The Mad Linguist
11th August 2003, 04:40 PM
The eternal issue of "defining an atheist" came up in a debate between Interesting Ian and Stimpson J Cat on Cleopatra's thread about atheists asking for proof of God.
Stimpson J Cat said:
... You will find several common usages of the term "atheism", which range from simple lack of belief, to positive assertions against specific conceptions of God, to positive assertions against all conceptions of God... Like I said, lack of belief is the most general definition. If somebody says they are an atheist, then that is about the only thing you can assume about their position... In other words, technically speaking, an atheist is anybody who is not a theist.
Interesting Ian, on the other hand, defended another definition of atheism:
Atheism is not a lack of belief in a "God" (that's agnosticism), but rather the positive belief that there is no "God". Moreover this "God" need not be personnel, and need not have created the Universe, nor need interact with it...... Atheism means a rejection (not lack of belief) of any reality or existent which could reasonably be labelled as "God". Not a lack of belief in a personal God, but a rejection of any God whether personal or impersonal. .... You cannot declare that those who believe in an impersonal God are atheists!
(apologies for taking both of you slightly out of context, but I’ve tried to capture your gists…)
This brought to my mind an issue which has troubled me for some time, but I didn’t want to derail that thread, so I’m starting this one instead.
As part of my training in formal linguistics, I’ve been taught about two important semantic theories – i.e. theories about how words mean what they mean, and how we decide what entities a word can apply to. I’ll explain them briefly.
A theory of meaning based on necessary and sufficient conditions defines an X in terms of a list of properties that an entity must possess to be considered an X. So for instance, if we are defining “bird”, we would say something like:
a) warm blooded
b) has feathers
c) toothless jaw
d) beak
and however many more we need to add until we’ve excluded everything that IS a bird and excluded everything that ISN’T a bird.
Another theory is based on the notion of a prototype. This is the idea that for each word, we hold in our heads a kind of an idea of a “good example” or “prototype” of the kind of things that that word can apply to. So for example, for “bird”, we hold in our heads a picture of a prototypical “bird” – maybe a duck, or a sparrow, or something like that – and other things are rated as “birds” or not based on how alike they are to our conception of the prototype.
Two key points must be made here. First, the category is graded; so some things are more “birdy” than others, there are good and bad examples of bird – ostriches, penguins and vultures might be notable bad examples. Secondly, the category has a fuzzy boundary – the set of “bird” shades slowly into the category of “not bird”. This is more obvious if you consider “cups” and “mugs” – two separate words, referring to drinking vessels of different dimensions, but experiments have shown that speakers have no clear cut-off point for when something is a cup and when it is a mug as the dimensions of one slowly change to the dimensions of the other.
Now these two theories don’t work together – by an approach based on necessary and sufficient conditions (NSC), fuzzy categories and good examples are nonsensical. Anything that is a bird is an equally good example of a bird – a penguin just as much as a robin. Something is either a bird or not a bird, with nothing between (yes, yes, archaeopteryx – all that shows is that the categories in our language aren’t designed to cope with extinct fauna).
As it turns out, the prototype theory (PT) is correct and NSC isn’t. There are problematic issues that can only be resolved using PT – for example, a question such as “Is the Pope a bachelor?” Clearly, by an NSC definition of “bachelor”, he must be: he is an adult unmarried male. But equally clearly, the Pope isn’t who you think of when you hear the word “bachelor”. You think of a young man, living alone in the marriage market. That is the prototypical bachelor. If the Pope is a member of the category “bachelor”, he’s a very marginal member – way out in the realm of “bad examples”.
There are various other arguments. But the upshot is that in human language, we use a system of meaning based on prototypes. This is pretty well established now in linguistics.
Going back to the point with which I began this excessively lengthy post, it seems to me that Ian and Stimpy’s differing definitions of “atheist” are demonstrably based on a use of PT and a use of NSC. Ian’s definition seems to be based on a prototype atheist – someone who rejects and refuses to believe in any God at all. Others are then labelled “atheist” or not depending on how similar they are to this prototype. (Note: prototypical does not equal typical, although sadly it often equals stereotypical.)
