Scientism: Part 4: Reductionism

This is the Fourth Part of a review of Science Unlimited? The Challenges of Scientism, edited by Maarten Boudry and Massimo Pigliucci. See also Part 1: Pseudoscience, Part 2: The Humanities, and Part 3: Philosophy.

Reductionism is a big, bad, bogey word, usually uttered by those accusing others of holding naive and simplistic notions. The dominant opinion among philosophers is that reductionism does not work, whereas scientists use reductionist methods all the time and see nothing wrong with doing so.

That paradox is resolved by realising that “reductionism” means very different things to different people. To scientists it is an ontological thesis. It says that if one exactly replicates all the low-level ontology of a complex system, then all of the high-level behaviour would be entailed. Thus there cannot be a difference in high-level behaviour without there being a low-level difference (if someone is thinking “I fancy coffee” instead of “I fancy tea”, then there must be a difference in patterns of electrical signals swirling around their neurons).

To philosophers, however, “reductionism” is about explanations and theories. It asserts something along the lines that high-level explanations can always be translated into low-level explanations, and that the low-level explanations are more important or more proper, and that ideally the high-level explanations could be dispensed with. I say “something along the lines” because this sort of eliminative reductionism is pretty much a strawman in that no-one (sensible) advocates it. And philosophers are right, in general it does not work.

But the scientific notion of ontological reductionism does work. At least, all of science assumes that thesis, and science works very well, producing unarguable and unmatched mastery of technology and engineering. Since adopting that thesis works so well we can be pretty sure that ontological reductionism (which philosophers may instead refer to as “supervenience physicalism”) is a true feature of the real world.

But the different meanings lead to miscommunication. “Scientism” is supposed to include a naive faith in reductionism, which the accuser would take to be the philosophers’ inter-theoretic reductionism. But those defending scientism are likely to think like scientists, and so hold only to ontological reductionism but not defend ideas of inter-theoretic reductionism. The latter might work in limited instances, but does not work in general.

Most physicists would agree. In his Reductionism Redux essay, Stephen Weinberg refers to inter-theoretic reductionism as “petty” reductionism, saying that it usually doesn’t work, while he regards ontological reductionism as a “grand” reductionism that underpins all of science. Similarly Sean Carroll defines reductionism as the idea that “objects are completely defined by the states of their components”, and says: “I could imagine hypothetical worlds in which reductionism failed … It’s just not our world”.

Carroll also says that one can “object to the claim that ‘the best way to understand complex systems is to analyze their component parts, ignoring higher-level structures’, but only if you can find someone who actually makes that claim”, and adds that: “nobody thinks that the right approach is to break a giraffe down to quarks and leptons and start cranking out the Feynman diagrams”.

So let’s see what the philosophers in Boudry and Pigliucci’s book make of the concept. Filip Buekens accepts Alex Rosenberg’s claim that “physics fixes all the facts” (by which Rosenberg means the supervenience thesis that the state of a complex system is completely specified if all its low-level physical properties are specified), but he demurs about the “much stronger claim” that “all other facts are ultimately explained by physics”.

He continues: “conceptual anti-reductionism holds that explanations employing psychological concepts cannot be replaced by explanatory strategies relying on physical concepts”. So one could not translate the concept “fear” into language about electrons and protons and their motions.

He’s right on the latter point, but it’s important to realise that explanations are not mutually exclusive. Explanations are always commentaries about some aspect of a system. They never describe the entirety of a system. And that means that multiple different explanations can be true at the same time.

The doctrine of supervenience says that one could — given an advanced Star Trek transporter device — exactly replicate a system from an exhaustive listing of every particle it contains (and the replicated system would manifest the same high-level properties including “fear”). But an “explanation”, being a commentary about aspects of a system, never contains enough information to do this. You could not feed “Tom was afraid of the dog” into the transporter and exactly replicate Tom and the dog from that alone.

The same holds for explanations used in physics. They also are reduced-information commentaries; physicists no more work with exhaustive listings of particles than psychologists do — they are too unwieldly and so impractical as to be useless. Thus, even in physics there are multiple higher-level concepts (such as “temperature”, “entropy”, “elasticity”, “ductility”, “conductivity”) that are properties of an ensemble, and which are not even defined at the lower level of single particles.

Since, for any system, there will always be many mutually-consistent and equally-true explanations, it follows that even if one develops explanations of high-level properties in terms of lower-level properties, these will never replace and do-away with high-level explanations, they will only add to and complement them.

Richard Feynman said that any good theoretical physicist knows six different ways of thinking about the same thing. Explanations at different levels of description are complementary ways of thinking about the same thing. They don’t replace or abolish other explanations, instead they must all be simultaneously true. And the different explanations are held together, coherently, not by the philosophers’ notion of inter-theoretical reductionism — not by translations between different explanations — but by the doctrine of ontological reductionism or supervenience.

As an aside here, philosophers use the weird term “special science” for sciences where inter-theoretic reductionism is held not to work, and by doing that they imply that it does work for at least some sciences, by which they usually mean physics or perhaps fundamental physics; they are wrong, it does not work even there, there are no “special” sciences since they are all “special”.

