## Reductionism

Not quite philosophy discussions, debates, various thought experiments and other topics of interest.

### Reductionism

BioWizard,

I guess that this stuff's hard to talk about, but here's kinda my attempt to put my view into words...

In general, I acknowledge that any and all mutually consistent logical systems can always be generalized into a single logic system, and so I adopt reductionism as my fundamental epistemic approach. For me, learning is the process of getting new information to test and update my general understanding.

I understand stuff by starting from my general model and reducing to whatever I'm trying to consider. For example, when reading academic studies, I reduce scope down to the domain of interest and apply the appropriate approximations to reproduce the models employed in the study. I can then read about their results to test and update my beliefs against their findings.

My learning process (as described above) is fairly fruitless when studies apply tons of approximations like are commonly found in the social sciences. The problem is that noisy results, such as those found by way of gross approximations, can't really reject or confirm my current beliefs nor form the basis for new beliefs. I suspect that some folks might think, "Well, just read lots of studies and pick up on a trend!", but even that doesn't work since the approximation techniques have major skews in their errors, leading to false trends.

From Ursa's closing remarks, I kinda got the gist that he doesn't hold a similar philosophical viewpoint that all mutually consistent models are always describable as reductions from a single, general understanding. I see only two ways in which his position could be maintained:
1. Simple ignorance to the fact that generalized systems can always be constructed when logic is consistent.
2. Allowance for inconsistent logic.
I'd prefer to think that it's ignorance, as ignorance is less objectionable than fallacy. Still, it's satisfying enough to get why someone wouldn't accept the concept of rigor across disciplines or/and within specific methodologies.

For me, the degree of rigor can always be assessed based on a consideration's construction from the general understanding. And then when I complain about a lack of rigor
• What I mean:
It was possible to construct that consideration using fewer reductions.
• Why I'm saying it (often at least one of the following):
1. I couldn't get useful information out of the consideration, e.g. a study, due to the noise resulting from the reductions used.
• I'm annoyed that those positing the consideration, e.g. the researchers writing up a study, reported conclusions not significantly supported by the consideration due to its reductions.

PS - This is probably largely a repeat of my prior post on this subject, but just trying to express the idea in a different way to see what communicates it best.
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Rigor

I'd add that reductions can be appropriate, so long as we understand the expected noise. We often teach new concepts using reduced considerations of those concepts.

Long-winded example; skippable
For example, in the recent About Orbital Hybridization thread, I was trying to answer a question from someone who probably didn't know anything about Quantum Chemistry in terms of it. To make this easier to swallow, I applied tons of reductions, including but not limited to:
1. Presented the wave function as though it were real (as opposed to being able to be complex, i.e. have non-zero imaginary values).
2. Focused on the hydrogen-like approximation in which all atoms had exactly one electron.
3. I told the poster that molecular orbitals are like hybrid orbitals that involve more than one atom. This was true-ish, but molecular orbitals are actually more complex since they involve multiple nuclei, making linear independence far more questionable. I didn't even mention this problem.
4. I used 1D and 3D Hilbert spaces throughout like they represented all spacial configurations considered by Quantum Mechanics. However, Quantum actually allows for far more exotic types of space.
5. I explicitly neglected gravity. Provided reasons for this one in the thread.
6. I applied the Born-Oppenheimer approximation throughout, i.e. I assumed that nuclei don't move. Didn't mention this one, either.
7. Didn't even attempt to consider solutions with outside potential affecting the system.
8. Didn't even mention the fact that we're talking about time-independent solutions, which was a major approximation.
9. Didn't even try to include relativistic effects.
10. Made a mistake in saying $r^{-1}$ instead of $r^{-2}$ for Coulomb's law. Fixed it later.
These reductions were selected to avoid confusion and explain the concept. The final consideration had tons of noise in it, and as Quantisierung noted, the consideration wasn't anywhere near analytical quality. However, I feel that I posited with enough rigor that the consideration provided a meaningful conceptual introduction.

Back on-topic
So like I mentioned to Ursa earlier, lacking rigor isn't necessarily a bad thing, and there can be reasons for it.

In general, the two most common problems with lacking rigor seem to be:
1. Useless: The lack of rigor is so gross that it invalidates any potential use for the findings.
2. Fallacious: Researchers draw conclusions beneath the level of noise.
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Reductionism

Natural ChemE » Tue Oct 06, 2015 5:08 pm wrote:BioWizard,

I guess that this stuff's hard to talk about, but here's kinda my attempt to put my view into words...

In general, I acknowledge that any and all mutually consistent logical systems can always be generalized into a single logic system, and so I adopt reductionism as my fundamental epistemic approach.

The wiki doesn't talk about reductionism as an epistemic approach. What do you mean?

mtbturtle
Banned User

Posts: 9742
Joined: 16 Dec 2005

### Re: Reductionism

mtbturtle » October 7th, 2015, 7:50 am wrote:The wiki doesn't talk about reductionism as an epistemic approach. What do you mean?
Epistemological Unities, "The Unity of Science" in the Stanford Encyclopedia of Philosophy, looks like a decent source.

To connect some terminology, focusing on vocabulary italicized in the linked article:
1. Their "reduction" is my "reduction".
2. Their "connectability" is my "mappability".
3. Their "approximation" is my "lossful simplification".
My "lossless simplifications" are perfect approximations, e.g. scope reductions and conversions to normal forms.
4. Their "derivability" is my ability to understand something (third paragraph in this post).
The above "Epistemological Unities" article is a good source for the overall concept.

