We have folks on both sides of this question, so it should make for an interesting discussion.
(I’m a ‘yes’, by the way.)
257 thoughts on “Is biology reducible to physics?”
Leave a Reply
You must be logged in to post a comment.
The Skeptical Zone
"I beseech you, in the bowels of Christ, think it possible that you may be mistaken."
Bruce,
Yes, essentially. And since we’re talking about reducibility in principle, not in practice, we can go whole hog and specify that the wavefunction encompasses the entire universe. Then any truth about biology should be expressible (in principle) as a statement about that single wavefunction and its evolution, if biology is in fact reducible to physics.
An easier-to-grasp analogy would be Conway’s Game of Life.
At the “biological” level, there are all kinds of interesting phenomena and laws. “Glider guns” shooting out “gliders” at a certain angle, “puffer-type breeders” giving birth to glider guns, etc. (Check out this cool animation of the “puffer-type breeder”.).
At the “physics” level, there are only four laws. From Wikipedia:
Yet these simple laws alone give rise to the puffer-type breeders, glider guns, and gliders of the “biological” level. Thus “biology” is reducible to “physics” in the Game of Life, and any emergent phenomena are only weakly emergent.
My question for KN is this: There are self-maintaining, autopoietic systems in the Game of Life (see my comment to Bruce above), yet the autopoiesis clearly reduces to the simple “physics” of that world.
If so, what justifies the claim that autopoiesis in the real world is irreducible to physics?
I would have thought it more correct to say:
“There is a model of self-maintaining, autopoietic systems in the Game of Life”.
It is non-obvious whether gliders reduce to physics. They are an artifact of human perception.
keiths:
Alan:
That doesn’t make sense to me. A “glider” in the Game of Life isn’t a model of something else. It’s an entity that maintains itself by following the laws of “physics” in its world. Likewise, we are entities that maintain ourselves by following the laws of physics in our world.
keiths:
Neil:
Yes, they are an artifact of human perception, and you’ll be pleased to be reminded that the motion is an example of indirect perception, not direct perception, since it is inferred, not actual. 🙂
Whether you think of it as an entity or a percept, it’s still an example of reduction. Take your pick:
1A. Gliders reduce to “physics” in the Game of Life.
1B. The phenomenon giving rise to the human perception of “glider” reduces to “physics” in the Game of Life.
Likewise:
2A. Cats reduce to physics in our universe.
2B. The phenomenon giving rise to the human perception of “cat” reduces to physics in our universe.
Keith: I see the challenge for reduction is that it goes the other way: you start with the gliders and their laws of interaction, then you have to re-express those concepts and their laws of interaction using only the four base laws.
How do you use only the words of those four laws only to express concepts like gliders and the laws of their interaction (eg that glider guns shot out three gliders). For example, I don’t see how one can say “above” or “below” without adding to the laws you stated. But I think you need concepts like that to specify shapes for the entities in the game of life level of the laws.
Of course, you can add those concepts to the laws you have already stated. But I don’t see that as a legal move in the reduction game as I understand you have defined it. You can only use terms needed for the laws, not add extras to help reduction by looking at the concepts of a higher level.
ETA: I see the reduction as you describe as closer to the idea that physics constrains biology.
Keith: I turned on the dials to eleven in that thought experiment so I was a little surprised you bit the bullet and said “yes” without qualification. Good for you: I admire someone with the courage to follow his intuitions to the logical limit.
My intuitions is different: I think the answer to my thought experiment is no. It’s for the same reason as my above post on the game of life: I don’t think the laws of physics would be rich enough to express the concepts of biology. Further, we have the additional challenge that biology is often expressed in terms of mechanisms via words and diagrams, not mathematical equations as physics uses. Nor are obvious ways to use math for those cases of biology.
In fact, I don’t think there is any general program of reduction in science if we consider reduction between two different sciences (I’m excluding within one science examples). I don’t see that as a problem as I think the physics constraint alone is enough to forbid downward causation or strong emergence.
