An exemplar of theory from sociology.

[Ursa Minimus]

Power Relations in Exchange Networks
Author(s): Barry Markovsky, David Willer and Travis Patton
Source: American Sociological Review, Vol. 53, No. 2 (Apr., 1988), pp. 220-236
Stable URL: http://www.jstor.org/stable/2095689



Theory has many definitions in sociology.

From the perspective of those in the example below (and many who have extended the theory since 1988), theory is: a set of interrelated statements that make causal claims which can be falsified. They have a philosophy of science consistent with people like Lakatos. https://en.wikipedia.org/wiki/Imre_Lakatos

These people also would tend to say the core goal of science is to develop and extend theory. Producing results is a technical matter, developing theory is a scientific matter.

Call the exemplar Network Exchange Theory (NET).

Why is NET a good exemplar case?

The theory is overtly explicit, and highly formalized. It identifies what it is addressing. This includes the scope of the theory. Just as chemistry is not expected to be very useful outside its scope (say, to explain architecture), we should evaluate the theory where it claims to apply. Within the scope.

It is a new theory. This means that we don't have to look at any other papers to get, fully, what the theory contains.

It is simple. The theory consists of 4 axioms, the first of which is an equation. I'm not sure, without checking, if the authors would consider the theorems as part of the core theory or not, but that really does not matter. They derive 4 theorems.

It takes on an existing theory with a long research tradition. Social Exchange Theory (SET). SET came out of a tradition than can be seen broadly as economic sociology. Critical tests are possible to determine which theory is better.

It holds up to testing, and does a better job that previous theory.

It is extended and improved by future research. Thus, the theory bears a lot of fruit.



So, because some here don't have access to the databases, I will post JUST the theory from the paper. But no more, so as to assure fair use for educational reasons on sociological theory. And I will post the screen shots separately to aid in any quoting people wish to do.

I will answer any questions on this theory, as I have time for.

I likely will ignore any questions about methods, or results. Not because such are not important, but rather because this would likely involve discussing other work in the area.

If you understand the theory, you should be able to construct a network and apply the correct GPI numbers to each node through the indicated iterative process. I will accept demonstrated ability to do this as sufficient proof that a person understands the theory.


So, on to NET:

[Ursa Minimus]

Click on the images to maximize.

Figure 1 will be the last image in the sequence from the body of the paper, followed by the proofs from the appendix.

sequencing of images clarified

[Ursa Minimus]

[Ursa Minimus]

[Ursa Minimus]

[Ursa Minimus]

[Ursa Minimus]

[Ursa Minimus]

That's the theory.

[Ursa Minimus]

FYI, I remember exactly three works in this area well, all earlier than 2000. Beyond that my knowledge is very slight in terms of either breadth or depth.

On this one, I actually forgot they tested the 2 exchange formulation in this paper, to show how the theory held beyond the single exchange condition. So while this came back to me on a re-read pretty quickly, don't think I am trying to set you all up or anything by pulling out something I have spent the past 20 years working on.

[Ursa Minimus]

Not all of what is considered theory in sociology is from a scientific perspective. So for more casual readers...

The posted exemplar is an example of modern scientific sociological theory. A classic example would be Durkheim, and his approach to studying suicide from 1897 where he tried to explain group patterns of suicide by the level of social obligations in the groups. https://en.wikipedia.org/wiki/Suicide_%28book%29

There are non-scientific theories in sociology. I would put Marx in this category. Certainly we could use Marx to make testable predictions, such as "the greater the class consciousness, the greater the union membership" or something similar. But such work is not what is done to develop Marxist theory, and rarely to test it.

Modern examples of "good" theory from this approach might be seen in the area of culture. Sociologically, culture is the "highest" level of aggregation, according to many. Scientific approaches to explaining cultural causes have largely failed to account for conflict and social change, as the assumption of cultural primacy in social causes tend to lead to conservative conclusions about lower levels of aggregation. IOW, cultural forces explain stability at lower levels when they are theorized to be the most controlling force in society, and can claim to explain gradual changes, but fail to explain rapid changes. Unless some hand waving is done.

