From Bacteria to Bach and Back

Part I: Turning Our World Upside Down

1. Introduction

• How come there are minds?
  • Minds evolved and created thinking tools that eventually enabled minds to know how minds evolved, and even to know how these tools enabled them to know what minds are.
  • What thinking tools? The simplest, on which all the others depend in various ways, are spoken words, followed by reading, writing, and arithmetic, followed by navigation and mapmaking, apprenticeship practices, and all the concrete devices for extracting and manipulating information that we have invented: compass, telescope, microscope, camera, computer, the Internet, and so on.
  • These, in turn, fill our lives with technology and science, permitting us to know many things not known by other species. We know there are bacteria. Even bacteria don't know there are bacteria.
  • Our minds are different. It takes thinking tools to understand what bacteria are, and we're the only species (so far) endowed with an elaborate kit of thinking tools.
• A birds-eye view of the journey:
  • Life has been around on Earth for around 4bn years. The first 2bn were spent optimizing the machinery for self-maintenance, energy acquisition and reproduction, and the only living things were relatively simple, single-celled entities - bacteria or their cousins, archaea: the prokaryotes.
  • Then an amazing thing happened. Two different prokaryotes collided and instead of one eating the other, it let it go on living, and, by dumb luck, found itself fitter, more competent in some way that mattered, than it had been before. This was perhaps the first successful instance of technology transfer. A fortuitous mutation almost never happens, but evolution depends on those rarest of rare events. This is the birth of the eukaryotes
  • Every living thing big enough to be visible to the naked eye is a multicellular eukaryote.
  • The Cambrian Explosion, which occurred over several million years about 530m ya, saw the sudden arrival of a bounty of new life forms.
  • The "MacCready Explosion", at the dawn of human agriculture, about 10k ya, transformed the terrestrial vertebrate biomass (excluding insects, other invertebrates, and marine animals). At the beginning, humans plus their livestock and pets make up only 0.1%, and now we make up 98% (mostly cattle). This explosion is based on three factors - population, technology, and intelligence (our so-called native intelligence depends on both our technology and our population numbers).
• Dennett identified the "romantic" and "killjoy" sides of the duel over the stature of animal minds. We are not the God-like geniuses we think we are, but animals are not so smart either, and yet both humans and other animals are admirably equipped to deal "brilliantly" with many of the challenges thrown at them.

2. Before Bacteria and Bach

• until there were systems that could be strictly called reproducing systems, the processes at work were only proto-evolutionary, semi-Darwinian, partial analogues of proper evolution by natural selection; they were processes that raised the likelihood that various combinations of ingredients would arise and persist, concentrating the feedstock molecules until this eventually led to the origin of life.
• A living thing must capture enough energy and materials, and fend off its own destruction long enough to construct a good enough replica of itself.
• The reverse-engineering perspective is ubiquitous in biology and is obligatory in investigations of the origin of life. It always involves some kind of optimality considerations: What is the simplest chemical structure that could possibly do x? Or would phenomenon x be stable enough to sustain process y?
• Orgel's second rule: "Evolution is cleverer than you are."
• Here is an example of a possible gambit in the origin of life:
  • It is tempting to assume that the very first living thing capable of reproducing must have been the simples possible living thing (given the existing conditions on the planet at the time).
  • Make the simples replicator you can imagine and then build on that foundation.
  • But this is by no means necessary. It is possible, and more likely, I think, that a rather inelegantly complicated, expensive, slow, Rub-Goldberg conglomeration of objets trouvés was the first real replicator, and after it got the replication ball rolling, this ungainly replicator was repeatedly simplified in competition with its kin.
  • Many of the most baffling magic tricks depend on the audience no imagining the ridiculously extravagant lengths magicians will go to in order to achieve a baffling effect. If you want to reverse-engineer magicians, you should always remind yourself that they have no shame, no abhorrence of bizarre expenditures for tiny effects that they can then exploit. Nature, similarly, has no shame - and no budget, and all the time in the world.
• Adaptionism is alive and well; reverse-engineering is still the royal road to discovery in biology.