For a common, practical application of it, Tree of Knowledge System, Wikipedia, seems decent enough.
However this particular application mostly focuses on connecting fields while the overall approach is far more general. The overall approach connects not just fields, but all considerations (fields, models, concepts, laws, axioms, etc.).

I have to say that looking for sources on this felt odd. Don't get me wrong, I entirely, 110% agreed with your request for a good source, and I also fully agree that the Wikipedia article on reductionism wasn't very helpful for what I was trying to say. Still, this sort of epistemic approach is so basic to how I operate as an intelligent being - so basic to my very existence - that finding sources on it felt like finding instructions on how to breath or walk or move one's arms. The sources that I did find made it out to be some kind of high-minded concept, but for me it's basically just how thinking works.

Heck I'm used to folks joking about it, e.g. xkcd posted
, which is basically kidding about how the wide spread acceptance of Tree-of-Knowledge-type views has led to common cultural egotism across disciplines.
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Reductionism

I found a book on this topic: Beyond Reduction, Steven Horst (2007).

Online abstract for Beyond Reduction by Steven Horst, wrote:Beyond Reduction: Philosophy of Mind and Post-Reductionist Philosophy of Science
Steven Horst

ABSTRACT
Contemporary debates in philosophy of mind‐between reductionists, dualists, nonreductive materialists, and eliminativists‐have been based upon the perception that mental phenomena like consciousness and intentionality are uniquely irreducible. The “explanatory gap” between mind and body seems to be an urgent and fascinating problem if one assumes that intertheoretic reductions are the rule in the special sciences, with the mind as the lone exception. While this debate was going on in philosophy of mind, however, philosophers of science were rejecting this very sort of reductionism: intertheoretic reductions are not ubiquitous but rare. This book argues that post‐reductionist philosophy of science poses problems for all the familiar positions in philosophy of mind and calls for a deep rethinking of the problematic. To this end, a new perspective, Cognitive Pluralism, is urged.

After reading the author's commented drafts (I'd suggest the outline if you want to skim it), I can tell ya that this guy isn't exactly the brightest of bulbs. It's basically a fringe theorist objecting to his own misunderstanding of the concept. Still, it at least discusses the concept from the perspective of a Philosophy professor, whatever his limitations might be.
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Reductionism

You might like this (and Quine, on holism, generally):

https://en.wikipedia.org/wiki/Duhem%E2%80%93Quine_thesis

TheVat

Posts: 7214
Joined: 21 Jan 2014
Location: Black Hills

### Re: Reductionism

Natural ChemE » October 6th, 2015, 4:08 pm wrote:
My learning process (as described above) is fairly fruitless when studies apply tons of approximations like are commonly found in the social sciences. The problem is that noisy results, such as those found by way of gross approximations, can't really reject or confirm my current beliefs nor form the basis for new beliefs. I suspect that some folks might think, "Well, just read lots of studies and pick up on a trend!", but even that doesn't work since the approximation techniques have major skews in their errors, leading to false trends.

NCE,

The most charitable interpretation I can give for the above is that you know enough of the language of social science statistics and research methods to say things, but not enough to be fluent. And you don't know that you are not fluent. So you say things that are laughably wrong to the "ear" of people who are fluent.

If you don't consider the difference between biased errors in measurement (affecting validity of the measures) and unbiased error (affecting reliability of a single measure), then I can see how you say what you do in the paragraph above. If you wish to discuss the matter of measurement error in detail, you need to keep this distinction, which I have mentioned MULTIPLE TIMES, in mind. If you don't do so, I will be unable to consider it as other than intellectual dishonesty in any future posts.

Skew from measurement error results from biased error. Unbiased error does not skew results, though it will reduce what can be said about the results. Sampling error from sample size is unbiased error.

There are ways to establish that estimators are unbiased. I suggest you google proving unbiased estimator and do some reading before attempting further discussion.

There are ways of measuring both reliability and validity. We can know relative levels of each for different measurement techniques. Google is your friend here as well.
Ursa Minimus

### Re: Reductionism

Ursa Minimus,

It's weird to me that you'd think that the methods employed in the social sciences are anything but basic. But, hey, maybe I'm wrong?

Let's pick a concrete sample to discuss. One of us needs to pick this sample. If I pick it, then nothing I say could ever prove my position because you might simply disregard it as a bad example. So, could you pick what you'd consider to be a good example of social science research?

Preferably:
1. Respectable due to at least one of the following:
1. published in a top-tier journal;
3. from a well-respected institution.
2. Complex. Ideally it should confuse a non-social scientist such as myself, as to demonstrate your point about fluency.
Do NOT hold back.
The harder, the better.
• Employs sampling, modeling, and data analysis techniques to arrive at a non-obvious conclusion.
Confirmation studies that just show what we already know, e.g. "We found that people generally don't like being punched in the face", wouldn't be too useful here. However, a study that quantifies exactly how much various demographics dislike being punched in the face would be non-obvious.
• Accessible. I have VPN connections to several major universities, so if you select an article from just about any mainstream journal, I should be able to read it. But if you pick one in some very obscure journal that most universities don't have a subscription to, then I'd be unable to read it.
• More than one would be nice, but I understand that we're all pressed time.
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Reductionism

Natural ChemE » October 6th, 2015, 4:08 pm wrote:
From Ursa's closing remarks, I kinda got the gist that he doesn't hold a similar philosophical viewpoint that all mutually consistent models are always describable as reductions from a single, general understanding. I see only two ways in which his position could be maintained:
1. Simple ignorance to the fact that generalized systems can always be constructed when logic is consistent.
2. Allowance for inconsistent logic.
I'd prefer to think that it's ignorance, as ignorance is less objectionable than fallacy. Still, it's satisfying enough to get why someone wouldn't accept the concept of rigor across disciplines or/and within specific methodologies.