While I can't really summarize the the work of https://en.wikipedia.org/wiki/Pierre_Bourdieu or https://en.wikipedia.org/wiki/Jean_Baudrillard easily, these would be two who would be considered cultural sociologists whose theories could be seen as attempts to explain downward causal forces in cultural terms. Foucault too, in many ways. Not very scientific, but considered theory worth paying attention to in the area of the sociology of culture.

Scientific approaches to culture tend to approach things from a macro-micro framework in sociology. This might be an individual/group, or individual/culture differentiation. Non-scientific modern theory tends to approach things from an agency/structure approach, where agency and structure are expected to be seen at all levels of the macro/micro spectrum.

[Ursa Minimus]

Most social science people come across is far from what is considered "pure" science. A good amount of it is more accurately described as using social science measures. This form of network exchange research broadly, and NET and the GPI in particular, is as close an example as you can get of "pure science" in the social sciences.

Another reason this is a good example is that the concept of "network" is not very domain specific. So there are many bases of knowledge that help to get what is being talked about. https://en.wikipedia.org/wiki/Network_theory And of course "exchanges" and "seeking higher profit" make sense at the common sense level, not just the social science level. So the thing being talked about is sensible to many from the start, even the theory that purports to explain the thing is not.

Another advantage of this, if people find it interesting, is that there is rapid development in the theory. Recall that this paper took on a theory called SET. A couple years later, SET showed weaknesses in NET for some networks, and showed that a revised version of SET predicted outcomes better.

By 1993 ( http://www.jstor.org/stable/2095966 ) NET was formulated so the GPI power numbers above were called "strong power", while another structurally identified type of power called "weak power" was also incorporated. And did better at predicting than SET.

This competition between research programs should sound familiar to anyone in any science.

By 1995 ( http://www.jstor.org/stable/2580626 ), they broadened the theory of the actor a bit, made the actor assumptions more complex, and showed how in simulations and experiments NET made more accurate predictions than five other network exchange theories. That's quite a bit of theory development, competition, and increasing accuracy in 7 years.

Changing the actor assumptions is not pushing the explanation of structural causes further (as in the 1993 paper), but it is still improvement in the accuracy of the theoretical predictions. The exemplar in this thread tried to get at structural effects by limiting the actor to the least complex assumptions that still make sense when talking about exchanges of value in networks where individual utility maximization is assumed: seek profit, exchange with best offer, raise offer when excluded, lower offer when included. Get rid of any of those, and we would have no reason to believe that actor would be reasonable as a simplification of actors in exchange networks.

For example, you could assume purely random action by actors, and try to claim that reveals the pure structural effect. Finding the outcomes of random action is pretty easy to do, and some tried to use that in their models as a base line. It turns out that starting from that basis is not very useful in terms of getting to explain outcomes with people. IIRC, some forms of exchange theory that led to SET started with this kind of assumption, maybe in the 1950s or 60? Sounds right at least.


In any case understanding this 1988 paper, one could easily skim those later two papers, and get the picture pretty well. Whether they are social scientists or not. I doubt any will be interested enough to do so at this time, but as for the future, who knows?

[Natural ChemE]

Ursa Minimus,

I'd like to thank you for showing a good example of a Sociology paper. This does look like one of the better works in the field, and since I feel that fields should generally be viewed in terms of their merits, this paper is a great look into Sociology.

A quick peek at the paper suggests to me that it's basically a framework based on graph theory in which nodes are social entities and branches are relationships. This is a pretty common framework through the human knowledge pool, so there're a lot of relationships that we can draw to other existing theories. These relationships can be useful since many of the tools developed for one framework can be applied across frameworks.

Some of the more obvious relationships include:

1. Steady state.
   There's no explicit time dependence.
   
   • May be asserted to be pseudo-steady state.
   • May be arrived at through a more rigorous approach by asserting a kinetic model and driving time .
     • A good example of this generalization is in how electric circuits are taught. Generally electric circuits are introduced as steady state graphs, then generalized to time dependence.
     • Chemical system kinetics also follows this path.
2. Graphically first-order.
   Directional relationships are functions of single nodes.
   