What I do see instead of a general program of reduction are cases where one science is used to help build and explain mechanisms which form part of the theory of another. For example, in neuroeconomics, behaviours observed in psychology experiments on choice are modeled using the neural activity in various brain areas. But the psychological framework of cause and effect remains for the overall behavior.
For much more in the above vein, see the Bechtel paper I linked early in the thread.
Bruce,
To be honest, my intuitions actually run (or at least used to run) in the other direction, but the scientific image won out over the manifest image. 🙂
As far as I can see, in-principle reduction can fail only if strong emergence exists, and so far I haven’t seen any persuasive evidence of that.
Keep in mind that we’re talking about the reduction of biology to physics as a whole, not the reduction of biology to the laws of physics. It’s an important distinction, because the laws of physics by themselves don’t supply a complete description of the world. To get a complete description, you need the laws, the initial conditions, and the results of any nondeterministic events — but those are all part of the physics.
I agree, but that’s for pragmatic reasons. Reduction is clearly impractical in most cases, but I think it’s still possible in principle.
Thinking about Wolfram’s automata, the problem isn’t that you can’t express higher level structures, but that you can’t know of their existence without running the program.
Bruce,
See my comment above. A successful reduction needn’t restrict itself to the laws alone. Initial conditions and nondeterministic events count too. What matters is that truths expressed in terms of the higher-level theory can also be expressed in terms of the lower-level theory.
Within the Game of Life, a “biological” truth like
…has an equivalent expression in terms of the “physics” — that is, you can define “glider”, “move”, “speed”, “direction”, “orientation”, “collide”, and “object” solely in terms of the “physics” and prove that the “biological” law is an inevitable outcome.
It’s stronger than that. It really is the idea that the truths of biology can be re-expressed as truths of physics.
petrushka,
But again, that is an in-practice limitation of the human mind, not an in-principle limitation on reducibility. To a sufficiently advanced mind, everything that happens within a cellular automaton would be the blindingly obvious result of the specified “physics”.
I disagree. Minds — at least the ones we know about — are physical structures that learn by trial and error. They can extrapolate from regularities, but emergent phenomena are, by definition, irregular. If they were derivable from principle, they would be lawful and regular, not emergent.
Weak or strong, emergence is pretty much defined as that which cannot be derived from regularity. A circular definition as stated, but in fact, we do not know of any way to predict all the properties of all combinations.
petrushka,
That’s not right. Google ’emergence’ and you’ll see.
keiths:
My mistake.
Since “reducible to physics” does not mean “reducible to the laws of physics” what does “reducible to physics” mean?
Does “reducible to physics” mean “not reducible to the laws of physics but yet still reducible to physics”? Whatever that means.
Physics without the laws of physics. How can biology be reduced to whatever
“physics without the laws of physics” means?
Is biology reducible to physics?
It’s a meaningless [nonsense] question.
We have folks on both sides of this question, so it should make for an interesting discussion.
Not really.
petrushka,
I think this comment of yours was intended for the “Compressed Sensing” thread, so I’ve responded there.
Mung,
Ha. I think what he is saying is “keithspeak”.
In many circles its more commonly referred to as gibberish. Its a common language used here.
This seems to be not so much a matter of richness as a matter of different ontologies – a problem that necessarily arises in any non-trivial instance of inter-theory reduction. If this issue were necessarily a show-stopper, then we would not even be talking about reduction – it wouldn’t exist as a concept.
In practice, the process of reduction includes a step where we translate between ontologies, i.e. we obtain a description of a system in one theory, then redescribe the same state by means of another theory. For example, a particular instance of a “glider” can be defined via the values of cells in some area. Conversely, the values of cells in an area can be interpreted as an instance of a glider.
I agree that “richness” is a vague term which is why I stuck with my personal intuition about it.
I also agree that the translation is an issue. My point about biology versus physics was that it would seem to be harder to translate between concepts expressed mainly in biological mechanisms and a physics expressed solely in mathematics. Again, that is simply an intution about how different the types of expression are. (I know one can model diagrams to some degree mathematically, but I am not sure if that type of modelling helps).
