You are viewing the historical archive of The Philosophy Forum.
For current discussions, visit the live forum.
Go to live forum

An Argument for Conceptual Atomism

quine March 11, 2017 at 05:41 12000 views 49 comments
Conceptual atomism is the view that concepts have no internal structure. I will offer an argument for conceptual atomism:

(1) Concepts can be composed.
(2) If concepts can be composed, then conceptual atomism is the case.
Therefore, (3) Conceptual atomism is the case.

The premise (1) shows compositionality of concepts. For example, 'brown' and 'dog' compose 'brown dog'. The total structure of the argument above is modus ponens, i.e. logically valid one.

Defense:
Since concepts are compositionable, if concepts had internal structures, then concepts would not be composed accurately. Suppose that 'brown' has internal sub-contents such as 'color', 'brighter-than-black', and so on. Then we cannot expect the combination of 'brown' and 'dog' into 'brown dog', since we should know their sub-contents, or qualities in order to compose 'brown dog', etc. Therefore, concepts must be atomic in order that compositionality is the case.

Comments (49)

SophistiCat March 11, 2017 at 06:41 #60204
I don't get it. Don't "composed" concepts have structure? If not, then I don't understand what you mean by having structure.
quine March 11, 2017 at 06:50 #60205
Concepts mean lexical concepts. The units are atomic concepts. The structure of 'brown dog' is divided into 'brown' and 'dog'. However, 'brown' itself has no internal-sub-structure. 'dog' itself has no internal-sub-structure, too. Other theories accept the thought that 'brown' or 'dog' has some structures, so they reject the thought that concepts are atomic. For example, some theories of concepts might say that 'dog' is divided into 'animal', 'quadrupedal', 'mammal', and so on. They are positive to accept the structures of concepts, and they are not conceptual atomist positions. Get it?

apokrisis March 11, 2017 at 10:39 #60211
Doesn't make sense. You must mean concepts can be combined but not decomposed. If concepts can be composed, then they would be composites. So you might be wanting to say that concepts are the atomistic, non-decomposable, units that can then be used to compose.

But atomism is disproven in physics now. And even in language use, it is clear that instead organisation and particularisation are created by contexts of constraint.

The word "dog" limits reference to a class of phenomenon fairly tightly. Talking of the brown dog restricts things even more.

So constraints can be combined reducing the freedom of the possible with any arbitrary degree of precision a situation might require. And there is an atomism of a kind in that composability. But concepts don't have to rigidly possess particular fixed properties as atomism implies. Instead they only need to restrict interpretations, and they don't even have to do that more than loosely much of the time.
mcdoodle March 11, 2017 at 10:40 #60212
I don't get it either. I gather it was a Fodor proposal though I remember the Puttnam talk about how he didn't know an elm from a beech.

What, for instance, is a Labrador? If it's (a) a dog, (b) with certain specifiable features, then it is a concept with, within it, relation between other concepts.
SophistiCat March 11, 2017 at 10:52 #60214
Quoting quine
lexical concepts


You'll need to explain what you mean by that.
quine March 11, 2017 at 12:13 #60225
Reply to SophistiCat
What you didn't get is what philosophers of concepts usually talk about. Lexical concepts mean simple vocabulary concepts. 'dog', 'brown', 'cat', 'stupid' are lexical concepts. 'brown dog', 'stupid cat' are concepts composed. Most theories of concepts say that lexical concepts have lower levels of contents: for example, 'dog' is divided into 'animal', 'quadrupedal', 'mammal', and so on. Conceptual atomism rejects that concepts (i.e. lexical concepts) have internal structures.
tom March 11, 2017 at 12:36 #60234
Quoting quine
(2) If concepts can be composed, then conceptual atomism is the case.


You haven't shown this. You haven't even shown that concepts can be composed or that they can't be composed under different conceptions of "concept".


Luke March 11, 2017 at 12:39 #60236
The second premise seems problematic. If lexical concepts can be composed of "lower level" concepts, then why must there be any termination to this composition? I don't see why mere composition should imply atomism. You appear to be assuming this, rather than demonstrating it.
quine March 11, 2017 at 12:40 #60237
Reply to tom
Do you not see that 'brown' and 'dog' compose 'brown dog'? Compositionality of concepts is one of the important characteristics of concepts.
What I am defending is that conceptual atomism is necessary for compositionality of concepts.
quine March 11, 2017 at 12:43 #60238
Reply to Luke
You misunderstood. I claim that concepts are atomic.
Luke March 11, 2017 at 12:53 #60240
I haven't misunderstood.
tom March 11, 2017 at 12:55 #60241
Quoting Luke
The second premise seems problematic. If lexical concepts can be composed of "lower level" concepts, then why must there be any termination to this composition? I don't see why mere composition should imply atomism. You appear to be assuming this, rather than demonstrating it.