   • May be arrived at by asserting a graphically higher-order framework, then limiting scope to first-order relationships.
     See also:
3. Logically first-order.
   The underlying logic system does not build itself.
   • Knowledge about first-order logic can be applied, including first-order logic for graphs.
   • May be arrived at from a more general framework by removing higher-order principles.
4. Non-stochastic.
   No randomness.
   
   • May be arrived at from a more general framework by assuming static values for all points of randomness.

It's pretty common when working in frameworks like this to keep using them when they're sufficient, then to extend them as necessary to account for phenomena that can't be fit into the basic framework itself. I'm sure that you're well aware of this, so I'd be curious, what sorts of generalizations are commonly applied to this framework in modern literature? For example, has there been a movement toward considering stuff like:

• kinetics; or
• stochastics; or
• higher-order relationships?

For applications, this framework's neat in that it can form a theoretical underpinning for welfare economics. For example, if sells some product in a free market, then obviously both sides perceive benefit from the transaction (or else wouldn't conduct it). But if the product is of very low value to and of very high value to , then the transaction can happen at a wide range of prices while still being agreeable to both parties; so how does that price get determined? As far as I've seen in my MBA studies, this isn't a well-answered question in the context of microeconomics (though consistent-pricing constraints help address it in macroeconomics), though I can see how your consideration of networks could go toward coming up with a consistent, predictive framework for answering it.

Noting generalizability should help answer this question since, at the macro-limit, the micro-framework should reduce to the known solutions of profit optimization that tend to work pretty well when we're talking commodities. This would be a consistency principle like how quantum mechanics reduces to classical mechanics at the macro-limit.

[Ursa Minimus]

Ursa Minimus,

I'd like to thank you for showing a good example of a Sociology paper. This does look like one of the better works in the field, and since I feel that fields should generally be viewed in terms of their merits, this paper is a great look into Sociology.

A quick peek at the paper suggests to me that it's basically a framework based on graph theory in which nodes are social entities and branches are relationships. This is a pretty common framework through the human knowledge pool, so there're a lot of relationships that we can draw to other existing theories. These relationships can be useful since many of the tools developed for one framework can be applied across frameworks.

Some of the more obvious relationships include:

1. Steady state.
   There's no explicit time dependence.
   
   • May be asserted to be pseudo-steady state.
   • May be arrived at through a more rigorous approach by asserting a kinetic model and driving time .
     • A good example of this generalization is in how electric circuits are taught. Generally electric circuits are introduced as steady state graphs, then generalized to time dependence.
     • Chemical system kinetics also follows this path.
2. Graphically first-order.
   Directional relationships are functions of single nodes.
   
   • May be arrived at by asserting a graphically higher-order framework, then limiting scope to first-order relationships.
     See also:
3. Logically first-order.
   The underlying logic system does not build itself.
   • Knowledge about first-order logic can be applied, including first-order logic for graphs.
   • May be arrived at from a more general framework by removing higher-order principles.
4. Non-stochastic.
   No randomness.
   
   • May be arrived at from a more general framework by assuming static values for all points of randomness.

It's pretty common when working in frameworks like this to keep using them when they're sufficient, then to extend them as necessary to account for phenomena that can't be fit into the basic framework itself. I'm sure that you're well aware of this, so I'd be curious, what sorts of generalizations are commonly applied to this framework in modern literature? For example, has there been a movement toward considering stuff like:

• kinetics; or
• stochastics; or
• higher-order relationships?

For applications, this framework's neat in that it can form a theoretical underpinning for welfare economics. For example, if sells some product in a free market, then obviously both sides perceive benefit from the transaction (or else wouldn't conduct it). But if the product is of very low value to and of very high value to , then the transaction can happen at a wide range of prices while still being agreeable to both parties; so how does that price get determined? As far as I've seen in my MBA studies, this isn't a well-answered question in the context of microeconomics (though consistent-pricing constraints help address it in macroeconomics), though I can see how your consideration of networks could go toward coming up with a consistent, predictive framework for answering it.

Noting generalizability should help answer this question since, at the macro-limit, the micro-framework should reduce to the known solutions of profit optimization that tend to work pretty well when we're talking commodities. This would be a consistency principle like how quantum mechanics reduces to classical mechanics at the macro-limit.

