The Big Picture: On the Origins of Life, Meaning, and the Universe Itself – Sean Carroll

Summary: Carroll argues in favor of a philosophical worldview he calls poetic naturalism. Poetic naturalism holds that there is a single universe which, at a fundamental level, behaves in a consistent and predictable way, but that there are many different, valid ways of talking about the universe. He supports this argument by discussing how we are able to learn about the universe (empiricism), how certain we can be of things (we can never be 100% or 0% certain of the truth of a statement, but we can come very, very close), what we know about the universe (at a fundamental level, the universe is highly predictable and likely deterministic), how new, complex phenomena can emerge from simpler systems, whether things like consciousness and human agency are consistent with a deterministic universe (as far as we can tell, they are), the limits of what science and reason can teach us (they can’t tell us how to build our moral systems, but they can warn us against the complacency of thinking that there exist fundamental moral values), and what we should do with this knowledge.

Thoughts: I found The Big Picture to be very compelling: while it challenged several of my beliefs, its largest effect has been to clarify my worldview: Carroll expresses ideas I have been grasping at but unable to clearly articulate, and explores some of their finer details and ramifications. The discussion of Bayesian reasoning was particularly valuable, as well as the idea that even if a category is invented, it still deserves to be thought of as “real” so long as it does a good job explaining the patterns we observe in the universe. Of all the topics Carroll raises and deftly handles, the issues in the book’s final section, “Caring”, are least well-addressed, and I’ll need to explore them further. That said, I’d recommend this book to anyone interested in understanding how the universe works.

(The notes below are not a summary of the book, but rather raw notes - whatever I thought, at the time, might be worth remembering.)

Carrol, Sean. 2016. The Big Picture: On the Origins of Life, Meaning, and the Universe Itself. Dutton.

Prologue

• 3-4: Poetic Naturalism
  • Naturalism: there’s just one world, the natural one
  • Poetic: there are multiple ways of talking about the world

Part One: Cosmos

1. The Fundamental Nature of Reality

• 10: Ontology and Epistemology
  • ontology: the study of the basic structure of the world
  • epistemology: how we obtain knowledge about the world
• 11: Roots of Naturalism: traces in Buddhism, some thinkers in ancient Greece and Rome, Confucius, Chinese thinker Wang Chong. “only in the last few centuries that the evidence in favor of naturalism has become hard to resist.”

2. Poetic Naturalism

• 17: ontologies can have fundamental categories and derived categories.
  • fundamental ontological categories: those at the deepest level of reality
  • derived ontological categories: categories that are invented because they are useful - they allow us to talk about higher-order phenomena.
    • usually, they’re fuzzy at the edges
• 17-18: ontologies can be rich (with many fundamental categories) or sparse (a small number of categories, or perhaps just one)
• 19: “one benefit of a rich ontology is that it’s easy to say what is ‘real’ - every category describes something real”. In a sparse ontology, things are more difficult.
  • Eliminativism argues that anything that isn’t a fundamental ontological category is not real, is illusory
• 19: In this book, Carroll argues that useful ideas that describe macroscopic reality deserve to be called “real”. Useful ideas are in contrast to various non-useful ideas, which don’t reflect/predict reality very accurately
• 19-20: poetic naturalism
  • Naturalism:
    • "1 - there is only one world, the natural world
    • 2 - the world evolves according to unbroken patterns, the laws of nature
    • 3 - the only reliable way of learning about the world is by observing it"
  • Poetic:
    • "1 - There are many ways to talk about the world
    • 2 - All good ways of talking must be consistent with one another and with the world.
    • 3 - our purposes in the moment determine the best ways of talking"
• 20-21: Carroll outlines three types of stories we can tell within poetic naturalism
  • 1. fundamental ones
  • 2. “emergent”, “effective” descriptions, valid within a limited domain
  • 3. stories related to values - these can’t be checked by observation/science: we’ve got to figure these out on our own

3. The World Moves by Itself

• 25: in physics, you can make extremely simplifying assumptions (frictionless surfaces, etc.) and get useful results; in fields like biology, economics, such extreme simplifications don’t usually generate useful results
• 26-27: Aristotle thought every motion had a cause. People such as John Philoponus (6th C.), Ibn Sina (aka. Avicenna, ca. 1000), and Jean Buridan participated in a change in thinking, where motion is not caused by anything, but rather conservation of momentum is just a property of matter
• 28: cosmological argument (e.g. Acquinas, borrowing from Aristotle): the universe happens because of an “unmoved mover” - an immortal and benevolent God that makes the world tick. Once we abandon the notion of motion requiring causes, and we accept conservation of momentum, this argument falls apart. “The universe doesn’t need a push. It just keeps going.”

4. What Determines What Will Happen?

• 32: as far as we can tell, the universe behaves somewhat like a computer: you take an input (the state of the universe, i.e.the positions and trajectories of each particle), it does a calculation (the laws of physics), and you get an output (the state of the universe at a later point in time). An idea like this was proposed in the first millennium BC by “Ajivika, a heterodox school of ancient Indian philosophy”
• 34: two laws of conservation that we observe in the universe: conservation of momentum and conservation of information
• 35: Laplace’s argument (“Laplace’s demon”) is that “perfect information leads to perfect prediction. Chaos theory argues that slightly imperfect information leads to very imperfect prediction.” Since the two talk about different things, i.e. perfect vs. imperfect information, the two aren’t contradictory.
  • j: digital information mitigates against chaos - since digital information is discretized, it is much easier for that information to be transmitted/processed/etc. perfectly
• 36-37: our best understanding/model of particle physics is deterministic, but our best model of, say, human behavior is not deterministic. The most useful model depends on the scale of the phenomenon we’re trying to talk about.