3. On the Origin of Reasons

• There are three strategies to adopt when trying to understand, explain, and predict phenomena:
  • The physical stance - is the least risky but also the most difficult; you treat the phenomenon as obeying the laws of physics, and use physics to predict what will happen next.
  • The design stance - is only for things that have been designed, either artifacts or living things or their parts, and have functions or purposes.
  • The intentional stance - works primarily for things that are designed to use information to accomplish their functions. It works by treating the thing as a rational agent, attributing "beliefs" and "desires" and "rationality" to the thing, and predicting that it will act rationally.
• Evolution by natural selection is not itself a designed thing, an agent with purposes, but it acts as if it were. It is a set of processes that "find" and "track" reasons for things to be arranged one way rather than another.
• The reasons found by human designers are typically (but not always) represented in the minds of the designers, whereas the reasons uncovered by natural selection are represented for the first time by those human investigators who succeed in reverse-engineering Nature's productions.
• Our human world of reasons grew out of a world where there were no reasons.
• Two meanings of the word "why":
  • What for - "Why are you handing me your camera?"
  • How come - "Why does ice float?" This is asking for a cause or a process narrative
• Evolution by natural selection starts with "how come" and arrives at "what for". We start with a lifeless world in which there are no reasons, no purposes at all, but there are processes that happen.
• A central feature of human interaction, and one of the features unique to our species, is the activity of asking others to explain themselves, to justify their choices and actions, and then judging, endorsing, rebutting their answers, in recursive rounds of the "why?" game.
• Our capacity to respond appropriately in this reason-checking activity is the root of responsibility. Those who cannot explain themselves or cannot be moved by the reasons offered by others, those who are "dead to" the persuasions of advisors, are rightly judges to be of diminished responsibility and are treated differently by the law.
• The "logical space of reasons" is bound by norms, by mutual recognition of how things ought to go. Wherever there are reasons, there is room and need for some kind of justification and the possibility of correction when something goes wrong. This normativity is the foundation of ethics.
• But there are two kinds of norms and corrections:
  • social normativity - concerned with social norms, practice, and collaboration
  • instrumental normativity - concerned with quality control or efficiency, the norms of engineering
• Natural selection is an algorithmic process, a collection of sorting algorithms that are themselves composed of generate-and-test algorithms.
• In the prebiotic or abiotic world (before life), there were cycles at many spatio-temporal scales: seasons, night and day, tides, the water cycle, and thousands of chemical cycles discoverable at the atomic and molecular level, gradually changing the conditions in the world and thus raising the probability that something new will happen.
• This led to differential persistence, some temporary combinations of parts hang around longer than others. The rich can get richer, even though they can't yet bequeath their riches to descendants.
• Differential persistence must then somehow gradually turn into differential reproduction.
• "Serendipity" is when something good happens randomly, while "clobbering" is when something bad happens randomly.
• Walls or membranes that are randomly persisted are serendipitous in that they allow internal cycles to operate for a time without interference, and we see the engineering necessity of membranes to house the collection of chemical cycles - the Krebs cycle and thousands of others - that together permit life to emerge.
• Before we can have competent reproducers, we have to have competetent persisters. We are witnessing an automatic (algorithmic) paring away of the nonfunctional, crowded out by the functional.
• There are reasons why the parts are shaped and ordered as they are and this is the birth of reasons. Through Darwinism about Darwinism, we see the gradual emergence of the species of reasons out of the species of mere causes, what fors out of how comes*
• Natural selection is an automatic reason-finders, which "discovers" and "endorses" and "focuses" reasons over many generations.
• If there happens to be a "difference that happens to make a difference" then we have the germ of a reason, a proto-reason, and when this is selected to persist longer, then we can see emerge the accumulation of function by a process that blindly tracks reasons.
• Reasons existed before there were reasoners. There are reasons why trees spread their branches but they are not, in any strong sense, the trees' reasons. They don't "have" the reasons and they don't need to have the reasons.
• Darwin didn't extinguish teleology - he naturalized it.
• Reverse-engineering in biology is a descendant of reason-giving-judging.
• The evolution of what for from how come can be seen in the way we interpret the gradual emergence of living things via a cascade of prebiotic cycles. Free-floating rationales emerge as the reasons why some features exist; they do not presuppose intelligent designers, even though the designs that emerge are extraordinarily good.