Natural ChemE,

I am glad you found it good, at least on the surface. It is a graph theoretical approach, which is one of the reasons I used it. Very general approach to many problems in many fields.

Generally speaking, the top journals in sociology (american journal of sociology, american sociological review, social forces, social psychology quarterly would be examples) do not publish things that do not critically test theory, or test multiple theories against one another. Theory = model. No matter how good the research, if there are not theory implications, they go to lower level journals. So you see this type of approach often in those journals. But rarely is anything in those journals starting at the beginning of a model's development, as in the exemplar case.

I don't have much time right now, to the point of having to do a 7am meeting today, but briefly:

The application of a network exchange approach to economics does seem like it makes sense. It would suffer from the problem of all economics, in that it will work to the extent that you can define value. Econ likes to say that if you can tell them what people value, they will tell you how people will behave. But specifying value hierarchies is a tough row to hoe. I do know that in the past people I have told about this theory in business often go right to economic models as an application. (I used to be on a floor of a building with the business departments on campus, so lots of cross talk).

Another implication for MBAs, especially if the theory is specified in a bit more detail. The theory implies that one can affect exchange structures by forming new network connections distant from other connections. It is obvious that if you deal with a partner, you affect them. Or a partner to that partner. But what if you pick the right partner of a partner of a partner of a partner and deal with them? It might change the resource flows in interesting ways. That might apply to trade, or to friendship groups, or any other group of individuals or organizations that can be seen in network exchange terms. This might be seen as speaking to the concept of "solidarity", which is a good general social science concept that goes across many fields.

Have people in this area used other types of models? (Kinetic, etc.) I have no idea, and would have to look at a lot to find out. Probably.

But of those types listed, I would say kinetics... no. I can't find much of anything on a search of "kinetic model" in the sociology databases. In any area of research.

So of course I thought about it. I do think I can come up with a kinetic model for a theory of social movements (resource mobilization theory), and that model might apply to things like "projects" in the abstract, which would be of more interest to those with MBAs.

If you want to play around with a kinetic model to explain "projects" that can account for success and failure, as well as speed of success or failure, I am sure you could come up with something similar to what I am thinking of. Testing such a model, that would be a huge undertaking methodologically. But theoretically, I can at least see ways to try that are not that complicated.

5 minute version. Assume resources impact the project and impart momentum. There will be some impulse function for different types of resources (people joining versus extra funding might act differently). Some impacts will add momentum, some will reduce it, depending on vector. We can also talk about the "friction" and what factors raise or lower that. So, does the project fit with, or go against, organizational culture? With, smooth, against, rough. We could assume different "terrain": downhill, flat, uphill, a series of hills, etc. And other factors might work on the terrain. So, maybe in business, "regulations" might make a project more or less uphill.

Something like that. Is that the type of kinetic model you might have been thinking about?

[Natural ChemE]

Ursa Minimus,

Back when we were talking about reductionism, an important point for me was that we're all basically computers. We can make predictive models/theories only because the phenomena that we're describing are approximately reproducible within our own brains.

For example, we can add numbers because our brains are somehow capable of constructing a structure (thought) that represents the sum of other structures (the numbers being added). If this weren't true, then we necessarily couldn't add numbers in our brains.

Tangential:
It's conceivable that our brains could spontaneously form a thought process that somehow correlates to external phenomena without a known mapping method. In fact, due to the wide variety of possible mappings, this is probably moreso the rule than an exception. However, in the absence of known mappings - whether that knowledge is explicit or not - the model is useless.

Then using methods like counting, we're able to predict things like, "If I count apples in a basket and add more apples, then I will count apples in that basket." This holds true exactly because apples in a basket persist additively in the exact same sense that the thoughts in our brain representing them do. Or at least over the scope of applicability; if other factors are in play, then we need to consider them. And if apples start suddenly disappearing one day, then obviously our thoughts don't correlate to reality as well as we'd like.

Theorists then have to balance three factors:

1. Accuracy: How well their model corresponds to reality.
2. Generality: How much of reality is described by their model.
3. Computability: How hard their model is to construct/employ.

