# Two approaches to science

• I
• Demystifier
In summary, the standard approach to theoretical physics is focused on making predictions that can be tested and measuring the results. The realist approach, on the other hand, focuses on understanding what the world is made of.f

#### Demystifier

Gold Member
As you may have noticed, I am obsessed with understanding the difference between two views of quantum mechanics, one of which can be called the "standard" view, and the other the "realist" view. The difference, of course, is very complicated, but I believe that the essence and origin of the difference must be simple. In this thread I am trying to explain the difference in terms of two simple schemes, corresponding to two approaches to theoretical physics as a science.

The standard approach:
1. Write down the equations.
2. Make the measurable predictions implied by the equations. That's the most important part to do if you aspire to be a scientist, rather than just a mathematician.
3. If you can, try to say what does it all tell us about what the world is made of. But if you are not sure, stay silent about it.

The realist approach:
1. Say what is the world made of.
2. Write down the corresponding equations. That's the most important part to do if you aspire to be a scientist, rather than just a philosopher.
3. If you can, try to make the measurable predictions implied by the equations. But if you are not sure, stay silent about it.

You don’t say where the equations come from in the standard approach.

You don’t say where the equations come from in the standard approach.
Sort of. I mean fundamental equations, such as those by Heisenberg and Schrodinger in their famous papers.

Sort of. I mean fundamental equations, such as those by Heisenberg and Schrodinger in their famous papers.
Given that there are only a handful of quantum interpretations, doesn’t that mean that there are only a handful of realists? The followers of the realists are merely following the realist papers and are thus standard In outlook.

I am unsure that you have properly differentiated the two groups.

Given that there are only a handful of quantum interpretations, doesn’t that mean that there are only a handful of realists? The followers of the realists are merely following the realist papers and are thus standard In outlook.
The followers of realists are themselves realists. By "standard" I mean the approach followed by the majority.

I don't think there is a clear demarcation line between the two approaches to science. All physicists must claim that their theories (somehow) describe the real world. And the inventory of a theory (to avoid the loaded term "ontology" ) can change with time. Fourier solved the correct heat equation describing some hypothetical "caloric" even before the first law of thermodynamics was formulated. And was Maxwell a realist? I think so. He arrived at useful equations for an elastic fluid, although this "ether" later turned out to be non-existent.

We interpret the macroscopic world around us in terms of objects, classical physics is "about" objects. And we think of these objects as composed of smaller objects. But at the microscopic level "objects" seem to make no sense. We expect an object to have properties, but these properties become uncertain, or even undefined. They pop into existence by some mysterious process called "measurement". I think the problem with quantum theory is that we think that it is about quantum "objects" and measurements. Like Maxwell's theory, quantum theory is extremely successful, even though quantum "objects" are at least as puzzling as the ether was in Maxwell's time. In the case of electrodynamics it "only" took four decades to solve the puzzle.

dextercioby, martinbn, gentzen and 2 others
The followers of realists are themselves realists. By "standard" I mean the approach followed by the majority.
You have not convinced me that your distinctions are the relevant ones and that they map “cleanly” onto the majority and minority viewpoints.

You have not convinced me that your distinctions are the relevant ones and that they map “cleanly” onto the majority and minority viewpoints.
In this thread, it was not my intention to convince anybody that this distinction is relevant, because I was doing this in many other threads. In this thread I took this distinction for granted and tried to explain where does this distinction come from.

In this thread, it was not my intention to convince anybody that this distinction is relevant, because I was doing this in many other threads. In this thread I took this distinction for granted and tried to explain where does this distinction come from.
Ok. This will be my last comment, so as not to derail the thread. I would argue that most realists are dissatified with some of the loose threads of the standard approach and searched for an alternative viewpoint. They do not start with “this is what the world is made of.”

dextercioby
The standard approach:
1. Write down the equations.
2. Make the measurable predictions implied by the equations. That's the most important part to do if you aspire to be a scientist, rather than just a mathematician.
3. If you can, try to say what does it all tell us about what the world is made of. But if you are not sure, stay silent about it.

The realist approach:
1. Say what is the world made of.
2. Write down the corresponding equations. That's the most important part to do if you aspire to be a scientist, rather than just a philosopher.
3. If you can, try to make the measurable predictions implied by the equations. But if you are not sure, stay silent about it.
I don't see much connection between the approaches discussed in QM foundations and the approaches of day-to-day scientists. But I am not a theoretical physicist. Is there perhaps a "case study" (outside of QM foundations) that exemplifies these two approaches? I hear the pentaquark is making some ripples.