4. The Strange Inversions of Meaning

• The world before Darwin was held together not by science but by tradition, through the trickle-down theory of creation from God, which Darwin replaced by the bubble-up theory of creation.
• Design space:
  • Skyhooks - float high in design space, unsupported by ancestors, the direct result of a special act of intelligent creation.
  • Cranes - are non-miraculous innovations in design space that enable ever more powerful lifting and efficient exploration of the space. Endosymbiosis is a crane, as are sex and language and culture.
• Turing showed that it was possible to design mindless machines that were absolutely ignorant, but that could do arithmetic perfectly, following "instructions" that could be mechanically implemented.
• This is "competence without comprehension" and Turing saw that it could provide a traversable path in design space from absolute ignorance to artificial intelligence.
• All the brilliance and comprehension in the world arises ultimately out of uncomprehending competences compounded over time into ever more competent - and hence comprehending - systems.
• This overthrows the pre-Darwinian mind-first vision of Creation with a mind-last vision.
• Darwin discovered evolution by natural selection, while Turing invented the computer, but he is one of the twigs on the Tree of Life who is, himself, an indirect product of the blind Darwinian processes.
• Distribution of expertise or understanding of this sort is a hallmark of human creative processes.
• "Ontology" is the set of "things" that an animal can recognize and behave appropriately with regard to, and equally the set of things that a computer program has to be able to deal with to do its job. Humans have extremely varied ontologies. Some believe in electrons and some believe in abominable snowmen, but there is a huge common core that is shared by all normal human beings from around 6 years old:
  • Manifest image - the things we use in our daily lives to anchor our interactions and conversations. For every noun in our everyday speech, there is a kind of thing it refers to. It comes along with your native language. It's the world according to us
  • Scientific image - populated with molecules, atoms, electrons, gravity, quarks. But even scientists spend most of their day in the manifest image.
• These two versions of the world that are now quite distinct were once merged or intertwined in a single ancestral world of "what everybody knows" that included all the local fauna and flora and weapons and tools and dwellings and social roles, but also goblins and gods and miasmas and spells.
• We can treat animals as having different ontologies without settling issues of whether they are conscious of them or simply the beneficiaries of designs that can be interpreted (by reverse engineers or forward engineers) as having those ontologies.
• A well designed elevator:
  • Has a kind of ontology. It is a good elevator if it interacts appropriately with its environment and its passengers. It uses variables to keep track of all the features of the world that matter to getting its job done and is oblivious to everything else.
  • It has no need to know what its ontology is or why - the rationale of the program is something only the program's designers have to understand.
  • Its prudent self-monitoring can be seen to be an elementary step towards consciousness.
• Even bacteria are good at staying alive, at making the right moves at the right times, but they have elevator-type minds, not elevated minds like ours. And these minds are the products of an R&D process of trial and error that gradually structured their internal machinery to move from state to state in a way highly likely - not guaranteed - to serve their limited but vital interests.
• But unlike the elevator there is noting at all that plays the rols of the labels or comments in a source program. There is nothing anywhere at any time in that R1D history that represents the rationales of it. But they can be discovered by reverse engineering - there is a reason why the parts are shaped as they are, why the behaviors are organized as they are, and that reason will "justify" the design (or an earlier design that has now become either vestigial or transformed by further evolution to serve some newer function.
• The elevator has replaced a human - the elevator operator - by a machine that "sorta" follows the same rules as the human. We humans often occupy this kind of intermediate level of consciousness where we have internalized or routinized through practice a set of explicit rules that we may then discard and even forget.
• The Manhattan project had a small number of intelligent designers who organized a massive group of people, most of whom knew nothing about what they were doing beyond their immediate tasks. The "need to know" principle means that it is possible to create very reliable levels of high competence with almost no comprehension for rather insulated tasks
• GOFAI can be seen in retrospect as an exercise in creating something rather Cartesian, a rationalistic expert with myriads of propositions stored in its memory, and all the understanding incorporated in its ability to draw conclusion. It relied on the comprehension of the designers to contrive systems composed of subsystems that were foresightedly equipped with exactly the competences they would need in order to handle the problems they might face.
• But modern deep-learning AI is bottom-up, using wasteful, mindless, less bureaucratic, more evolution-like processes of information extraction.
• Top-down intelligent designing works, but it is responsible for much less of the design in our world than is commonly appreciated.
• Comprehension, far from being a Godlike talent from which all design must flow, is an emergent effect of systems of uncomprehending competence; natural selection on one hand, and mindless computation on the other.