5. Reasons Why

• 39: “Not only do we seek order and causation, we favor fairness as well.” Melvin Lerner identified the “just world fallacy,” noting that people tend to blame victims for the unfortunate things that happen to them.
• 40: Principle of Sufficient Reason: “For any true fact, there is a reason why it is so, and why something else is not so instead”. Carroll argues that this is a flawed principle: many facts have reasons, but some facts are “brute” - “things that are simply true, with no further explanation possible” - basic parts of our ontology.
• 40-41: deduction vs. induction vs. abduction
  • deduction: begin with a set of axioms and derive conclusions from them
  • induction: start with examples, and generalize about a wider context
  • abduction: start with how we think the world works, take an observation, and try to decide the range of possible explanations that are consistent with all the facts we have. related to bayesian reasoning
• 43: the laws of physics don’t function along the lines of “cause” and “effect” - they are “simply patterns that connect what happens at different places and times” - on the level of particles, past and future are the same
• 44: the arrow of time (entropy increasing) is not a fundamental property of the laws of physics.
• 44: questions like “why are there three dimensions of space” or “why was entropy low near the big bang” may have answers, but they may not - they could be brute facts.

6. Our Universe

7. Time’s Arrow

• 54: It’s better to think of the world as ekinological - from “start” or “departure” - rather than teleological, directed towards a future goal
• 55: a pendulum moves in exactly the same way forwards in time as backwards in time. In fact, all physical motions are reversible.
• 57: entropy increasing, time’s arrow, is observable on a macro scale. The motions of all the particles can in principle be reversed, but certain macroscopic features are highly unlikely to come about on their own (temperature gradients, an intact eggshell, etc.)
• 58: equilibrium = state of maximum possible entropy. in a system in equilibrium, time’s arrow doesn’t exist.
• 59: all differences between past and present, as far as we can tell, can be traced back to the fact that entropy was low when the big bang happened, and the universe is slowly shifting towards equilibrium.

8. Memories and Causes

• 61-62: The current universe has only one possible set of past states, based on the principle that information is conserved. But we don’t have perfect information, so we try to narrow down all the possible past states of the universe that are consistent with the one we observe. We use the Past Hypothesis, that entropy was higher in the past, to narrow down these possibilities.
  • e.g. we observe a broken egg. Any number of things could have led to that collection of molecules being in that state, but since we’re familiar with one particular pattern of entropy increasing (i.e. intact eggs becoming broken), we conclude all possible past universes consistent with our observation include an unbroken egg which was dropped, thrown, or otherwise broken.
• 63: on a raw physical level, there are no “causes” - the universe arrived at its current state because the universe evolves along consistent patterns, but there’s no reason to prioritize one moment over any other as a “cause”.
• 65: on a macro level, how can we establish causation? modal reasoning: we compare what actually happened to what could have happened, based on hypothetical universes where some detail is different.
  • say we leave a whole cake in an empty room, a person walks in with a knife, and a moment later walks out, and now we observe that the cake is in two pieces. In a hypothetical universe where the person didn’t walk into the room the cake, we would observe it in just one piece. So, we can conclude that the person caused the cake to become sliced.
  • (refer to the work of David Lewis, “one of the most influential twentieth-century philosophers whom non-philosophers have never heard of.”
  • how do we know that the person slicing with the knife caused the cake to become sliced, and not the other way around? Leverage: We can imagine many ways for the cake to become sliced (I could slice it, someone could have put it in a guillotine, etc…), but we couldn’t have observed an unsliced cake after the person pressed the knife into the cake. That’s why causality goes one way, and not the other.

Part Two: Understanding

9. Learning about the World

• 70: What is probability really? Frequentists argue that probability is shorthand for how frequently things would happen if a trial was run an infinite number of times. Bayesians argue that probability a statement of your belief that something will happen.
• 70-71: if we’re being honest/careful, we hold none of our beliefs with 100% certainty - every belief is held with some degree of belief. The technical term for degree of belief is credence
• 72: abduction: trying to find the “best explanation”, by starting with a prior credence, and updating that credence as information comes in based on compatibility with each original possibility.

10. Updating our Knowledge

• 77-78: (Credence in Proposition X given observation D) is proportional to (likelihood of observation D if proposition X is true) * (prior credence in proposition X)
• 78-79: for every question, there are a range of possible factual explanations, all of which have a non-zero, non-one likelihood of being true. Bayesian statistics allow us to work with these likelihoods quantitatively.
• 79-80: occam’s razor: in some cases, it’s tricky gauge simplicity, but in some cases, it’s easy to gauge relative simplicity.
  • E.g. consider three explanations of planetary motion
    • 1. the motion of planets is governed, pretty closely, by the laws of motion/gravity described by Newton
    • 2. the motion of planets follows Newton’s laws only by coincidence, but each planet is actually guided by an angel
    • 3. most of the planets follow Newton’s laws, except for Mercury, which is guided by an angel
  • no matter what credence we assign to propositions 1) and 2), we must necessarily assign a lower credence to proposition 3)
• 80: One possible objection to Bayesian approach is its reliance on subjective priors. But there’s no alternative to starting with subjective priors. If different people with different priors follow a Bayesian approach in good faith, their credences will converge
• 80-81: Bayesian approach breaks down when people assign a credence of 0 or 1 to a view. This is a practical problem in reasoning with other people, when no matter what the evidence says, it will make no impact on their beliefs.
• 81: Carroll argues that “evidence that favours one alternative automatically disfavors others”. J: Is this necessarily true? if you have a range of similar explanations, and evidence comes in that generally supports them but doesn’t really specify between them, does that not boost credence of all of them while decreasing credence in dissimilar explanations?

