What philosophy buffs have against Science buffs

[Kriminal99]

That kind of ends the discussion don't you think? :shrug:

 

 

 

That would be an argumentum ad populum. It proves it's a data point, not that it's relevant data.

 

EXACTLY. What is the perspective you bring to the question which makes you look at this dimension? A typology reflects the biases of it's creator.

 

Let's stay with buttons, and get off the racial issue for now.

 

We have five buttons. When I organize it, I will begin my typology with the dimension of button-ness which I believe is most likely to answer whatever question I have about buttons. (Is there a relationship between number of holes and material?) First I divide it into shell buttons and plastic buttons. I now have three shell and two plastic buttons. I notice that each of the plastic buttons have four holes, while the shell buttons have two.

 

I draw the conclusion that the material strength of plastic is greater than shell, and therefore allows more holes. (I don't know if this is true without a test)

 

We go back to five buttons now. First I divide the buttons by color. The red buttons from the white buttons. I notice that the red buttons have four holes and the white buttons have two.

 

I conclude that the red blouse had four hole buttons and the white blouse had two hole buttons. (I don't know if this is true without a test.)

 

Back to five buttons. This time we divide by the number of holes. I have three with four holes and two with two holes. I conclude that it is harder to paint buttons with two holes. (I don't know if this is true without a test.)

 

Okay, we've now established how when using typologies to draw conclusions my own ideas influence the conclusions I will draw. The same holds true with race based typologies.

 

I am not arguing that statistical racial differences do not exist, or that there is no measurable difference between "black people" and "white people" (Although, it's a debatable point - I doubt you could draw that as accurately as you imagine.)

 

When you BEGIN with the division of race, your typology will necessarily be correlated to race.

 

If you BEGIN with the division of neighborhood, your typology will necessarily be correlated to neighborhood. The typology with which you begin informs the results.

 

In all cases, your typology must be tested from the outside. In all cases, a four hole shell button, or a red button with two holes, or a white plastic button pokes a hole in the typology.

 

To reiterate - the construction of a correlating typology does not mean the differences are intrinsic.

 

TFS

 

No it isn't an argument ad populum. I never said it was relevant data to a third party. What I said is that every person is capable of gathering data for themselves and does so and that this shows that correlations between superifically defining characteristics of race and other traits are REPRODUCABLE. Its up to each person to make reasonable conclusions from the data they personally collect, but there are plenty of such conclusions to be made.

 

Careful reasoning would never permit the drawing of such conclusions as you have laid out. I said MAYBE the buttons with more holes are made with a stronger material. If I had other information to point to that fact only then would I draw such a conclusion. Not necessarily scientifically tested, but more information.

 

The point here is that a correlation is a correlation both before and after your understanding the connection between the two. Correlations do not happen out of sheer coincidence. In the button example it could just be that a certain button maker both likes to use a certain material and have buttons with a certain number of holes and this could explain the correlation. This information would allow you to then draw the conclusion that a certain button was made by that person or not. You do not say that the correlation is meaningless because you are afraid of what the reason for the correlation MIGHT be.

 

So

 

A) If most shell buttons you see have 2 holes, then you would be better off guessing that a button made out of shell has 2 holes than not - unless you have reason to believe that the population you sampled from is signifigantly different then what you are trying to predict (for example all previous shell buttons were made by Person A, and then you here that Person B also makes shell buttons and you still don't know if its the strength of shell or the style of manufacturer that is responsible for the hole correlation)

 

:) If your other experience (of any kind as long as you reason carefully in gathering it) gives you reason to believe that shell is a weaker material in the very sense that would make shell buttons with many holes problematic then there would be no problem in connecting the two...

[TheFaithfulStone]

You're not getting it Krim, although I agree with Q that we're off topic.

 

(and I had to look up prolix - nice one!)

 

THE POINT.

 

If you begin your division of people by race, then correlations with race will emerge. This is true of ALL divisions. If you begin your division of people by hair color, or age, or ANYTHING, correlations with that factor will emerge.

 

Correlation does not equal causation! Which variable is independent? If we didn't know from outside experience, we could take the correlation between poverty and race and say that being poor causes an increase in melanin.