5. The Evolution of Understanding

• Human designers start with a goal (which may be refined or abandoned along the way) and work top-down, with the designers using everything they know to guide their search for solutions to the design problems they set for themselves.
• Evolution, in contrast, has no goals, no predefined problems, and no comprehension to bring to the task.
• How could a slow, mindless process build a thing that could build a thing that a slow mindless process couldn't build on its own? A process with no Intelligent Designer can create intelligent designers who then design things that permit us to understand how a process with no Intelligent Designer can create intelligent designers who then design those things.
• An organism's "umwelt" is the behavioral environment that consists of all the things that matter to its well-being.
• "Affordances" are the relevant opportunities in the environment of any organism: things to eat of mate with, openings to walk through or look out of, holes to hide in, things to stand on, and so forth.
• Organsims can be the beneficiaries of design features that imply ontologies without themselves representing those ontologies.
• Biology is reverse-engineering, and reverse-engineering is methodically committed to optimality considerations. Bacteria don't know they are bacteria, but they respond to other bacteria in bacteria-appropriate ways and are capable of avoiding or tracking or trailing things they distinguish in their umwelt.
• In software engineering, there is a reason why debugging cannot be completely automated: what counts as a bug depends on all the purposes (and sub and sub-sub-purposes) of the software, and specifying in sufficient detail what those purposes are is, at least for practical purposes, the very same task as writing debugged code in the first place!
• Design revision in Nature must follow the profligate method of releasing and test-driving many variants and letting the losers die, unexamined.
• Evolution explores the "adjacent possible".
• Natural selection is full of bugs. Organisms are filled with all-but-undecipherable "spaghettit code" of undisciplined programmers, but the free-floating rationale of the whole system is clearly good enough for practical purposes.
• When does comprehension emerge?
• We are right to adopt the intentional stance to understand the benefits derived from competences, but these competences can be provided by the machinery without any mentality intruding at all.
• We can say that organisms with spectacular competences but without comprehension, are "gifted".
• When there isn't enough stability over time in the selective environment to permit natural selection to "predict" the future accurately (when "selecting" the best designs for the next generation), natural selection does better by leaving the next generation's design partially unfixed. Learning can take over where natural selection left off, optimizing the individuals in their own lifetimes by extracting information from the world encountered and using it to make local improvements.
• Costly-signalling theory, where an animal does something to deceive or distract a predator does not need comprehension. These animals cannot choose to deceive, they simply do it automatically in certain circumstances due to a "knee-jerk reflex".
• Comprehension is not the source of competence or the active ingrediant in competence - instead, it is composed of competences.
• The illusion that understanding is some additional, separable mental phenomenon is fostered by the aha! phenomenon, or eureka effect.
• Comprehension comes in degrees, but even at the highest levels of competence, comprehension is never absolute. All comprehension is sorta comprehension from some perspective.
• We count on experts to have deep "complete" understanding of difficult concepts we rely on every day, only half-comprehendingly, and language is the capacity to transmit, faithfully, information we only sorta understand.
• The Beatrix Potter syndrome, or intentional stance towards animals works whether the rationales it adduces are free floating or explicitly represented in the midst of the agents we are predicting.
• Whatever is going on in the animal's brain has the competence to detect and respond appropriately to the information in the environment. But the intentional stance just gives the specs for the mind and leaves the implementation for later.
• We idealize everybody's thinking, and even our own access to reasons, blithely attributing phantom bouts of clever reasoning to ourselves after the fact. Asked "Why did you do that?", the most honest thing to say is often "I don't know, it just came to me," but we often succumb to the temptation to engage in whig history, not settling for how come but going for what for.
• Four grades of competence:
  • Darwininian Creatures - Have predefined and fixed competences created by the R&D of evolution. They are born hard-wired, knowing all they will ever know, they are gifted but not learners.
  • Skinnerian Creatures - Have, in addition, the ability to adjust their behavior in reaction to "reinforcement". They start out with some "plasticity". They more or less randomly generate new behaviors to test the world and those that get reinforced are more likely to recur in similar circumstances in the future.
  • Popperian Creatures - Look before they leap. They extract information about the cruel world and keep it handy, so they can use it to pretest hypothetical behaviors offline. Eventually they must act in the real world, but their first choice is not random, having won the generate-and-test competition trial runs in the internal environment model. The "habit" of "creating forward models" of the world and using them to make decision and modulate behavior is a fine habit to have, whether or not you understand it.
  • Gregorian Creatures - Their Umwelt is well stocked with thinking tools, both abstract and concrete. Only with them do we find the deliberate introduction and use of thinking tools, systematic exploration of possible solutions to problems, and attempts at higher-order control of mental searches. Only we human beings are Gregorian creatures, apparently.
• The smartest animals are not "just" Skinnerian creatures but Popperian creatures, capable of figuring out some of the clever things they have been observed to do. They engage in exploratory behavior. They need not know that this is the rationale for their behavior, but they benefit from it by reducing uncertainty. The fact that they don't understand the grounds of their own understanding is no barrier to calling it understanding, since we humans are often in the same ignorant state about how we manage to figure out novel things.
• Some animals, like us, have something like an inner workshop, a portable design-improvement facility.
• An unconscious mind is no longer seen as a contradiction in terms. The puzzle today is what is consciousness for (if anything)?
• Animals, plants, and even microorganisms are equipped with competences that permit them to deal appropriately with the affordances of their environments. There are free-floating rationales for all these competences, but the organisms need not appreciate or comprehend them to benefit from them, nor do they need to be conscious of them. In animals with more complex behaviors, the degree of versatility and variability exhibited can justify attributing a sort of behavioral comprehension to them so long as we don't make the mistake of thinking of comprehension as some sort of stand-alone talent, a source of competence rather than a manifestation of competence.

