Macroscale Minds

by Evan Clark4 min read9th Mar 20189 comments

1

Personal Blog

The original China Brain thought experiment imagines that each person in China uses a cell-phone or other device along with a list of numbers and instructions to simulate one neuron in the human brain. For the purposes of this post I will be assuming that a China Brain is analogous enough to a human brain - perhaps supposing an increased number of simulated neurons per person - to generate a conscious entity capable of thought.

When I first heard about this, it totally fascinated me, but I saw no obvious way to go forward, until I read Meditations on Moloch and Person-moment Affecting Views. I am taking a lot of obvious inspiration from the former (such as the alliteration) but the latter was a necessary piece as well.

If we grant that a thoroughly interconnected and efficiently organized nation can generate a mind, then what happens as we adjust the dials on connectedness and organization?

As a quick sanity check, we already know that there exists a sliding scale of consciousness - the animal kingdom. It seems reasonable to think that dolphins and chimps are closer to humans on the Great Scale of Being (that was a pun and not an ethical assertion) and that ants and bees are probably somewhere on there, just significantly further to the left. So, in principle, if a China Brain - which for our purposes will include any arrangement of multiple human beings sufficient to simulate a single conscious entity, not just one that simulates human brains - is somewhere on the right next to humans, we should be able to to at least consider where we would place other, more diffuse or chaotic groups of humans.

In the spirit of simulating certain human brains, let's skip better examples and jump straight to America -

Is America a mind? A good reason to suspect that the answer is no is that the actual parts are constantly shifting. In the interest of simplifying the question and dissolving that objection, "America" shall refer to all residents of the United States of America who are reasonably likely to continue to be residents in the near future. The problem no slightly less ill-defined, let us consider some points in favor of an affirmative response:

  1. A majority of America's parts are aware of and accept the existence of America as a meaningful unit
  2. Parts of America are constantly commenting on desirable and undesirable traits of America and reflecting on the nature and properties of it. Ahem
  3. The subset of parts of America that exerts the greatest disproportionate influence on the whole of America is regularly redefined by a function that seeks to approximate the vector average of the interests of all of a significant fraction of all other parts.
  4. Parts of America deliberately converge on common understandings of phenomena, even when the actual details do not warrant such convergence, and encourage and train other parts to do so
  5. The aforementioned disportionately powerful subset (i.e. the federal government) presents a public face as a single unified being with a single unified interest, and interacts with other nations on that basis
  6. Most parts (be they people or low-level organizations) are reasonably capable of contacting a large number of other parts

While this is all true, and while the mind explanation might serve as a decent abstraction, the similarities break down fairly quickly. America is just not focused, united, or stable over time to really warrant a comparison to human beings, and while it is capable of self-reference, the extent of the self-reference is relatively shallow. Another problem might be the miniscule extent of goal-orientation. America is best modeled as thousands of competing interests with occasional agreement that certain ideas are even worthy of thought, not a single being with a clear goal. Lastly, even though Americans are connected, they are not connected enough (yet) to form bonds as a nation that approach interpersonal bonds in strength.

As counterintuitive as it might be, however, I would label America as a sort of conscious being, given that it displays some goal-oriented behavior, self-awareness and capacity for directed self-modification, and a national identity (if an unstable and inconsistent one).

(I chose America for the sake of ease of writing, I think this all generalizes)

So in summary, relaxing constraints on the China Brain experiment to match real life produces nations which can only weakly be described as constituting minds, in the same way that simplifying the architecture of the human brain produces animals which are only barely sentient (such as small insects, etc).

I grant that this analysis is cursory and informal at best, but I would like to get this discussion going. What levels of organization in humans, if any, are mind-like? And what about non-humans, such as ant colonies and beehives? (The above pretty optimistically assumes that any other human will read this post)

1

9 comments, sorted by Highlighting new comments since Today at 9:52 AM
New Comment

Are you familiar with Searle’s “Chinese Room”[1] thought experiment? The concept you describe is almost identical to several of the variations on that hypothetical scenario that came up in the prominent discussions of Searle’s paper when it was published—and is disposed of in the same way as those were. I refer you to the relevant section of Daniel Dennett & Douglas Hofstadter’s book, The Mind’s I (or, for a quicker read, to this old Slate Star Codex comment thread, which briefly covers a part of what Dennett & Hofstadter say).

One relevant matter brought up in the aforementioned discussions is that of scale. The Wikipedia article “List of animals by number of neurons” may be instructive, here. The current population of the United States is about 327 million. Of the animals listed in that Wikipedia article, the one with closest to that number of neurons is the common parakeet. So, if we imagine every single person in America (including babies, etc.) being organized in such a way as to give rise to a mind-like structure (connectome), then it would seem that the resulting mind would be about as “smart” or “conscious” as a parakeet. Not very impressive! But, of course, neurons fire at about 200 Hz, and humans can’t really do anything that quickly, much less make cell phone calls and follow instructions or what have you—so it would be more like a very, very slow parakeet.