I can’t help thinking that much philosophical travail against inter-theoretic reductionism is misplaced, in the sense of attacking doctrines that no-one holds. Stephen Pinker’s essay on the humanities aroused fears of a hostile take over that reduces the humanities to a mere adjunct of science. But, as Russell Blackford explains in his contribution to the volume, that is a mis-reading of Pinker, who is instead arguing for a consilience in which different styles of approach complement each other.

Similarly, Taner Edis expounds a scientism that “highlights continuities in the various ways we produce knowledge, and weaves the products of our knowledge-seeking enterprises into a naturalistic overall picture”, saying that: “this scientism is harmless: it seeks connections and coherence, not intellectual conquest”.

The most direct condemnation of reductionism in the book is by Mariam Thalos. Declaring reductionism “the enemy”, she argues against sociobiology and the claim that, because human brains have evolved “therefore biology explains human behaviour too, utilizing principles of natural selection”. But that’s true, it does!

Thalos, however, suggests that accepting this idea “would with one stroke sweep away all competing models of human behaviour”. Interpreting that narrowly, yes it would sweep away competing models — those that are incompatible with the sociobiology perspective. But it would not sweep away complementary models and explanations — those that are different from but compatible with the evolutionary perspective, and which are equally true.

Thalos generalises her argument: physics explains the behaviour of physical bodies, humans are made of physical stuff, therefore physics explains human behaviour. “Whence, biology, as such, is made irrelevant.”

To this she adds that if we accept that “physics explains human behaviour, utilizing physical principles”, then “we are explicitly denying the need for biological theory as independently valuable in the enterprise of scientific explanation”. Thus, to Thalos, only one type of explanation can be valid. Physical systems can only be explained in physical terms; biological systems can only be explained in biological terms.

No! This is a rejection of the “grand reductionism” that is the very soul of science. Complex systems (such as humans) need explanations at all levels of analysis. We should develop explanations of humans in physical terms, and in chemical terms, and in bio-chemical terms, and in biological terms, and in evolutionary terms, and in psychological terms, and in sociological terms, and in the languages of the humanities. All such explanations complement each other and mesh into a grand, consilient picture. The different explanations don’t compete with or displace each other, they complement each other. They must all be mutually compatible and mutually build to an overall grand picture in which they are all true.

That follows from the doctrine of scientific or ontological reductionism, which holds everything together because it tells us that all these different explanations are about the same ontological stuff; they are reporting different aspects of the same ensemble. And that is the consilient grand picture of science and of scientism.

This is not a merely philosophical point, it is eminently practical. Given multiple explanations about the same stuff, we then need to ensure that they are fully compatibile, and investigating that is the central driving force of science. Ensuring that the explanations in physical terms mesh seamlessly with the explanations in chemical terms, and with the bio-chemical explanations, the biological explanations, and the evolutionary and psychological explanations, is exactly how science makes progress.

Any field that wants to stand aside from that process risks turning itself into a parochial fiefdom prey to fads and ideologies (a current example being areas of sociology that totally ignore the genetic underpinnings of human behaviour), and deprives itself of the best tool that science has.

20 thoughts on “Scientism: Part 4: Reductionism

  1. Pingback: Science Unlimited, Part Three: Philosophy | coelsblog

  2. Brent Meeker

    A good essay. I have long advocated a view of epistemological view of science in the broad sense. Schematically is would look something like: physics->chemistry->biology->evolution->experience->culture->language->mathematics->physics->… It can’t be quite to simple of course. But the basic idea is that explanations are just following the arrows, either forward or backward, until to come to something you already understand or know how to investigate. So science can be done at every level and part of it is making the connections “->”. One difference, which may just be semantic, is that you refer to ontological reductionism; which usually refers to the idea that the reduction level defines what really, really exists. I think that’s a bad idea, and the reason I emphasize epistemological reductionism. In my view the integers exist in mathematics just as quarks exist in physics and charity exists in culture. We can explain charity in terms of biology and evolution, but that doesn’t mean charity is made of biology and evolution.

    Reply
    1. Coel Post author

      Hi Brent, maths will be part of a future installment of the series, but for now:

      In my view the integers exist in mathematics just as quarks exist in physics …

      What does “exist” mean as used there?

    2. Brent Meeker

      They are constituents on whose properties and relations we can agree. So they are part of a domain of knowledge we use to explicate the world and our experience. But they are in a different domain or level than physical objects which are, as Dr Johnson would say those that kick back when you kick them.

    3. Coel Post author

      Hi Brent,

      So they are part of a domain of knowledge we use to explicate the world and our experience.

      If one says numbers “exist” one is generally taken to mean they exist independently of humans or of our knowledge of the world. Being part of our world-model is different, no-one would deny that they are part of our world-model.

    4. Brent Meeker

      The mathematicians I know are Platonist on Monday thru Friday, and fictionalists on Saturday and Sunday. As a physicist, I’m with Sean Carroll when he quips, “All human progress has come from studying the shadows on the cave wall.”

  3. verbosestoic

    I do intend to get back to you on the morality discussion in the near future (again, things have been busy and they take me a long time to write), but I would like to comment on this.