11. Is It Okay to Doubt Everything?

• 87: Cartesian doubt: Descartes asserted that there is no way to be absolutely certain that the world around us is real. Carroll argues that he is correct in asserting this.
• 89: Bolzmann Brain: assume the universe is infinitely old. At some point, particles will arrange to form, say, the entirety of the Milky Way as it is now. But it is incredibly more likely that particles come together to form a human being, with memories encoded in their brain. In fact, it’s incredibly more likely than that for just a brain, encoded with memories, to form.
  • How do we know that each of us is not a Bolzmann brain that has momentarily and randomly blipped into existence? We can’t know for certain, but if it were true, we would expect the world to immediately dissolve into chaos - but with each moment where that doesn’t happen, we can update our priors to strongly disfavor that explanation.
• 90: Wittgenstein, in On Certainty: “From its seeming to me—or to everyone—to be so, it doesn’t follow that it is so. What we can ask is whether it can make sense to doubt it.” Carroll: “Put conversely, something might conceivably be true, but there might not be any point in assigning much credence to it.”
• 91: Carroll argues that if a particular worldview leaves us paralyzed (for example, that we’re in a Bolzmann brain, or that some demon is deceiving us), we’re free to give low credence to it (at least until/unless evidence comes in that suggests we should give it more credence)
• 92: the cognitive instability of the Bolzmann Brain hypothesis: “You use your hard-won scientific knowledge to put together a picture of the world, and you realize that in that picture, it is overwhelmingly likely that you have just randomly fluctuated into existence. But in that case, your hard-won scientific knowledge just randomly fluctuated into existence as well; you have no reason to actually think that it represents an accurate view of reality. It is impossible for a scenario like this to be true and at the same time for us to have good reasons to believe in it. The best response is to assign it a very low credence and move on with our lives.”

12. Reality Emerges

• 94: Emergence: a feature of a system is emergent if it is 1) not part of the system’s fundamental ontology and 2) useful in describing the system at a more macroscopic level.
• 96: it’s important to recognize that each theory has its own domain of applicability - take it outside of that domain, and it’s likely not to provide useful explanations. Within its domain of applicability, however, a valid theory is autonomous - it doesn’t need to refer to anything outside of itself.
• 99: because of this autonomicity, every theory can be compatible with many theories that look at things on a smaller scale. This doesn’t work the other way around, though - we could, in principle, deduce a theory at a higher level of emergence based on a valid lower-level theory.
• 99: Why study emergent phenomena? Usually, they’re way more computationally efficient.
• 101: Phase transitions (e.g. water melting, then boiling) - transitions where something moves from one domain of applicability to another (e.g. modelling a substance’s behavior based on it being solid vs. liquid vs. gas)
• 102-103: Carroll draws a comparison between phase transitions and Thomas Kuhn’s idea of paradigm shifts.

13. What Exists, and What Is Illusion?

• 107: to look up: Philip Anderson’s 1972 article “More is different” - argues that it can be useful to have multiple, overlapping models of a phenomenon.
• 110: Carroll suggests (or appears to suggest) that it’s worth taking the idea of “strong emergence” seriously - the idea that there are some phenomena that must be described on their own terms, that an emergent model cannot (even in principle) be derived from a full understanding of the more microscopic levels of our model.
  • j: i’m not sure what to think of this - maybe I missed his refutation of this point somewhere along the way? I suppose it’s worth being open to the possibility of this being true, but I think that if some emergent phenomenon was discovered that could not be predicted based on our lowest-level model, it would simply be evidence that there’s a problem with that model…
• 113: When working with multiple models at different scales, be careful of borrowing terminology from one model and transplanting it outside of its domain of applicability. E.g. a person can want something, even though none of the atoms that they are made of wants the thing.

14. Planets of Belief

• 115-119: Carroll suggests the metaphor of “planets of belief”. Philosophers like Descartes have looked for a “bedrock” belief upon which everything else is built, but we now recognize that there’s no fundamental belief that cannot be questioned (in the same way there is no bedrock beneath a planet - it’s held together by its own gravity). If large collections of beliefs cohere with each other (i.e. form a planet), that suggests that each of those beliefs is worth putting more credence in. Of course, there is always the possibility that a new, incompatible observation will hit our planet of belief, making us have to reevaluate our belief system.
• 120: planets of belief develop their own “defence systems”, namely cognitive biases. e.g. self-serving bias - we gravitate towards theories that flatter ourselves - and confirmation bias - we only accept new ideas that are compatible with our current beliefs.
  • confirmation bias can lead to the backfire effect - by having to justify our belief when challenged, we come to hold that belief even more strongly
• 120: Carol Tavris and Elliot Aronson’s “Pyramid of Choice”: “Imagine two people with nearly identical beliefs, each confronted with a decision to make. One chooses one way, and the other goes in the other direction, though initially it was a close call either way. Afterward, inevitably, they work to convince themselves that the choice they made was the right one. They each justify what they did, and begin to think there wasn’t much of a choice at all. By the end of the process, these two people who started out almost the same have ended up on opposite ends of a particular spectrum of belief”
  • j: as an antidote to this in yourself, it’s worth noting your credences when you make decisions, just to remind yourself how uncertain you were at the outset.
• 121: to fight against cognitive biases, you can assign lower credences to things you want to be true, and weigh evidence that’s incompatible with your worldview more heavily than evidence that’s compatible.