Stimpy seems to be taking a much more NSC approach: he looks at the group of entities referred to as atheists, sees what they have in common, and from this devises a list of necessary conditions – which I will state succinctly as “lacking belief in the God of Theism” – because that is all that atheists have in common.
Now, I must admit I have a great deal more sympathy to Stimpy’s definition than Ian’s. When I say I am an atheist, I mean no more than that I don’t believe in the God of Theism. But this goes directly against the way that meaning actually works. Let me demonstrate this. If we define “atheist” to mean “human being who is not a theist” (which is the same thing as “human being lacking belief in the God of Theism”), then we can see that this includes not only the people you’d expect it to include, but also the Buddha, Albert Einstein (since the pantheistic God is not the theistic God), Hal Bidlack (since the deist’s God is not the theistic God), Mahatma Gandhi (since, if I understand Hinduism correctly, their God is not the theistic God), Augustus Caesar (a polytheist: many nonomnipotent gods is not the single theistic god), the believers in an impersonal God mentioned by Interesting Ian in that other thread, and – to cut a long story short – all kinds of people who we would not normally expect to be defined as atheists.
Therefore, if we create necessary and sufficient conditions based on all the people who say (justifiably) that they are atheists, we end up with a definition that includes other people who were not in the original group. This is obviously not ideal.
In contrast, Ian’s definition of “atheist” is based in the mode of thinking that we know to actually underlie how meaning is used in language. While his definition has the failing that it characterises some (many?) atheists incorrectly (as expected under PT: not all Xs possess all the relevant attributes of the prototypical X), he does not run into the problem of his definition either a) excluding persons to whom the word X is actually applied or b) including persons to whom the word X has never been applied.
But let us consider the nature of philosophical argument. Arguments based in logic often utilise syllogisms which are very much dependant on the NSC way of thinking. Science itself is, I think, quite possibly dependant on this type of classification. Consider the following:
1) A bird is an animal with feathers and a toothless beak.
2) This animal has feathers and a toothless beak.
3) Therefore this animal is a bird.
This argument is, I belief, valid: given the premises in 1) and 2) one cannot avoid the conclusion in 3). But note that this is a mode of thinking that is implicitly based in an NSC approach to meaning. Could such an argument ever handle the “good examples” and “bad examples” necessary for PT? Could it cope with categories whose edges are fuzzy in an unknown fashion? I cannot see how it would. Science, it seems to me (though I am pushing well beyond my area of expertise with this statement) is also dependent on the capability to make harsh categorisations: something is either an X or not an X, and there is no notion of reference to a prototypical X.
I see this as an inescapable problem. The system of meaning that our language runs on is incompatible with the system of meaning needed to argue logically or to utilise science. Stimpy’s definition of “atheist” is the one you would need to use in a philosophical argument, I think; yet Ian’s is the one which has been constructed in the way that everyday meanings are actually constructed in language. In a logical argument, it might well turn out that Hal Bidlack (sorry to keep using you as an example, Hal) is, technically speaking, an atheist according to the NSC definition. But in everyday terms he rejects the label, and with good reason: his beliefs do not resemble those of the prototypical atheist to any significant degree. There seems to be a gulf opening up between the philosophical-scientific language and the everyday language – and this seems far more fundamental than any concern over the technical terminology or jargon that philosophers and scientists may use.
So my question is this: our everyday definitions, negotiated socially with other speakers of our language, are based on the fuzzy logic of prototype theory. But our philosophical arguments must be built on clear definitions using necessary and sufficient conditions. Can we ever reconcile the two? Can a philosophical argument be translated successfully to everyday language? (Is “translation” even the right word for the kind of thing we’d need to happen here?) Can any logical debate ever be successfully undertaken without departing from the everyday way of making meaning? And if we do depart from everyday ways of making meaning, how do we make our logic relevant to life again?