The philosophical concept of emergence is relevant here.

http://plato.stanford.edu/entries/properties-emergent/

Durkheim, the first person to hold a chair in Sociology, said that sociology is the study of social facts.

Social facts are produced by individual actions, singly and in concert, this is true. However, he defined social facts as things that are external to and coercive of the individual. So Durkheim said that they take on a life of their own once created. Emergent properties develop.

Social facts are both material (buildings, law as written, etc) and non-material (norms, cultural values, etc).

He also said social facts should be explained by other social facts, and that is what sociology should do. Using statistics. His classic on Suicide does this, showing how higher numbers of social obligations in a group decreases suicide rates in that group. A social factor has causal effect beyond individual level causes. And that idea has been tested, a great deal, in many contexts. For over a hundred years in fact. Seems to hold up pretty well.

Here is a simple example of a a social reality that exerts causal influence that cannot be reduced to characteristics and choices of actions of individuals. And a method to test those causal influences that results in not having to worry about measurement error.

Social networks. The patterns of connections between people are a social fact in Durkheim's terms. They are external to us, and coercive of us in that they have effects on us whether we want them to or not.

Assume that there are different social structures out there for exchanging. And suppose you limit what you are looking at to exchanges where all people in the network can make one deal at a time with one other person. There are a series of exchanges over time.

Can we explain the outcomes based purely on personality characteristics? Decisions made by individuals? Or is there some property of the network beyond the individual that alters outcomes INDEPENDENT of the individuals in the network?

Imagine two different networks:

A-B-C

A-B-C-D

Those in the positions in the network know who they can deal with, but nothing else about the network shape. They don't know the connections of their connections.

Suppose I set up a lab to test the effects of the network. I assign people randomly to the positions. This will control for all variations between people, and to the extent I have a higher N, I will have more control over ALL individual variations. So I will have some random error there, but no biased error. If I needed to reduce my random error, I just add more cases of course.

I have no measurement error at all for network position, or deals made, or "profit" in those deals.
I have all that recorded exactly.

Clear? The hypothesized cause is network structure, and I have no measurement error at all for the proposed causal mechanism. I have no biased error, as random assignment to experimental position in the network wipes those out by design. I have no reason to believe that any differences in people are patterned by network position assignment. And good reason to believe such patterned differences are not present. And more reason to believe that with increasing sample size.

What happens?

Well, in A-B-C what happens is that B has high power from the structural position, and gets better and better deals over time.

In A-B-C-D, everyone ends up with equal power, and the positions all have equal profit. In fact, the network splits, into two subnetworks: A-B and C-D. Which should obviously have not structural advantage, given the symmetry.

Can these network effects be reduced to individual acts of individual people?

No.

Sociology looks for exactly this kind of effect as a core principle of the field.

There are competing theories that try to explain these effects of structure. They calculate power from the network shape using equations based on the network properties, and where networks under those exchange conditions will "break", among other things. The math involved will work on ANY network of ANY shape with ANY number of positions.

The prediction of where networks break is the bigger deal imo, and has far more application than the accuracy of profit predictions. But the theories have a high degree of accuracy in prediction for profit. If "deals" average 10 units, the theories based purely on network structure accurately predict mean outcomes by position to somewhere in the hundreths of unit. Or at least they did last time I looked at that literature.

So feel free to try to decompose that effect of social structure to the individual, or lower, level. If you so desire.
Ursa Minimus

### Re: Reductionism

Natural ChemE » October 7th, 2015, 11:21 am wrote:Ursa Minimus,

It's weird to me that you'd think that the methods employed in the social sciences are anything but basic. But, hey, maybe I'm wrong?

Let's pick a concrete sample to discuss. One of us needs to pick this sample. If I pick it, then nothing I say could ever prove my position because you might simply disregard it as a bad example. So, could you pick what you'd consider to be a good example of social science research?

Preferably:
1. Respectable due to at least one of the following:
1. published in a top-tier journal;
3. from a well-respected institution.
2. Complex. Ideally it should confuse a non-social scientist such as myself, as to demonstrate your point about fluency.
• Employs sampling, modeling, and data analysis techniques to arrive at a non-obvious conclusion.
Confirmation studies that just show what we already know, e.g. "We found that people generally don't like being punched in the face", wouldn't be too useful here.
• Accessible. I have VPN connections to several major universities, so if you select an article from just about any mainstream journal, I should be able to read it. But if you pick one in some very obscure journal that most universities don't have a subscription to, then I'd be unable to read it.
• More than one would be nice, but I understand that we're all pressed time.

Already gave you an example (we were cross posting), but not a link. Try the explanation first.

Generally, any quantitative research article in American Journal of Sociology, American Sociological Review, or Social Forces will be at the highest level. Qualitative research as well, but there is a lot less of that in those journals, and the issues of qualitative data analysis will be very far from what you know.