15. Accepting Uncertainty

• 126: mathematics and logic prove that, given a certain set of axioms, certain statements logically follow, and this would be true in all possible universes. Science, in contrast, teaches us about things that are true in our universe, but could conceivably be true in an alternate universe.
  • j: cf. thinking in time vs thinking out of time?
• 126: a basic thing to recognize about doing science (i.e. learning about how the universe works): knowledge is never perfect.
• 128: black holes of belief: when you holds on to your beliefs so strongly that they collapse into a singularity (i.e. you hold a belief with 100% conviction, and no insight/evidence could possibly change your mind)
  • one example of a black hole of belief is religious conviction
  • scientists make assumptions about how the universe works - these are parts of their planet of belief - but each of these assumptions is open to re-evaluation if necessary.
• 129: some propositions are so likely to be false that, even though they can never be 100% proven, we can safely ignore them. “We take 'I believe x’ not to mean ‘I can prove x to be the case,’ but rather ‘I feel it would be counterproductive to spend any substantial amount of time and effort doubting x.’”

16. What Can We Know about the World without Actually Looking at It?

• 132: “‘Theorem’ doesn’t imply ‘something that is true’; it only means ‘something that definitely follows from the stated axioms.’”
• 133: you may hear that science is only concerned with “natural” processes (as opposed to “supernatural” ones), but this isn’t correct - science is concerned with finding the truth, and if there is a phenomenon that could only be explained by supernatural causes, one would need to accept that explanation if one was doing science. “Science isn’t characterized by methodological naturalism but by methodological empiricism… Science is a technique, not a set of conclusions.”
• 135: in contrast to empiricism is the tradition of rationalism - that we can “[learn] about the world by reason alone, without any help from observation”

17. Who Am I?

• 141: Eliminativism (the tendency to dismiss many things as illusory) and its opposite, essentialism (the tendency to “take certain categories as immovable features of the bedrock of reality”). Some people take an essentialist approach to, for example, gender.
• 142: “A person who is biologically male but identifies as a woman isn’t thinking to themselves, ‘Male and female are just arbitrary categories, I can be whatever I want.’ They’re thinking, ‘I’m a woman.’ Just because a concept is invented by human beings, it doesn’t imply that it’s an illusion. Saying, ‘I am a woman,’ or just knowing it, is absolutely useful and meaningful.”
  • j: This makes sense. It’s worth thinking about how this applies to other issues like race etc.

18. Abducting God

• 147-148: Imagine a world where miracles happen frequently, or where religious traditions sprang up around the world and all arrived at the same beliefs, or a world without random suffering, or the souls of the dead regularly interacted with the living, etc. “In any of those worlds, diligent seekers of true ontology would quite rightly take those aspects of reality as evidence for God’s existence. It follows, as the night the day, that the absence of these features is evidence in favor of atheism.”

Part Three: Essence

19. How Much We Know

• 157: David Hume on when to accept reports of miraculous events: “we should accept such a claim only if it would be harder to disbelieve it than to believe it.”
• 157-158: even if we discover a deeper theory to explain reality than the Core Theory, that will not “dispove” the Core Theory - it will still be able to accurately model phenomena within its domain of applicability (in the same way that Newton’s laws of motion still predict things in their domain of applicability)

20. The Quantum Realm

21. Interpreting Quantum Mechanics

• 166: the quantum measurement problem - a particle in a superposition that’s observed will collapse into one state or another, but what does “observed” actually mean? Physicists still aren’t sure how to deal with this question, but they have a few approaches:
  • there’s a deeper way the universe works, where, if we figured it out, we would be able to predict the results of quantum experiments.
  • 167: quantum mechanics don’t actually exist, and we can only talk about the results of experiments (!?) (the antirealist approach - Niels Bohr seems to have taken this stance (!?))
  • 168-169: “When you make a measurement of a particle to see whether it’s spinning clockwise or counterclockwise…, the wave function doesn’t collapse into one possibility or the other. It evolves smoothly into an entangled superposition, part of which has ‘the particle is spinning clockwise’ and ‘you saw the particle spinning clockwise,’ while the other of which has ‘the particle is spinning counterclockwise’ and ‘you saw the particle spinning counterclockwise.’” (the Many-Worlds approach)
• 170: in the Many-Worlds approach, all the universes are there from the beginning, at least in principle. "It’s all the other versions of quantum mechanics that have to work to get rid of the extra worlds"
• 171: common to all interpretations of quantum mechanics: “what we see when we look at the world is quite different from how we describe the world when we’re not looking at it.” j: what?