Sorry about the length of this, but I wanted to have something substantial to say before I started my first thread :)
The Mad Linguist
Stimpson J Cat said:
... You will find several common usages of the term "atheism", which range from simple lack of belief, to positive assertions against specific conceptions of God, to positive assertions against all conceptions of God... Like I said, lack of belief is the most general definition. If somebody says they are an atheist, then that is about the only thing you can assume about their position... In other words, technically speaking, an atheist is anybody who is not a theist.
Interesting Ian, on the other hand, defended another definition of atheism:
Atheism is not a lack of belief in a "God" (that's agnosticism), but rather the positive belief that there is no "God". Moreover this "God" need not be personnel, and need not have created the Universe, nor need interact with it...... Atheism means a rejection (not lack of belief) of any reality or existent which could reasonably be labelled as "God". Not a lack of belief in a personal God, but a rejection of any God whether personal or impersonal. .... You cannot declare that those who believe in an impersonal God are atheists!
(apologies for taking both of you slightly out of context, but I’ve tried to capture your gists…)
This brought to my mind an issue which has troubled me for some time, but I didn’t want to derail that thread, so I’m starting this one instead.
As part of my training in formal linguistics, I’ve been taught about two important semantic theories – i.e. theories about how words mean what they mean, and how we decide what entities a word can apply to. I’ll explain them briefly.
A theory of meaning based on necessary and sufficient conditions defines an X in terms of a list of properties that an entity must possess to be considered an X. So for instance, if we are defining “bird”, we would say something like:
a) warm blooded
b) has feathers
c) toothless jaw
d) beak
and however many more we need to add until we’ve excluded everything that IS a bird and excluded everything that ISN’T a bird.
Another theory is based on the notion of a prototype. This is the idea that for each word, we hold in our heads a kind of an idea of a “good example” or “prototype” of the kind of things that that word can apply to. So for example, for “bird”, we hold in our heads a picture of a prototypical “bird” – maybe a duck, or a sparrow, or something like that – and other things are rated as “birds” or not based on how alike they are to our conception of the prototype.
Two key points must be made here. First, the category is graded; so some things are more “birdy” than others, there are good and bad examples of bird – ostriches, penguins and vultures might be notable bad examples. Secondly, the category has a fuzzy boundary – the set of “bird” shades slowly into the category of “not bird”. This is more obvious if you consider “cups” and “mugs” – two separate words, referring to drinking vessels of different dimensions, but experiments have shown that speakers have no clear cut-off point for when something is a cup and when it is a mug as the dimensions of one slowly change to the dimensions of the other.
Now these two theories don’t work together – by an approach based on necessary and sufficient conditions (NSC), fuzzy categories and good examples are nonsensical. Anything that is a bird is an equally good example of a bird – a penguin just as much as a robin. Something is either a bird or not a bird, with nothing between (yes, yes, archaeopteryx – all that shows is that the categories in our language aren’t designed to cope with extinct fauna).
As it turns out, the prototype theory (PT) is correct and NSC isn’t. There are problematic issues that can only be resolved using PT – for example, a question such as “Is the Pope a bachelor?” Clearly, by an NSC definition of “bachelor”, he must be: he is an adult unmarried male. But equally clearly, the Pope isn’t who you think of when you hear the word “bachelor”. You think of a young man, living alone in the marriage market. That is the prototypical bachelor. If the Pope is a member of the category “bachelor”, he’s a very marginal member – way out in the realm of “bad examples”.
There are various other arguments. But the upshot is that in human language, we use a system of meaning based on prototypes. This is pretty well established now in linguistics.
Going back to the point with which I began this excessively lengthy post, it seems to me that Ian and Stimpy’s differing definitions of “atheist” are demonstrably based on a use of PT and a use of NSC. Ian’s definition seems to be based on a prototype atheist – someone who rejects and refuses to believe in any God at all. Others are then labelled “atheist” or not depending on how similar they are to this prototype. (Note: prototypical does not equal typical, although sadly it often equals stereotypical.)