Last edited:
"ether" later turned out to be non-existent
No, later turned out that explanation without "ether" is technically simpler. And the same can be said about micro "objects" in QM. Standard QM does not talk explicitly about objects, because it's technically simpler. For practical purposes, we can regard "ether" and micro "objects" as non-existent. Nevertheless, their existence is not ruled out, and it is possible to formulate the theories with them, but it makes things more complicated technically.

No, later turned out that explanation without "ether" is technically simpler.
Actually, in the later evolution of physics the ether was replaced by something with higher symmetry, and it's now called "vacuum".
Standard QM does not talk explicitly about objects, because it's technically simpler.
I'm not with you. Can you learn QM and never hear about photons and electrons?? That's the kind of quantum objects that I had in mind.

vanhees71, martinbn and gentzen
I think I approximately see the distinction Demystifier tries to line out.

The "standard approach" settles with - after in principle lots of state and process tomography - a describing the "black box" with input/output pattern of what we control and observer, and this characterizes the system. To try to invent an ontology with kinematical structure of the "inside" that as per the ontolgoy, naturally "explains" the observed patterns, are not considered meaninful, as it may not be unique, or possible to determined, so we settle with the behaviour, encoded by equations and abstract mathematical representations. And this then becomes an effective theory, and is therfore hard to generalize, this is the con side

The "realist apparoch" would then insist seeking to construct a "theory" for the black box itself, that does reproduce and "explain" the descriptive picture we alread know from the tomography. If this works, this seems to be nice, and would be easier to generalize to OTHER similar situations, but the question is, how does one come up with the candidate ontologies? Maybe some sort of abduction to best explanation? And how does one "prove" when the ontology is "correct"? This is a similar question you face in an agent learning; how do you "prove" or deduce that the agents inference is "optimal"? I think you don't, learning does not progress by deduction, not even after "filtering" the inductions with falsficitation tests. I think this gets lots of people poppian itch though.

Ok. This will be my last comment, so as not to derail the thread. I would argue that most realists are dissatified with some of the loose threads of the standard approach and searched for an alternative viewpoint. They do not start with “this is what the world is made of.”
I wonder how Demystifier would categorize the people that does both. They both stick onto the "standard approach" to QM, AND claim that the world is made of little strings? Isn't that a little bit of both?

/Fredrik

Demystifier
I wonder how Demystifier would categorize the people that does both.
In any categorization of people into different types, any real individual is a mixture of all types. For example, we can divide people into introverts and extroverts, but nobody is 100% introvert or 100% extrovert. The same applies to your question.

Last edited:
Peter Morgan
I'm not with you. Can you learn QM and never hear about photons and electrons?? That's the kind of quantum objects that I had in mind.
You said: "But at the microscopic level "objects" seem to make no sense. We expect an object to have properties, but these properties become uncertain, or even undefined."
Are you saying that standard QM talks about objects that seem to not make sense? Or are you just using typical Copenhagen double talk?

Of course standard QM talks about photons and electrons. But it does not explicitly talk about them as objects in the classical sense, just as you said. So I have no idea why you are not with me, when I said essentially the same thing as you did.

Last edited:
gentzen
Actually, in the later evolution of physics the ether was replaced by something with higher symmetry, and it's now called "vacuum".
True. In fact, the meaning of the word "vacuum" is not unique in physics. In condensed matter it often means a bunch of atoms in the state without collective excitations, which is very close to the original notion of "ether".

gentzen
You said: "But at the microscopic level "objects" seem to make no sense. We expect an object to have properties, but these properties become uncertain, or even undefined."
Are you saying that standard QM talks about objects that seem to not make sense?
Yes, that's exactly the point. And sorry for echoing "typical Copenhagen double talk". Electrons and photons are not "objects in the classical sense", but what are they? I think it's misleading to the extreme to call them any kind of "object" at all. For me the word "photon" (more precisely, of course, the photon propagator) describes just the correlation between short-lived currents in the "source" and the "detector", emission and absorption events. I don't assume that there is anything that "travels" from A to B, expanding into space, and then instantaneously collapsing to a point. I think this is more in line with the actual QED formalism: a minimal statistical interpretation with a minimal ontology. Merely statistics of events distributed in spacetime. I consider myself a realist, and my take on the Bell-type experiments is the opposite of that of @vanhees71. The absorption and emission processes in these experiments are real and definite, and the description (theory) is decidedly non-local. Talk of "quantum objects" with indefinite but entangled "properties" is much more confusing than simply stating (describing) correlations.