The research I talked about on exchange networks (a good search term) also will show up in Social Psychology Quarterly. Which is top level, but a specialty journal and not a generalist one like the other three. You find anything in those, I will say it has gone through the best vetting process. If you find recent things, it will be at the current state of knowledge in the area being researched. So don't expect me to defend stuff from the 1950s, please.

But I do suggest you stick to the main point of this thread, and if you want to talk about a specific article, post in social science. Because we will be covering a lot of ground would be my guess.
Ursa Minimus

### Re: Reductionism

Ursa Minimus » October 7th, 2015, 11:37 am wrote:
Let's pick a concrete sample to discuss.

http://www.jstor.org/stable/2095689

This is probably the best place to start for the network example I posted. Note that it is 1988, so the accuracy is not as nailed down as later refinements. And the network breaks are further refined later. And a huge amount of work has happened since then, much of which I have not even glanced at.

It requires a lot less literature knowledge than the later ones, which might cite 20 other papers in the area. It is, to say the least, explicit.

Knock yourself out if you like.
Ursa Minimus

 Natural ChemE liked this post

### Re: Reductionism

Ursa Minimus,

Awesome, I can split this thread later to discuss the complexity of the social sciences using that paper as an example. In particular, I'd like to ask you why you think this stuff's confusing to someone who finds far more competitive fields to be quite accessible.

So, what would you like for me to answer? For example, are you asking me to demonstrate how social networks are describable as reductions from a general understanding?

Just to avoid a false disagreement, I don't argue that all social science studies are useless or/and critically flawed. What I have said is:
1. A lot of social research could be improved by better modeling techniques.
• Yes, I get bias estimation and all of that. My complaint isn't that I don't understand how they're doing it, but rather that they're not doing it well enough. This is, I think that most social scientists don't know enough math/statistics/modeling to be optimally effective at what they're trying to do.
• I also don't get why social researchers don't bother to improve. Surely many of them could learn non-linear modeling if they tried, right? Why aren't a large number of them seizing on the opportunity to stand above their peers, raising their field to a new height?
2. A lot social research is flawed.
• But not all; I'm not dismissing the entire field as pointless. Still, when I read an article about how a bunch of papers in some social field weren't reproduced well or at all, my reaction is mostly just, "Duh." And then I do get confused, but only because I don't get why that's news to anyone.
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Reductionism

Natural ChemE » October 7th, 2015, 12:21 pm wrote:Ursa Minimus,

Awesome, I can split this thread later to discuss the complexity of the social sciences using that paper as an example. In particular, I'd like to ask you why you think this stuff's confusing to someone who finds far more competitive fields to be quite accessible.

Why? Many, many examples. For example...

If one makes confused statement about biased measurement error and unbiased measurement error time after time....

And for another, anyone who uses IQ scores to measure "competitiveness" (and who thinks I am going to click a blog link for this discussion) has likely not thought about the term they are using in depth. Competitiveness could be measured as #seeking/#positions, a nice ratio measure. I don't see how IQ affects anything in that ratio, which is a far better definition of competition for positions than whatever you are using with IQ.

Do you need more math for some fields? Yep. Do you need more of other skill in other fields? Yep. Nothing to do with competition there.

And....Philosophy majors have the highest GRE scores in verbal and analytical. Not so good on math, but top in 2 of three categories. Physics, not so good on verbal, and on analytical about where philosophy is on math. So, what does that mean if we use that instead of IQ for your claim about "competitiveness"?

I could go on, but I have a feeling if that doesn't do it, nothing will.
Ursa Minimus

### Re: Reductionism

Natural ChemE » October 7th, 2015, 12:21 pm wrote: This is, I think that most social scientists don't know enough math/statistics/modeling to be optimally effective at what they're trying to do.
[*]I also don't get why social researchers don't bother to improve.

Please explain how you have enough knowledge to claim anything about "most" social scientists, their level of knowledge, and what would be optimal for what they are trying to do.

Social science, like all science is constantly improving. There are entire journals that publish ONLY on methods, and each publication proposes an improvement. They tend to use math and everything.

The question for any given technique is not if it is perfect, but if it works well enough for the purposes at hand. I don't need a scalpel to slice a sausage, because such precision does not matter for the task at hand. I would not want a kitchen knife for neurosurgery, it is too gross a tool. Think of regression(s) in this way, good enough for the kitchen, not good enough for brain surgery.

And none of this has anything to do with reductionism.
Ursa Minimus

### Re: Reductionism

Ursa Minimus,

I kinda feel like I'm shooting fish in a barrel here. Like, that article I linked uses GRE scores to project IQ scores. Then you go on to say that GRE scores are important while IQ scores are moot. I'm guessing you didn't know that they correlated, or..? Anyway, you can just look at the GRE scores instead if you prefer 'em. Both metrics agree.

But if it helps, you were right about Philosophy - those guys did alright, with their verbal reasoning making up for their weak math! They almost matched physicists! Just not sure why you'd point this out when it's Sociologists we're talking about, and their GRE and IQ scores are both pretty bad as far as academia is concerned.

And, yeah, "less competitive" was my way of saying something politely as opposed to precisely. My apologies, just.. I'm literally saying that people in your field tend to have lower cognitive abilities, which I believe leads to a lower quality of work. More precise terms for this just seemed rude. I'm really trying to discuss this with you out of academic interest - not to win a pissing contest on the internet - so I'm making an effort to be respectful.

Anyway, let's stop harping on sociologists - I'm always curious to learn about my own failings!
Ursa Minimus » October 7th, 2015, 2:31 pm wrote:If one makes confused statement about biased measurement error and unbiased measurement error time after time....
I didn't realize that I've been making a lot of confused statements time after time. Could you point them out so that I can address my own confusion?