That does not prove full reduction is impossible. I doubt we will ever know, because I don’t think anyone cares about whether it is possible or impossible for whole sciences to be reduced to a different science except maybe some philosophers.
For the examples in Betchel and my understanding of your examples, what happens instead is reduction on an as-needed, localized basis for modeling and explaining specific classes of phenomena, eg psychological choices by neural activity or memory by biochemical changes related to neural activity.
Bechtel characterized many of those reductions as mechanisms, with the cause/effect structure of the reduced theory preserved in the interactions of the mechanism with the context. So it is not a complete, pure reduction in that case.
But I don’t think specific examples of success after a lot of hard work are enough to conclude that all the laws of biology could be deduced from the laws of physics and some bridge laws for boundary conditions and terms. First, one would have to look at cases where the local reduction attempt failed.. Second, one would have to some idea of how much work would be needed to be done overall.
ETA: I should say that there are psychologists who think the whole fMRI program is “bullshit” and does not succeed in any sort of reduction. They subscribe to the ecological school that KN also supports.
Keith: Sure you can, but then you have to introduce more conceptual machinery than those four laws (even allowing for the addition to those four laws of definitions of basic terms already in those laws, like neighbor).
Adding those concepts ad hoc to the lower level science solely to do the reduction is not permissible according to the model I thought you had in mind.
For example, can you define any given shape that might appear in the theory (ie gliders, etc) without adding something about the orientation of cells beyond is/is not neighbor? What could a glider gun shoots a glider mean, using solely those four laws and definitions of the terms they contain with the restriction that such definitions are the ones minimally needed to make the laws work?.
Keith: That’s helpful. I had thought you had the standard Nagelian approach in mind.
Namely, deduce the laws of (say) biology from those of physics by logical deduction. To do logical deduction between propositions about different concepts, one needs bridge laws to cover translation from biology and physics of terms and of boundary conditions (boundary conditions would be needed to constrain terms which have different meanings in different coditions in biology).
But if you don’t mean Nagelian reduction, then I am not sure what you DO mean by “the reduction of biology to physics as a whole” or by “It really is the idea that the truths of biology can be re-expressed as truths of physics.” (from your other comment to me).
For example, given an initial state in biology, a final state in biology, and GIVEN the corresponding states in physics, do you just mean that there are laws in physics that are all that are needed to predict and explain how the initial state expressed in physics got to the final state as expressed in physics? (ignore randomness for this).
If you mean more, can you explain what more you mean?
BTW, it is very important the the translation from biological states to physics states be GIVEN. If you needed to derive one from the other, you’d need the bridge laws, I think.
Bruce,
Don’t forget that the grid is already part of the “physics” in the Game of Life. Assuming a Cartesian x-y style labeling system, “neighbor” would be defined something like this:
For any cell (x,y), (x+1, y) is a neighbor.
For any cell (x,y), (x, y+1) is a neighbor.
…and so on.
Since orientation is already a part of the grid itself, you are not importing the concept. Orientation is primary, and “neighborness” is a derivative.
But if you wanted to, you could make “neighbor” the primary concept and derive the grid and the concept of orientation from it. Imagine that each cell has an arbitrary but unique name: Bob, Sue, Fred, Eugene Goostman, etc. By simply specifying who is a neighbor of whom, you could imply the grid and the concept of orientation. (You could also specify a lot of weird neighbor relations that wouldn’t map onto a grid at all, but those would not be instances of the Game of Life.)
“Glider gun shoots a glider” is just a particular succession of local grid states. Once you’ve specified the initial conditions and the laws of “physics”, the rest follows inevitably. In fact, a Game of Life “physicist” could discover glider guns without even recognizing them as distinct entities. Where she would see a succession of local grid states, the “biologist” would see a distinct entity shooting out other distinct entities. Same phenomenon, but different descriptions.