Of course standard QM talks about photons and electrons.
Maxwell thought of light as waves, and for him waves traveling without any medium that carries them must have been as unthinkable as QED without electrons and photons is for us. But these concepts carry too many classical connotations that we should get rid of. For many it seems self-evident that the correlations in Bell-type experiments must be caused by something traveling from the source to the detectors. It's a simple, "obvious" explanation, but it gets us into the quantum quagmire, and I prefer to stay away from it.

Electrons and photons are not "objects in the classical sense", but what are they?

and what are made of ?

......

and what are made of ?
You think electrons and photons are "stuff"? That they must be made of something?
I don't. For me they do not exist in this sense. For me they are patterns of events in spacetime.

Peter Morgan
You think electrons and photons are "stuff"? That they must be made of something?
I don't. For me they do not exist in this sense.
For me they are patterns of events in spacetime.

They are process ?

They are process ?
Not in the sense of something being processed.
But this is drifting away from @Demystifier's initial topic.

Yes, that's exactly the point. And sorry for echoing "typical Copenhagen double talk". Electrons and photons are not "objects in the classical sense", but what are they? I think it's misleading to the extreme to call them any kind of "object" at all. For me the word "photon" (more precisely, of course, the photon propagator) describes just the correlation between short-lived currents in the "source" and the "detector", emission and absorption events. I don't assume that there is anything that "travels" from A to B, expanding into space, and then instantaneously collapsing to a point. I think this is more in line with the actual QED formalism: a minimal statistical interpretation with a minimal ontology. Merely statistics of events distributed in spacetime. I consider myself a realist, and my take on the Bell-type experiments is the opposite of that of @vanhees71. The absorption and emission processes in these experiments are real and definite, and the description (theory) is decidedly non-local. Talk of "quantum objects" with indefinite but entangled "properties" is much more confusing than simply stating (describing) correlations.
Indeed, here you are contradicting the mathematical facts of local relativistic QFT. Interestingly you precisely describe the meaning of the formalism, i.e., that the observables are encoded in autocorrelation functions (Green's functions) of observables with the meaning of detection probabilities. That's the meaning of quantum fields, and photons are pretty special states, i.e., Fock states of asymptotically free electromagnetic fields. Their charactristic is that they can be registered once or not at all. That's what makes them "quanta" in the original sense of Planck, and indeed what's localized is not the photon, which has not even a position observable at all but the detector (or the atom within the detector with which the photon interacted to finally lead to its detection).

The formalism is built such that the observations ("detector clicks") are local, i.e., space-like separated registration events cannot be causally connected. The theory is not "realistic" in Bell's sense, because as in any QT also in relativistic local QFT the observables of a "quantum object" (and for me photons, electrons, atoms, etc. are indeed objects, because they can be prepared in quantum states, as is obvious from the fact that we can do experiments with high precision with them) only take determined value, if they are prepared in a corresponding state. I also don't see any problem with the notion of an "object" in the case of entangled quanta. Then the corresponding single-particle states are strongly correlated and the single-particle observables are indetermined, but the state as a whole is still an object with clearly defined properties, which are encoded in the state they are prepared in.

Of course, what's real are indeed the "detector clicks", and all we can predict about them with QFT are the probabilities for a detector to click at its position at any time after the preparation of the "object" observed.
Maxwell thought of light as waves, and for him waves traveling without any medium that carries them must have been as unthinkable as QED without electrons and photons is for us. But these concepts carry too many classical connotations that we should get rid of. For many it seems self-evident that the correlations in Bell-type experiments must be caused by something traveling from the source to the detectors. It's a simple, "obvious" explanation, but it gets us into the quantum quagmire, and I prefer to stay away from it.
The correlations in Bell-type experiments are due to the preparation of the systems in entangled states. The strong correlations are already there, when the system is measured. It's not the measurement that causes the correlations, which is clearly demonstrated by the fact that the correlations are observed also when the observation events are space-like separated. There were even experiments done, where the choice of the measured observables on the separate parts of the system was made at spacelike separation, i.e., also here any causal effect of the choice of the measured observables at the separated places is excluded.

Within QED, what "propagates in a wavelike way" are the probability distributions for photon detections. That's not different from classical electrodynamics, where what we observe in optics are also the intensity distributions (energy densities) of the electromagnetic field. The fields themselves manifest by the interactions with the charged matter making up the detectors we use to observe or measure them (including our eyes). There is not so much difference between classical field theories and quantum field theories in this respect. The difference is however that there are situations, that can only be described by quantum and not by classical electrodynamics (spontaneous emission, quantum beats, single photons, entanglement,...).