Honestly I think that you're mistaking my criticism of studies having error bias for my not understanding the concept. But that line of logic doesn't really work, so perhaps you mean something else?
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Reductionism

Ursa Minimus » 07 Oct 2015 12:24 pm wrote:Can these network effects be reduced to individual acts of individual people?

No.

Sociology looks for exactly this kind of effect as a core principle of the field.

There are competing theories that try to explain these effects of structure. They calculate power from the network shape using equations based on the network properties, and where networks under those exchange conditions will "break", among other things. The math involved will work on ANY network of ANY shape with ANY number of positions.

The prediction of where networks break is the bigger deal imo, and has far more application than the accuracy of profit predictions. But the theories have a high degree of accuracy in prediction for profit. If "deals" average 10 units, the theories based purely on network structure accurately predict mean outcomes by position to somewhere in the hundreths of unit. Or at least they did last time I looked at that literature.

So feel free to try to decompose that effect of social structure to the individual, or lower, level. If you so desire.

Yes assigning people randomly to the nodes creates a destructive interactions between person-to-person differences, therefore eliminating bias from individual variation (simply stated, individual bias cancels out). But you're forgetting that it also creates constructive interactions between the traits that ARE shared amongst them, thereby giving rise to the network edges (whether or not you know the exact mechanisms). The fact that a stable network structure emerges is proof of that. If instead of people, you assign hamsters to your nodes, do you think your network will maintain the same structure?

Yes there is feedback between the structure of the network and the behavior of the individual person. But your network develops from people's original states, and feedback to the individuals isn't antithetical to reducibility (think different cultural structures, for example). Complex is all you can say about it, with any certainty.

If you thought that was a good example of "irreduciblity", you'll need to try again. And if you thought this kind of problem is unique to social science, you need to read up on cell signaling, interactomics, and so on, and the new math being developed for them. Maybe you'll be less likely to miss such obvious and elementary facts about networks and won't so readily confuse complexity with irreducibility.

Controllability of complex networks : http://www.nature.com/nature/journal/v4 ... 10011.html

If you just want to "fit" models cause it's easier, that's OK. But don't fool yourself into thinking that's all there is (and all there can be) to it...

BioWizard

Posts: 12071
Joined: 24 Mar 2005
Location: United States
Blog: View Blog (3)

### Re: Reductionism

Natural ChemE, the thread might not be on rigor but ad hominems are still unwelcomed.

mtbturtle
Banned User

Posts: 9742
Joined: 16 Dec 2005

### Re: Reductionism

mtbturtle » October 7th, 2015, 4:10 pm wrote:Natural ChemE, the thread might not be on rigor but ad hominems are still unwelcomed.

As you'd recall, ad hominem is pretty much, "You're [some negative quality], therefore your argument is wrong." This is distinct from, "A group of knowledge workers with a relatively low cognitive ability will tend to operate in a relatively simple domain of knowledge."

'course then there's the politeness issue, and I'm trying really hard to be nice here. I'm having to balance honesty against politeness. It's not a fun trade-off to have to make.
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Reductionism

Natural ChemE » Wed Oct 07, 2015 4:16 pm wrote:
mtbturtle » October 7th, 2015, 4:10 pm wrote:Natural ChemE, the thread might not be on rigor but ad hominems are still unwelcomed.

As you'd recall, ad hominem is pretty much, "You're [some negative quality], therefore your argument is wrong." This is distinct from, "A group of knowledge workers tends to produce lower quality work due to their significantly lower abilities". The first is a common logical fallacy while the second is completely valid.

'course then there's the politeness issue, and I'm trying really hard to be nice here. I'm having to balance honesty against politeness. It's not a fun trade-off to have to make.

Right you aren't making an argument nor responding to one, you are just being rude, insulting. That also is unwelcome.

mtbturtle
Banned User

Posts: 9742
Joined: 16 Dec 2005

### Re: Reductionism

mtbturtle » October 7th, 2015, 4:41 pm wrote:Right you aren't making an argument nor responding to one, you are just being rude, insulting. That also is unwelcome.

Oh, my response didn't seem connected to the topic? Okay, gotcha, I can elaborate.

What I was trying to respond to what Ursa's repeated assertions that non-social scientists are ill-equipped to understand social science. From where I sit, that's absurd because social science doesn't really have anything unique about it - it's not "different" so much as just "easy". Everything's really simplified.
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Reductionism

Natural ChemE » Wed Oct 07, 2015 4:53 pm wrote:
mtbturtle » October 7th, 2015, 4:41 pm wrote:Right you aren't making an argument nor responding to one, you are just being rude, insulting. That also is unwelcome.

Oh, my response didn't seem connected to the topic? Okay, gotcha, I can elaborate.

What I was trying to respond to what Ursa's repeated assertions that non-social scientists are ill-equipped to understand social science. From where I sit, that's absurd because social science doesn't really have anything unique about it - it's not "different" so much as just "easy". Everything's really simplified.

I thought Social Science was really complex, difficult to deal with but I really don't see how any of this is relevant. X has a lower IQ, therefore their theories are wrong? Say what! How does your IQ equip you to evaluate Social Science?

