Fallacy of Composition & Determinist Argumentation

The fallacy of composition is a pattern of erroneous reasoning, whereby an inference on an existent as a whole is based on facts pertaining to the existent's components. The following argument is an instance of this fallacy.

"Atoms are invisible. Elephants are made up of atoms. Therefore, Elephants are invisible."

This scheme of argumentation is common in a certain line of arguments held by determinists -- generalized as follows:

"The behavior of atoms, particles, and matter are metaphysically determined. Our brains and bodies are comprised of atoms, particles, and matter. Therefore, our behavior is metaphysically determined."

The determinist argument proceeds by inferring about the nature of an existent as a whole, based on a fact about the existent's constituents. This argument is indeed fallacious, for it commits the fallacy of composition, and consequently the conclusion does not follow necessarily. Hence, the argument is invalid, or at best, weak.

(Edited by Warren Chase Anspaugh on 7/19, 9:56pm)

Post 1

Monday, July 20, 2009 - 6:02am

Yes, Warren, there are cases in which it is a fallacy to carry a trait of the elements of a composite to the composite as a whole. But there are other cases in which it is not. The nature of the trait has to be examined in its particular character to see whether conveying it from an element to a certain sort of composite formed from such elements is fallacious or not. (See here.)

In your target case, that means the character of determinism in cellular activities needs to be examined closely, then closely whether there results the same character of determinism in collective neuronal activities, then in brain subsystem activities, then in brainy animal activities. For some of this task of characterization, I can offer the sections �Physical Determinism� and �Organic Determinism� from my essay �Volitional Synapses� here.

Post 2

Monday, July 20, 2009 - 11:59am

Thanks for the additional reading material, Stephen. I look forward to reading it. I also agree with your caveat that the "fallacy of composition" is not necessarily fallacious in every instance, such as the following argument:

"All parts of this toy are made of plastic. Therefore, this toy is made of plastic."

This inference is different compared to the aforementioned line of determinist argument, in terms of verifiability or clarity. Whereas a toy can be dissected, observed, and verified as entirely plastic, no simple procedure is available (at least, to my knowledge) for a validation of the determinist's argument that free will (or volition) is not attributable to our existence.
In light of this, it seems perfectly appropriate to criticize the determinist argumentation on the grounds that it rests upon a part-whole relationship (all parts are metaphysically determined, therefore the whole..) in a certain model of how the human creature is constructed. Given that this particular determinist model has not been validated as factually correct, it is subject to criticization that it rests on the fallacy of composition.

Post 3

Monday, July 20, 2009 - 12:23pm

Stephen,

Is there a clear test for determining when the fallacy of composition applies? Intuitively, certain attributes of components are attributable to the whole, e.g., having color, wealth, limited duration, mass. But then certain attributes of components are *not* attributable to the whole, e.g., not having color (being invisible), pliability, speed, breakability.

I can't tell whether Warren is right that the determinist argument he mentions falls prey to the fallacy of composition.

Jordan

Post 4

Tuesday, July 21, 2009 - 2:51am

Jordon,

I do not know a general criterion by which we sometimes know that a particular trait of a constituent also applies to the composite; we sometimes know that a particular trait of a constituent does not apply to the composite; and we sometimes do not know whether a particular trait of a constituent applies to the composite. In his Introduction to Logic, when he begins to discuss informal fallacies, such as the fallacy of composition, Irving Copi observes that in logic the name fallacy is reserved for those errors of reasoning that bear some resemblance to forms of correct reasoning.

Warren's diagnosis of this particular argument for determinism as a fallacy is correct. One arguing for determinism must take it to be unknown (not agreed to be known) whether the particular trait in question applies to the whole because it applies to the parts. To then argue in a form that resembles cases in which we know that a particular trait of a constituent also applies to the composite is a fallacy in exactly the narrow sense that Copi set out. In addition to committing the fallacy of composition, I think this argument for determinism also commits the fallacy of begging the question. Serious arguments for determinism, such as in Ted Honderich's A Theory of Determinism, do not use the argument Warren is dismissing.