    Having had experience with reductionism since it’s widely commented on wrt consciousness, I think that philosophers in general are a bit more sophisticated in their view of reductionism than you give them credit for. They segment out what you might consider eliminative reductionism from reductionism as a whole (I was indeed once chided for confusing the two). Eliminativist views are the ones that say that you don’t need the explanations at the higher level and should simply use the ones at the lower level. While you say that that is generally a strawman view, it actually isn’t. In consciousness, there at least were a number of people who claimed that there are no mental states and all we should really refer to are neuron firings. In my examination of Rosenberg, he also seems to be one as he says that everything outside of the equations of science is just a story we tell ourselves. So some people do argue that we should eliminate the higher level explanations in favour of the true ones at the lower levels, so it isn’t a strawman. It’s just not really reductionism per se.

    For reductionism, the base claim is that you can map the higher level explanations/phenomena onto the lower level explanations/phenomena, but it doesn’t necessarily mean that the terms or explanations at the higher level are meaningless or shouldn’t be used, or that the lower level explanations are to be preferred. But you do need to be able to map every higher level phenomena to a lower level phenomena, or else it can’t be reduced, and that has to be done in a way that the two are identical and that you can, in some sense, say that what it means to be something at the higher level is to have that sort of implementation (although that might be shaky and actually isn’t necessary for the philosophical objections).

    So it isn’t a surprise that, as you note, philosophers do think that sometimes reductionism works. The common examples are the moves from biology to chemistry to physics. You can come up with bridge laws that map all the phenomena from the higher levels to the lower ones. This doesn’t mean and has never meant that we should just use the physics terms when doing biology — that would be eliminativism — but it DOES mean that there’s no phenomena in biology that cannot be equated, through the bridge laws, to phenomena in physics.

    So, then, what about the “special sciences”? These are ones where the argument is that you cannot find bridge laws that would reduce all of its phenomena to the lower level phenomena, and generally to physics. The example would be psychology, for one simple reason: if we could find bridge laws that would reduce all psychological phenomena to biological phenomena, then it would mean that they are in a real sense identical to biological phenomena, but then no non-biological entity could ever had psychological states. Thus, no artificial intelligence could ever really have psychological states, since it won’t have those biological entities. Thus, it looks like the brain IMPLEMENTS psychological states, but that then requires us to have a view of psychological states that’s independent of the biology of the brain. So it’s entirely possible, then, to have even a biological brain that can’t or doesn’t implement every psychological state, which would break reductionism. Also, attempting to reduce psychological states to biological states seems fairly useless since what we’re really interested in is that biology-neutral explanation so that we can know if a computer can ever have psychological states.

    So there’s more to it than you say here, and more reasons to think that even science doesn’t really use the sort of reductionism that you talk about and that even that sort of reductionism runs into issues with some phenomena.

    Reply
    1. Coel Post author

      Hi verbose, after a hectic week, some replies:

      In my examination of Rosenberg, he also seems to be one as he says that everything outside of the equations of science is just a story we tell ourselves.

      I find that Rosenberg often phrases things in the most provocative way he can, but if you interpret him properly then his claims are not as outrageous as they might seem. In your sentence, a lot rides on the word “just”. One could make a case that all of our concepts, *including* the equations of science, are “stories that we tell ourselves”, stories that are helpful and necessary for operating in the world. Thus everything we say is not “reality itself”, but is our (partially-correct) story about reality.

      For reductionism, the base claim is that you can map the higher level explanations/phenomena onto the lower level explanations/phenomena, …

      If by “map” you mean something akin to bridge laws, then that is the philosophical concept of inter-theoretic reductionism. Again, this is a different thesis from how scientists think, which is about ontological reductionism. (Ontological reductionism could be true and useful, even if we humans were totally incapable of mapping some explanations from one level to explanations at another.)

      The common examples are the moves from biology to chemistry to physics. You can come up with bridge laws that map all the phenomena from the higher levels to the lower ones.

      Well actually you can’t, if by “bridge law” one means something fairly succinct. For one thing, there is a huge amount of historical-accident contingency in how biology actually is, and that cannot be captured in bridge laws.

      So, then, what about the “special sciences”? These are ones where the argument is that you cannot find bridge laws that would reduce all of its phenomena to the lower level phenomena, and generally to physics.

      I think that philosophers are perhaps underestiming the difficulties in having bridge laws.

      As an aside, philosophers tend to think that *scientists* think in terms of bridge laws, and sometimes accuse scientists of being naive in not realising the difficulties. But its the opposite! Scientists don’t think like that, they think in terms of *ontological* reductionism.

      Thus, bridge laws don’t work even *within* physics. For example, there is no bridge law linking the mass of the proton to the masses of the constituent parts of the proton.

      All you can do is take the brute-force simulation approach: simulate the low level (e.g. simulate all the component parts, with all the messy complications), and you can laboriously compute the outcome. But that approach rests on ontological reductionism, it is not about inter-theoretic bridge laws.

      The example would be psychology, for one simple reason: if we could find bridge laws that would reduce all psychological phenomena to biological phenomena, then it would mean that they are in a real sense identical to biological phenomena, but then no non-biological entity could ever had psychological states. Thus, no artificial intelligence could ever really have psychological states, since it won’t have those biological entities.

      As above, I’m not defending bridge laws, but I don’t find that argument convincing. Or at least, it only holds for such a narrow conception of what is allowable as a “bridge law” that bridge-law reductionism then obviously does not work.

    2. verbosestoic

      I find that Rosenberg often phrases things in the most provocative way he can, but if you interpret him properly then his claims are not as outrageous as they might seem.