Good theorists try to achieve reasonable accuracy over as general a scope as possible while maintaining reasonable computability. And since perfect prediction of all reality is an unobtainable ideal, we generally have to apply a large succession of reductions,

1. approximations: reduce accuracy;
2. scope reductions: reduce generality;

until some viable (computable) model is derived.

This process of theorizing has dramatically shifted in very recent history due to the changing nature of computability. This is, we're now able to compute much, much more than before due to the astronomical improvements in electronic computational methods. Old-fashion models should be rederived, avoiding approximations and scope reductions that used to be necessary whenever possible. Practically speaking, this means that stuff like linear regressions and hand-calculable correlations go out the window unless they're already super-accurate.

Personally I'm a big fan of retro-engineering the theory-derivation process. I try to figure out how theories were formed, then relax approximations given all sorts of computational approaches afforded by very recent innovations in electronic computation. So when I see a paper like this, I think,

Gee, I really understand why the author put together this map of connections. They were trying to make something that was complex enough to make some predictions while removing anything that wasn't absolutely necessary to maintain some level of predictability.

However, I'm a computer guy! I've got all of these awesome computers, so I'm just gonna redo their work with fewer approximations 'cause my computers will handle the complexity. This is obviously the best way to expand on prior work.. why isn't everyone doing this more?

I guess that folks must not know about these awesome tools that they can use instead of gross approximations. I guess that this makes sense, too, since they're changing so fast and not everyone likes computers as much as I do. Time to spread awareness on teh interwebs!

So, I guess exactly what I'd do with models like this would depend on what phenomena I'm trying to predict. And once I decide what phenomena I want to predict, I'd start with a general world view, then start making reductions/approximations.

For example, I'd ask myself if I really need to consider how the phenomena happens over time; I'd really like to avoid removing time from the model unless I can be reasonably sure that the real-world phenomena is reasonably reproducible as a timeless thing. And since computers exist, I don't particularly care if that makes this model more complex; I'd rather maximize accuracy within computational limits.

This is pretty much how I see the process of theorizing and how I'd like to see existing theories updated.

[Natural ChemE]

I need to take about a week off the forums, so just one last thing before I forget to mention it.

Artificial neural networks (ANN's) are great examples of three relevant topics in this thread:

1. They're graphs like the 1988 paper, though they can be used in a self-updating scheme to arbitrarily high order. I'd note that self-updating would mean using reflective programming to modify themselves, not merely the normal optimization process used to train ANN's. This is a contrast against the logically-first-order qualification I'd noted in my first reply.
2. ANN's are derived in an explicitly reductionist manner. The name "artificial neural network" clearly reflects that they're meant to be neurons in a computational sense in an attempt to mimic brains. Should they ultimately prove to be unable to reproduce human intelligence, then some simplifying aspect of their derivation can be relaxed. An extreme example of such a relaxation would be rejecting the position that thought is a computational process conductible on computer chips, leading future ANN researchers to instead grow biologically-derived brains in Petri dishes (which, admittedly, would be gross). However, I'd note that all current signs indicate this isn't a problem; ANN's are pretty much just limited by current technology, but that's a problem that's resolving itself.
   Tangential:
   Modern approaches to AI are more general than ANN's. While ANN's could suffice if we had better computers, more general approaches resolve this weakness.
3. ANN's are enabled by computational tools that have only been recently created and are continuously improving. As these tools improve, the scope of their application as well as their general desirability increases.

[Ursa Minimus]

Ursa Minimus,

Back when we were talking about reductionism, an important point for me was that we're all basically computers. We can make predictive models/theories only because the phenomena that we're describing are approximately reproducible within our own brains.

A very psychological approach. But keep in mind sociology does not care about individuals, but rather the effects of social structures on individuals. Still, let's take this idea and see if it breaks down when we start thinking beyond a single individual.

Computers require languages, and languages are used to create algorithms.

Some algorithms are harder to write in a given language, others are easier.

Are language and algorithms bootstrapped from the "computer"? Nope, they come from outside. They are a social fact, in that they are external to, and coercive of the individual.