Post 5

Tuesday, July 21, 2009 - 10:16am

Hi Stephen,

I found an answer to my question. Here's the short version: It is proper to ascribe to the whole a property shared by all of its parts where (1) said property is absolute as opposed to relative, and (2) the property is unaffected by the arrangement of the whole. Here's the long version:

STEP 1

Examples of absolute properties include being green, having mass, being circular, being made of iron, ocurring in 1975. These properties have meaning without comparison to other properties. Examples of relative properties include being cheap, tall, big, strong, heavy. The meaning of each of these entails comparison, e.g., cheap implies that other stuff is more expensive; tall, that other stuff is shorter; big, smaller; strong, weaker; heavy, lighter.

Relative properties shared by all parts cannot be ascribed of the whole. For example, if all parts of a chair are cheap, that doesn't necessarily mean the chair is cheap, too. The fallacy of composition always applies to attempts to ascribe to the whole any relative property shared by all parts.

Absolute properties of parts are different. They might be ascribed of the whole. For instance, if all parts of the chair are made of wood, it's safe to say the chair is made of wood. Here, the fallacy of composition doesn't apply. But if we say all parts of the chair are rectangular (also an absolute property) that doesn't necessarily mean the chair is rectangular. Here, the fallacy of composition does apply. Onto step 2.

STEP 2

Having rendered relative properties of all parts ineligible for ascription to the whole, we now need to discern different kinds of absolutes. Some absolutes of all parts remain true regardless of the arrangement of the whole, i.e., whether the whole is summative or integral, a collection or a gestalt. These absolute properties of parts are considered independent from the nature of the whole. Green and wooden are two examples of independent absolutes. If all the parts of X are green or wooden, then X is green or wooden. X could be a chair (an integrative whole) or a bunch of blocks (a summative collection). Here again, the fallacy of composition does not apply.

The same is not so with absolutes that are dependent on the nature of the whole. Take the absolute property of edibleness. If all the parts of a dish are edible, that doesn't necessarily mean the dish is edible. Why? Because the arrangement (combinations of flavors) of the whole matters when it comes to edibleness. Whether a whole is edible depends on how the whole is arranged. The fallacy of composition applies!

Just In Case

Rounding out this thought, relative properties shared by all parts -- even those relative properties that are independent of the nature of the whole -- still cannot be ascribed of the whole. For instance, just because all parts of X are lightweight doesn't necessarily mean X itself is lightweight, even though its lightness doesn't depend on whether the X is a chair or just a pile of sticks.

So there you have it! One may ascribe to the whole a property shared by all parts only if the property is absolute and independent.

Application to Determinism Argument

Does the fallacy of composition apply to the argument that all particles are determined; we are composed of particles; therefore, we are determined.

First, I think being determined is an absolute. It's not the type of property whose meaning entails a comparison. Either one is determined or one is not. There's no degree, no relativeness, to the property. So it passes step 1. Agreed?

Second, is being determined dependent or independent of the nature of the whole? Is being determined like green or wooden or having mass (independent), or is it like rectangular or edible (dependent). Can we mess up the arrangement of a whole and yet keep it as determined? I do feel like I'm missing something here -- and I wasn't expecting this answer, to be honest -- but the answer I'm seeing is that being determined is independent. If all parts of X are determined, then X is determined, whether X is a chair (integrative whole) or bunch of blocks (summative collection). Because being determined is absolute and independent, the fallacy of composition does not apply.

Jordan

Post 6

Wednesday, July 22, 2009 - 9:54am

I agree with Stephen in his Post #4. Strictly speaking, the fallacy of composition is a logical fallacy -- a formal fallacy. Therefore, it does not depend on the particulars of the argument, but on the form of the argument.

I.e., the parts of a thing have this characteristic; THEREFORE, the whole has it.

It is the "therefore" -- the claim that the conclusion follows from the premises -- that accounts for the fallacy. Remember, there is a difference between whether the conclusion of an argument is true and whether it is valid.

Suppose one were to argue: ""The behavior of atoms, particles, and matter are subject to the law of causality. One's body is composed of atoms, particles, and matter. Therefore, the behavior of one's body is subject to the law of causality."