22. The Core Theory

• 174: Bosons vs. Fermions:
  • Bosons “can pile on top of each other to create force fields, like electromagnetism and gravity” (e.g. photon, graviton)
  • Fermions take up space - only one can be in a given location at a given time. E.g. neutrons, protons, electrons

23. The Stuff of Which We Are Made

• 178: Scientific theories vary in terms of how powerful they are. A powerful theory, in this sense, is a restrictive theory: it tells us lots of things about what cannot happen - one can start with few assumptions and draw reliable and wide-ranging conclusions.
• 183: We’re certain that we know all the particles that are involved in everyday life. Any particle that hasn’t been detected yet must either:
  • interact with ordinary matter so weakly that it is rarely produced,
  • be so heavy that it can’t be produced by our most powerful particle accellerators, or
  • decay incredibly rapidly

24. An Effective Theory of the Everyday World

• 187: an effective theory: one that models the features of a system that you care about while stripping away features that you don’t care about.
• 188: effective theories are compatible with many possible theories that describe the system at a more microscopic level. - example of emergence
• 191: for earlier theories—Newton’s, Laplace’s—we didn’t originally know how far down the theory held. But for effective field theory, we do know how far it extends - we know, based on experiment, that the theory holds as long as you’re below a certain energy and above a certain distance.
• 192: some emergent theories can be dismissed based on knowledge of a more microscopic theory (Carroll offers the example of spoon-bending with one’s mind), but many emergent theories are very difficult to derive from a lower-level theory.

25. Why Does the Universe Exist?

• 198: one interpretation of quantum theory is that time is an emergent phenomenon. “Quantum mechanics allows us to consider universes that are fundamentally timeless, but in which time emerges at a course-grained level of description.”
• 200-201: as far as we know, all the conserved quantities in the universe add up to zero:
  • electromagnetism: positive and negative charges balance out
  • the energy in the universe is balanced out by gravity - if mass is spread uniformly in the universe, the universe has zero energy.
  • so, it is physically possible for the universe to have had a beginning, when there was no mass and no energy
• 203: “There may be no ultimate answer to the ‘Why?’ question. The universe simply is, in this particular way, and that’s a brute fact. Once we figure out how the universe behaves at its most comprehensive level, there will not be any deeper layers left to discover.”

26. Body and Soul

27. Death Is the End

• 215: if anyone wants to claim that there is something beyond physical reality, they have to account for how that thing interacts with the particles of which we are made, i.e. explain why the Core Theory is incomplete, and how it must change.
• 219: people used to think that life was a property that some matter had. Nowadays, we recognize that it’s a process.
• 219: earlier theories of essential substances:
  • élan vital - proposed by Henri Bergson
  • Phlogiston - flammable substances were thought to have it, and it was released by burning
  • Caloric - thought to account for heat - caloric would flow from a hot substance toward a cold one
• 220: experiments have been done to test out-of-body experiences (where one feels they are floating above and looking down on themselves) - researchers will hide an object in a hospital room that could only be seen from above. No one has yet observed such an object during an out-of-body experience.

Part Four: Complexity

28. The Universe in a Cup of Coffee

• 227: Another useful definition of entropy: if there are many possible ways for particles to be arranged where it would still look the same (macroscopically), it’s high-entropy. if there are few possible ways, it’s low entropy.
• 229: Imagine mixing a cup of coffee which had been carefully set up to have cream sitting clearly above coffee. to begin, low-entropy (you can’t exchange bits of cream with bits of coffee, without changing the appearance). As you begin to stir is, complexity appears (swirls), and eventually, the whole thing becomes a homogeneous mixture of cream and coffee (high entropy). As far as we can tell, complexity can only happen at intermediate levels of entropy, and only while a system is travelling from a low- to a high-entropy state.
• 232: “algorighmic” or “Kolmogorov” complexity: the minimum length of a computer program to output a given binary string. If a string is random, the simplest program would be print([string])
• 233: complexity doesn’t evolve in all cases: if there are only small-scale interactions (cream/coffee molecules interacting with their immediate neighbours), things tend to form a cline - the development of complexity seems to require at least some large-scale interactions (e.g. a spoon that begins to stir the coffee)
• 235: Complexity relies on the Past Hypothesis - if everything were already high-entropy, then the only way interesting features could be generated is through random fluctuations (which would happen extremely infrequently). complexity requires the flow of a system moving toward equilibrium

29. Light and Life

• 239: Schrödinger’s definition of life:
  • “When is a piece of matter said to be alive? Wen it goes on ‘doing something,’ exchanging material with its environment, and so forth, and that for a much longer period than we would expect an inanimate piece of matter to ‘keep going’ under similar circumstances.” -hinges on the much longer
• 240: another way of thinking about entropy. “Free energy” (aka. useful energy) is slowly converted to disordered energy. A system with lots of disordered energy is high entropy
  • 241: another way to put the 2nd law of Thermodynamics: over time in a closed system, free energy tends to be converted to disordered energy
• 242: we’re lucky to live in an environment that’s very low-entropy: lots of free energy comes at us from only one direction (the sun), and the earth radiates disordered energy (IR radiation) in all directions

30. Funneling Energy

• 247: An important difference between designed machines and biological systems: “Design tends to be specific, and brittle… Biological organisms, which have develoged over the years with no specific purpose in mind, tend to be more flexible, multipurpose, and self-repairing.”

31. Spontaneous Organization

• 252: self-organization: when a large system, made up of smaller subsystems, falls into an orderly configuration or behavior, even though all of the subsystems work independently, and aren’t programmed with any specific goal in mind.
  • j: think boids, Conway’s game of life
• 254: many computer programs used to demonstrate self-organization aren’t accurate analogs of the universe, since they aren’t reversible, and information isn’t preserved. j: E.g. conway’s game of life, there are many ways to create any particular still life.
• 255: lipids and phospholipids tend to self-organize into micelles and bilayers. an an aqueous solution, their hydrophobic tails “want” to cluster together (physically, they tend to not remain in orientations where they contact water molecules) - on this more macroscopic level, we can see how it is useful to think of “wanting” - volition is an emergent phenomenon.
• 257-258: a Markov blanket: in order to accurately model a component of a system, you need only look at (A) all the things that affect with the components, plus (B) all the things that are affected by the individual components, plus © all the things that affect (B)s. Karl Friston argues that a cell membrane (or an individual human brain) can be thought of as a Markov blanket.