Indeed, here you are contradicting the mathematical facts of local relativistic QFT.
I wasn't making any mathematical statement, so it could have been only contradicting what you construe as a hard physical fact from a piece of the formalism. I'm aware that there is only one definition of "locality" that you consider the correct one, and there is no point discussing this yet again as in hundreds of other threads.

The theory is not "realistic" in Bell's sense, because as in any QT also in relativistic local QFT the observables of a "quantum object" (and for me photons, electrons, atoms, etc. are indeed objects, because they can be prepared in quantum states, as is obvious from the fact that we can do experiments with high precision with them) only take determined value, if they are prepared in a corresponding state.
It's okay if you think that quantum theory is best expounded in terms of "state preparation" and "measurement". I don't share this view. I think it carries too much metaphysical baggage. (In spite of your dislike of philosophy, you seem to have become quite accustomed to it. )

dextercioby and Simple question
I wasn't making any mathematical statement, so it could have been only contradicting what you construe as a hard physical fact from a piece of the formalism. I'm aware that there is only one definition of "locality" that you consider the correct one, and there is no point discussing this yet again as in hundreds of other threads.
You shouldn't make statements about a theory, which clearly contradict its mathematical construction.
It's okay if you think that quantum theory is best expounded in terms of "state preparation" and "measurement". I don't share this view. I think it carries too much metaphysical baggage. (In spite of your dislike of philosophy, you seem to have become quite accustomed to it. )
State preparation and measurements are done by experimental physicists but (at least nowadays) never by philosophers ;-).

PeroK
You shouldn't make statements about a theory, which clearly contradict its mathematical construction.
You shouldn't mistake a feature of a brick that was used in the construction of a house as the defining characteristic of that building. But as I said already, it's almost pointless to continue this discussion.

physika
I'm aware that there is only one definition of "locality" that you consider the correct one, and there is no point discussing this yet again as in hundreds of other threads.
I think the resolution to this neverending confusion is something like this.

In the effective view of the world from the perspective of any actual agent, where the information it holds had to be communicated, "locality" is there be construction as in QFT as well.
=> In this view, the challenge is to understand the mechanism of the correlation, given the entanglement. And as we know, ignorance mecahnism as per bell does not work. You can also ignore it, and just presume there is a mechanism that we don't understnad.

But in the ontological god view of the world from the perspective of the set of all "non-communicated" bare views of all potential observers, there is not "spacetime" relation in this coding, so there i no locality, but otoh it's not a problem either. It's what i hinted at here https://www.physicsforums.com/threa...omputation-theory.1052331/page-5#post-6891161
=> In this view, the challenge is to explain how macroscopic spacetime relations (between all the intrinsich ontological views emerge). You can also ignore it and just presume that it somehows must work out.

It seems to me both views just avoid different things, the question is, in which view do we think the problem most easily solved? And what is least painful to ignore. It's how I select my stances, and we make different choices.

/Fredrik

For me the word "photon" (more precisely, of course, the photon propagator) describes just the correlation between short-lived currents in the "source" and the "detector", emission and absorption events. I don't assume that there is anything that "travels" from A to B

Of course, what's real are indeed the "detector clicks", and all we can predict about them with QFT are the probabilities for a detector to click at its position at any time after the preparation of the "object" observed
If there's nothing "real" propagating between the source and the detector, why do experimentalists need all those fibre-optic links between them?

More on topic, I think everyone has an ontology. But the camps for me would be:
• A) those who suggest an ontology beyond the observations
1. Those who believe the equations themselves are the "stuff" that exists beyond the observations
2. Those who believe the equations are only a tool for performing calculations and do not necessarily correspond to real things in the world
• B) Those who believe observations are all that exists and it's forbidden to talk about what may be there beyond observations
I think you can't straddle more than one of these categories and stay consistent.

A good ontology may lead to testable predictions that ultimately could improve our theories and understanding of nature. A bad ontology can lead to stagnation if it doesn't give any ideas about where to look for new things or discourages further investigation. It's obvious how Camp B can lead to stagnation, but the pitfalls of A1 can't be ignored either.