The proposition forming the conclusion is certainly true -- the behavior of one's body is indeed subject to the law of causality -- but does that admittedly true proposition follow from the premises? The answer is, no. The conclusion is invalid -- not false but invalid -- because the argument commits a logical fallacy -- the fallacy of composition.

- Bill

Post 7

Wednesday, July 22, 2009 - 10:37am

Bill,

Stephen said it was an *informal* fallacy, which it is. The validity of the argument depends on the content. This is *never* the case with formal fallacies.

Jordan

Post 8

Wednesday, July 22, 2009 - 3:00pm

Jordan,

I was perhaps a bit too hasty in my stated agreement with Stephen. After re-reading his post, I'm not sure that he would agree with my take on this. I now tend to think that he wouldn't. I realize that "the fallacy of composition" is widely regarded as an "informal" fallacy, on the grounds that it depends on the content of the argument instead of simply on the form. So, the conventional view would be that the example I used in Post #6, viz. --

"The behavior of atoms, particles, and matter are subject to the law of causality. One's body is composed of atoms, particles, and matter. Therefore, the behavior of one's body is subject to the law of causality." --

does not in fact commit that fallacy, as I claimed it does.

So, I am taking exception to the conventional view, because I would say that despite the obvious truth of its conclusion, the law-of-causality inference is still guilty of the fallacy of composition. It still commits that fallacy, because the conclusion cannot be inferred directly from the premises. In other words, it doesn't follow simply from the fact that the behavior of atoms, particles and matter are subject to the law of causality, the behavior of one's body as composed of these things is therefore subject to it. To infer that the behavior of one's body is subject to the law of causality, one needs a different premise, e.g., the premise that everything is subject to the law of causality, because causality (i.e., identity) is inherent in existence. In other words, one would need something like the following argument for the conclusion to be considered a valid inference: "Every existent is subject to the law of causality; one's body is an existent; therefore, one's body is subject to the law of causality."

So, I am saying that, unlike the fallacy of ambiguity, for example, which does depend on the content of the argument, the fallacy of composition does not. One can never simply infer that what is true of the parts is therefore true of the whole.

- Bill
(Edited by William Dwyer on 7/22, 3:02pm)

(Edited by William Dwyer on 7/22, 3:08pm)

Post 9

Wednesday, July 22, 2009 - 3:51pm

Hi Bill,

What's wrong with the idea I put forth in post 5? It sure seems that an absolute, structure-independent property shared by all constituent parts is properly ascribable to the whole. Philosopher Nelson Goodman calls these properties "expansive." Such expansive properties include green, wooden, existing in 1975, being inside the garage, and to my eyes, being caused or determined.

Jordan

Post 10

Wednesday, July 22, 2009 - 4:35pm

Hi Jordan,

Yeah, I think you're correct. I may be off base on this. It's certainly true that if all parts of a chair are made of wood, then the chair itself is made of wood. Given my previous argument, I would have to say that this statement is fallacious, because it attempts to infer that the chair is made of wood simply from the fact that its parts are made of wood. But, of course, we know that if the parts are made of wood, the chair itself has to be made of wood. We know the statement is true.

The only question is whether or not this is a valid logical inference that can stand on its own. It could be objected, along the lines of my previous post, that we need another premise to make the argument completely valid. The premise would be that if all of a things parts are composed of a particular material, then the thing itself is composed of that material.

So, the fully fleshed out argument would be as follows:

A thing whose parts are made of a material is itself made of that material.
A chair's parts are made of a particular material, namely, wood.
Therefore, the the chair itself is made of wood.

I'm not sure that this latter formulation is necessary for full validity, but I think it might be. If it isn't, then I stand corrected.

- Bill

Post 11

Wednesday, July 22, 2009 - 6:33pm

Jordan, you cannot address the problem of how the limiting concept determinism affects free will until you define what free will is; both what will is, and what it is that makes will free.

Will is obviously an emergent property. We speak of living people as having volition. We do not speak of dead corpses having volition. The fact that a corpse does not have free will, or will at all, does not trouble the voluntarist. Obviously volition is a property of the whole that depends upon the proper arrangement of the parts. Will is no more a property of the parts than the roundness of a rubber ball is the property of rubber molecules. You can't start with the parts of a rubber ball and argue that their roundness or lack of roundness has any relevance to the roundness of the ball. Indeed, the concept roundness really doesn't even apply on the molecular level. Applying the notion of determinism (a limit on the possible types of causation evident in the will) to objects which do not have the capacity to express volition (to make choices between alternative ends) is either an example of begging the question, or stealing the concept.