Human language has all sorts of issues that computer languages do not. Meaning shifts, it is not fixed. What shifts meaning? Context, which is a social factor. Meaning is intersubjective, basically, in human interaction.

Take a simple word for something in the world. How about what "watches and clocks" do as they move. In English, they are said to run. In Spanish, they walk. Do you think this shows cultural differences in how time is treated? Well, sure. I may go into detail, mañana. For now though, I have work to do, and time is money after all. Not something to waste, as it rushes by.

(I hope what I did just there is clear.)

You know what Germans say? Clocks "function".

Rechtsschutzversicherungsgesellschaften is a word used in the normal course of events in Germany. It strings together other words, and means "insurance companies providing legal protection." The tendency of German speakers to be literal shows up a lot in the language itself. Which might be one reason why Germans have a reputation for a lack of humor. The "switch at the end" that makes much of humor result in a laugh is hard to construct in a literal language. Don't grow up making such switches, and the literal meanings will stop the humor from being "gotten".

So clearly, the very language people use, which they get from the world they live in from birth, alters how they look at, and treat, time. Or anything else for that matter.

And what about algorithms? Where do people as computers get their algorithms? You have calculated a path through life, and made decisions to achieve your goals. But those decisions are conditioned by a great deal that you have learned from others. And their advice to you has been conditioned by the world in which they live.

So, if you were born into the Sudan, would you have the same algorithms? Nope. If you were born into a highly individualistic culture, would your algorithms be the same as if you were born into a more collectivist culture? Nope.

Sociology is built on the assumption that there are causal processes that are not reducible to individual psychology, or biology. Sociological theory tries to identify such causal processes, and show how they have effects over and above individual level variables. The individual is a control variable, in other words, in sociological research. Even things that seem individual, like gender, are not individual level. Gender refers to "cultural assumptions about sex differences". And those assumptions, and the social conditions that contain them, do seem to have causal effect. In societies with more gender inequality, women speak in higher pitched tones, and men prefer such tones. In societies with more gender equality, women's tones are lower, and men do not find that unattractive.

Biology does not explain that, psychology does not explain that very well, but sociology does. "Societal inequality" explains many things, across many different societies. It is a useful concept in terms of causality, and measures such as the GINI coefficient measure it pretty well for making cross cultural comparisons. https://en.wikipedia.org/wiki/Gini_coefficient

So, socialization is the process of society getting in your head. There are theories in sociology that try to explain identity (not just control for identities), and they put the causal locus not in the individual, but in the social forces that shape them. The individual mind is a version of society carried around in someone's head, and we use that model to act in the larger world. Different societies literally produce different minds from this perspective. Sociologists model the individual as a product of society, not the other way around.

Because that is what sociology is all about.

[BioWizard]

Hey Ursa,

I have some questions, if you don't mind. I'm typing from my phone again, so I apologize for typos.

I understand everything you said about the interplay between society/culture and the individual mind. In fact, I would say no biological system develops in isolation from its environment. The genetic code only sets the limits on what biology can produce, but not the exact outcome. Thus reducing the outcome requires that the interplay between the organism and the environment be considered. If you grow up doing lots of manual labor, your muscles will adapt, hypertrophy, and store more glycogen. The same genes in you would've produced a less lean individual in the abscence of much physical work. Your brain biology will set the limits on your mental development (for example you'll never be able to beat computers, or certain individuals, at certain tasks), but the exact mind you end up developing will be an interplay between your brain and the interactions with its environment, including both perceptual interaction as well as biochemical interactions (drugs will impair you, toxins may affect your mind irreversibly, good nutrition may improve its functions, etc). When we talk about reducibility there, we don't necessarily mean that the outcome needs to be entirely reducible to the state of the organism at t=0, in isolation from everything else around it throughout the course of its temporal evolution. After all, a system needs to be reduced to its components, and not considering the environmental component is missing parts of the system. If you're trying to explain why only one of two identical twins developed heart disease in his 20s, first things you might consider are differences in diet, exercise habits, life style, and so on. If you find that it was caused by excessive cosumption of steroids by the afflicted twin, due to running into the wrong crowds at the gym, would that mean that you failed to reduce the etiology of the disease? Because the steriods weren't inside the body of the twin when he was born?