 

Again - "race" is a culturally dictated typology. The typology that we construct around "race" tells us certain things about people, yes.

 

For a biologist, who wants to answer different questions, the typology doesn't provide meaningful answers.

 

The begin state of your typological division will inform the conclusions you draw.

 

I question whether we're even talking about the same thing any more.

 

Simply put, as long as you're looking for racial differences, you'll find them.

 

TFS

[Kriminal99]

No we are not off topic, but nice escape attempt.

 

Correlation is more signifigant than causation

 

Correlations will appear when you start with race because things ARE correlated with race. To say race is meaningless is to say that race is 0% correlated with everything. That a person of a given race is equal likely to have or not have any trait. Correlation is plenty meaningful in of itself.

 

How is it that you think the human mind even defines causality? How about two events that are highly correlated and one happens after the other? If 99% of people with dark skin absorb sunlight better than people with light skin and 85% of african americans have dark skin then .85 *.99 african americans absorb sunligh better. If 100% of biscuit eaters smack roaches on sight and 75% of georgians eat biscuits then 75 % of georgians smack roaches. Maybe you would feel comfortable saying the dark skin causes better absorbtion of sunlight but that biscuit eating doesn't cause roach smacking. Causation is not a relevant issue to the roach who sees a georgian coming. (Pretending the roach had a sense that measured incoming people as georgian or not such that this correlation was more relevant to him then any correlation between roach smacking and other percievable attributes)

 

Why worry about causation at all?

 

The only signifigance of talking about causality has to do with predicting correlations when you can't simply measure or accepting that correlations you have already experienced are not some kind of fluke that will dissapear along with some 3rd factor that was consistent amonst the population for which the correlation was measured. For these two cases you just have to reason carefully. This whole topic can be put in terms of the generality problem of induction. If the roach sees an incoming georgian wearing a shirt with a giant picture of a cockroach, he might consider this as the most relevant indicator instead.

 

The person originally addressed by me was making different arguments

 

One thing that has been repeatedly ignored is that the person whom I was making an example of did nothing remotely resembling what any of you are claiming as far as claiming race was meaningless in a biology context (even if he was, he would still be wrong but thats not what he was doing). He was arguing for the abolishment of race as a social concept because he was claiming it had been disproved by science. And his tactics for attempting this were no different than tactics many scientists use to support less absurd but no less inaccurate claims.

[TheFaithfulStone]

Why shouldn't we abolish race as a social concept? The only reason we find racial differences is because we have "race" as a top tier division on our typology of "people."

 

If you divided people into "smart and dumb" and then "rich and poor" and then "long hair and short hair" you would end up with an entirely different distribution than if you were to divide them "black and white", "educated or not","tans fast"

 

As for ignoring causation in favor of correlation? I really don't know how to respond to that. If we do that, we get to a scary place, really, really quickly. What causes a higher incidence of poverty among African Americans? What causes more out of wedlock births? You aren't going to solve a problem or come up with answers if all you know is correlations and probabilities.

 

That's just... illogical - to not even worry about cause and effect.

 

TFS

[Kriminal99]

If you are trying to define race as only the color of someone's skin (in america) as opposed to that skin color and all correlated traits, the answer is obvious (especially to any average person rather than a "scientist") and I already touched on it in the cockroach example.

 

Most easily obtained trait

 

We have a bunch of factors that are correlated with each other and one of them is skin color. Skin color is the least difficult to percieve, the most obvious. If we were intelligent bats and a certain grouping "sounded different" when hit with sonar than others, and that group had a number of other correlated factors we would use that grouping instead. And if say this medium allowed us to most easily recognize differences in bone structure from person to person, then we might still end up with mostly the same members in our grouping as we do now with african americans/africans. There might be a few people that would be in different groupings depending on our main indicator. But most likely we would have the same group with all the same correlated factors.

 

All groupings that can be obtained and have correlated factors are useful

 

But even if people were split up completely differently from how they were now, if there were correlated traits to whatever means we used to split them up then the grouping would still be meaningful.

 

The fact that there are any correlated traits at all to whatever we use to group is what makes that grouping meaningful, and how easily we can recognize members of that group is why we choose that grouping over another.