32. The Origin and Purpose of Life

• 268: “DNA is good at storing information, and proteins are good at performing biochemical functions; RNA is able to do both, although it’s not as good at either one”
• 268: Ribozymes (n.b. different from ribosomes) can both catalyze their own synthesis and catalyze the synthesis of proteins. Wow!

33. Evolution’s Bootstraps

• 273-274: Richard Lenski’s experiment in evolving E. coli bacteria. Started them out in a particular solution; every day, 10% of the bacteria are transferred into a fresh solution. Over time, bacteria evolved the ability to metabolize citrate.
• 276-277: this ability to metabolize citrate didn’t evolve directly; rather, at some point near generation 20000, a random mutation (genetic drift) occurred that set the stage for the later evolution of citrate metabolism.

34. Searching through the Landscape

• 282: it’s easy to conceptualize natural selection as a search for peaks in a landscape across two dimensions, but it’s worth remembering that natural selection’s search landscape involves as many dimensions as there are genes. It can take a long time for species to climb to the top of a hill.
  • 282: large parts of the lanscape can be relatively flat, meaning that genetic drift accounts for most change over time
  • 282-283: also, the landscape is time-dependent: the heights and locations of high points are constantly shifting
• 288-289: Michael Behe’s idea of “irreducable complexity” - an argument that some things are so complicated, and involve so many interdependent parts, that they must have been designed top-down rather than evolved. He offers the example of a mousetrap and an eye. Turns out both can slowly evolve (and in fact did, demonstrated by Joachim Dagg’s study of historical changes in mousetrap design)
• 290: Octopuses don’t have blind spots: the nerves for their photoreceptors are behind their retinas, not in front of them as in mammals

35. Emergent Purpose

• 293: Truly new emergent phenomena should not be seen as surprising - as matter gets organized in new ways, new phenomena emerge. Carroll argues “purpose” is one of these emergent phenomena. Is “purpose” a useful concept for describing how the world works within some domain of applicability? Then it exists.
• 297: macroscopic information (the kind contained in a book, or on a hard drive) differs from Laplacean information, the kind at the deepest level of reality. Macroscopic information is context-dependent: it emerges when it reliably correlates with other aspects of the world (e.g. someone seeing the word “giraffe” and tending to think of a giraffe)
• 301: in the preceding pages, Carroll looks at many possible ways an extra-natural force (i.e. God) could influence the world. If God does do this, he must do it in very subtle ways indeed, in order for them to be compatible with the Core Theory.
  • “It is unclear why God would place such a high value on acting in ways that human beings can’t notice. This approach [asserting that God could guide evolution through quantum indeterminacy] reduces theism to the case of the angel steering the moon [c.f. chapter 10]… You can’t disprove the theory by any possible experiment, since it is designed precisely to be indistinguishable from ordinary physical evolution.”

36. Are We the Point?

• 303: Carroll states that fine-tuning - that the fundamental constants and laws of the universe are fine-tuned to allow for the evolution of life - is “the most respectable argument in favor of theism.”
• 305: if the values of universal constants were different, it’s hard for us to imagine how life could evolve. But it’s extremely hard to derive chemistry or biology from the first principles of the Core Theory in the current universe. Who knows what complex and exotic phenomena might emerge?
• 306: The anthropic principle: "If we live in a world where conditions are very different from place to place, then there would be a strong selection effect on what we will actually observe about that world: we will only ever find ourselves in a part of the world that allows for us to exist."
• 307: if we invoked the anthropic principle just to explain away problems of fine-tuning, it wouldn’t be very satisfactory. But the existence of a multiverse is a prediction made by several other theories - theories that were designed for other purposes
• 308: String theory predicts multiple dimensions, but they are hidden, they are packed into tiny areas that cannot be directly observed. There are many possible ways these dimensions could be bundled - each called a compactification - and each corresponds to a different possible set of values for universal constants.
• 311: theism doesn’t do a great job as a predictive theory: concepts of “God” are very fuzzy, allowing us to tweak our theories to match the universe we actually observe. Equations are more restrictive, and give us less freedom to adjust our theories post hoc.

Part Five: Thinking

37. Crawling into Consciousness

• 318: the transition from water to land may have contributed to the emergence of human consciousness. In water, light is absorbed more quickly than in air, so there is more evolutionary pressure to be able to see a thing and react quickly, rather than to see something and plan to move towards or away from it. (creatures above water can see things many miles away, but creatures in water, only tens of meters)
• 321-322: Daniel Kahneman’s two modes of thought:
  • System 1: automatic, fast, intuitive
  • System 2: conscious, slow, rational
  • System 2 has only a little bit of control over System 1.
• 323: Endel Tulvig’s two kinds of memory
  • semantic memory: recalling facts
  • episodic memory: recalling personal experience

38. The Babbling Brain

• 333: a system’s critical point: the moment when a system is changing from one state to another. “Criticality can be thought of as a sweet spot between boring order and useless chaos”.
  • 332-333: Carroll argues that consciousness can only occur during criticality