      If I recall correctly, Rosenberg flat-out says that the equations are what are real and have real meaning and the stories aren’t and don’t. There’s not much room to “properly interpret” him there, and the comment isn’t about him being right — the view is ridiculous — but is instead about some people indeed making the strong sort of claims that you say no one makes, which is what philosophers are reacting to. Which leads to …

      If by “map” you mean something akin to bridge laws, then that is the philosophical concept of inter-theoretic reductionism. Again, this is a different thesis from how scientists think, which is about ontological reductionism.

      As stated above, I was indeed talking about the inter-theoretic reductionism the philosophy talks about a lot, showing that not only do some people advocate for it, but that it actually works on occasion. Ontological reductionism, as far as I can tell from looking around, would apply to entities and not fields, as the scientific reductionism that I came across talked about describing an entity by its component parts. This, then, wouldn’t apply to any case where someone was trying to say, for example, that you could reduce psychology to biology, since that is reducing fields as a whole and not entities. It’s certainly true that scientists often break down entities into their component parts to build up and develop explanations, but that’s not the sort of reductionism that philosophers are worried about, nor would they, in general, call it scientism in general to do so. What they are concerned about is attempts to reduce fields to lower fields, often accompanied by claims that the lower fields are more certain or useful. The arguments around consciousness are prime examples of that sort of move, and a debate that is and has been happening for a long, long time.

      So to call your scientific/ontological reductionism reductionism and to relate it directly and contrast it with the philosophical, inter-theoretical kind of reductionism is misleading, because that’s not the debate is about. Philosophers usually won’t oppose your kind of reductionism, but will note that a lot of scientists and scientifically-minded philosophers are actually using the kind that you are dismissing. So your argument would be more dismissing those sorts of scientistic arguments than showing that philosophers are interpreting reductionism wrong. You’d reject inter-theoretic reductionism, but then would also have to reject the arguments they make using it.

      Well actually you can’t, if by “bridge law” one means something fairly succinct.

      No one has ever said that it has to be succinct or even completely specified. All bridge laws do is show how you can get from one level of explanation to another, leaving nothing out from the higher level. As stated, there can be concepts that are much more clear and useful at the higher level that aren’t directly mapped to things at the lower level, but for that sort of reductionism — that I repeat many scientifically minded people argue for — you should be able to work it out if you try, even if once you do so you’ve lost what was interesting about them.

      I think that philosophers are perhaps underestiming the difficulties in having bridge laws.

      Since philosophy, at least the last time I checked, all pretty much agreed that we have then for at least the move from chemistry to physics and almost certainly from biology to chemistry, I think you’re overestimating how difficult they are. Reducing biology to chemistry and then chemistry to physics is pretty much uncontroversial, and is the standard example of how reductionism would work. Since philosophers, as you know, almost never agree on anything, to oppose that is going to require some very solid evidence and argumentation, and all you’ve done, really, is talk about another sort of reductionism instead.

      As an aside, philosophers tend to think that *scientists* think in terms of bridge laws, and sometimes accuse scientists of being naive in not realising the difficulties.

      I’ve never seen that tendency, except in cases where scientists wander into philosophical discussions, push for inter-theoretic reductionism, and insist that there have to be bridge laws that let that work, and suggest ones that don’t cover the phenomenal without leaving major portions out. See consciousness, for example, where the bridge laws tend to leave qualia out completely and assume that it’s probably something about neurons that does it, despite that entailing the odd concept that nothing without neurons can have qualia, which either makes qualia not required for consciousness or means that AIs can’t be conscious. So, yeah, it’s harder than they think.

      Then again, I don’t recall ever seeing a GENERAL argument that bridge laws were more difficult to find than people thought … other than yours in this comment. Do you have an example?

      Scientists don’t think like that, they think in terms of *ontological* reductionism.

      I’ll concede that for scientists inside their own field. When trying to cross fields, though, that’s not the reductionism that they use, pretty much by definition.

      Thus, bridge laws don’t work even *within* physics. For example, there is no bridge law linking the mass of the proton to the masses of the constituent parts of the proton.

      Bridge laws are only relevant to inter-theoretic reductions between fields. So trying to do what you’re doing here with bridge laws would be totally wrong. And scientists trying to reduce between fields do use bridge laws, and again it does generally work between the hard sciences.

      OK, so we “reduce” psychology to biology; but then we can go further and reduce biology to physics, so we’ve reduced psychology to physics. But if we can do that then we can reduce artificial life to physics. So “psychological states” are properties of “physical stuff” which of those implementations we go with, so there’s no problem. A biological implementation and an artificial implementation are just two different implementations of physical stuff, both giving psychological states.

      The problem here is that psychological entities and theories would, at that level, have to be identical, but then in dealing with a specific case we not only would be doing two different reductions with two radically different sets of bridge laws, we would HAVE to have that to make them work. But we can reduce a number of different fields to physics (not everything would have to get there through chemistry), and we would differentiate the fields, at least in part, on having different bridge laws to get to physics. So why should we consider, then, those two “psychologies” identical? But you’d have to if you wanted to claim that AI psychology and human psychology are the same field. This, then, precludes claiming to reduce psychology to biology.

