresponds with axial age and tableware.
"Another issue is that he doesn’t provide a mechanism of self-discovery."
Rise of infectious disease load (corresponding to the plagues of Egypt).
Remember, consciousness needs calories. Europeans in the 19th century frequently suggested the Northern Chinese of their day had no consciousness.
Get on the network -the bicameral mind is a product of human posture:
https://eharding.substack.com/p/why-does-russian-physical-therapy
As I've gotten more curious about id/ego/superego after reading Snake Cult, I came across this bit from Wikipedia which seems very relevant to the connection between pronouns and recursive sapience.
> Freud himself used the German terms das Es, Ich, and Über-Ich, which literally translate as "the it", "I", and "over-I". The Latin terms id, ego and super-ego were chosen by his original translators and have remained in use.
There's a lot of room for studying convergence through the lens of biology. Assuming that the tongue, diaphragm and vocal cords of humans for 50,000 years were functionally identical over time it could be that the similarities in language were due to convergence from optimization constraints. In other words peoples over time separately arrived at the same linguistic norms because physical biology only allows so many ways to optimize.
ChatGPT responds to questions with "I" answers, though, often like "Sorry, I can't answer that as I am just an AI, a tool for humans..." How do we differentiate?
Well, the AI didn't invent the word. I'm concerned about the first person who communicated to another about this internal experience we call "I". Parrots using pronouns also doesn't imply they have internal experiences.