39. What Thinks?

• 338: the idea of a philosophical zombie: a person who looks and acts just like a normal person, but has no inner experience, no qualia
• 342: imagine we find out exactly how one particular neuron in your brain, and swap it with an artificial one that behaves exactly the same way (in terms of inputs and outputs). Would your brain still be conscious? What if we substituted out two neurons, or half of them, or all of them?
  • substrate independence: in the same way that Conway’s Game of Life behaves the same way whether implemented in C or in Lisp or manually on a Go board, “many different substances could embody the patterns of conscious thought.”
• 343: Carroll argues that “consciousness” and “understanding” are concepts that humans invent in order to create more efficint, useful descriptions of the world (in the same way we invented “entropy” and “heat”)
• 344: “you” are not the sum of all the particles in your body, but rather the pattern of the interactions between all those particles. Consciousness probably works the same way.
• 345: although fundamental particles don’t have intent, it can be really useful to talk about higher-order phenomena (like human beings) having intention.

40. The Hard Problem

• 350: qualia: the subjective experience of how a thought/sensation feels to us.
• 354: Is my experience of blue the same as your experience of blue? Strictly speaking, no - your experience of blue is an way of talking about all the interactions between particles in your brain, not my brain. But I have good reason to believe that the neurons/particles in my brain are arranged similar to those in yours, so my experience of blue is probably pretty similar to yours.

41. Zombies and Stories

• 361: a strong-reductionist approach to explaining subjective experiences is to say that they have strict physical correlates - e.g. Hilary Putnam’s assertion that pain is the firing of c-fibers (parts of the nervous system that carry pain signals).
  • 362: pain does correlate with the firing of c-fibers in humans. But it’s an unreasonable leap to say that the firing of c-fibers is pain - what about an artificial creation with the same connectome (i.e. artificial c-fibers)? What about animals that have very different nervous systems from humans?

42. Are Photons Conscious?

• 368: plants transfer energy via particles in quantum superpositions during photosynthesis; certain phosphorus atoms in ATP may become quantum-entangled.
• 370: Gödel’s incompleteness theorem doesn’t exactly say that there are true statements that can’t be proven. Rather, it says that in a sufficiently powerful, consistent formal system, such statements exist. Consciousness is not such a system, but by using Bayesian reasoning, we can be pretty darn sure of the truth of a statement.

43: What Acts on What?

• 373: One theory’s success is not proof that another theory is wrong; we lose faith in a theory if it either 1) doesn’t do a good job describing observed phenomena or 2) is internally incoherent
• 374: logical difficulties can arise when we use the vocabulary that’s relevant in one domain of applicability to describe another domain of applicability: we can describe a phenomenon in multiple ways, but not by mixing and matching bits of these different theories.
• 375-376: we double back to Strong Emergence to entertain the idea of downward causation: the idea that large-scale phenomena can influence the behavior of elements at more microscopic levels. In principle, downward causation could occur, but if we observed this (we haven’t yet), a more reasonable explanation than “the higher-order phenomenon is influencing the behavior of lower-level components” is “the lower-level theory has a smaller domain of applicability than we previously thought.”

44: Freedom to Choose

• 379: Compatibilism - in philosophy, the idea that free will is compatible with a deterministic (or at least impersonal) universe.
• 380: “When a feature of our current state implies (given the Past Hypothesis, and all else being equal) something about the past, that’s a memory; when a feature of our current state implies something about the future, that’s a cause of some future effect. The small differences in a person’s brain state that correlate with different bodily actions typically have negligible correlations with the past state of the universe, but they can be correlated with substantially different future evolutons. That’s why our best human-sized conception of the world treats the past and the future so differently. We remember the past, and our choices affect the future.”
• 380-381: Carroll offers one popular definition of free will as “the ability to have acted differently.” With perfect information about the universe’s quantum state, one doesn’t have free will. But none of us has access to that information, so given the information we have about any past situation, it’s quite plausible that we could have acted differently.
• 381: the idea of libertarian freedom (a philosophical concept, as distinct from political libertarianism) states that human agency introduces indeterminacy into the world. It would imply that humans can break the laws of physics just by thinking.
• 384: Carroll predicts that we’ll have to get more and more used to the determinacy of human action as psychology continues to advance, but believes that our best model of human behavior will retain some element of personal autonomy.

Part Six: Caring

45. Three Billion Heartbeats

• 389: “The best way to talk about ourselves is as thinking, purposeful agents who can make choices. One of those choices, unavoidably, is what kind of life we want to lead.”
• 392: Life is characterized by change, by riding a wave in the current from low entropy to high entropy. “In human terms, the dynamic nature of life manifests itself as desire.”
  • desire gets a bad rap - there are philosophical traditions that state that one can achieve true happiness if one eschews desire completely - but it’s unsustainable.
  • curiosity is a form of desire; “desire is an aspect of caring: about ourselves, about other people, about what happens to the world.”