All that said, why is the interplay between the mind and society an argument for the irreduciblity of the system? Maybe our alternate views are caused by what each of us is considering as the complete system under consideration?

As an example of extreme simplicity, you can't explain the behavior of hydrogen atoms in water without incorporating the properties of oxygen, just as you can't reduce the behavior of salt ions in solution without the solvent (salts behave very differently in vacuum). So why would we be able to reduce the social behavior of humans without the social milieu? And why does that necessarily mean that social behavior is irreducible? Or that the social milieu has no underlying structure with certain rules that govern its evolution? I personally see sociology as a critical component of a possible reductionist approach down the line rather than being at odds with it, since it tells us what we may need to know about the "milieu".

[BioWizard]

I should elaborate that, by extension of what I said above, I do agree that sociology is unlikely to be entirely reducible to human biology. Just like it's not possible to reduce the structure of a beaver dam entirely to beaver biology. The beaver is but one component in the system that gives rise to, and erodes, the dam (which I suppose could also be modeled kinetically, while paying heed to the dynamic interaction between the beaver and its dam).

My thought here is that:



and



By the way, I make those comments and ask those questions specifically because you've said that you believe in the utility of a scientific approach to this kind of problems. So I'm curious to know more about how all these things fit together from your perspective.

Also, really cute new avatar.

[Ursa Minimus]

All that said, why is the interplay between the mind and society an argument for the irreduciblity of the system? Maybe our alternate views are caused by what each of us is considering as the complete system under consideration?

As an example of extreme simplicity, you can't explain the behavior of hydrogen atoms in water without incorporating the properties of oxygen, just as you can't reduce the behavior of salt ions in solution without the solvent (salts behave very differently in vacuum). So why would we be able to reduce the social behavior of humans without the social milieu? And why does that necessarily mean that social behavior is irreducible? Or that the social milieu has no underlying structure with certain rules that govern its evolution? I personally see sociology as a critical component of a possible reductionist approach down the line rather than being at odds with it, since it tells us what we may need to know about the "milieu".

Bio,

I think the sociological view of identity, self, and the mind is a topic that is not easy to talk about with those not in the know. It's weird, and it tends to use psychological terms in non-psychological ways. IOW, it has a lot of domain specific knowledge that does not seem domain specific. But in brief:

Why look at a computer running software when you can look at the source code installed on many computers? Does the specific architecture of a PC matter? Sure, sometimes. But the source code is the big thing to consider.

From this perspective, culture offers a great deal of the "source". So, if we consider that English has a lot of "black" words that are negative (and have little or even no historical connection to race) and "white" words that are positive (again, unconnected to race largely)... as in this classic essay:

http://www.femi.pwp.blueyonder.co.uk/race.htm

Language and Culture

An integral part of any culture is its language. Language not only develops in conjunction with a society's historical, economic and political evolution; it also reflects that society's attitudes and thinking. ...

...A Short Play on "Black" and "White" Words

Some may blackly (angrily) accuse me of trying to blacken (defame) the English language, to give it a black eye(a mark of shame) by writing such black words (hostile). They may denigrate (to cast aspersions; to darken) me by accusing me of being blackhearted (malevolent), of having a black outlook (pessimistic, dismal) on life, of being a blackguard (scoundrel)‑which would certainly be a black mark (detrimental fact) against me. Some may black‑brow (scowl at) me and hope that a black cat crosses in front of me because of this black deed. I may become a black sheep (one who causes shame or embarrassment because of deviation from the accepted standards), who will be blackballed (ostracized) by being placed on a blacklist (list of undesirables) in an attempt to blackmail (to force or coerce into a particular action) me to retract my words. But attempts to blackjack (to compel by threat) me will have a Chinaman's chance of success, for I am not a yellow‑bellied Indian‑giver of words, who will whitewash (cover Lip or gloss over vices or crimes) a black lie (harmful, inexcusable). 1 challenge the purity and innocence (white) of the English language. I don't see things in black and white (entirely bad or entirely good) terns, for I am a white man (marked by upright firmness) if there ever was one. However, it would be a black clay when I would not "call a spade a spade," even though some will suggest a white man calling the English language racist is like the pot calling the kettle black. While many may he niggardly (grudging, scanty) in their support, others will be honest and decent‑and to them I say, that's very white of you (honest, decent).

