I recently had the pleasure of reading "The Weirdness of the World", by philosopher Eric Schwitzgebel. The central thesis, that every theory of consciousness and our place in the universe contains some aspect(s) which defies common sense, is so fascinating and far-reaching, that I'd almost recommend the book solely on the basis of the ideas contained within it. Covering real world and significant implications of artificial intelligence (AI), different types of consciousness, and even the wilder (though perhaps less practical) implications of things like living in an infinitely large universe, the book is at heart current analytical philosophy complete with experiments, counter argumentation from prominent contemporary philosophers, and interesting end notes. However, Schwitzgebel's arguments are so well argued, nuanced, and complete that they are quite compelling. Add to that his clarity of thought and grounded, personable writing style that make the whole thing very engaging. I had a really hard time putting it down!

Download ww-cover.jpg

In my last blog post (https://world.hey.com/cipher/theory-of-everyone-by-michael-muthukrishna-a-book-review-f34a3e18) I reviewed "A Theory of Everyone", by Michael Muthukrishna. I was impressed by the thought-provoking ideas in the first half of that book, some of which have real applications about how humans learn and innovate, but found the theory to be very vague on the edges and his prescriptions in the second half of the book to be under-developed and not particularly inspiring. After reading "The Weirdness of the World" I've come to realize that Muthukrishna's Theory of Everyone lacks a strong foundation and that this fundamentally limits what can be done with it. In particular, Muthukrishna refers to "collective brain thinking" and "higher levels of cooperation", but ignores the philosophical and moral implications. Muthukrishna wants us to think and plan ahead, but his Theory of Everyone doesn't give us the tools we'd need.
 
In contrast, Schwitzgebel's "The Weirdness of the World" digs very deeply into foundational concerns about how we think about consciousness (human and otherwise), and takes these concerns to fascinating places. Indeed, to the very edge of what we know and how that should inform our approach to near future developments. In particular, in chapter eleven he develops a cogent analysis of artificial intelligence (AI) and why it's so important that we think about and plan ahead for potentially conscious machines.

Still, for those thinking Integrative Information Theory (IIT) or some other theory has nailed what consciousness is and are looking for confirmation, you will not find it in The Weirdness of the World. Instead you'll find a beautiful and beautifully argued series of interrelated essays that will broaden your mind and enrich your ideas, possibly even changing them in some way, regardless of what you knew, or what you thought you knew, before you started reading it.

Let's start by returning to The Weirdness of the World's central thesis: every theory of consciousness and our place in the universe contains some aspect(s) which defies common sense. More specifically, this is broken down into two separate, but related theses:

Helpfully, he develops a "taxonomy of weirdness":

The "theoretical wilderness" concept is key. He puts it succinctly: "one of the core ideas of this book, is that the metaphysics of mind is a theoretical wilderness. Something bizarre must be true about the relationship between mind and world, but we can't know - at least not yet, not with our current tools - which among the various bizarre possibilities is correct."

From there, Schwitzgebel explores compelling arguments, as well as counter arguments, for this so-called theoretical wilderness, beginning with chapter three, If Materialism Is True, the United States Is Probably Conscious (note: from here on out I will bold the chapter number and title when I first introduce them). All, or most, of this chapter is available online here: https://faculty.ucr.edu/~eschwitz/SchwitzAbs/USAconscious.htm. Like the entire book, it is a fun read. In this case, he uses not only real world examples, but also considers some thought experiments involving theoretically possible aliens: Sirian Supersquids and Antarean Antheads.

For me personally, chapter three was an important wake up call that materialism, while tending to hew closely with empirical science, can lead to wild consequences. I won't go into each argument here, but suffice it to say that he articulates arguments and counter arguments that various flavors of materialism lead us to, giving credence to the idea of group consciousness and that the country of the United States represents one such group. In my non-expert estimation Schwitzgebel makes a compelling case that even if empirical science uncovers foundational aspects of consciousness and leads to a winnowing of the field of competing theories, the bizarreness of the implications will likely remain. Including the possibility that groups of conscious entities are, above and beyond the constituent entities, themselves conscious.