Until this is explored I don't see how one can start with the negatively defined notion determinism and get anywhere useful. What the will is is the fundamental question. The Objectivist has to start there.

Post 12

Thursday, July 23, 2009 - 10:13am

Ted,

My concern was whether the fallacy of composition applies to the argument Warren gave. I found that it did not. Your discussion of will has no bearing on that finding. That is, whether the fallacy of composition (or division) applies to free will has no bearing on whether it applies to determinism. Or: A whole's emergent properties do not preclude its expansive properties.

Jordan

Post 13

Thursday, July 23, 2009 - 1:06pm

Jordan, I found your research helpful and useful on its own part, and I sanctioned your post. But I simply fear that the dialog in which you and Warren are engaged premature. Warren's opening post seems entirely correct and unobjectionable as it stands.

Your post five, however interesting, leads you to a mistaken conclusion. You said "Because being determined is absolute and independent, the fallacy of composition does not apply." The problem here is that because you have not defined what will is, you are taking determinism as if it is a unequivocal primary. Why do you say that determinism is absolute and independent? I assume, since you do not make yourself explicit, that you are thinking along these lines: The motions of particles are governed by Newtonian mechanics. Newtownian mechanics is time reversible and, given enough knowledge, future states of a system are fully predictable from early states, therefore, since brains are made of particles, and their actions are uniquely predictable based on knowledge of prior states in an absolute and independent way, there is no fallacy in asserting determinism.

Now this may not be your argument, but I think it is not unreasonable to infer something like this argument is implicit in what you said. So I am disagreeing with that implicit argument, not you personally.

And the argument is rife with problems. First and foremost is the implicit argument that the motions of particles is dependent upon their prior motions alone, and not the nature of the particles. Particles, whether they are chickens or atoms, have an internal nature. Drop some chickens from the side of a building. Newtonian physics will predict that they will impact the ground at a speed based upon their acceleration due to gravity. But if some of those chickens are alive, they will generate internal motion that will wreak havoc with your ability to predict their time of impact. The same for atoms. Atoms are not inert substanceless forms. Atoms have an internal nature too. At least some, if not all atoms can decay radioactively. Such decay (while I myself doubt it is metaphyscially random) is for all extents and purposes epistemologically random - we can't predict it. This decay will also cause what is in effect internal motion which, while it does obey the conservation of mass/energy, makes it impossible to absolutely predict their motion since their is no way to have perfect knowledge of their prior states.

But we need not even go into random nuclear decay to see that there is a problem with holding "determinism" as absolute and independent. In just the same way that we cannot say that because Newtonian mechanics is absolute and independent that chickens are gravitationally determined, we cannot say a priori that since the motions of particles are predictable, the choices of men are determined.

Indeed, what does "determined" even mean in this context? We don't assert that chickens violate or are not governed by gravity. We come up with the Aristotelean notion of self-generated internal motion, and integrate that with our knowledge of gravity to understand the mechanics of dropped chickens. Likewise, we need to begin with defining what the will is, before we can integrate it with our notions of the motions of particles or other types of causation of animal motion.

I think it is fruitful to examine a hierarchy of organisms and their behaviors. If we start with a million bacteria each with the same genetics and placed in the same nutrient gradient they will all move in the same direction based on their nature and the external stimulus. Yet if we start with two human twins and place them in identical beds, they will not necessarily exit the beds at the same time or on the same side. The nature of simple organisms is such that they can pretty much be governed in their actions by on/off yes/no mechanisms. But in the case of a human waking from sleep, external stimuli are no sufficient to determine whether he will jump from bed, wake up slowly, hit the snooze, or decide to pull a John Lennon. External stimuli are not sufficient to prompt a man to put on his right shoe first or his left. But he must choose some option in order to forward his own life, even if no option on its own is sufficiently compelling to force him to act. Whether we choose to wear the blue shirt or the green one is not determined by a chemical gradient in the air. Yet we can indeed act, we can indeed choose, even when external circumstances are insufficient to compel or determine our actions in one way or another. We do have the ability to choose between the blue and the green shirt when no external cause demands that we do so. This capacity is what I call volition.