The truth or validity of theories, etc. does not depend or relate to IQ.

mtbturtle
Banned User

Posts: 9742
Joined: 16 Dec 2005

### Re: Reductionism

Natural ChemE » Wed Oct 07, 2015 10:41 am wrote:
mtbturtle » October 7th, 2015, 7:50 am wrote:The wiki doesn't talk about reductionism as an epistemic approach. What do you mean?
Epistemological Unities, "The Unity of Science" in the Stanford Encyclopedia of Philosophy, looks like a decent source.

To connect some terminology, focusing on vocabulary italicized in the linked article:
1. Their "reduction" is my "reduction".
2. Their "connectability" is my "mappability".
3. Their "approximation" is my "lossful simplification".
My "lossless simplifications" are perfect approximations, e.g. scope reductions and conversions to normal forms.
4. Their "derivability" is my ability to understand something (third paragraph in this post).
The above "Epistemological Unities" article is a good source for the overall concept.

For a common, practical application of it, Tree of Knowledge System, Wikipedia, seems decent enough.
However this particular application mostly focuses on connecting fields while the overall approach is far more general. The overall approach connects not just fields, but all considerations (fields, models, concepts, laws, axioms, etc.).

I have to say that looking for sources on this felt odd. Don't get me wrong, I entirely, 110% agreed with your request for a good source, and I also fully agree that the Wikipedia article on reductionism wasn't very helpful for what I was trying to say. Still, this sort of epistemic approach is so basic to how I operate as an intelligent being - so basic to my very existence - that finding sources on it felt like finding instructions on how to breath or walk or move one's arms. The sources that I did find made it out to be some kind of high-minded concept, but for me it's basically just how thinking works.

Heck I'm used to folks joking about it, e.g. xkcd posted
, which is basically kidding about how the wide spread acceptance of Tree-of-Knowledge-type views has led to common cultural egotism across disciplines.

Read the SEP - could you pull out which parts of it apply to what you are talking about? It's not obvious from the section you referenced.

mtbturtle
Banned User

Posts: 9742
Joined: 16 Dec 2005

### Re: Reductionism

mtbturtle » 07 Oct 2015 05:29 pm wrote:I thought Social Science was really complex, difficult to deal with but I really don't see how any of this is relevant. X has a lower IQ, therefore their theories are wrong? Say what! How does your IQ equip you to evaluate Social Science?

The truth or validity of theories, etc. does not depend or relate to IQ.

Mtb,

I think what's been said/suggested was that the systems that social science attempts to study are complex, while the approaches that social science uses tend to incorporate a large number of simplifications, and may therefore be considered relatively simple/simplistic compared to what gets done in other scientific fields.

I agree though that average IQ for scientists in the field isn't all that relevant. Especially with how interdisciplinary research is done nowadays. I mean, it's not necessarily the molecular biologists themselves who are developing all the math for their network models. Relatively few individuals tend to possess that kind of crossdiscipline prowess, particularly between the life/biological and computational/mathematical fields.

BioWizard

Posts: 12071
Joined: 24 Mar 2005
Location: United States
Blog: View Blog (3)

### Re: Reductionism

BioWizard » Wed Oct 07, 2015 7:51 pm wrote:
mtbturtle » 07 Oct 2015 05:29 pm wrote:I thought Social Science was really complex, difficult to deal with but I really don't see how any of this is relevant. X has a lower IQ, therefore their theories are wrong? Say what! How does your IQ equip you to evaluate Social Science?

The truth or validity of theories, etc. does not depend or relate to IQ.

Mtb,

I think what's been said/suggested was that the systems that social science attempts to study are complex, while the approaches that social science uses tend to incorporate a large number of simplifications, and may therefore be considered relatively simple/simplistic compared to what gets done in other scientific fields.

On what basis would any of us have for such comparisons?

I agree though that average IQ for scientists in the field isn't all that relevant. Especially with how interdisciplinary research is done nowadays. I mean, it's not necessarily the molecular biologists themselves who are developing all the math for their network models. Relatively few individuals tend to possess that kind of crossdiscipline prowess, particularly between the life/biological and computational/mathematical fields.

all that relevant? it's not relevant at all.

mtbturtle
Banned User

Posts: 9742
Joined: 16 Dec 2005

### Re: Reductionism

mtbturtle » 07 Oct 2015 08:04 pm wrote:On what basis would any of us have for such comparisons?

On the basis of things mentionef in this thread, the rigor thread, the (un)reliability of published psycology research thread, the insistance that social systems are irreducible (which could mean that the field can not provide mechanistic insights), and so on.

all that relevant? it's not relevant at all.

Well yes, that's what that entire paragraph said, in several different ways. Are you going to argue with me on punctuation next? Maybe I should've added some exclamation marks to emphasize just how not relevant I thought it is?

That said (again), I am a lot more inclined to think that if a field falls behind (on reliability/reproducibility/quality/etc) it's because of the intellectual/technical laziness of its scientists rather than anything to do with their average (or combined) IQ.

And yes, I do find constantly going with what's "good enough" and being frequently nonchalant about the underlying mechanisms (or using complexity to invoke anti-reductionist philosophies and argue that they are virtually unknowable) to be scientifically lazy. Why don't I just do the same about the tremendously complex biological sigaling networks I study and call it a day. It'll make it a lot easier to just sit around and fit data. It'll also free up a significant amount of my time that I can then use to pointificate to scientists from other fields about all kinds of stuff.

BioWizard

Posts: 12071
Joined: 24 Mar 2005
Location: United States
Blog: View Blog (3)

### Re: Reductionism

mtbturtle » October 7th, 2015, 5:29 pm wrote:How does your IQ equip you to evaluate Social Science?