 

Best indicator for what we are trying to determine that we can afford

 

If we were capable of directly percieving traits like brain structure or bank account size then we would group by those instead. In fact in every day life when we do feel we are able to recognize those factors we use them to better answer related questions. To use skin color instead when we have access to that information is what I consider to be the type of racism that is wrong.

 

On the other hand if say if you cannot acertain that information on everyone due to cost limitations, and you want to spend the money to acertain such info only on the most likely candidates, guess which correlated factor you are going to use? The most easily obtained... or at least the most effective of the most easily obtained.

 

Causation - Beyond human comprehension

 

To start off with I will reiterate that correlation is simply a more general form of causation and therefore causation is not seperately nescessary. What makes someone say A caused B? When A is highly correlated to B happening subsequently.

 

Want to try and use a concept of causation that can be seperated from the human mind? Ok, you asked for it. When you ask what caused something the REAL answer is beyond your comprehension. Unless of course it is simply understood that when you ask it, you really mean what part of the cause can you change to alter the outcome.

 

If you kick a ball, it flies through the air. But your boot isn't the only link in the chain that caused the flying ball. There's the laws of physics, the design of the ball etc. But if your a little kid that doesn't know any better and never experienced differences in those factors all you know is if you kick the ball it flies through the air. If you try to wack the ball with a samurai sword, then you might learn that the object you collide with it is a signifigant part of the cause.

 

If you are a physicist you know there are tons of things you can do to alter what happens to the ball. Changing weights, air pressure, materials etc etc.

Beyond that there are probably things that a physicist of tommorow could change to alter the outcome that the physicist of today could not. Do we worry about these unkown factors when answering the question of cause? How could we if we don't know whether or not factor x is a signifigant part of the "cause" or even what factor x is?

 

So even if you change the definition of causation to "what can you change to change the outcome?" The answer is still more than any answer you will ever give or recieve to such a question.

 

The concept is flawed and has no place in careful analysis. Better to just collect correlations and when you happen to come across a controllable factor A that is highly correlated with an unwanted outcome when all else remains the same just alter the factor A to acheive what you want.

[TheFaithfulStone]

The concept is flawed and has no place in careful analysis.

 

:)

 

Your argument against cause and effect is thinly disguised sophistry. Something CAUSES the ball to move. If you alter the cause you alter the effect. Whether it's your boot, or the ball, or whatever, the ball isn't going to move on it's own. Whether it's the atoms in your boot, or whatever, the action is always repeatable under the same conditions.

 

I fail to see how you can be either a philosopher OR a scientist if you do not accept cause and effect. It's the basic axiom of logic.

 

You haven't done much to disprove causality either, you've just said "It's beyond our ability to comprehend."

 

Let me give you a fer'instance.

 

Being being African American is highly correlated with born out of wedlock. source (actually vice versa

 

So, if we don't examine what the root cause of this, we can solve the problem by eliminating the correlation!

 

Eliminate black people, eliminate out of wedlock births! Voila, problem solved.

 

It is beyond obvious that race has nothing, nothing, nothing to do out of wedlock births. But they are highly correlated.

 

You ignore cause and effect at your own peril.

 

There might be a few people that would be in different groupings depending on our main indicator.

 

What if your indicator was voice timbre? What if your indicator was a variable that is entirely independent of race? Your divisions would shake out entirely different wouldn't they?

 

TFS

[Qfwfq]

There is hardly a point in continuing so much discussion of race, in this manner.

 

It is really a matter of saying that race is a term for taxonomic groupings within a same species, which is itself a rather arbitrary thing. There has always been disagreement about this type of classification within homo sapiens and today differences are hardly recognized to justify divisions into race, only into ethnic groups. Historically, classifications have been proposed with as few as three (Cuvier) and as many as eleven (Pickering) races, while there are many more ethnic groups. One of the most recognized groupings was that of Blumenbach, who describes five categories: Caucasian (white), Mongolian (yellow), Ethiopian (negro), American (red) and Malayan (brown).

 

There are however far less differences between these than there are between different breeds of dogs, horses or oxen. Animal races may arise artificially (due to breeders) or in nature due to geographic separation for a great number of generations, this hasn't so clearly happened with our species. There is, however, ongoing debate as to whether or not neanderthalis was really a separate species, some hope to read its genome to enable a comparison and work out the compatibility. It is possible that we could have been either races of the same species or species similar enough to give sterile offspring.