Perhaps the concept "dog" is actually made up of all the atomic conceptions of a dog. We could list a few of these atomic conceptions:

1. The black dog belonging to the butcher with three legs.
2. A poodle
3....

So, let's fabricate a "modus ponens"

1. Concepts are composed
2. If concepts are compositions, conceptual atomism is not the case
3. Therefore conceptual atomism is false.

Anyone convinced?
tom March 11, 2017 at 12:58 #60243
Quoting quine
Do you not see that 'brown' and 'dog' compose 'brown dog'?


No. Do you think "quantum mechanics" and "general relativity" compose to "quantum mechanics general relativity"?
tom March 11, 2017 at 13:01 #60244
A question:

Is there a set of all atomic concepts?
quine March 11, 2017 at 13:08 #60245
Reply to tom
Your example is not valid one. Of course, 'tuna', 'salad', 'tom' do not compose 'tuna salad tom' conceptually.
tom March 11, 2017 at 13:17 #60247
Reply to quine

That used to be my nick-name at school, to differentiate me from the tom who didn't like tuna salad.

quine March 11, 2017 at 13:22 #60249
Reply to tom
My friend has a nickname called 'quantum mechanics general relativity'.
SophistiCat March 11, 2017 at 13:25 #60250
Quoting quine
What you didn't get is


...pretty much everything. The thing is that your argument is so simple structurally that it is either trivial or the real meat of the argument needs some unpacking.
Luke March 11, 2017 at 13:26 #60251
You state that "Most theories of concepts say that lexical concepts have lower levels of contents". And, like you, these other theories also claim that concepts can be composed. Yet they reach the opposite conclusion to you. Why?
quine March 11, 2017 at 13:36 #60253
Reply to Luke
Many theories of concepts say that concepts have lower level structures, and that concepts can be composed. Conceptual atomism says that concepts can be composed only when they are atomic.
tom March 11, 2017 at 13:38 #60254
I'm still intrigued whether there is a set of atomic concepts or not.
Luke March 11, 2017 at 13:46 #60256
Reply to quine
Again, I understand this, but you are just asserting conceptual atomism rather than arguing for it.
tom March 11, 2017 at 14:06 #60263
Quoting Luke
Again, I understand this, but you are just asserting conceptual atomism rather than arguing for it.


It is however mildly ironic that the atomic concept of an atom isn't atomic any more.
quine March 11, 2017 at 14:18 #60265
Reply to Luke
I did, but you forgot.
jkop March 11, 2017 at 17:06 #60281
What is 'Snoopy'? Is a concept of the fictional dog atomic, or a composition?
quine March 11, 2017 at 20:45 #60298
'Snoopy' and 'dog' are independent atomic concept respectively.
apokrisis March 11, 2017 at 21:07 #60303
Quoting quine
'Snoopy' and 'dog' are independent atomic concept respectively.


In what sense are either not composed of multiple instances of experience?

You might mean that words, to behave in "the right way" - act like a conjunctive arithmetic - would have to be treated as atomistic particulars ... even when they are actually general umbrella terms. But it is clear that being atomistic is not what word meanings actually are.
Luke March 11, 2017 at 22:08 #60311
Reply to quine

You have said both:

"Many theories of concepts say that concepts have lower level structures, and that concepts can be composed. Conceptual atomism says that concepts can be composed only when they are atomic."

"(2) If concepts can be composed, then conceptual atomism is the case."
quine March 11, 2017 at 23:18 #60322
Reply to apokrisis
It's about semantics of concepts. Informational semantics says that meanings of concepts are information.
apokrisis March 12, 2017 at 00:37 #60332
Reply to quine So what does that mean for your notion of the atomism of concepts. Are they informationally closed or open?

Uttering "dog" is an atomistic act. As a noise we might hear, we know its not someone saying log or bog or dig or dob. But as a concept, dog seems informationally open. We could either generalise or particularise its semantics by adding or subtracting information (or constraints).

Are we talking about a dog or the dog? The Platonically ideal canine, or that little mutt snapping at our ankles?

Atomism kind of works because we can leave the contexual information implied and unsaid. But that doesn't make word interpretations foundationally atomistic. It is of the essence of words that their boundaries of reference are porous or vague. Atomism is relative as semantics is essentially open even if syntax does try to create a closed compositional structure that is meant to keep all the meanings trapped securely inside the spoken sentence.
quine March 12, 2017 at 00:48 #60335
Reply to apokrisis
According to informational semantics, concepts are mental particulars representing information about objects. For example, concept 'dog' is a mental particular representing information about real dogs.
apokrisis March 12, 2017 at 01:48 #60342
Reply to quine Is that really what information semantics claims? To me it seems the kind of semiotic, constraints-based, approach I've described.

So meaning is use. The utterance of "dog" stands as a sign mediating a habit of relation with the world. Thus while the vocal act is particular, it only has meaning by virtue of the coordination it might bring between a mind and the world (the world including other minds that might have an interest in being likewise oriented to the world beyond themselves).