At the end of the book the author includes an appendix of additional counter arguments to group consciousness from philosophical luminaries such as Dennett and Chalmers. I highly recommend it.

In chapter four, 1% Skepticism, Schwitzgebel takes examples from his own life to explore whether we should assign any credence to bizarre and dubious theories and, if so, what implications that might have. His argument here is that if the question of the relationship of our mind to the cosmos remains in a theoretical wilderness, then shouldn't we ascribe some amount of credence, not zero and not infinitesimal, to some of the wild ideas? "1% skepticism" refers to being a little bit skeptical (more or less 1%, in fact, by Schwitzgebel's reasoning) that one's own idea of how the world works might be radically, and meaningfully, wrong.

In this chapter, two things stood out to me. First, 1% skepticism, in my mind, turned into a bit of a skepticism of the gaps argument. This is my own idea, not Schwitzgebel's. For example, one consequence of 1% skepticism is that we might expect greater scientific knowledge to be associated with a lower credence to skepticism. That is, the more reliable and consistent and detailed our knowledge about and ability to predict the world around us, the less likely it is that we are radically incorrect about our relationship to the cosmos. If that is true, then an ancient human, if taught modern analytical philosophy but somehow without any additional knowledge about the world, might ascribe a much higher credence to skepticism than say a human in 2024. For example, an ancient human encountering a rare solar eclipse may ascribe a 50% credence to skepticism about their relationship to the cosmos. As we move through history to the present, the built up knowledge about the world would slowly at first, then more quickly in the 20th Century, drive the credence in skepticism down eventually to zero.

Unfortunately, my idea of skepticism of the gaps is an over simplification. There are aspects of the various theories about our relationship to the cosmos which, in my opinion, we'll likely never be able to fully pin down. Ultimately, that undermines my skepticism of the gaps idea and lifts Schwitzgebel's 1% skepticism even higher in terms of reasonableness. 

The other big thing that stood out to me in this chapter is that he justifies his conversion to agnosticism due to 1% skepticism (part eleven, "A Defense of Agnosticism"). Not only did this make me stop and consider my own beliefs, but it also brought up a specific idea which Schwitzgebel brings up later in the book in relation to a different thing entirely. What it brought up for me is the idea that belief is binary: either I believe in something or I don't. I can't kind of believe in something, otherwise I'd always believe in everything or at least most things which have greater than infinitesimal credence to me. In that way I consider myself an atheist: someone who lacks belief in any particular god or religion. I can't rule out the possibility that there's a god completely, but to me I remain an atheist.

Perhaps the difference in my position compared to Schwitzgebel is that he is defining atheist differently. I don't believe he clarifies his definition of atheist in the book. Additionally, I suspect there is a deep history of philosophical examination of this topic with which I am unacquainted. Indeed, on Schwitzgebel's website I see a bibliography of numerous articles about belief which he has penned: http://www.faculty.ucr.edu/~eschwitz/. At any rate, in chapter 10, Schwitzgebel suggests consciousness could be binary in the same way I think of belief, "If one is a little conscious, one is conscious." Later on the same page, "In the abstract, it is appealing to suspect that consciousness is not a dichotomous property and that garden snails might occupy the blurry in-between region. It's a plausible view that ought to be on our map of antecedent possibilities. However, the view requires conceiving of a theoretical space -- in-between consciousness -- that has not yet been well explored." Perhaps in-between faithfulness has been explored. Regardless, I had fun thinking about all of it! I really appreciate books that can inspire and enable such fruitful thinking.