Stimpy seems to be taking a much more NSC approach: he looks at the group of entities referred to as atheists, sees what they have in common, and from this devises a list of necessary conditions – which I will state succinctly as “lacking belief in the God of Theism” – because that is all that atheists have in common.
Now, I must admit I have a great deal more sympathy to Stimpy’s definition than Ian’s. When I say I am an atheist, I mean no more than that I don’t believe in the God of Theism. But this goes directly against the way that meaning actually works. Let me demonstrate this. If we define “atheist” to mean “human being who is not a theist” (which is the same thing as “human being lacking belief in the God of Theism”), then we can see that this includes not only the people you’d expect it to include, but also the Buddha, Albert Einstein (since the pantheistic God is not the theistic God), Hal Bidlack (since the deist’s God is not the theistic God), Mahatma Gandhi (since, if I understand Hinduism correctly, their God is not the theistic God), Augustus Caesar (a polytheist: many nonomnipotent gods is not the single theistic god), the believers in an impersonal God mentioned by Interesting Ian in that other thread, and – to cut a long story short – all kinds of people who we would not normally expect to be defined as atheists.
Therefore, if we create necessary and sufficient conditions based on all the people who say (justifiably) that they are atheists, we end up with a definition that includes other people who were not in the original group. This is obviously not ideal.
In contrast, Ian’s definition of “atheist” is based in the mode of thinking that we know to actually underlie how meaning is used in language. While his definition has the failing that it characterises some (many?) atheists incorrectly (as expected under PT: not all Xs possess all the relevant attributes of the prototypical X), he does not run into the problem of his definition either a) excluding persons to whom the word X is actually applied or b) including persons to whom the word X has never been applied.
But let us consider the nature of philosophical argument. Arguments based in logic often utilise syllogisms which are very much dependant on the NSC way of thinking. Science itself is, I think, quite possibly dependant on this type of classification. Consider the following:
1) A bird is an animal with feathers and a toothless beak.
2) This animal has feathers and a toothless beak.
3) Therefore this animal is a bird.
This argument is, I belief, valid: given the premises in 1) and 2) one cannot avoid the conclusion in 3). But note that this is a mode of thinking that is implicitly based in an NSC approach to meaning. Could such an argument ever handle the “good examples” and “bad examples” necessary for PT? Could it cope with categories whose edges are fuzzy in an unknown fashion? I cannot see how it would. Science, it seems to me (though I am pushing well beyond my area of expertise with this statement) is also dependent on the capability to make harsh categorisations: something is either an X or not an X, and there is no notion of reference to a prototypical X.
I see this as an inescapable problem. The system of meaning that our language runs on is incompatible with the system of meaning needed to argue logically or to utilise science. Stimpy’s definition of “atheist” is the one you would need to use in a philosophical argument, I think; yet Ian’s is the one which has been constructed in the way that everyday meanings are actually constructed in language. In a logical argument, it might well turn out that Hal Bidlack (sorry to keep using you as an example, Hal) is, technically speaking, an atheist according to the NSC definition. But in everyday terms he rejects the label, and with good reason: his beliefs do not resemble those of the prototypical atheist to any significant degree. There seems to be a gulf opening up between the philosophical-scientific language and the everyday language – and this seems far more fundamental than any concern over the technical terminology or jargon that philosophers and scientists may use.
So my question is this: our everyday definitions, negotiated socially with other speakers of our language, are based on the fuzzy logic of prototype theory. But our philosophical arguments must be built on clear definitions using necessary and sufficient conditions. Can we ever reconcile the two? Can a philosophical argument be translated successfully to everyday language? (Is “translation” even the right word for the kind of thing we’d need to happen here?) Can any logical debate ever be successfully undertaken without departing from the everyday way of making meaning? And if we do depart from everyday ways of making meaning, how do we make our logic relevant to life again?
Sorry about the length of this, but I wanted to have something substantial to say before I started my first thread :)
The Mad Linguist