      So what we’d be after is an “implements” relation, where we argue that biology and neurology reflects a specific IMPLEMENTATION of psychology. But that’s not any kind of reductionism, even ontological reductionism. We can implement bubble sort in Java and in C++, but would never claim that this allows us to claim to have REDUCED bubble sort to Java or C++, nor that bubble sort is made up of Java or C++ components. Carrying that on further, we can see that while we currently compile those languages into machine language, they also don’t REDUCE to machine language, and aren’t MADE UP OF machine language, because we can easily conceive of systems that don’t use machine language but implement Java and C++. So neither form of reductionism would seem to work there.

      And note that the explanation I just gave you for psychology and why it can’t be reduced fits in very will with the functionalist and algorithmic views of consciousness and intelligence, and so gets some credibility from that as that’s generally more successful than the alternatives.

    3. Coel Post author

      No one has ever said that it [ a bridge law] has to be succinct or even completely specified. All bridge laws do is show how you can get from one level of explanation to another, leaving nothing out from the higher level.

      I confess to not fully understanding what is involved in a “bridge law” (which is a philosophers way of thinking, and not part of how scientists tend to think about such things). It is obvious that we need novel concepts at the higher level that are not present at the lower level (and thus not part of the lower-level description). An example is a metal, where the ensemble has properties (conductivity, ductility) that are not part of the lower-level description. How does a bridge law deal with that?

      Since philosophy, at least the last time I checked, all pretty much agreed that we have then for at least the move from chemistry to physics and almost certainly from biology to chemistry, I think you’re overestimating how difficult they are. Reducing biology to chemistry and then chemistry to physics is pretty much uncontroversial, …

      Plenty of philosophers (e.g. Massimo Pigliucci) have stated that they don’t think that biology can be reduced to physics in the bridge-law, inter-theoretic reduction sense, and thus they regard biology (and indeed everything other than physics) as a “special science”.

      If you think that biology can be uncontroversially reduced (in that sense) to chemistry and thence to physics, can you point me to an account of how this is done, and an account of the relevant bridge laws?

      Then again, I don’t recall ever seeing a GENERAL argument that bridge laws were more difficult to find than people thought … other than yours in this comment. Do you have an example?

      I’d need a clearer account of what does and doesn’t count as a “bridge law” to produce a general counter-argument.

      And scientists trying to reduce between fields do use bridge laws, and again it does generally work between the hard sciences.

      To make this discussion concrete, can you point me at accounts of this (since, again, this way of thinking seems alien to me as a scientist, unless I’m misunderstanding what a bridge law amounts to).

    4. verbosestoic

      I confess to not fully understanding what is involved in a “bridge law” (which is a philosophers way of thinking, and not part of how scientists tend to think about such things).

      So let me start by pointing out that this isn’t actually true. The sort of armchair philosopher that you tend to disagree with the most — like me — aren’t very interested in reductionism of any sort, and so aren’t all that interested in bridge laws. For my part, I only mentioned the ones that I did because from everything I’ve studied about reductionism those were the ones that were accepted without much debate. But it tends to be scientists and scientifically-minded philosophers who try to do reduction in philosophy. The reason that reductionism is often associated with scientism is that it’s a common way for those sorts of people to try to turn problems that seem philosophical into ones that are more scientific and so more amenable to the scientific methods and scientific fields. In the morality discussion, you yourself perform a sort of reduction where you reduce moral, aesthetic and emotional concepts to neurology and then say that what’s interesting is at that level and not at the higher levels of ethics, philosophy and psychology. That’s a strongly eliminativist reduction, although not all reductions need to be eliminativist.

      Armchair-style philosophers aren’t interested in reduction those concepts, theories and fields to anything else because they have no desire to make them scientific and tend to think that it works better to deal with them as they are. It’s the scientifically-minded ones that want to reduce it to science or something more scientific to use the methods of science on them.

      It is obvious that we need novel concepts at the higher level that are not present at the lower level (and thus not part of the lower-level description). An example is a metal, where the ensemble has properties (conductivity, ductility) that are not part of the lower-level description. How does a bridge law deal with that?

      I’ll try to answer this and give a notion of bridge laws and how they work at the same time. Again, I’m a more traditional. armchair philosopher and so not really a reductionist, so you can try to read more about it at this link: https://plato.stanford.edu/entries/reduction-biology/#TheoRedu . That one seems a bit opaque, though, relying on equations and the like, which tend to make my eyes glaze over.

      Anyway, the name bridge laws is more impressive than they actually are. Essentially, bridge laws are just the mappings from the concepts/phenomena in one field to another. So for chemistry, what we would note is that all chemical reactions react the way they do because of the properties of the atoms that we are joining together, and that this explains which molecules will form and which chemical solutions we make and which we can’t. So we could reduce chemistry to physics doing that. However, there is no concept of a molecule or a solution in physics. To replace those concepts with the physical description would be long-winded and inefficient, so we’d like to keep those concepts for work in chemistry. But those concepts have no meaning or use in physics, and so we have no reason or wish to add them there. Thus, even though we can reduce chemistry to physics, we don’t want to eliminate the field or concepts of chemistry because they are useful for the work that field is doing and aren’t useful for physics. Thus, we perform a reduction and yet maintain the field and its terms.