It's just normal Science, plus I've read tons of journal articles on it. There's really not much more to it.

Same to you, by the way. Don't let pretentious pricks scare you off or make you feel stupid. You seriously could learn Quantum Mechanics this week if you'd like. Heck, I'd help ya if you wanted; you'll need a few nights to sleep to let some of the stuff you learn to sink in properly, but it's really not that bad. Also it's super fun - given your interest in learning, it could be a neat project, right?

Anyway, that's how this junk works. Ursa doesn't get to say who can and can't understand social science - it's a simple subject with no special tricks. There's nothing behind the pretense.
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Reductionism

mtbturtle » October 7th, 2015, 7:36 pm wrote:Read the SEP - could you pull out which parts of it apply to what you are talking about? It's not obvious from the section you referenced.

I'm not really sure what points to pull out since I'm not sure what you're interested in. I'll kinda try to select points most obviously related to what we'e been talking about. My goal is to form a direct bridge to the entry to make it more accessible.

I'll just bold the more immediately relevant parts.
The Unity of Science, Stanford Encyclopedia of Philosophy wrote:3. Epistemological Unities
3.1 Reduction
Philosophy of science consolidated itself in the 1950s around a positivist orthodoxy roughly characterized as follows: a syntactic formal approach to theories, logical deductions and axiomatic systems, with a distinction between theoretical and observational vocabularies, and empirical generalizations. Unity and reduction may be introduced in terms of the following distinctions: epistemological and ontological, synchronic and diachronic. The specific elements of the dominating accounts will stand and fall with the attitudes towards the elements of the orthodoxy mentioned above. Reductionism must be distinguished from reduction: reductionism is the adoption of reduction as the global ideal of unified structure of scientific knowledge and a measure of its progress. As before, I will consider methodological aspects of unity as an extension of epistemological matters, insofar as methodology serves epistemology.

Two formulations by logical positivists in the United States about the ideal logical structure of science again placed the question of unity of science at the core of philosophy of science: Carl Hempel's deductive-nomological model of explanation and Ernst Nagel's model of reduction. Both were fundamentally epistemological models, and both were specifically explanatory. The emphasis on logical structure makes unity of explanation and reduction chiefly of the synchronic kind. Nagel's model of reduction is a model of scientific structure and explanation as well as of scientific progress. It is based on the problem of relating different theories as different sets of theoretical predicates.

Reduction poses two requirements: connectability and derivability. Connectability of laws of different theories requires meaning invariance in the form of extensional equivalence between descriptions, with bridge principles between coextensive but distinct terms in different theories.

Nagel envisaged two kinds of reductions: homogenous and heterogeneous. When both sets of terms overlap, the reduction is homogeneous. When the related terms are different, the reduction is heterogeneous. Derivability requires a deductive relation between the laws involved. In the quantitative sciences, the derivation often involved taking a limit. In this sense the reduced science is considered an approximation of the reducing new one.

Neo-Nagelian accounts have attempted to solve Nagel's problem of reduction between putatively incompatible theories. Here are a few:
Nagel's two-term relation account has been modified by weaker conditions of analogy and conventions, requiring it to be satisfied not necessarily by the two original theories, T1 and T2, which are respectively new and old and more and less general, but by the modified theories T1′ and T2′. Explanatory reduction is strictly a four-term relation in which T1′ is “strongly analogous” to T1 and corrects, with the insight that the more fundamental theory can offer, the older theory, T2, changing it to T2′. He also required that the bridge laws be synthetic identities, in the sense that they be factual, empirically discoverable and testable, rather than conventions (Schaffner 1967; Sarkar 1998).[2] The difficulty lay especially with the task of specifying or giving a non-contextual, transitive account of the relations between T and T′ (Wimsatt 1976).

An alternative set of semantic and syntactic conditions of reduction bear a counterfactual interpretation. For instance, syntactic conditions in the form of limit relations and ceteris paribus assumptions have the function of explaining why the reduced theory works where it does and fails where it does not (Glymour 1969).

A different approach to reductionism acknowledges a commitment to providing explanations but rejects the value of the focus on the role of laws. This approach typically draws a distinction between hard sciences such as physics and chemistry and historical sciences such as biology and social sciences, and claims that laws that are in a sense operative in the hard sciences are not available, or play a more limited and weaker role. The rejection of empirical laws in biology has been argued on grounds of dependence on contingent initial conditions (Beatty 1995), and as matter of supervenience (see the entry on supervenience) of spatio-temporally restricted functional claims on lower level molecular ones, and the multiple realization (see the entry on multiple realizability) of the former by the latter (Rosenberg 1994; Rosenberg's argument from supervenience to reduction without laws must be contrasted with Fodor's physicalism about the special sciences about laws without reduction (see below and the entry on physicalism); for a criticism of these views see Sober 1996). This non-Nagelian approach furthermore assumes that explanation rests on identities between predicates and deductive derivations (reduction and explanation might be said to be justified by derivations, but not constituted by them; see Spector 1978, also for the articulation of this view in physics). Explanation is provided by lower-level mechanisms; their explanatory role is to replace final why-necessarily questions (functional) with proximate how-possibly questions (molecular). A similar line of argument has been proposed as a defense of the explanatory power of the lower level, the disjunction of supervening bases without reduction (understood in the Nagelian sense of requiring bridge laws with one-to-one mappings between predicates at different levels and the derivation of the reduced laws or theories from the reducing ones)(Kincaid 1997). On this view lower-level explanations of, say, biological or sociological properties and events might fail to capture and discriminate causal properties and patterns and yield false causal inferences. Most, if not all, lower-level theories are not less localized or qualified ceteris paribus than higher-level ones. Lower-level descriptions can still get their explanatory relevance in the following limited form: from the heuristic asymmetry that makes the higher-level valuable and ineliminable by answering higher-level questions centered on higher-level salient descriptions and groupings of phenomena; lower-level explanations, then, would proceed by targeting specific token representations of events or properties, not more general types that are realized by and supervene on the more specific tokens. The discussion here enters the conceptual metaphysical arena I discuss below. One suggestion to make sense of the possibility of the supervening functional explanations without Nagelian reduction is a metaphysical picture of composition of powers in explanatory mechanisms (Gillette 2010). The reductive commitment to the lower level is compositional, from epistemological analysis and metaphysical synthesis, but not derivational; we infer what composes the higher level but we cannot simply get all the relevant knowledge of the higher-level from our knowledge of the lower-level (see also Auyang 1998).