Part II: From Evolution to Intelligent Design

6. What is Information?

• Did the information age begin:
  • When people began writing things down, drawing maps, and otherwise recording and transmitting valuable information they couldn't keep in their heads with high fidelity?
  • When people began speaking and passing on accumulated lore, history, and mythology?
  • Over 530m ya, when eyesight evolved during the Cambrian Era, triggering an arms race of innovation in behavior and organs that could respond swiftly to the information gathered from the light?
  • When life began - even the simplest reproducing cells survived thanks to parts that functioned by discriminating differences in themselves and and their immediate surroundings?
• Dennett focuses on "Semantic Information", which is so important to us that we want to be able to use if effectively, store it without loss, move it, transform it, share it, hide it.
• Memory can be conceived as an information channel, just as subject to noise as any telephone line.
• Analog to digital converors (ADCs), are analogous to the sensitive cells that accomplish transduction on the out input edges of the nervous system, though the conversion in brains is not into bit strings, but neuronal spike trains.
• McCulloch and Pitts, in 1943, demonstrated the logical possibility of a general purpose representing-and-learning-and-controlling network made out of units that performed simple, nonmiraculous, clueless tasks - a comprehender of sorts made of merely competent parts.
• The brain is certainly not a digital computer running binary code, but it is still a kind of computer.
• Economic information is whatever is worth some work.
• Survival depends on information, on differential and asymmetric information: I know some things you don't know.
• Semantic information is design worth getting - design always involves R1D work of some kind, using available semantic information to improve the prospects of something by adjusting its parts in some appropriate way.
• One can actually improve one's design as an agent in the world by just learning useful facts. All learning, learning what and learning how, can be a valuable supplement or revision to the design you were born with.
• Semantic information is a distinction that makes a difference, a difference that makes a difference.
• Information in general is that which justifies representational activity, that which determines form.
• Misinformation and disinformation are dependent or even parasitic kinds of information. Disinformation is the designed exploitation of another agent's systems of discrimination, which themselves are designed to pick up useful information and use it.
• Most of what anybody knows is adaptively inert, but cheap to store, and the bits that do matter, really matter.
• Advertisers and propagandists seek to build "outposts of recognition" in other agents minds.
• In natural selection, R&D happens, designs are improved because they all have to "pay for themselves" in differential reproduction, and Darwinian lineages "learn" new tricks by adjusting their form. They are, then, in-formed, a valuable step up in local design space.
• In the same way, Skinnerian, Popperian, and Gregorian creatures inform themselves during their own lifetimes by their encounters with their environments, becoming ever more effective agents thanks to the information they can now use to do all manner of new things, including developing new ways of further informing themselves. The rich get richer. And the richer and richer, using their information to refine the information they use to refine the information they use to refine the information they obtain by the systems they design to improve the information available to them when they set out to design something.
• Useful information is a descendent of JJ Gibson's affordances.
• Information is that which is selected.
• What semantic information can be gleaned from an event depends on what information the gleaner already has accumulated.
• Offspring inherit a manifest image with an ontology of affordances from their parents and are born ready to distinguish the things that are most important to them.
• Evolution is all about turning bugs into features, and turning noise into signal
• Where does all the information in DNA come from? From the gradual, purposeless, nonmiraculous transformation of noise into signal, over billions of years.
• Information is always relative to what the receiver already knows.
• Remembering is not simply retrieving some thing that has been stored in some place in the brain.
• Don't acquire and maintain what doesn't pay for itself. More is not always better. Intentional mind-clearing or unlearning is not an unusual phenomenon.
• So much of the semantic information that streams into our heads each day is not worth getting.
• Within an organism the information-transmitting channels tend to be highly reliable.
• The brain's job in perception is to filter out, discard, and ignore all but the noteworthy features of the flux of energy striking one's sensory organs.
• Semantic information:
  • Is valuable - misinformation and disinformation are either pathologies or parasitic perversions of the default cases.
  • Its value is receiver-relative and not measurable in any nonarbitrary way but can be confirmed by empirical testing
  • The amount carried or contained in any delimited episode or item is also not usefully measurable in units but roughly comparable in local circumstances
  • Need not be encoded to be transmitted or saved.
• Utility or function counts against a creation, since copyright is intended to protect "artistic" creation.
• You have to be informed to begin with, you have to have many competences installed, before you can avail yourself of information. How are humans so much better at extracting information from the environment than any other species?
• We have many more affordances - a hardware store is a museum of affordances.
• Information is information, not matter or energy. No materialism that does not admit this can survive at the present day.

7. Darwinian Spaces: An Interlude

• Darwin talked of "the infinite complexity of the relations of all organic beings to each other and to their conditions of existence".
• Evolution by natural selection is change in a population due to:
  • variation in the characteristics of members of the population,
  • which causes different rates of reproduction, and
  • which is heritable.
• Whenever all three factors are present, evolution by natural selection is the inevitable result, whether the population is organisms, viruses, computer programs, words, or some other variety of things that generate copies of themselves one way or another.
• Darwin discovered the fundamental algorithm of evolution by natural selection, an abstract structure that can be implemented in different materials or media.
• Darwin refutes essentialism, the ancient philosophical doctrine that claimed that for each type of thing, each natural kind, there is an essence, a set of necessary and sufficient properties for being that kind of thing. But in fact there is no principled way of drawing a line between related things.
• A Darwinian space is a 3D array to show 3 variables and see to what extent a process is pure Darwinism, quasi-Darwinian, proto-Darwinian, or not Darwinian at all.
• Evolutionary processes are themselves evolutionary products and as a result emerge gradually and transform gradually.
• Looking at this is "Darwinism about Darwinism".
• We could for example look at the relationship between:
  • Fidelity of heredity. Evolution depends on high-fidelity copying but not perfect copying, since mutations (copying errors) are the ultimate source of all novelty.
  • Dependence of realized fitness differences on intrinsic properties. The differences in fitness between members of a population may depend on "luck" or "talent" or any combination in between. When luck is dominant, you can have genetic drift, when some random feature gets boosted to fixation.
  • Continuity (smoothness of fitness landscapes). Natural selection is a gradual process and depends on blindly taking "small steps". When the landscape is "rugged" (rapidly changing), evolution is next to impossible since small steps are uncorrelated with progress or even maintaining one's fitness.
• De-Darwinizing is when a lineage that evolved for generations under paradigmatic Darwinian conditions moves into a new environment where its future comes to be determined by a less Darwinian process.
• The developmental process that wires up your brain is a de-Darwinized version of the process that evolved the eukaryotes. There are many neurons in your brain at birth and only those that make the right connections are saved, but they just happen to be in the right place at the right time.
• The origin of life (from the abiotic world to bacteria) is a set of processes that went from pre-Darwinian to proto-Darwinian to Darwinian.
• You can look at cultural evolution using a Darwinian space with the axes:
  • Growth vs reproduction - eg the Roman Catholic Church grwos but seldom spawns descendants these days, while the Hutterites are designed to send of daughter communities whenever a community gets big enough. Religions are large complex social entities. Words are more like viruses, simpler, unliving, and dependent on their hosts for reproduction.
  • Cultural vs genetic - Trust is (mainly) a cultural phenomenon
  • Internal complexity
• An inverted Darwinian space shows Darwinian at the base and intelligent design at the opposite extreme, with the following axes:
  • Bottom-up vs top-down - Human culture started out profoundly Darwinian, with uncomprehending competences yielding various valuable structures in roughly the way termites build their castles, and then gradually de-Darwinized, becoming ever more comprehending, ever more efficient in its ways of searching design space. As human culture evolves, it fed on the fruits of its own evolution, increasing its design powers by utilizing information in ever more powerful ways.
  • Comprehension
  • Random vs directed search
• All the real cultural phenomena occupy the middle ground, involving imperfect comprehension, imperfect search, and much middling collaboration.
• We are the only species so far that has developed explosively cumulative culture. Culture has obviously been a good trick for us, but what barriers have stood in the way of other species developing it?