I hadn't intended this post to become a chapter-by-chapter review, but here we are at chapter five, Kant Meets Cyberpunk. You're probably wondering what's so cyberpunk about eighteenth century philosopher Immanuel Kant? I was too. Kant came up with something called transcendental idealism, which Schwitzgebel points out is a "historically important alternative to materialism." In chapter three, materialism led us to the wild possibility that the United States is itself conscious. Competitors to materialism, it turns out, can lead to wild things as well. In this case the simulation hypothesis. "Simulation hypothesis" is the informal name for those theories that the world itself is a simulation. There are various types of these theories which Schwitzgebel mentions in a number of different places in the book. Of course, the general ideas have been popularized by movies like The Matrix and by some prominent celebrities. If you haven't see it, I recommend Welt am Draht (World on a Wire), a two-part German miniseries from 1973 which is an early fictional story about the simulation hypothesis: https://www.imdb.com/title/tt0070904/.  While the idea feels recent thanks to video games and virtual reality, I believe the hypothesis was actually first considered in earnest by philosophers and scientists in the mid 20th century, as a natural consequence of theoretical computer science.

Here I'm just going look at how Schwitzgebel defines transcendental idealism, how that can inform simulation theories, and some interesting unanswered questions this brought up to me. Here is how he defines it:

You can probably see where he's going with this.  Transcendental idealism's claim that we cannot know things as they are in themselves leaves room for fundamental reality to be radically different than we perceive it to be. Pointing out a straight-forward, simple example Schwitzgebel says, "It is conceivable that intelligent, conscious beings like us could spend most or all of their lives in a shared virtual reality, acting upon empirical objects laid out in an immersive spatial environment, possibly not realizing that they have biological brains that aren't part of the same spatial manifold." The possibilities abound and the author leads us through some more.

This chapter stimulated some questions in my mind: some people have built working computers within the video game Minecraft (examples: https://www.businessinsider.com/virtual-computers-built-inside-minecraft-2015-2) (tutorial: https://minecraft.fandom.com/wiki/Tutorials/Redstone_computers) so could our brains be simulated in a virtual environment? If so, would it be conscious? Is it possible to conceive of and understand radically different realities that are fundamentally non-computable or non-spatial? Is the answer to that question related to the structure of our brains or the structure of our minds or both?

Not directly related, but this chapter also reminded me of my desire to create an immersive virtual environment with four spatial dimensions. As with any virtual environment it would have to be projected onto a two-dimensional stereoscopic field within virtual reality headsets in order to give us the perception of three dimensional depth (so the fourth dimension would remain flattened). A somewhat novel user interface to enable travel along the additional spatial dimension of virtual space would need to be implemented (this could be a gesture, or additional mouse, or keyboard motion). I see no other practical limitations and it sounds really fun to me. Considering chapter eight of The Weirdness of the World (see below), I wonder how quickly humans could adapt to such an environment...

The next chapter gets very personal and grounded. Indeed, the chapter is dedicated to a family member. Nevertheless, chapter six's ideas are as lofty and thought-provoking as ever: Experimental Evidence for the Existence of an External World. I won't go into too much detail here, but I will say that the personal nature of what the author goes through is both deeply touching and illustrates an important, and very human aspect of philosophy. A majority of the chapter is taken up in a series of experiments that the author conducts on himself with the help of a graduate student, Alan Tonnies Moore. The basic idea is that if the external world does not exist, that is, some form of extreme solipsism is in effect, then it should not be possible that some other aspect of reality can outperform oneself at various computational and memory access tasks. If some part of reality can outperform oneself, then it stands to reason that there is a lower credence to the claim of solipsism and a higher likelihood that those other things exist in some meaningful way externally to oneself. I don't want to spoil the chapter for you, but I'll say that the experiments involve things like calculating pi, recalling random words, and playing chess.

The problem with reviewing this book is that I loved it and I want to tell you about everything in it! I will do my best to keep things concise. Chapter seven, Almost Everything You Do Causes Almost Everything (under Certain Not Wholly Implausible Assumptions); or, Infinite Puppetry is all about the weirdness of living in an infinitely large universe. I really enjoyed this chapter, because it made me think about cosmology and quantum mechanics, but in new ways. For this, Schwitzgebel worked with theoretical physicist Jacob Barandes.