The preceding is of course a white lie (not intended to cause harm), meant only to illustrate some examples of racist terminology in the English language.

By taking such meanings deeply embedded in the English language as the starting point for thoughts of individuals, some research results are not at all surprising. Such as:

http://goizueta.emory.edu/profiles/docu ... P_2015.pdf

Racial labels often define how social groups are perceived. The current research utilized both archival and experimental methods to explore the consequences of the “Black” vs. “African-American” racial labels on Whites' evaluations of racial minorities. We argue that the racial label Black evokes a mental representation of a person with lower socioeconomic status than the racial label African-American, and that Whites will react more negatively toward Blacks (vs. African-Americans). In Study 1, we show that the stereotype content for Blacks (vs. African-Americans) is lower in status, positivity, competence, and warmth. In Study 2, Whites view a target as lower status when he is identified as Black vs. African-American. In Study 3, we demonstrate that the use of the label Black vs. African-American in a US Newspaper crime report article is associated with a negative emotional tone in that respective article. Finally, in Study 4, we show that Whites view a criminal suspect more negatively when he is identified as Black vs. African-American. The results establish how racial labels can have material consequences for a group.

Let me offer a non-mind based example. Bureaucracy. Would you say bureaucracies are reducible to the individuals in the bureaucracy? In the sense of "people do stuff", sure you can make that case. In terms of exerting causal effects? Nope. Which you feel every time you start to fill out forms, before you ever submit them, or ask a person about how to fill them out and submit them.

[Ursa Minimus]

I should elaborate that, by extension of what I said above, I do agree that sociology is unlikely to be entirely reducible to human biology. Just like it's not possible to reduce the structure of a beaver dam entirely to beaver biology. The beaver is but one component in the system that gives rise to, and erodes, the dam (which I suppose could also be modeled kinetically, while paying heed to the dynamic interaction between the beaver and its dam).

My thought here is that:



and



By the way, I make those comments and ask those questions specifically because you've said that you believe in the utility of a scientific approach to this kind of problems. So I'm curious to know more about how all these things fit together from your perspective.

Also, really cute new avatar.

Bio,

Sociology was founded on the assumption that social effects are not reducible to psychological/biological effects (your first sense). That's core to the discipline, and so sociologists look to find such effects. And look to show how they are not reducible to psychological/biological processes, often by using psych/bio factors as control variables. A good amount of soc looks to explain what SEEMS to be reducible on the surface to show it really isn't (showing what is claimed to be your second sense really is the first sense).

However, there are approaches in sociology that are not oriented towards giving total primacy to social structures in causality. The "biopsychosocial" view takes as a starting point that we must consider biological, psychological AND sociological processes to explain humans doing anything. This makes sense as an approach, theoretically, and has an advantage methodologically. If we can explain 1/3 of the variance in dependent variables through social processes, and 1/3rd from psych, and 1/3rd from biology.... then we have a societal TOE.

RE: new avatar. I thought a less "this bear will bite your head off" image might be a good idea. :D

[BioWizard]

Bio,

Sociology was founded on the assumption that social effects are not reducible to psychological/biological effects (your first sense). That's core to the discipline, and so sociologists look to find such effects. And look to show how they are not reducible to psychological/biological processes, often by using psych/bio factors as control variables. A good amount of soc looks to explain what SEEMS to be reducible on the surface to show it really isn't (showing what is claimed to be your second sense really is the first sense).

Ursa,

Do you think that the field of sociology - given the development of social constructionism (on one end), behavioral neuroscience (on the other end?), and psychobiology / evolutionary psychology (somewhere in between?) - will continue to hold this unidirectional core position indefinitely? What about your own personal opinion on the matter? I know you said that there are approaches that seem to be inching away from giving total primacy to social structures. Do you notice a trend of them becoming more or less frequent over time?