8. Brains Made of Brains

• A bacterium can discriminate a few vital differences to make itself at home in its tiny Umwelt.
• Swift control is the key competence of mobile organisms, so nervous systems, with a headquarters, are obligatory. Brains are control centers for dealing swiftly and appropriately with the opportunities and risks - the affordances - of a mobile life.
• Brains are designed by natural selection to have, or reliably develop, equipment that can extract the semantic information needed for this control task.
• Other mammals, and birds, can afford to be altricial in contrast to precocial; they are designed to be fed and protected by parents through a prolonged infancy, picking up semantic information that doesn't have to come through their genes and doesn't have to be learned by unsheltered trial and error in the dangerous world.
• Brains develop competences, including the meta-competences needed to acquire and hone further consequences.
• Turing was the epitome of a top-down intelligent designer.
• Computer programming is top-halfway-down design; the grubby details of the "bottom" of the design is something you can ignore (unless the program you are writing is a compiler.
• Complex evolvable systems (basically all living, evolvable systems) depend on being organized "hierarchically": composed of parts that have some stability independently of the larger system of which they are parts, and that are themselves composed of similarly stable parts composed of parts. A structure - or a process - need be designed only once, and then used again and again, copied and copied not just between an organism and its offspring, but within an organism as it develops. As Dawkins has observed, a gene is like a toolbox subroutine in a computer.
• In the genome, there is a vertebra-making subroutine, a finger-making subroutine, and eyelid making subroutine, each of these are modular tasks.
• The developing organism sorta understand the commands of its genes the way a von Neumann machine sorta understands its machine language instructions - it sorta obeys them.
• Bottom-up R&D is Darwin's strange inversion, but brains are not exactly like digital computers:
  • Brains are analog and computers are digital.
  • Brains are parallel and computers are (mainly) serial. - The brain's architecture is massively parallel, with a vision system about a million channels wide, but many of the brain's most spectacular activities are (roughly) serial, in the so-called stream of consciousness, in which ideas, or concepts or thoughts float by not quite in single file, but through a von Neumann bottleneck of sorts.
  • Brains are carbon based (protein etc) and computers are silicon.
  • Brains are alive and computers are not?
• Deacon argues that, by divorcing information processing from thermodynamics, we restrict our theories to basically parasitical systems, artifacts that depend on a user for their energy, for their structural maintenance, for their interpretation, and for their raison d'être. It is important, he claims, that a brain be made of cells that are themselves autonomous little agents with agendas, chief of which is staying alive, which spawns further goals, such as finding work and finding allies. His insistence on making brains (or brain substitutes) out of living neurons might look at first like some sort of romanticism - protein chauvinism - but his reasons are practical and compelling.
• The hardware of existing digital computers depends critically on millions (or billions) of identical elements. Neurons, in contrast, are all different, and they get organized not through bureaucratic hierarchies, but by bottom-up coalition-formation, with lots of competition.
• What do neurons want? Do they have nano-intentionality, agency? They want the energy and raw materials they need to thrive. Neurons are, like yeast and fungi, highly competent agents in a life-or-death struggle, in the demanding environment between your ears, where the victories go to those cells that can network more effectively, contributing to more influential trends at the levels where large-scale human purposed and urges are discernible.
• A neuron is always hungry for work; it reaches out exploratory dendritic branches, seeking to network with its neighbors in ways that will be beneficial to it.
• Top-down intelligent designs depend on foresight, which evolution utterly lacks. Evolution's design are all in a way retrospective - this is what worked in the past.
• Variable selective environments, because of their unpredictability, favor the selection of incomplete designs, along with mechanisms to tune the design to suit the circumstances, exploitable plasticity or "learning".
• Brains are more like termite colonies than intelligently designed corporations or armies.
• An organism's Umwelt is populated by two R&D processes:
  • evolution by natural selection and
  • individual learning of one sort or another.
• An organism is floating in an ocean of differences, a scant few of which might make a difference to it. Born to a long lineage of successful copers, it comes pre-equipped with gear and biases for filtering out and refining the most valuable differences, separating the semantic information from the noise.
• Bayesian hierarchical predictive coding - a method of calculating probabilities based on ones prior expectation:
  • Given that your expectations based on past experience (including the experience of your ancestors as passed down to you) are such and such (expressed as probabilites for each alternative), what effect on your future expectations should the following new data have? What adjustments in your probabilities would it be rational for you to make?
  • It is a normative discipline, purportedly prescribing the right way to think about probabilities.
  • Computer reading of handwriting involves a cascade of layers in which the higher layers make Bayesian predictions about what the next layer down in the system with "see" next. When the predictions proves false, they then generated error signals in response that lead to Bayesian revisions, which are then fed back down toward the input again and again, until the system settles on an identification.
  • Practice makes perfect, and over time these systems get better and better at the job, the same way we do - only better.
• Hierarchical, Bayesian predictive coding is a method for generating affordances galore - we expect solid objects to have backs that will come into view as we walk around them, we expect doors to open stairs to afford climbing, and cups to hold liquid.
• The network doesn't sit passively, waiting to be informed, but constantly makes probabilistic guesses about what will come next and treating feedback about its errors as the chief source of new information to guide its next round of guesses.
• In visual pathways, for example, there are more downward than upward pathways, more outbound that incoming signals. The brain's strategy is continuously to create "forward models" or probabilistic anticipations, and use the incoming signals to prune them for accuracy.
• When the organism is in deeply familiar territory, the inbound corrections diminish to a trickle and the brain's guesses, unchallenged, give it a head start on what to do next.
• This is descended from "analysis by synthesis"
• In a Bayesian network, silence counts as confirmation. Whatever the higher levels guess counts as reality by default in the absence of disconfirmation.
• These are expectation-generating fabrics with a remarkable competence they don't need to understand. They don't need to express or represent the reasons they track; like evolution itself, they "blindly" separate the information wheat from the chaff and act on it. Reasons are not things in their ontologies, not salient items in their manifest images.
• And these systems come, via cultural evolution, a whole new process of R&D, less than 1m years old - that designs, disseminates, and installs thinking tools by the thousands in our brains (and only ours), turning them into minds - not "minds" or sorta minds, but proper minds.
• One kind of neuron, the von Economo, or spindle cell, is found only in animals with very large brains and complex social lives: humans and other great apes, elephants and cetaceans.
• Brains are computers:
  • composed of billions of idiosyncratic neurons that evolved to fend for them selves,
  • their functional architecture is more like a free market that a "politburo" hierarchy where all tasks are assigned from on high
  • They are composed of Bayesian networdks that are highly competent expectation-generators that don't have to comprehend what they are doing.
  • Our kind of comprehension is only made possible by the arrival quite recently of a new kind of evolutionary replicator - culturally transmitted information entities: memes.