      So, onto biology. As best as I can understand it, the main idea would be that all biological processes are chemical processes — think digestion — and so can be reduced to chemistry, which can be reduced to physics as per the above. I don’t know what the objections to biology being able to be reduced are, but if I had to guess I’d make one of two guesses:

      1) Not all biological processes are at least neatly chemical processes (think of using your muscles, for example). So those can’t be reduced to chemistry, or at least not easily. It’s possible they could be reduced to PHYSICS, though, directly. But then we’d have a problem similar to that I mentioned for psychology to neuroscience: I can’t reduce all of biology to chemistry, but certainly would have to reduce SOME of it that way and then to physics, while the rest would have to go directly to physics. But that means that I can’t reduce all of biology to chemistry, but since I’m reducing biology in two different ways I also can’t say that I can reduce it directly to physics either just because at the end it ends up there. So biology cannot be neatly reduced to any other field.

      2) Animal behaviour, if part of biology, doesn’t reduce to chemistry and may not reduce to physics. But then that’s more similar to psychology than most other forms of biology.

      Beyond that, I don’t know why someone might claim biology as a special science. If you have an example that might help.

    5. Coel Post author

      Hi verbose,

      However, there is no concept of a molecule or a solution in physics. To replace those concepts with the physical description would be long-winded and inefficient, so we’d like to keep those concepts for work in chemistry. But those concepts have no meaning or use in physics, and so we have no reason or wish to add them there. Thus, even though we can reduce chemistry to physics, we don’t want to eliminate the field or concepts of chemistry because they are useful for the work that field is doing and aren’t useful for physics.

      That is not a description of bridge-law, inter-theoretic reduction, as I understand the terms (though of course I could be wrong). My understanding is that bridge-law reductionism is a much stronger concept than that.

      The concept of a “bridge law” was introduced by Ernest Nagel, and wiki says: “He was the first to propose that by positing analytic equivalencies (or “bridge laws”) between the terms of different sciences, one could eliminate all ontological commitments except those required by the most basic science”.

      The phrase “analytic equivalencies” is strong. It means you can map concepts in one science to equivalent concepts in the other science; and you can do the mapping “analytically”. This is explicitly at odds with your above account (e.g. “But those concepts have no meaning or use in physics”). Under bridge-law inter-theoretic reduction, the concepts *would* have a meaning and use in the lower-level science. They would simply be differently phrased. But you could translate from one phrasing to the other by analytical mappings.

      That is a rather strong thesis. It also does not work. It is not how *scientists* think about these things (it can’t be, since it doesn’t work, and science is pragmatic).

      To a scientist, reductionism is ontological. That means that if you perfectly simulated the low-level behaviour of a system (e.g. the physics), then the high-level behaviour (e.g. the chemistry) would be manifest. In other words, if you looked at the output of a perfect simulation, then you could see that it was behaving in accord with chemistry textbooks.

      But there is no claim to analytical mappings between the physics and the chemistry, no claim that all concepts in chemistry have an equivalent concept in physics, that can be arrived at on the other end of an analytical mapping. The suck-it-and-see, brute-force simulation approach may be the most-concise “mapping” between the two levels that is possible.

      Now, again as I understand it, most philosophers say that bridge-law inter-theoretic reductionism does not work in general. (Thus, it does not work between chemistry and physics, or between biology and chemistry.) They are entirely right in that.

      But, philosophers often make three assumptions that are also wrong. One is that they think that, by “reductionsm”, scientists generally mean bridge-law reductionism (they don’t, they only mean ontological reductionism). The second is that they think scientists *do* think that bridge-law reductionism works (because they hear scientists talk about “reductionism”, and interpret that as bridge-law reductionism), and then they conclude that scientists are native, since they know that it doesn’t work.

      The third is that they think that bridge-law reductionism *does* work *within* physics (namely, that a concept in one level of physics will always have an analytical mapping to concepts in another level of physics); that’s wrong, it doesn’t; bridge-law reductionism does not work at all.

    6. verbosestoic

      The phrase “analytic equivalencies” is strong. It means you can map concepts in one science to equivalent concepts in the other science; and you can do the mapping “analytically”. This is explicitly at odds with your above account (e.g. “But those concepts have no meaning or use in physics”). Under bridge-law inter-theoretic reduction, the concepts *would* have a meaning and use in the lower-level science. They would simply be differently phrased. But you could translate from one phrasing to the other by analytical mappings.

      Actually, no, not in the sense you mean. There IS a form of reductionism that wants to eliminate the terms at the higher level and say that all you need to use are the concepts at the lower level, but that’s a very strong form of reductionism … and I was chided in a philosophy class — in Cognitive Science — for assuming that reductionism must be eliminative.

      What we want for an analytic mapping is a MEANING sort of mapping. So, as an example, “Water is H2O” is an analytic mapping, because once we understand what water is, and what H2O is, then we know that water maps to H2O. But that doesn’t mean that we would say that no one should ever use the word “water” anymore, anywhere. Water makes far more sense as a term in everyday language than H2O does, even though they mean the same thing.

      Okay, so let’s return to my reduction of chemistry to physics. A solution, once we understand the mechanism, would just mean that the atoms in that area are in the state that means that they dissolved together. But saying it that way is long-winded and inefficient, and not particularly interesting in and of itself. We’d still like to talk about a solution without having to reference the specific configuration of atoms. But we don’t want to introduce the term “solution” into physics, because that’s not of any interest to what physics normally does. They don’t really CARE about the details of a solution, just about the atoms. So while we know that a solution in chemistry just IS a particular configuration of atoms, there’s value in talking about that configuration as a solution and using the chemical terms rather than having to eliminate them and talking only about the terms from physics. So we reduce chemistry to physics with those sorts of mappings, but don’t ELIMINATE chemistry in doing so.