 

P. S. In this case, if we weren't different species we certainly were different races!

[TheFaithfulStone]

It is really a matter of saying that race is a term for taxonomic groupings within a same species, which is itself a rather arbitrary thing.

 

EXACTLY.

 

Taxonomy itself is kind of arbitrary.

 

Aristotle classified animals on how they traveled. Linnaeus, who invented the binomial nomenclature classified them on morphological similarity. (I think) Now, species taxa are based on common descent.

 

Which one of these is the "right" way to do it? They are all logical, valid divisions.

 

Classifying organism by common descent won't tell you whether they swim or fly at a glance, which won't tell you about morphological similarities, which won't tell you about common descent.

 

When making taxonomy or typologies the divisions you make at the beginning inform the things you find out at the end.

 

The classifications of things are not (necessarily) intrinsic in those things. Classification of people by race is an extrinsic category.

 

S'all I'm saying.

 

TFS

[Kriminal99]

:P

 

Your argument against cause and effect is thinly disguised sophistry. Something CAUSES the ball to move. If you alter the cause you alter the effect. Whether it's your boot, or the ball, or whatever, the ball isn't going to move on it's own. Whether it's the atoms in your boot, or whatever, the action is always repeatable under the same conditions.

 

I fail to see how you can be either a philosopher OR a scientist if you do not accept cause and effect. It's the basic axiom of logic.

 

You haven't done much to disprove causality either, you've just said "It's beyond our ability to comprehend."

 

Let me give you a fer'instance.

 

Being being African American is highly correlated with born out of wedlock. source (actually vice versa

 

So, if we don't examine what the root cause of this, we can solve the problem by eliminating the correlation!

 

Eliminate black people, eliminate out of wedlock births! Voila, problem solved.

 

It is beyond obvious that race has nothing, nothing, nothing to do out of wedlock births. But they are highly correlated.

 

You ignore cause and effect at your own peril.

 

 

 

What if your indicator was voice timbre? What if your indicator was a variable that is entirely independent of race? Your divisions would shake out entirely different wouldn't they?

 

TFS

 

I did plenty to address the concept of cause and effect you just didn't read or respond to it. BTW no it is not a basic axiom of logic. Logic is deduction and induction neither of which involve cause and effect.

 

Cause and effect as a concept is a proper subset of the concept of correlation

 

Cause and effect can be modeled using only correlation, as the concept of cause and effect is realized in the human mind by recognizing correlations between events and subsequent events

 

When someone talks about cause and effect, what do they really mean?

Many, perhaps infinite, concurrent factors are correlated with a given event subsequently occuring. (The concurrent factors being what you attempt to refer to as cause) No one ever answers the question "What caused X?" by referring to all these concurrent factors which include factors they are not even aware of. Instead the question "What caused X? is more like asking "Which of the potentially infinite concurrent factors can we alter to change the outcome?" or maybe just "Which of these factors can we percieve?"

 

Now to use your example...

 

Well you COULD reduce out of wedlock births by eliminating a race correlated with them...

 

Yes, if most out of wedlock births were black people then eliminating black people logically would remove the most out of wedlock births. Of course removing black people isn't really an option, nor do I personally care about out of wedlock births. It is a clear demonstration of simple reasoning though.

 

How does cause and effect matter here again?

 

If they had nothing to do with each other then they wouldn't be correlated

You are 100% wrong that race has nothing to do with out of wedlock births. If it didn't have anything to do with it, there would be no correlation between out of wedlock births and race.

 

Playing with your concept of cause and efffect

Why do you think it sounds funny to say race causes out of wedlock birth? Perhaps because race is not a consious entity? Because it cannot be changed by a consious entity? What if white people made black people correlated with out of wedlock birth by mistreating them... why did the white people do it? Answer: Because they were black, so race did cause it after all. On the other hand how do you know that their biological factors do not predispose them to out of wedlock births? You don't. Pretending this were true, maybe you would want to say that their biology only predisposes them towards it if whites treat them a certain way. Maybe your point would be the whites can't complain if they are contributing to it even if there are racial differences.