9. The Role of Words in Cultural Evolution

• The evolution of the evolution of culture is from:
  • Profoundly Darwinian processes - involving minimal comprehension, bottom-up generation of novel products by random search processes, to
  • Processes of intelligent design - comprehending, top-down generation of novel products by directed search.
• Words are the best example of memes
• Other species have some rudiments of cultural evolution that are not transmitted genetically between the generation but rather are ways of behaving that depend on the offspring's perception of the elders' behavior - an "instinct to learn" things like nest building or singing.
• Many animal behaviors that were thought to be genetically transmitted "instincts" have proven to be "traditions" transmitted between parent and offspring.
• We are the only species so far to have richly cumulative culture and this is primarily due to language. We have had a sustained population growth unprecedented by any other vertebrate.
• Our genes haven't changed very much in the last 50k years, and the changes with have seen are driven by new selection pressures created by human cultural innovations, such as cooking, agriculture, transportation, religion, and science.
• The widespread adoption of a new way of behaving creates a one-way ratchet: once almost everybody is eating cooked food, the human digestion system evolves, genetically, to make it no longer practical - and then no longer possible - for humans to live on a raw diet.
• One of the facts of life, both genetic and cultural, is that options become obligatory. A clever new trick that gives is users a distinct advantage over their peers soon "spreads to fixation", at which point those who don't acquire it are doomed.
• Eccentricities of a few members of the population bezcome a species necessity, embodied in an instinct.
• There is no law obliging people to have a credit card or a cellphone, but it is increasingly inconvenient to not have them.
• Words are the lifeblood, the backbone, the DNA of cultural evolution.
• Languages evolve like species, and/but there is widespread anastomosis, whereby what had been distinct lineages join together, as words jump between languages.
• In bacteria and other unicellular organisms, genes are often traded or shared by a variety of processes independent of vertical gene descent (reproduction). Similarly, etymologies (descent lineages) for words are more secure than the descent of the languages in which they are found, because of horizontal word transfer between languages.
• Charles Sanders Peirce's type/token distinction:
  • "Word" is a word, and there are three tokens of that word in this sentence.
  • Tokens can also be silent events in your brain. These brain-tokens will not look like "word" or sound like "word (they're brain events and it's dark and quiet in there), but they will no doubt by physically similar to some of the events that normally occur in your brain when you see of hear "word".
  • There are lots of intermediate cases of words - definite specific words - tokened in our minds without going into the distinction between spoken or written, heard or seen. And there also seems to be "wordless" thinking where we don't even go to the trouble of "finding" all the words, but just tokening their bare meanings.
  • Internal tokens seem to resemble external tokens, but this is because they make use of the very same neural circuitry we use to detect the resemblances and differences between external tokens, not because this neural circuitry renders copies of what it identifies.
  • Any process that makes a new token of a type from an existing token of a type counts as a replication, whether or not the tokens are physically identical or even very similar. Tokens of words are all physical things of one sort or another, but words are, one might say, made of information, like software, and are individuated by types, not tokens, in most instances.
• Words are structures in memory that are autonomous in the sense that they must be independently acquired (learned). They are items of information, and other informational structures include stories, poems, songs, slogans, catchphrases, myths, techniques, "best practices", schools of thought, creeds, superstitions, operating systems, web browsers, Java applets.
• Informational structures comes in various sizes from large novels to shorter poems and traffic signs.
• Words have, in addition to the visible or audible parts of their tokens, a host of informational parts (making them nouns and verbs, comparatives and plurals, etc).
• Words are autonomous in some regards; they can migrate from language to language and occur in many different roles, public and private.
• A word, like a virus, is a minimal kind of agent: it wants to get itself said. Every token it generates is one of its offspring. The set of tokens descended from an ancestor token form a type, which is thus like a species.
• Some of a word's offspring will be private utterances: its human host is talking to herself, maybe even obsessively rehearsing the word in her mind, over and over, a population explosion of tokens, building an ever more robust niche for itself in a brain. And it may well be that many more internal tokenings - offspring - are born outside our conscious attention altogether. At this very moment, words may be replicating competitively in your head as inconspicuously as microbes replicate in your gut.
• The problem with introspection is that it acquiesces in the illusion that there is an inner eye that sees and an inner ear that hears - and an inner mind that just thinks.
• How do words get themselves installed in infant brains? Children learn about seven words a day, on average, from birth to age six, by which time they have a vocabulary of about 15k words