The reduction is successful because the computer doesn’t have to know about glider guns or the “biological” laws governing them in order to simulate them successfully. Glider guns emerge from the blind application of the laws of “physics”. “Biology” reduces to “physics” in the Game of Life.
keiths:
Bruce:
I haven’t read Nagel, but what you describe is very close to my view.
Let’s take the following “biological” truth in the Game of Life:
The “bridge laws” tell us how to translate biological references like “glider gun in a specified orientation” into local grid states like “a state in which cell (x,y) is alive, cell (x+1, y) is dead, cell (x, y+2) is alive”, etc. There would be a corresponding bridge law for translating grid states back into glider references.
There are bridge laws (I dislike the term, because they are really mappings, not laws, but I think we’re stuck with it for historical reasons) for processes as well as entities. The description “shooting out gliders” at the biological level has an equivalent at the physical level: it’s just a particular succession of local grid states.
As I see it, the essence of the reduction claim is this:
1) Take a biological initial state;
2) Apply the relevant biological laws;
3) Arrive at a biological final state.
4) Take the same initial biological state;
5) Apply a bridge law to translate it into a physical initial state;
6) Apply the relevant physical laws;
7) Arrive at a final physical state;
8) Apply a bridge law to translate the final physical state into a final biological state.
If biology reduces to physics, then the final biological state will be the same in steps 3 and 8.
Discovering workable bridge laws is part of the project of reduction. To say that biology is reducible to physics is simply to say that there exists a set of bridge laws for which the eight-step process I outlined above will always work.
–E.W. Hall, 1932
walto,
In the context of this thread, what do you see as the relevance of Hall’s discussion of relevance?
keiths,
Not sure. It suggests to me that the data of one field of study may be irrelevant–and so irreducible–to another. But I don’t know, myself.
Anyhow, it’s got a good beat and you can dance to it. It’s from one of Hall’s first publications.
walto,
Well, that’s often true. Anthropology will never be reduced to aerodynamics, for example, because their respective domains barely overlap. Good luck to anyone trying to explain Arawak family structure in terms of the Reynolds number.
The reduction of biology to physics seems much more promising, on the other hand, because all of the things that biology studies are ultimately physical. Unless strongly emergent phenomena exist in biology (or somewhere between physics and biology, perhaps in chemistry), I claim that biology is reducible to physics — in principle.
In some literature on reduction a distinction is drawn between type-type reduction and token-token reduction. The former is the Nagelian reduction between the laws of two different theories by way of bridge laws. The latter is what I had in mind: a redescription of specific phenomena, or events, where two different theories are applied to describe/predict the same observations. More on this in the SEP articles The Unity of Science and Reductionism in Biology (see also Types and Tokens).
To me, Nagelian type-type reduction never seemed like a particularly credible idea: my intuition doesn’t tell me that it ought to be possible, and I haven’t seen any persuasive arguments for why it ought to be possible. But I have a strong intuition for token-token reduction, and the arguments for it are basically the arguments against strong supervenience.
Biology may be reducible to physics, but can we aggregate physics to biology with no prior knowledge of biology?
In the practical sense, probably not; I mean, we can barely even aggregate biology to biology. In principle… here is an interesting example: It turns out that biological organisms are very efficient dissipative systems: they produce more entropy per unit mass than even the cores of stars. Non-equilibrium thermodynamics thus retrodicts that such systems are likely to arise in our universe (see Eric Chaisson‘s research).
Keith: There is no grid as part of the four laws that I can see. They work, eg, with the proper definition of neighbour and a linked list of cells with pointers to the neighbours.
In the real world, physics covers many more phenomena than biology but those phenomena are much simpler than biology.
Game of life is simplification we can try to use, but I think an analogy to real world biology/physics fails unless we let the physics (ie the four laws) be something more than the biology (the emergent patterns in the grid). Sticking to the four laws as you quoted from Wiki is my an attempt to do that. No need for a 2D grid: they work for abstract linked lists, 1D grids, etc with the right defn of neighbor.
If we assume the four laws are only meant to apply to the patterns we see on a 2D grid, then the analogy I was trying to use won’t work.