 

Conclusions about cause and effect

The point is, you don't have any clue what you are talking about when you mention cause and effect. You have some cryptic subconsious definition for it that has the basic form of "an event that is highly correlated with another event happening subsequently" and contains pieces of ideas like "Can be altered by a concerned person" or "Was altered by a concerned person".

 

This unspecific and unobjective concept is completely unnecessary when you are already talking about correlations.

 

Different Indicators

 

I already answered your question about different indicators. Yes some indicators might cut people into different groups. You just use whatever indicator you have access to. And when different values of that indicator were correlated with other traits, you use those correlations as well.

[TheFaithfulStone]

Many, perhaps infinite, concurrent factors are correlated with a given event subsequently occuring.

 

Maybe - but just because something doesn't have ONE cause doesn't mean that causality is a waste of time.

 

The basic structure of a logical argument is "If P then Q." I fail to understand how that is NOT causality.

 

How does cause and effect matter here again?

 

Let us pretend for a moment that you would like to eliminate out of wedlock births. What steps would you take?

 

If it didn't have anything to do with it, there would be no correlation between out of wedlock births and race.

 

Correlation does not equal causation. This is a logical fallacy. You want the Latin? This doesn't outright mean you're wrong, but it does mean your explanation is lacking.

 

TFS

[Kriminal99]

if p then q refers to deductive reasoning not causation. Q being true is something that is inherently part of p being q and therefore q is always 100 % necessarily true when p.

 

The motivation for trying to reduce causation into correlation comes from an approach to reason that views the human mind (including its limitations) as the source of all human reason. This requires us to first look at how we are subconsiously handling concepts before we use them so that our reasoning is more precise. The method for doing so is simply deducing the precise algorithm that determines our use of a concept like cause using thought experiments. That way we can reason more precisely instead of stumbling around with a vague concept like causation.

 

So lets do a few on causation. If A Causes first b and then c to occur, b would be correlated with c but not as strongly as a is. If b is truly not a cause of c, then both b should be capable of occuring without C subsequently occuring as well as c occuring without b previously occuring. Otherwise b would seem like a cause of c. If you say that b could occur without c after or c without b before but it just never does, then your sample is not representitive of the population you are trying to predict and your correlations are invalid.

 

If either a or d causes first b and then c, then it might be possible that b is better correlated with c than a. However it is again less correlated than (a or d) which is the true cause of c.

 

Feel free to try and break down the connection between cause and correlation. I think it is impossible because cause IS correlation in the human mind. IE thats how we determine cause. But to disprove it all you need is a situation where cause is less correlated with an event than a non cause.

[TheFaithfulStone]

But to disprove it all you need is a situation where cause is less correlated with an event than a non cause.

 

Having trouble thinking of one. If a cause (by your defintion) is something that is 100% correlated with something, how can anything be MORE correlated than 100%?

 

I didn't say that it isn't possible to get causality wrong, but just because you may miss it, or there may be hidden variables, I don't think that means it doesn't exist.

 

In your example, if A causes B and then C, I think that C & B would both be equally measurable with an occurrence of A.

 

Is it possible for a cause not to have it's intended effect? That would be thermodynamic miracle you'd need to find. An example of an effect without a cause could be (and probably is) an example of a different cause, same effect. (For example, stubbing your toe on a rock and on a chair leg are different causes, but they both have the same effect - you hurt your toe.)

 

TFS

[Qfwfq]

P ==> Q

 

Now, there are wo possibilities to consider. One is that of P being a cause of Q and without exception. Another is that of Q being a necessary condition for P. These are quite different possible explanations for observing the implication. There may be partial causation in either way, there may be no causation but just a non-prevention.

[Kriminal99]

Having trouble thinking of one. If a cause (by your defintion) is something that is 100% correlated with something, how can anything be MORE correlated than 100%?

 

I didn't say that it isn't possible to get causality wrong, but just because you may miss it, or there may be hidden variables, I don't think that means it doesn't exist.

 

In your example, if A causes B and then C, I think that C & B would both be equally measurable with an occurrence of A.