WernerQH
A good ontology may lead to testable predictions that ultimately could improve our theories and understanding of nature. A bad ontology can lead to stagnation if it doesn't give any ideas about where to look for new things or discourages further investigation. It's obvious how Camp B can lead to stagnation, but the pitfalls of A1 can't be ignored either.
Agreed.
More on topic, I think everyone has an ontology. But the camps for me would be [...]
Agreed. But I have difficulties identifying my "camp".
those who suggest an ontology beyond the observations
I don't think observations "dictate" an ontology -- every experiment is carried out with a specific interpretation in mind. Things like "phlogiston", "caloric", or "ether" were once thought to be real and have been replaced with something more "refined". Maxwell's electrodynamics is an instructive example. Maxwell surely thought of the ether as real, and used it as a scaffolding to arrive at equations similar to those in elasticity theory and hydrodynamics. Nowadays we no longer need to think of an electric field as stress in an elastic medium, and consider it "real" even though it has lost its material basis. The ether has been discarded and replaced with something with higher symmetry, the vacuum. And of course the vacuum (or ground state) is an important ingredient in field theories (and thus in most physicists' sense "real").
Those who believe the equations themselves are the "stuff" that exists beyond the observations
I don't believe that equations can be "stuff". To me they must always refer to something that exists happens in the real world. Call me an utterly naive realist, if you like. :-)
If there's nothing "real" propagating between the source and the detector, why do experimentalists need all those fibre-optic links between them?
Indeed, QED without photons is as unthinkable for us as electrodynamics without an ether was for Maxwell. But I think we should make that effort. And get rid of the "scaffolding" that we think QED still requires. The "photon" that comes out of the fibre-optic link is not the same as the one that went in. In his lectures, Feynman explains the origin of the refractive index as repeated absorption and emission processes. I don't consider the "traveling" photon real, but the short-lived microscopic currents that are induced in the medium. The word "traveling", and of course also the term "propagator", suggest continuity that is contradicted by the evidence of detector clicks.

If there's nothing "real" propagating between the source and the detector, why do experimentalists need all those fibre-optic links between them?

More on topic, I think everyone has an ontology. But the camps for me would be:
• A) those who suggest an ontology beyond the observations
1. Those who believe the equations themselves are the "stuff" that exists beyond the observations
2. Those who believe the equations are only a tool for performing calculations and do not necessarily correspond to real things in the world
• B) Those who believe observations are all that exists and it's forbidden to talk about what may be there beyond observations
A.-2
The Epistemic View.

Those who believe the equations are only a tool for performing calculations and do not necessarily correspond to real things in the world
That applies to the wave function. Something that evolves continuously and deterministically according to the time-dependent Schrödinger equation cannot faithfully describe the abruptness and randomness that we observe in the real world. Hence the uneasy mix of unitary evolution with measurements.

If there's nothing "real" propagating between the source and the detector, why do experimentalists need all those fibre-optic links between them?
Of course, there's something real propagated, i.e., the electromagnetic field. According to QFT everything is described by quantized fields.
More on topic, I think everyone has an ontology. But the camps for me would be:
• A) those who suggest an ontology beyond the observations
1. Those who believe the equations themselves are the "stuff" that exists beyond the observations
2. Those who believe the equations are only a tool for performing calculations and do not necessarily correspond to real things in the world
Whatever it might be, what's "beyond the observations" it's not, what's studied within the natural sciences. With this you clearly go into other realms of human experience than objective observations of phenomena.
• B) Those who believe observations are all that exists and it's forbidden to talk about what may be there beyond observations
It's pretty likely that there is something beyond what we can currently observe with our contemporary means of technology, e.g., "dark matter".
I think you can't straddle more than one of these categories and stay consistent.

A good ontology may lead to testable predictions that ultimately could improve our theories and understanding of nature. A bad ontology can lead to stagnation if it doesn't give any ideas about where to look for new things or discourages further investigation. It's obvious how Camp B can lead to stagnation, but the pitfalls of A1 can't be ignored either.
Of course, it's always worth aiming at ever more sensitive detectors/measurement devices to discover "new stuff".

gentzen
Of course, there's something real propagated, i.e., the electromagnetic field.
So you are finally outing yourself as a realist!

Structure seeker, Demystifier and gentzen
@vanhees71, how does that square with your previously expressed view that the Bell-type experiments force us to give up realism in favour of locality? If the quantum field is real and in a definite state at all times, a naive physicist would expect this to apply also to the polarization state of a photon. Shouldn't this be derivable from the state of the quantum field? But somehow the polarization becomes real (definite) only when photons encounter detectors, in a peculiar way that you insist on calling local.

Demystifier
But the polarization state for the entangled photons is clearly determined by the preparation procedure. I don't know, what you are looking for.

But the polarization state for the entangled photons is clearly determined by the preparation procedure.
It is not. Not if you think of the entangled photons as two photons, each with its own polarization state. I understand that you insist on describing the two photons as a single system, but I fail to see what you accept as "real". The two photons have indefinite polarization before the detection, but acquire a definite polarization state at the moment they cease to exist?

Simple question and Demystifier