Post 14

Thursday, July 23, 2009 - 2:25pm

Hi Ted,

Thank you for the sanction. I suppose in figuring out whether "being determined" is absolute and independent, we need to have a better idea of what it means to "be determined" in the first place. I think we can understand the term without reference to "will," even though an understanding of that term would be imminently useful in its own right.

Traditionally, "being determined" just means that the past, coupled with the laws of nature, entail every truth about the future. For this view, we needn't evoke Newtonian mechanics, nor need we limit causal efficacy only to external stimuli or even to strictly antecedent causation, nor need we part ways with Rands notion that what a thing is *determines* what it will do. Being determined remains absolute (no need to compare it) and independent (the structure of a whole doesn't affect its "determinedness"), hence expansive. I'm not sure if this is how Warren used the term.

Jordan

Post 15

Thursday, July 23, 2009 - 2:42pm

For clarification, what I meant by "metaphysically determined", is what I interpret as the meaning intended by determinists who follow such lines of reasoning -- that the "particles" are PREdetermined, due to an antecedent chain of cause-and-effect reactions. Which does seem to rest on an essentially Newtonian model of existence. At least, that's what I intuit.

Post 16

Thursday, July 23, 2009 - 4:53pm

I'm not sure, with that clarification, it affects the finding of post #5. The property of having an antecedent cause-effect chain is still expansive. It might be immune from fallacy of composition but vulnerable on other fronts. Perhaps this is not unlike the property of, say, the alleged property of being a product of a god's will, which is also an expansive property, hence exempt from the fallacy of composition, but nonetheless seriously vulnerable to loads of other criticisms.

Jordan

Post 17

Thursday, July 23, 2009 - 8:36pm

Imagine that you had a pocket universe, with a tiny little big bang and a bunch of particles resulting. While the mathematics might be more than anyone could afford to do, in theory you could exactly model this universe and present an accurate picture of it at any point "t" in time. One could, for example, create an identical duplicate, with the same exact starting conditions and simply observe and simply record at various times, knowing that the next itteration would do exactly the same things. The Heisenberg Uncertainly Principle would require creating a separate universe for each observation of course. (Whether one could actually do the math is irrelevant to the issue of determinism.)

Now, however, imagine that you impacted this universe. Let's say it was created in your laboratory using something like the big collider in the EU. So now you have this wormhole into your pocket universe, and you send something thru the wormhole into it at point "t2." From that instant on, you would have to report that the events in the pocket universe were now consisting of the original trajectories and interactions as altered by your intervention.

Maybe you created this pocket universe for the purpose of manipulating it toward a specific result, as part of an experimental physics series. Now you are perhaps in the position of actively responding to the permutations that you have caused. I.e., YOU are a determining factor in the events of this universe. From the position of someone inside the universe, it would be silly to try to deny that YOU were a causational factor, in addition to the masses, attributes and trajectories of the original particles.

But how does that differ in causational principle from actually exists now between you and the rest of this existing universe?

And, if there is no difference for the purposes of discussion, then what about the observer inside our completely describable universe who evolved into sentience sufficient to ask questions about determinism even before we tweaked his universe? How can we BOTH say that we can precisely model this universe and thereby predict exactly what will happen at any t(n) and ALSO ascribe "free will" to him? Or, if we can't ascribe it to him, because we can precisely predict his actions, then how can we ascribe it to US? Couldn't a similar interaction be going on from outside our own pocket universe?

I hope that this illustrates the problem of the concept of "free will."

Post 18

Friday, July 24, 2009 - 3:58am

And, if there is no difference for the purposes of discussion, then what about the observer inside our completely describable universe who evolved into sentience sufficient to ask questions about determinism even before we tweaked his universe?
........................

Common error here - sentience is the ability to feeling pleasure and/or pain... sapience is the ability to think, to reason... it is from being sapient that morals are derived, not sentiencey...

Post 19

Friday, July 24, 2009 - 6:22am