Biology has to be reducible to physics because birth of universe and thence evolution of life was a physics phenomenon.
Keith: Thanks for the details of your thinking.
I have not read Nagel either (except for the bat stuff): my understanding comes from the summaries in SEP and text books. It’s been 50 years since Nagel wrote about this stuff and I want to see how his ideas have held up.
Not that well among philosophers is my reading.
Alex Rosenberg has a nice intro to philosophy of science covering that and much else that might interest you. There is also a summary in the Bechtel paper I linked earlier.
He seems to share many of your views on reductionism, although the text itself is neutral on those viewpoints.
Thanks for clarifying your thinking.
The token-token reduction is the same type of reduction I had in mind for the second of questions I asked Keith. I think the issue is: given the token expressed in biological language (eg the frequency of specific genotypes in some population), what do you need to get the corresponding token in physics. Can you even do that in principle?
I’m afraid I have not worked through the subtleties involved in strong versus weak supervenience. I should do that as a refresher in predicate logic some day.
Were you around for the thread which covered the Ladyman and Ross book Everything Must Go. They trash all metaphysics which tries to work with the supervenience idea, at least as it is expressed in Kim and those that use the terms similarly.
Not the same Nagel.
Dumb question. How do we know we know all the laws? My understanding is that we make up the laws. Laws are curve fits. Newton’s being a prime example. They fit the data at hand, but not necessarily all the data we may find.
To say one science must reduce to another implies we know all the laws. I would argue that if we can’t derive biology or psychology from physics and can’t derive emergent phenomena, it is because our understanding of the laws is incomplete.
OK, thanks, I should have checked that more closely.
ETA: actually, it was pretty dumb of me not to take ages into account since the bat Nagel is still working today, eg the book on evolution. Then again, there’s Putnam.
I think that that having a complete set of physical laws is another part of the dirt that is swept under the rug by saying “in principle”.
Bruce,
You’re still assuming that the “physics” is completely described by the laws, but it isn’t.
Newtonian physics isn’t completely described by its laws. There is no law that says “space exists and has three dimensions”, for example. Likewise, the Game of Life is not fully described by its laws, because there is no law that says “the grid exists and has two dimensions.”
However, any accurate and complete description of the Game of Life will tell you that it takes place on a two-dimensional grid, just as any accurate and complete description of Newtonian physics will tell you that it takes place in three-dimensional space.
petrushka,
That’s not a dumb question at all. It’s an important one.
One thing we can be sure of is that we don’t know all the laws. If we did, then we would already have reconciled quantum mechanics and relativity, for instance.
The claim of reducibility is provisional and based on our current understanding of biology and physics, which is why I wrote this:
The moment someone convincingly establishes the existence of a strongly emergent phenomenon, or of downward causation, that claim goes onto the trash heap.
Rich,
Yes, in the following sense. We can derive every possible biology by looking at initial states (on all the planets throughout the universe, for example) and applying the laws of physics to see where they lead. You’ll get a lot of biologies, and Earth’s biology will be one of them.
You could do a similar thing in the Game of Life. Exhaustively test all starting conditions, applying only the laws of “physics”, and you will uncover every possible “biology” in the Game of Life.
Of course both of the above are possible only in principle, not in practice (at least once the GoL grid reaches a certain size).
Bruce,
If we had to have a complete set of laws before talking about reduction, we would never talk about reduction!
While it seems that biology would be totally and most completely modeled as a grand fractal wavefunction, my concern is that in taking this level of analysis as sufficient description, one’s model ends up poorer for tossing out the cruder levels of analysis “above.” Maybe at a god’s eye perspective these cruder aspects would flow from the purely informatic bottom level, being identical to it, but as humans it seems our model of biology will always be most rich if its analysis includes data specifically pertinent to concepts like populations, fitness, kin selection, etc. Zoom in all the way, and you toss out concepts that could be useful in informing your model.
If we had to say what biology is reducible to, we would never talk about reduction!