10. The Meme's-Eye Point of View

• What are memes made of? They are a kind of way of behaving (roughly) that can be copied, transmitted, remembered, taught, shunned, denounced, brandished, ridiculed, parodied, censored, hallowed.
• Memes are ways: ways of doing something, or making something, but not instincts (which are a different kind of ways of doing something or making something). The difference is that memes are transmitted perceptually, not genetically. They are semantic information, design worth stealing or copying, except when they are misinformation, which, like counterfeit money, is something that is transmitted or saved under the mistaken presumption that it is valuable, useful/
• Words are the best examples of memes.
• Repetition is a key ingredient in creating new affordances. Multiple copies of anything tend to enable your pattern-recognition machinery to make yet another copy, in the recognizer, and thus a meme can get spread.
• Words are high on reproduction versus growth, high on culture versus genetic, and low on complexity.
• Once words are secured as the dominant medium of cultural innovation and transmission, they do begin to transform the evolutionary process itself, giving rise to new varieties of R&D much closer to the traditional, mythical ideal of intelligent design
• Ideas, practices, methods, beliefs, traditions, rituals, terms. These are all informational things that spread among human beings.
• A meme is any culturally-based way.
• Three conceptions of memes:
  • Competence without comprehension - Human comprehension - and approval - is neither necessary not sufficient for the fixation of a meme in a culture.
  • The fitness of memes. Memes thus have their own reproductive fitness, just like viruses.
  • Memes are informational things. They are “prescriptions” for ways of doing things that can be transmitted, stored, and mutated without being executed or expressed.
• Natural selection of memes can do the design work without any obligatory boost from human, divine, or group comprehension.
• Even the meanings of words can evolve by processes quite outside the ken of those using the words, thanks to differential replication. The fact that changes in cultural features can spread without notice is hare to account for. Memes provide an alternative vision of how culture-borne information gets installed in brains without being understood. The default presumption of folk psychology is that people, and even “higher” animals, will understand whatever is put before them.
• “ One could then say, with complete rigor, that it is the sea herself who fashions the boats, choosing those which function, and destroying the others.”
• No comprehension is required even if it probably accelerates R1D processes more often than it retards them.
• The meme perspective covers the whole spectrum of mutualist, commensal, and parasitical symbionts.
• Only a tiny minority of the trillions of viruses that inhabit each of us right now are toxic in any way. Do we need some viruses in order to thrive? We certainly need lots of our memes.
• Many memes, maybe most memes, are mutualists, fitness-enhancing prosthetic enhancements of our existing adaptations (such as our perceptual systems, our memories, our locomotive and manipulative abilities).
• The prospective of parasitical memes exploiting that infrastructure is more or less guaranteed.
• What does fitness means in the context of evolutionary biology? Not health or happiness or intelligence or comfort or security, but procreative prowess.
• We are the only species that has managed to occupy a perspective that displaces genetic fitness as the highest purpose, the summum bonum of life. We are the only species that has discovered other things to die for (and to kill for): freedom, democracy, truth, communism, Roman Catholicism, Islam, and many other meme complexes (memes made of memes
• We are the persuadable species, not just learners, not just trainable, but also capable of being moved by reasons represented to us, not free-floating. We often have reasons for what we do, in this sense: we have articulated them to ourselves and have endorsed them after due consideration. The individual human being’s capacity to reason, to express and evaluate logical arguments, arises out of the social practice of persuasion. Our skills were honed for taking sides, persuading others in debate, not necessarily getting things right.