Usually when people bring up an infinite multiverse they neglect the idea of living in a single universe which is infinite. Interestingly, Cosmologist Will Kinney's book An Infinity of Worlds does cover this from a scientific perspective and Schwitzgebel looks at it from a philosophical perspective (I highly recommend reading both as they are both excellent and have very little overlap). Just a single infinitely large universe leads to counter-intuitive possibilities in a number of ways and Schwitzgebel builds on some ideas in order to arrive at the particularly bizarre one suggested in the title: that our actions can cause almost everything thus leading to infinite puppetry: where our actions cause the actions of an infinity of other conscious beings similar to ourselves. (Note: I'm not doing his arguments justice; Schwitzgebel is careful and nuanced, constructing his arguments in a clear way over many pages which naturally I can't replicate here).

Schwitzgebel and Barandes justify this based on the idea of quantum fluctuations. This might sound like quantum bullsh*t (I recommend Physicist Chris Ferrie's book Quantum Bullsh*t as a short and fun read), but they aren't making the fallacious arguments of a crank here. While quantum fluctuations are far more likely at the very small scales of elementary particles, large scale quantum fluctuations are possible. If the universe is infinite in size and lasts for an infinite amount of time, then it turns out there is plenty of chance for entire star systems or even galaxies to pop into existence, perhaps ones that are nearly identical to our own. This is the bizarre nature of thinking about infinity. Things that are so unlikely to occur on the space and timescales we are aware of that we would consider their chance of occurring in billions of years infinitesimally small, are actually almost certain to occur at some point in infinity. Indeed, the age and size of the universe that we can observe is a tiny, nearly infinitesimal slice of such an infinite universe. Billions is almost nothing when compared to infinity.

In particular, what struck me in their arguments was the development of an infinite regress of puppetry and how the most likely case, that we ourselves are puppets, is the same kind of conclusion I'm familiar with for certain versions of the simulation hypothesis. In the case of simulation, if it is possible to create conscious entities in a simulated environment, and if those simulated conscious entities are able to do the same thing themselves, then there could be an infinite nesting of simulations where each simulation leads to a new simulation contained within it. If that were true, then it would mean that we are likely, statistically, somewhere in the middle of that chain of infinite simulations, not in the original reality where the chain began. So too with infinite puppetry. If infinite puppetry were true, then statistically we'd likely be puppets ourselves.

A quick plug for my own writing: I interviewed cosmologist Will Kinney last year (2023) about his book An Infinity of Worlds: Cosmic Inflation and the Beginning of the Universe and I also interviewed physicist Chris Ferrie about his book Quantum Bullsh*t: How to Ruin Your Life with Advice from Quantum Physics (if you prefer, the long version of my Quantum Bullsh*t review is on my blog here: https://world.hey.com/cipher/a-review-of-the-book-quantum-bullsh-t-6972ab3c).

Earlier chapters of the book were meant to ease us into thinking about consciousness and give us the breadth of weirdness that underlies the theoretical wilderness in which we humans currently seem to find ourselves. Those encompassed Part I: Bizarreness and Dubiety, Part II: Peculiar Possibilities, and Part III: the Size of the Universe.

Part IV: More Perplexities of Consciousness, begins with chapter eight which is a bit of a reset in order to dig deeper into consciousness and real world implications. Chapter eight is called "Consciousness, Innocent and Wonderful". Schwitzgebel points out some of the difficulties in discussing the concept of consciousness and provides readers with examples of how we might accidentally unlearn a more straight-forward definition of conscious as we proceed to more complex situations. To that end, he aims to develop a "theory-neutral definition of consciousness" that "fills an important gap in the field of consciousness studies" and covers three desiderata for his definition: 

He then defines consciousness by examples in the following areas: sensory and somatic experiences, conscious imagery, emotional experience, thinking and desiring, dream experiences, other people, and negative examples. He stresses the importance of negative examples which he feels have been overlooked in the literature. 