Are presentations by philosphers at professional conferences normally just them reading their published paper to the audience?
The reason I ask: in trying to get a better understanding of Brandom (to understand KN’s book), I saw he had a YT video on Intentionality. But it was simply him reading one of his densely-worded papers. That was followed by a Huw Price commentary, which (after a picture of a beach in Australia for the intro), also turned into him reading a paper. And I’ve seem Kitcher do the same.
Are all philosopher presentations like that? If so, why not skip that part, assume everyone has read the paper, and go right to the Q&A?
In my management career in IT, I’ve written a lot of reports and then given presentations on them. Simply reading the report would definitely be a career-limiting move (low evaluation of oral communication skills for starters). In presentations I have seen at other business IT conferences, the same applies: people who tried to just read reports soon lost their audience to the doors.
I can think of two possible explanations for this difference:
1. Philosophy departments cannot afford the big bucks businesses have to send their managers to courses teaching presentation skills.
2. If you cannot understand a presentation when it is simply the underlying paper read out verbatim to the audience, then you do not have the right stuff to be a philosopher.
I haven’t been to a ton of colloquia, but my sense is that the presenters are generally reading stuff that HASN’T been published yet (but sometimes soon will be). They’re getting the kinks out. The APA won’t accept submissions of already published stuff for talks.
I noticed Dennett has a 2006 paper (pdf) detailing his perceptions of the differences between him and Brandom, based on his understanding of Brandom’s 1994 Making it Explicit.
Dennett starts by saying he mainly agrees with Brandom’s description of meaning as being “the deontic score-keeping game of asking for and giving reasons” in linguistic communities. So in that sense, I suppose he does have access to the same resources as Brandom, although I don’t think he uses them directly the areas of work he currently focuses on.
He says their differences start with understanding where the original intentionality of linguistic communities comes from. Where is the source to ground the norms that allow members of communities to do the score keeping to determine meaning? That source cannot be simply the attitudes of members of the community: for those attitudes are themselves normative and so that explanation leads to a regress.
Dennett says Brandom grounds them in linguistic communities this way: One violates a norm by committing a social transgression. Norms are grounded by saying they are social transgressions. So original intentionality is a property of the community in that sense, not the individual.
Dennett, on the other hand, grounds norms in proper functioning of individuals. Proper functioning means functioning as designed. In the case of living entities like us, that means as designed by mother nature: something is functioning properly if it is fulfilling the role selected by evolution to contribute to fitness. As usual, he points out that that approach via design by evolution means there is no principled reason to use “original”.
Dennett also says that Brandom has to rely on social ie behavioral conditioning to explain how children learn the norms of their linguistic communities. Whereas he, with the subpersonal approach, can tell a better story by going beyond a purely behavioral description of how the whole person acts to a psychological and eventually biological explanation (he cites Millikan for the details of how this approach works in detail).
Dennett concludes by re-iterating that he accepts much of Brandom’s description of how meaning works via the action of linguistic communities but thinks Brandom would do well to add Dennett’s approach for explaining naturally the source of the grounding the norms of the individuals in those communities.
Thanks.
Of course, it is the “reading” part that would get you low grades on management presentation courses. You are supposed to be a more effective speaker if you talk extemporaneously, using outline notes to refer to if needed. Also using PowerPoint*, of course, preferably with graphics and pictures or very short, bulleted points on the slides.
Both reading and memorization then speaking are discouraged.
A philosopher might call those the norms of public speaking. The management professionals who teach those courses would likely have no idea what the philosopher was talking about.
———————————————————-
(*) Management consultant humor:
A good question is one I don’t know the answer for. I’ll go find out.
A very good question is one I do know the answer for. Here it is.
An excellent question is one where I have a PowerPoint presentation which gives the answer. Let me plug in my laptop.
Ah. I don’t think that’s frowned on in philosophy colloquia at all. I take it people want to hear pre-publication papers–just as they’re expected to appear–and get an early chance to tee-off on them. I just wish I could get away with it in the classroom. It would be sooo much easier if I could just read my lectures.