 

Is it possible for a cause not to have it's intended effect? That would be thermodynamic miracle you'd need to find. An example of an effect without a cause could be (and probably is) an example of a different cause, same effect. (For example, stubbing your toe on a rock and on a chair leg are different causes, but they both have the same effect - you hurt your toe.)

 

TFS

 

No that is not my definition. The definition of cause is determined by what you (the thinker) thinks sounds like a cause in conjunction with what is logically possible considering those attributes.

 

To me, for something not to cause an event, that event should be able to occur without the non-cause, and the non-cause should be able to occur without the event. That in addition to some rules regarding the nature of correlation and statistics is all I used to reduce causation to correlation.

 

What do you mean by equally measurable? B and C would be less correlated with each other than with A because they can occur sometimes without each other.

 

For instance if I have a siege weapon called a catanon that fires a cannon ball and catapults a rock, then the rock should be able to hit with the cannon missing and the cannonball should be able to hit without the rock hitting. If you say "they could but they just didn't" then you have violated an assumption of statistics such that none of your correlation measurements are valid.

 

What other attributes do you think cause has?

 

An effect without a cause? What are you talking about here? I do not believe in effects without causes.

 

@ Q.... if you are talking about the "If p then q" statement used in logic as TFS was, there is no room for it to model causation or anything other than a situation where q being true is part of p being true.

 

Causation is a weaker connection between events that has nothing to do with logic. As in if I kick the ball, it will fly through the air. Well... maybe... if x physical conditions are true including ones we may not be aware of (given generality problem of induction) etc etc.

 

If I kick the ball, then it will move : Still not logic. If the ball is made of granite it may not.

 

If I kick the ball, then I will swing my foot forward and collide my foot with the ball. This is the definition of kick and therefore "if p then q" has been used correctly.

 

Human thought can be broken down into one of two types, induction and deduction. If p then q refers to deduction. Causation always utilizes induction, and of induction we can never be sure. So while as you say a cause that brought on an event without exception might confuse this distinction, it can't happen because we can never be sure of the "without exception" part and therefore the connection cannot be as strong.

[Qfwfq]

If the ball is made of granite, you might get yourself a fractured foot! :hihi:

 

That is somewhat what I meant to say. I don't know how you understood my post. I did not mean to say that (P ==> Q) ==> causation, I said it might be explained by it but might not, there are different possible cases. Clouds in the sky don't always cause rain, but this doesn't remove that they may. Can it rain when there's not a trace of cloud in the sky? A logician or a mathematician will write rain ==> cloud, without meaning to say rain causes cloud but instead a necessary condition. In this case the condition is necessary because it is the cause and the only possible one, unless you call any falling water "rain" regardless of its origin.

 

Logicians also call P a sufficient condition. A cause which cannot be excepted would be an example of a sufficient condition for its effect. Consider the case of necessary and sufficient condition: P <==> Q, which is cause and which is effect?

[Qfwfq]

So while as you say a cause that brought on an event without exception might confuse this distinction, it can't happen because we can never be sure of the "without exception" part and therefore the connection cannot be as strong.

The same goes for an effect having no other cause!

 

So you can never be sure of any implication! As far as physical law goes, of course. Only in a formal system, where you start from axioms and a theorem can be certain, does it make sense to consider P ==> Q as certain. If you are fairly sure that putting a bullet through my brain will be an unexceptable cause of my death, you might write bullet ==> dead, but of course this is mere induction, something which Popper discards as being pure myth, you never know there might be a black swan somewhere in the universe!

[Kriminal99]

Can you think of an example of a cause that cannot be excepted that is not part of the definition of the effect? For instance in the rain cloud example, you demonstrated that rain contains as part of it's defintion that it came from a cloud in which case the "if then" connection is a purely deductive one. You demonstrated this when you said "unless you consider falling water from any source rain".

 

So it might be that that a "if then" logic connection parallels a causation connection, but only because you previously defined the effect as being a result of the cause. If you haven't defined it so, as you say the proposed connection between the cause and effect is only based on induction which can always be wrong.

 

In the case of necessary and sufficient condition, I think that we would simply define whichever event occured first as the cause and the subsequent event the effect. As I believe our use of the concept of cause is driven by subconsious use of an algorithm which involves only order of events and correlation.