You talk about mental particulars (concepts, ideas, impressions?) representing information. But informational semantics seems to be making the same point that they don't. What signs buy as the mediating mechanism is the possibility of an externalised habit of coordination rather than an internalised state of information.

So "dog" as a concept is not the representation of a doggy essence - something closed and internally represented - but instead the anchor of a habit of coordination between self and world. Doggyness characterises all we might do, think and say in a "doggy situation". So as a "concept" it is creatively open ended - while also being still usefully or pragmatically constrained enough to function as a lexical unit, a syntactic element, of some "propositional" sentence.
Luke March 12, 2017 at 03:30 #60346
Reply to quine

Do you have no comment to make on your inconsistent statements:

1. "Many theories of concepts say that [...] concepts can be composed."

2. "If concepts can be composed, then conceptual atomism is the case."
quine March 12, 2017 at 04:51 #60353
Reply to Luke
Many theories of concepts say (1) concepts have lower level structures, and (2) concepts can be composed, "both at the same time".
The premise 2 above is different from 'many theories of concepts'. The premise 2 does not say (1) concepts have lower level structures, but says (2) concepts can be composed.

Do you see the difference???
Luke March 12, 2017 at 05:00 #60354
I see the difference, but you said: "If concepts can be composed, then conceptual atomism is the case."

You have stated that there are opposing (non-atomistic) "theories of concepts" which also assert that concepts can be composed. This makes your premise 2 false.
quine March 12, 2017 at 07:06 #60368
Reply to Luke
Conceptual atomism asserts (1) concepts can be composed, and (2) concepts can be composed only if concepts are atomic. (2) can be restated as: (3) concepts *cannot* be composed if concepts have lower level structures (i.e. they are not atomic).
More precisely:
(1) If concepts are not atomic, then concepts cannot be composed.
(2) Concepts can be composed.
Therefore, (3) concepts are atomic.
Modus Tollens.
Luke March 12, 2017 at 07:32 #60373
Reply to quine

You're contradicting yourself. You said earlier: "Many [opposing] theories of concepts say that concepts have lower level structures, and that concepts can be composed."

Now you are saying: "If concepts are not atomic, then concepts cannot be composed."
quine March 12, 2017 at 07:41 #60375
One thing is a criticism on opposing theories, another is a report of which opposing theories state. There's no trouble.
Luke March 12, 2017 at 07:54 #60377
Okay, justify this new claim. Why are all of the opposing theories wrong? Why can't non-atomic concepts be composed?

You said earlier that "some theories of concepts might say that 'dog' is divided into 'animal', 'quadrupedal', 'mammal', and so on." So, why isn't this an example of "dividing" the concept of 'dog' into other concepts? That is, why don't those other concepts compose the concept of 'dog'? Why can't this be done?
quine March 12, 2017 at 09:08 #60382
Quoting Luke
So, why isn't this an example of "dividing" the concept of 'dog' into other concepts?


It's because dividing concepts into lower levels causes to make compositionality of concepts fail. Infinite regress of dividing concepts into parts and parts and parts...
Luke March 12, 2017 at 09:47 #60383
Reply to quine

How does it fail?
quine March 12, 2017 at 10:02 #60384
Quoting quine
Infinite regress of dividing concepts into parts and parts and parts...


This is how.
Luke March 12, 2017 at 10:18 #60385
I don't follow how the compositionality of concepts fails.
quine March 12, 2017 at 10:48 #60387
Reply to Luke
Conceptual atomism model:
'Brown' and 'dog' compose 'brown dog'. It's easy and simple.
Opposing theories model:
'Brown' is composed by 'color', and 'brighter-than-black', and so on. 'Dog' is composed by 'animal', and 'quadrupedal', and mammal', and so on. ... 'Brighter-than-black' is composed 'lighter-than-black' and 'color' and so forth. ... 'animal' is composed by 'living thing' and 'biologically constructed' and so forth. ... ... ... ... and 'quark' is composed by 'physical' and 'small particle', and so on. Is there an end?

Compare two stories.
Luke March 12, 2017 at 11:01 #60388
What if there is no end? Why should that imply that the compositionality of concepts fails?
quine March 12, 2017 at 11:21 #60389
Reply to Luke
That is not a question. Explanations are sufficient. I recommend you to go back to this thread to make up what you need to get.
Luke March 12, 2017 at 11:26 #60391
Reply to quine

It certainly is a question, but if you're unable to answer it then perhaps you should reconsider your argument.
quine March 12, 2017 at 11:40 #60394
Reply to Luke
My answers are certainly answers. If you are unable to accept this fact, go to the grave.
ernestm March 12, 2017 at 22:06 #60454
the problem with atomism is discussed in another thread on theseus' ship.