The Loose Friendship of Visual Experience and Reality, chapter nine, explores "whether human conscious visual experience resembles underlying reality" using a well-known example from the US: the convex passenger side mirror on vehicles which have the government mandated phrase, "objects in mirror are closer than they appear" printed on them. Central to this exploration is the idea of multiple veridicalities: "According to Multiple Veridicalities, there is more than one distinct way that an object can veridically -- that is, truthfully, accurately, and without distortion -- appear to us in vision (or any other sensory modality)." The author proceeds to make a strong case that the convex passenger side mirror is no less truthful than other ways to view reality. Part of this relies on fascinating case studies of people who have worn perception distorting lenses on their eyes and seemingly fully adapted to them.

Next we come face to face (or eye to tentacle?) with the theoretical wilderness when chapter ten asks, Is There Something It's Like to Be a Garden Snail? Or, How Sparse or Abundant Is Consciousness in the Universe? Significantly, Schwitzgebel's arguments are first grounded in a survey of the surprising biological complexity of snails. It's easy to forget, but life on Earth tends to share a great deal of DNA. The striking similarities and subtle differences between snails and other lifeforms, including humans, are important. This is where the author's definition of consciousness and exhortations to remain innocent and full of wonder become vital. While reading this chapter, and even while writing this now, I find myself unlearning our definition of consciousness from chapter eight, and instead making unjustified leaps in logic.

Schwitzgebel seems to further anticipate my unlearning and makes his argument very clear and careful. He breaks it down into three main categories: yes they are conscious, no they are not conscious, or gong. Gong is a category which is broken down further into three sub-categories: the question itself is poorly constructed, it's undecidable, or it exists on a spectrum or is a matter of degree (i.e., it's not a yes or no question). Schwitzgebel does a fantastic job arguing each of these categories which in the end highlights the considerable extent of the theoretical wilderness that surrounds the topic of consciousness. 

Another way of approaching arguments about snail consciousness is to consider how sparse or abundant consciousness might be in the universe. To that end, Schwitzgebel identifies five dimensions of sparseness or abundance. These five dimensions help us further refine the perplexities of consciousness. For example, the second dimension he identifies is the state of consciousness which refers to the question of whether consciousness is a constant or variable state. The refrigerator light error illustrates this. One might assume that the refrigerator light is constantly in the state of being on, because every time one opens the refrigerator door and looks in the light is on. Consciousness could be like that as well. One might assume one is constantly conscious (except perhaps when anesthetized or in a dreamless sleep), but maybe checking in on one's own consciousness turns the state of consciousness on. That would imply that humans are only sometimes conscious, even while awake.

An underlying issue is the nature of where our information about consciousness comes from, the "species-specificity of verbal and introspective evidence", as Schwitzgebel puts it. One imagines that the exploration of consciousness would be different if we could get detailed self-reports from non-humans we think might be conscious. Additionally, the theoretical wilderness leaves us with a serious "common ground problem". Schwitzgebel explains how researchers approach the nature of consciousness from such divergent initial premises that there's little common ground from the start. I won't say more about this chapter, like the rest of the book it's a must-read, but you can probably guess where Schwitzgebel comes down on the question of whether or not garden snails are conscious.

We arrived at something of a crossroads in chapter ten when discussing the garden snail's consciousness. Theoretical wilderness can seem to leave us lost and un-grounded. Yet, time marches on and we must make decisions with the information we have. In chapter eleven, The Moral Status of Future Artificial Intelligence, we are shown both that there is a way forward and also that we need to act soon (although he's less optimistic that we will actually do so). Central to his argument in this chapter is the idea that AI may have debatable personhood at some point in the near future. In my opinion, while it remains possible we'll never develop conscious AI, given the recent technological progress on Large Language Models (LLMs) and other forms of machine learning, it seems reasonable to expect that we will reach the point of having AI that are debatably conscious (related to debatable personhood) before we have found our way out of the theoretical wilderness with consensus on a solid theory of consciousness. Schwitzgebel seems to agree. Schwitzgebel defines debatable personhood this way, "an AI might have debatable personhood in two distinct ways: It might be debatably conscious, or alternatively it might indisputably be conscious but not meet the required threshold in every dimension that viable theories of personhood regard as morally relevant. Furthermore, these sources of dubiety might intersect, multiplying the difficulties."