But of course that’s quite silly. In reality, the organizational structures composed of humans are too loosely connected, and not connected in anything like the right ways, and not cohesive enough, and not made up of enough parts, etc., to be plausibly mind-like. So I think my answer to the question—“What levels of organization in humans, if any, are mind-like?”—has to be “none of them”.

[1] What is it with using China and Chinese as examples of these things??

Here, for anyone not already familiar with it (and maybe some who are), is Searle's original article along with commentary from a bunch of smart people and then Searle's response to their comments. (The article was published in Behavioral and Brain Sciences, which for each article solicits such commentary and then publishes them together.) The commenters include Daniel Dennett, Jerry Fodor, Douglas Hofstadter, Benjamin Libet, Marvin Minsky, Richard Rorty, and Roger Schank; it's quite a star-studded list.

I think the reason for using Chinese is not so much that it has a reputation for being "mechanical and analytic" as Evan suggests, as that it has a reputation for being exceptionally incomprehensible to those who don't know it, including in particular John Searle. He wanted a language for which the scenario he described would make it clearly impossible for him to have any inkling what the symbols he was pushing around mean.

Are you familiar with Searle’s “Chinese Room”[1] thought experiment?

Yes. As I believe the provided link makes clear, the China Brain is related both historically and obviously conceptually to the Chinese Room.

So, if we imagine every single person in America (including babies, etc.) being organized in such a way as to give rise to a mind-like structure (connectome), then it would seem that the resulting mind would be about as “smart” or “conscious” as a parakeet. Not very impressive!

On the contrary, this is incredibly impressive. Regardless, the point still stands: even a parakeet mind slowed down by millions of times is still, in some sense, a parakeet mind. I actually address the numerical point specifically, but to restate it, I believe humans are capable of simulating 100 or more neurons at once.

But of course that’s quite silly. In reality, the organizational structures composed of humans are too loosely connected, and not connected in anything like the right ways, and not cohesive enough, and not made up of enough parts, etc., to be plausibly mind-like

Well, I do agree that America is not even as sentient as a comparatively simple animal (for the exact reasons you mentioned), however I believe that there are a sufficient number of properties (complex but relatively consistent connections, self-reference and modification, etc.) that it shares in common with minds that deserves at least the label of mind-like.

[1] What is it with using China and Chinese as examples of these things??

Well the Chinese Room presumably uses the written Chinese language because it has a reputation for being mechanical and analytic. (This is really only little bit true in my experience, 可是我的中文很不好 so don't ask me).

For the China Brain, it is probably due to China's large population, as well as a play on the Chinese Room.

I agree with you that groups of humans are capable of phenomenal consciousness but I guess the question is how much can we do with that?

there was a controversial speaker at my university last year, I went to check it out and accidentally ended up joining in the protest. There were some moments when the crowd became mob-like and nearly rioted; when I was a part of the more momentous moments, it was like I was on autopilot.

In reaaaally simplistic terms, consciousness is subjective experience, and there's an argument that all of us in the crowd were experiencing and responding to the same things, that we had the same goal, and all felt and acted accordingly as one.

also, experiences that facilitate bonding seem to have a similar connectedness in them. "we were all a part of the same thing" can often be said with a bit of nostalgia, maybe there's something related there~

Two or maybe three years ago I suggested at a CFAR reunion that close-knit tribes / communities of humans, rather than individual humans, might be 1) alive / thinking in some important sense and 2) the natural unit of moral value. I wasn't able to communicate my intuitions in favor of this very well (and I still can't); among other things, I had in mind the tines from Fire Upon the Deep.

I feel like tine-ish levels of cooperation can be achieved in online games (like EverQuest) where everyone knows their role and the role of everyone else. It's harder to achieve and even to imagine in very dynamic environments with shifting roles.

I’ve done a great deal of high-end raiding (i.e., difficult group challenges, where success relies on perfect cooperation and synergy as well as on individual competence) in World of Warcraft. Based on my experiences, I can say that the comparison to tines, and any suggestion that there is any sort of “group consciousness” in any meaningful way, are severely misplaced.

As you can see, I similarly struggled to communicate my ideas. Probably more than you did, however.

Two or maybe three years ago I suggested at a CFAR reunion that close-knit tribes / communities of humans, rather than individual humans, might be 1) alive / thinking in some important sense and 2) the natural unit of moral value
  1. I am not sure that small groups of humans are complicated enough in their interactions to form a collective mind capable of thought.
  2. It seems like tribe-centered moralities have a poor track record, but that obviously assumes a metric for evaluating moral success that you might dispute.