A more general characterization is of reductionism as a research strategy. On this methodological view reductionism can be characterized by a set of so-called heuristics (non-algorithmic, efficient, error-based, purpose-oriented, problem-solving tasks) (Wimsatt 2006): heuristics of conceptualization (descriptive localization of properties, system-environment interface determinism, level and entity-dependence), heuristics of model-building and theory construction (model intra-systemic localization with emphasis of structural properties over functional ones, contextual simplification, external generalization) and heuristics of observation and experimental design (focused observation, environmental control, local scope of testing, abstract shared properties, behavioral regularity, context-independence of results).

It's awfully wordy for such a simple topic, but that's how these sources operate I guess.
Natural ChemE
Forum Moderator

Posts: 2744
Joined: 28 Dec 2009

### Re: Reductionism

Natural ChemE » Wed Oct 07, 2015 8:21 pm wrote:
mtbturtle » October 7th, 2015, 5:29 pm wrote:How does your IQ equip you to evaluate Social Science?

It's just normal Science, plus I've read tons of journal articles on it. There's really not much more to it.

articles on what "it"?

Same to you, by the way. Don't let pretentious pricks scare you off or make you feel stupid. You seriously could learn Quantum Mechanics this week if you'd like. Heck, I'd help ya if you wanted; you'll need a few nights to sleep to let some of the stuff you learn to sink in properly, but it's really not that bad. Also it's super fun - given your interest in learning, it could be a neat project, right?

Anyway, that's how this junk works. Ursa doesn't get to say who can and can't understand social science - it's a simple subject with no special tricks. There's nothing behind the pretense.

It's not about pretense, it's about training. Reading some journals articles is not going to give you or me that training. A high IQ is not going to give you the proper training. You wouldn't accept somebody that read some journal articles coming into the science forums and making sweeping generalizations about your methods, would you? You might rightly wonder if they know anything about the actual methods you use and why you choose one over the others. You also might recognize simple errors made because they are well read but untrained.

mtbturtle
Banned User

Posts: 9742
Joined: 16 Dec 2005

### Re: Reductionism

Mtb,

Can someone trained in several scientific disciplines really be considered "untrained" when the subject matter shifts to social science? Should we ask Ursa to stop talking about linear regressions, network modeling, and other things because those were developed by statisticians and mathematicians, not by social scientists?

If Ursa is using tools that we all use and are familiar with in our work, and if he's putting his null hypotheses and methodologies in clear terms, what is the unsurmountable barrier for someone who is familiar with the methodologies and can pubmed the previous literature to evaluate the work? Once you learn how to do science, the activation energy for shifting your subject matter goes down by a lot. Science isn't the way it used to be 100 years ago. Almost everybody uses the same tools nowadays, and we all have free access to online journals through our universities. It boils down to how quickly we're able to evaluate a study, not to whether or not we're able to do it. Ursa hasn't mentioned a single technique, approach, or example that I (and presumably NCE) aren't trained/experienced in, or haven't encountered something similar to in our research. Same applies for the studies I've encountered.

Is there really no pretense here?

BioWizard

Posts: 12071
Joined: 24 Mar 2005
Location: United States
Blog: View Blog (3)
 Natural ChemE liked this post

### Re: Reductionism

BioWizard » October 7th, 2015, 6:51 pm wrote:
I agree though that average IQ for scientists in the field isn't all that relevant.

Better to compare at the tails, but even that is a bad idea.

IQ is valid for measuring performance on IQ tests.

IQ is not perfectly reliable in the test-retest sense. So there is error. Hopefully unbiased error.

I can be convinced that IQ scores are valid measures of limited types of ability.

Given the literature, I do not see that IQ scores have been proven a measure of innate ability or intelligence. Not when I know of simple manipulations in education that have produced 20 point increases in IQ in a year. See Pygmalion in the Classroom, data on the 1st and 2nd grade students. And which showed the spikes in IQ in remedial classes can result in more negative teacher evaluation. Plus, the predictive value of IQ falls off quickly as you introduce a few obvious control variables.

Given that the literature shows lots of bias in IQ scores, anyone who uses them should address this if they want to be taken as less than a crank.

Did you know that when IQ tests were being developed, an early version of one had women scoring higher than men? The researchers, of course, changed the test to get rid of this obviously false finding.
Ursa Minimus

Next