46. What Is and What Ought to Be

• 395: David Hume noted that using deductive reasoning, one can never derive an “ought” statement from a collection of “is” statements. In naturalism, “is” statements are all there is.
• 396: valid arguments vs sound arguments
  • valid: the conclusion follows logically from the premises
  • sound: the argument is valid, plus the premises are true.
• 398: many people have tried over the years to derive an “ought” from "is"es. In every case, it involves introducing a hidden “ought” premise that seems so unobjectionable as to be mostly unnoticeable.
• 401: it’s worth noting that “is” statements can be falsified, reasoned about using Bayesian approaches, etc. That’s not true of “ought” statements.
• 401-402: you can still use logical reasoning to work with “ought” statements, but we have to accept certain “ought” statements as axioms in order to do so.
  • “There is an entire form of logical thought called instrumental rationality, devoted to answering questions of the form ‘Given that we want to attain a certain goal, how do we go about doing it?’”
• 402: Carroll states that, based on what we know about the world, there is no objective grounding for moral statements. Based on this fact, “we should recognize that our desire for an objective grounding for morality creates a cognitive bias, and should compensate by being especially skeptical of any claims in that direction.”

47. Rules and Consequences

• 405-406: ethics vs. meta-ethics. Ethics is about deciding what is right and wrong. Meta-ethics asks what it means to say something is right or wrong, and examines what we need to consider when creating ethical systems.
  • naturalism can’t tell us about ethics, but it can tell us about meta-ethics: it notes that ethical systems are not discovered but rather created by human beings, and should be evaluated on those grounds.
• 406: consequentialism vs deontology
  • a consequentialist approach evaluates actions based on their results
  • a deontological (etymology is distinct from “ontology”) approach evaluates actions based on whether the action itself is right or wrong.
  • System 1 is responsible for deontological impulses, while System 2 is engaged with approaching things consequentially
• 408: to add to consequentialism and deontology: virtue ethics doesn’t evaluate actions, but worries about who a person is.
• 408: a theory is “useful” to a scientist if it does things like providing insight into the behavior of a system, or making accurate predictions based on minimal input.
• 409: “A moral framework is ‘useful’ to a poetic naturalist to the extent that it reflects and systematizes our moral commitments in a logically coherent way.”
  • follows how people normally reason about morality: we have a feeling about right and wrong, and we try to systematize it.
• 410: moral constructivism: recognizes that morality originates in individuals, societies. But they aren’t arbitrary: people can talk/reason about their ethical systems, and work to improve them, putting together a systems that works for many people.
  • as distinct from moral relativism, which suggests that morality is created within a group of people cannot be judged from outside that group.
• 411: David Hume stated that “reason is, and ought to be, the slave of our passions”. Reason can help us get what we want, but our passions tell us what we want.
  • as we’ve learned about how the universe works, we can see that Hume was right: morals can’t be derived from reason alone, or from God, or from nature.

48. Constructing Goodness

• 412-413: Utilitarianism is the best-known consequentialist ethical approach. Sometimes runs into problems - how do you quantify utility? Also, it tends to value the lives of those that are far away from us as equal to those who are near to us, which runs counter to many humans’ moral intuitions.
• 413: “Psychologists have suggested that moral reasoning in general, and deontological reasoning in particular, functions primary to rationalize opinions that we reach intuitively, rather than leading us to novel moral conclusions.”
• 414: Carroll suggests that a practical solution may be to create moral codes that combine bits and pieces of various approaches. E.g. “soft consequentialism” - which mostly evaluates actions based on their consequences, but also the actions themselves to some degree.
  • j: this appeals to me intuitively, but thinking about it logically, I’m doubtful this is the best I can do. My current thinking is that people should employ an extremely-large-scale utilitarianism, zooming out to try to predict the consequences of their actions, habits etc. far into the future.
• 418: “Poetic naturalism doesn’t tell us how to behave, but it warns us away from the false complacency associated with the conviction that our morals are objectively the best… We don’t need an immovable place to stand; we need to make our peace with a universe that doesn’t care what we do, and take pride in the fact that we care anyway.”

49. Listening to the World.

• In lieu of ten commandments of poetic naturalism, Carroll offers "Ten Considerations: a list of things we think are true, that might be useful to keep in mind as we shape and experience our own ways of valuing and caring about our lives."
  • 1. Life isn’t forever
  • 2. Desire is built into life
  • 3. What matters is what matters to people
  • 4. We can always do better
  • 5. It pays to listen
  • 6. There is no natural way to be
  • 7. It takes all kinds
  • 8. The universe is in our hands
  • 9. We can do better than happiness
  • 10. Reality guides us
• 422: Mathematical proofs can be logically perfect; scientific discoveries are messier, drawn out affairs; moral advances are even messier.
  • “It may seem strange to claim the existence of moral progress when there isn’t even an objective standard of morality, but that’s exactly what we find in human history. Progress comes, not from new discoveries in an imaginary science of morality, but from being more honest and rigorous with ourselves—from uncovering our rationalizations and justifications for behavior that, if we admit it, was pretty reprehensible from the start. Becoming better people is hard work, but by sifting through our biases and being open to new ideas, our ability to be good advances.”
• 426: in thinking about meaning in life, it’s worth distinguishing between synchronic and diachronic meaning.
  • synchronic: depends on your state of being in the current moment
  • diachronic: about the journey; depends on the progress you feel you’re making towards your goals
• 426: a class of cognitive biases: “positive illusions”: beliefs people hold that are not true, but that make them happy.
  • 427: Carroll argues that while “illusions can be pleasant…, the rewards of truth are enormously greater.”

50. Existential Therapy

Further Reading

• would be worth checking out:
  • Dennet, D. C. (2014). Intuition Pumps and Other Tools for Thinking. W. W. Norton.
  • Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
  • Gottschall, J. (2012). The Storytelling Animal: How Stories Make Us Human. Houghton Mifflin Harcourt.