Part of the way out of this situation is developing consensus around a theory-neutral definition of consciousness (as the author attempts in chapter eight). From there, we need a clear articulation of the idea of debatable personhood (as the author attempts in this chapter eleven) as well as to discuss the different types of possible AI consciousness based on our current understanding. To that end, Schwitzgebel outlines the moral status of subhuman, humanlike, superhuman (possibly deserving of greater rights), and divergent AIs. A divergent AI is one that operates in a way that is significantly different than a human. With those things defined and agreed upon we should be able to develop a consensus around policies to handle situations involving artificial entities that have debatable personhood. Developing the consensus is the part that Schwtizgebel is not optimistic we can do. Instead, we will develop artificial entities with debatable personhood long before policy catches up to it, leading to significant suffering for humans, the artificial entities we create, or both.

For example, the Full Rights Dilemma arises when we give entities with debatable personhood full rights. Schwitzgebel asks us to consider a situation in which two humans and six artificial entities (with debatable personhood) are trapped in separate rooms in a burning building. In this situation only one or the other group can be saved before the building collapses. If we ascribe full rights to these artificial entities then we would need to save the six artificial entities first, leaving the two humans likely to die. However, if we later determine that artificial entities of this type are not, in fact, conscious or do not actually meet the threshold for personhood, then we made a tragic mistake and saved unworthy artificial entities instead of the two humans.

One response to the Full Rights Dilemma is to head it off with the "design policy of the excluded middle": "According to this policy, we should either go all in, creating AIs we know to be persons and treating them accordingly, or we should stop well enough short that we can be confident that they are not persons." At first that sounds reasonable enough, however, Schwitzgebel points out that "despite the appeal of this policy as a means of avoiding the Full Rights Dilemma, there is potentially a large cost." Namely, a significant slow down of technological progress on AI. The ramifications of such a slow down could be quite large indeed. Progress developing other technologies, medicines, etc. that may have benefited from the banned forms of AI could lead to unnecessary human suffering on a large scale.

As I write this, artificial intelligence is a very hot topic. Schwitzgebel recently wrote a couple of blog posts about aspects of AI (here: https://schwitzsplinters.blogspot.com/) and he references Theoretical Computer Scientist Scott Aaronson's blog in his book's references (here: https://scottaaronson.blog/) regarding a fascinating and cogent post by Aaronson about Integrative Information Theory (IIT). If you are into that kind of stuff I highly recommend checking those blogs out. That reminds me, in my longer review of Physicist Chris Ferrie's book Quantum Bullsh*t on my blog (here: https://world.hey.com/cipher/a-review-of-the-book-quantum-bullsh-t-6972ab3c) I point out that even Large Language Models (LLMs) can help us understand something about ourselves:
 

Thinking about consciousness with Schwitzgebel makes me think of what physicists have said about quantum mechanics. Physicist Niels Bohr reportedly said, "If you can fathom quantum mechanics without getting dizzy, you don't get it." Along the same lines Physicist Richard Feynman: "I think I can safely say that nobody understands quantum mechanics." Consciousness seems to fit in that part of reality which is tantalizingly close to our understanding, yet evades commonsense upon examination. Like quantum mechanics, it is important that we continue to make progress on refining our understanding of consciousness. However, unlike quantum mechanics we don't currently have a rigorous mathematical framework to draw upon when our commonsense fails us.

In the final chapter, chapter twelve, Weirdness and Wonder, Schwitzgebel makes a case for an open, as opposed to closed, philosophy:

As I write this review I'm on the 44th floor of an apartment building dog sitting for a friend. I'm looking out of a window onto a city constructed by, and full of, humans. I see rain moving in from the West. Lolla, the dog, paws at me, because she needs to go out. Millions of conscious beings coexisting right here around me, living out our lives in a universe that is likely infinite. Over time philosophy and empirical science will shed further light on the nature of consciousness, but Schwitzgebel has me convinced that whatever we learn will be weird and likely to lead to additional open questions worthy of skepticism. Thinking about our place in the cosmos makes me feel strange. After reading this book, that only seems natural.