      Another example would be the difference between solids, liquids and gasses. We know that all that means at the level of physics is the distance between the atoms/molecules, but at the level of physics that doesn’t have any special significance, but fields that care about them will still want to differentiate them. Again, there’s an analytic reduction there as once we understand what it means to be a gas we know that that’s how the molecules and atoms are going to be spaced, and vice versa, but one is not eliminated by the other.

      I think you, like me, are getting confused by eliminativist reductions. We can reduce one field to another without eliminating it.

      To a scientist, reductionism is ontological. That means that if you perfectly simulated the low-level behaviour of a system (e.g. the physics), then the high-level behaviour (e.g. the chemistry) would be manifest. In other words, if you looked at the output of a perfect simulation, then you could see that it was behaving in accord with chemistry textbooks.

      If this is the limit of what science can do, then things would be problematic. Surely one could EXPLAIN why it was that putting, say, atoms in that state produced the chemicals at the level of chemistry, and that requires more than being able to simulate it and see that it works out. It would have to be the case that it is in a real sense NECESSARILY the case that if you put the atoms in that state you will see those results at the chemical level. If it was possible to put the atoms in that state and NOT see the chemical result, then your entire premise is shot down. So this means one of two things: either the phenomena is emergent and we know that it will always happen but can’t ever say why, or else we can say that because we know WHY it is the case that that particular configuration of atoms produces that chemical result. And as soon as we have that explanation, we will almost certainly have a mapping for at least THAT phenomena to show analytically why the chemical phenomena REALLY IS that phenomena in physics. And that would be a bridge law.

      And you don’t want to retreat to emergence since that IS something that many philosophers and others doubt strongly.

      Now, again as I understand it, most philosophers say that bridge-law inter-theoretic reductionism does not work in general. (Thus, it does not work between chemistry and physics, or between biology and chemistry.) They are entirely right in that.

      Unless things have changed dramatically in the past, say, ten years, speaking as someone who has taken the concept in multiple philosophy courses from different professors this is entirely false. Most of the philosophers and literature I’ve read says that it does work, at least for chemistry and physics.

      One is that they think that, by “reductionsm”, scientists generally mean bridge-law reductionism (they don’t, they only mean ontological reductionism).

      When scientists wander into philosophical debates, that is indeed generally what they mean, especially since many of them are eliminativists which can never be done with the simpler notion of reduction.

      The second is that they think scientists *do* think that bridge-law reductionism works (because they hear scientists talk about “reductionism”, and interpret that as bridge-law reductionism), and then they conclude that scientists are native, since they know that it doesn’t work.

      Since they are trying to do that sort of reductionism and never argue with philosophers when they say that you can do that for the hard sciences, that’s not an unfair categorization. And, in general, the criticism from philosophers — see the debates over consciousness for prime examples — is not that it doesn’t or can’t work, but that the scientists tend to assert that it MUST work, even when it’s clear that it makes no sense.

      The third is that they think that bridge-law reductionism *does* work *within* physics (namely, that a concept in one level of physics will always have an analytical mapping to concepts in another level of physics); that’s wrong, it doesn’t; bridge-law reductionism does not work at all.

      I’ve never seen philosophers talk about doing that inside physics. That would be a job for physics. They DO talk about it between chemistry and physics, but also aren’t that interested in it and it seems to work.

      And let me reiterate: I am a typical armchair philosopher, so not particularly scientifically oriented. I NEVER attempt reductions. I only ever engage with them when scientismists and scientifically-minded philosophers try to reduce a field to one of the hard sciences, especially in an eliminativist way. So your assigning this to philosophers is a rather bizarre contention. Philosophers don’t care because they see no inherent benefit to linking philosophical concepts to scientific ones. Scientismists, on the other hand …

      Not quite. I said that the interesting thing from the point-of-view of meta-ethics is that that reduction *can* be done. There’s lots of higher-level stuff about applied ethics that is also of interest.

      No, you were using that to justify saying that moral statements JUST WERE aesthetic statements, so that you would never have to talk about or justify moral statements as moral statements. But that they would be the same at the level of neurology, as I pointed out there, doesn’t mean that the distinctions at the higher level aren’t valid or meaningful. In the parlance of reductionism, you were trying to reduce meta-ethics to neuroscience in an eliminativist way, removing all moral terms and asking us to AT LEAST talk only in terms of aesthetic terms, which you could only justify at the level of neuroscience. You didn’t ask us to only talk about neuroscience terms, but DID say that that was where the interesting results and concepts were.

    7. Coel Post author

      Just an addendum:

      In the morality discussion, you yourself perform a sort of reduction where you reduce moral, aesthetic and emotional concepts to neurology …

      Yes …

      … and then say that what’s interesting is at that level and not at the higher levels of ethics, philosophy and psychology.

      Not quite. I said that the interesting thing from the point-of-view of meta-ethics is that that reduction *can* be done. There’s lots of higher-level stuff about applied ethics that is also of interest.

    8. Coel Post author

      Just to amplify on why I don’t find this argument convincing:

      if we could find bridge laws that would reduce all psychological phenomena to biological phenomena, then it would mean that they are in a real sense identical to biological phenomena, but then no non-biological entity could ever had psychological states. Thus, no artificial intelligence could ever really have psychological states, …

      It’s very much a “philosophers” argument in the sense of being an “argument by attaching labels” (“biological stuff” etc). I can counter it as follows. OK, so we “reduce” psychology to biology; but then we can go further and reduce biology to physics, so we’ve reduced psychology to physics. But if we can do that then we can reduce artificial life to physics. So “psychological states” are properties of “physical stuff” which of those implementations we go with, so there’s no problem. A biological implementation and an artificial implementation are just two different implementations of physical stuff, both giving psychological states.

  4. Robin Herbert

    I think your approach to this as some sort of a duel is getting in the way.

    When Nagel proposed his definition of ‘reductionism’ he was doing so in the interests of clarification. He was saying, in effect, ‘in order to answer questions about reductionism it is useful to be precise about what we mean by the term. Here is one thing we might mean by it …’

    Nagel actually begins the chapter by stipulating that scientists are not claiming that reductionism in this sense is true and that most are saying that reductionism in this sense is not true.

    So how is that a straw man? He has provided a definition and made it clear that he is not attributing this definition to scientists.

    And then Fodor comes along and says “Reductionism, in the Nagel sense is not the case”. Again, what is not to agree with there?

    And he was not accusing scientists of holding that view, he was accusing the Logical Positivists of holding that view (he was wrong in any case, the LP’s didn’t hold that view)

    So, again, where is the straw man?

    I would have thought that this was useful, so that when we ask if reductionism is true we can be clear and say ‘reductionism in sense A is not the case, but in sense B it is the case’

    Reply
    1. Coel Post author

      Hi Robin,
      You may have a fair point on what Nagel originally said. However, whatever Nagel originally said, a lot of philosophers since then and nowadays have taken Nagel’s proposal as synonymous with “scientific reductionism” or even with “reductionism” overall, and have assumed that science attempts to attain that sort of reductionism, and/or that scientists typically think that such reductionism holds.

      For example, the wiki page on “special sciences” (https://en.wikipedia.org/wiki/Special_sciences ) says that Fodor has “argued that the special sciences … are not reducible to physics”, without any mention that he is only dealing with only one possible form of reductionsm. OK, it’s a wiki page, so not authoritative, but such accounts are common. For that reason, philosophers and physicists often totally misunderstand each other, making presumptions as to what they other means by “reductionism”.

      Overall, the prominence in philosophy of Nagel’s proposal and Fodor’s response to it is totally unwarranted given that it always was unrelated to how science actually proceeds. Why, for example, have a term “special sciences” that relates only to one form of reductionism, rather than being about science or reductionism more generally?

    2. Robin Herbert

      There is a lot in what you say. I have just read a philosopher’s blog where he says that parking regulations cannot be reduced to physics without talking about what sort of reduction he is talking about, and uses this as a kind of general argument against Materialism.

  5. Robin Herbert

    Also, you enlist Weinberg on the side of ontological reductionism, but here are his own words:

    ‘One can illustrate the reductionist worldview by imagining all the principles of science as being dots on a huge chart, with arrows flowing into each principle from all the other principles by which it is explained. The lesson of history is that these arrows do not form separate disconnected clumps, representing sciences that are logically independent, and they do not wander aimlessly. Rather, they are connected, and if followed backward they all seem to branch outward from a common source, an ultimate law of nature that Dyson calls “a finite set of fundamental equations.'”

    First of all, that does not sound like ontological reductionism.

    Secondly, if there is a difference between that statement and Nagel’s definition then it is lost on me. Nagel ‘s point is if there was a logical dependence between them then, by definition, this could be stated in a set of statements (even if only in principle).

    So I would say that the statement by Weinberg above is wrong

    Reply
    1. Coel Post author

      On that Weinberg essay, other parts are also important in interpreting him correctly, including his distinction between “grand” reductionism and “petty” reductionism. But that quote is indeed ontological reductionism. It follows from the statement: exactly replicate the low-level aspects of a system, and the high-level aspects would be manifest.

      It follows from that that “these arrows do not form separate disconnected clumps, representing sciences that are logically independent”. It follows from the above statement of ontological reductionism that one can always say: “because {low-level} therefore {high-level}”. That will always be one valid explanation (there can also be other forms of explanation of course). Further, the “common source” comment is again just a statement that, given the complete low-level description, all else is entailed.

      Nagel ‘s point is if there was a logical dependence between them then, by definition, this could be stated in a set of statements (even if only in principle).

      Again, let’s take the fundamental statement of ontological reductionism: completely replicate all low-level aspects of a system and the high-level aspects are necessarily entailed and manifest.

      That statement ties everything together. That means that the different sciences are not “logically independent”. However, it does not require any “bridge laws” between the levels of description. It may be that there is nothing beyond the suck-it-and-see approach: “completely and exactly replicate the low-level and yes, I can indeed see that the system is now manifesting the high-level behaviour”. Thus the only linkage statement that is required for this thesis to hold is: given {complete low-level description} then {high-level description}.

      That last is a vastly more general linkage than Nagel’s proposal, and is not vulernable to Fodor’s critique.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s