Blame David Hume

[Mindy Newton]

Isn't concept-formation inductive and concept-application deductive? We generalize, and then we apply those generalizations to specific cases, right? And those processes are, respectively, induction and deduction -- at least, in Rand's use of the terms. [emphasis added]

Does she anywhere else use the terms in so vaguely non-technical a way?

I looked in the Lexicon. There's no entry for "Deduction." The entry for "Induction and Deduction" cites exactly the quote from pg. 28 of ITOE, no other reference.

I think she was seeing a "family resemblance" -- generalizing and applying -- and using the terms "induction" and "deduction" loosely in that context. (I'm being charitable. ;-))

Ellen

___

I don't understand how the terms are being used loosely, vaguely, or "non-technically" (unless that was supposed to be "vaguely non-technical." Do you limit induction to enumeration?

--Mindy

[Ellen Stuttle]

I don't understand how the terms ["deduction" and "induction"] are being used loosely, vaguely, or "non-technically" (unless that was supposed to be "vaguely non-technical." Do you limit induction to enumeration?

I mean by "induction" reasoning from a group of non-contradicted particular observations "that X (obtains)" to the universalized proposition "that X (obtains)" (whether by virtue of hypothesized intervening causal mechanisms or not).

I've noticed, however, that often what Objectivists mean by "induction" -- or "inductively" -- is simply generalizing from experience. I'm in process of reading (I've almost finished it) a recent (2007) paper by a prominent Objectivist, industrial psychologist Edwin A. Locke, in which "induction" and "inductively" are used just that way. (See * below.)

This sort of vaguely broad usage can lead Objectivists down paths of error: For instance, into thinking that Hume, in denying the validity of inductive reasoning, was claiming that one can't learn anything from experience. Worse, into thinking that Rand's doctrine of "contextual certainty" answers Hume's "logical problem of induction." It doesn't. It accedes that Hume was right. Rand's "contextual certainty" doctrine holds that one could arrive at a generalization which is non-contradicted in one's current context of knowledge but which could turn out to be inadequate to an expanded context of knowledge. Thus her "contextual certainty" isn't in disagreement with the point in regard to the tentativeness of our generalizations which in other circles is called "scientific skepticism." In that respect it uses a different label for the same thesis (or one might say a new bottle for old wine).

--

* For instance, Locke writes:

"The Case for Inductive Theory Building"

Journal of Management, Vol. 33 No. 6, December 2007 867-890

DOI: 10.1177/0149206307307636

© 2007 Southern Management Association. All rights reserved.

pg. 882

Concepts are formed inductively, from observing reality. The facts discovered about members of the class are generalized to all members, including those not yet seen.

The first sentence illustrates the vague use of "inductively." The second I find a rather remarkable statement, given the obvious discrepancy with Locke's own brief description in the same paper of Rand's theory of concept-formation, in which description he says:

pg. 881

[his emphasis]

But how do you form the concept chair when in reality every chair is in some way different from every other (e.g., even "identical" chairs will differ somewhere by a millimeter [also of course in location -- my add][?] Her [Rand's] original insight was that, although the shape and dimensions of chairs entail a range of measurements, in forming the concept, the measurements are omitted. They are assumed to exist but are not stated.

How then could "facts discovered about members of the class" be "generalized to all members, including those not yet seen," given that, in particular measurements, each member of the class is said to differ from all others? From what he goes on to say, I have to assume he's talking narrowly about causal "facts discovered about members of the class."

He writes (repeating the above two sentences and continuing):

pg. 881

[his emphasis]

But how do you form the concept chair when in reality every chair is in some way different from every other (e.g., even "identical" chairs will differ somewhere by a millimeter [also of course in location -- my add][?] Her [Rand's] original insight was that, although the shape and dimensions of chairs entail a range of measurements, in forming the concept, the measurements are omitted. They are assumed to exist but are not stated. How then do we discover causal generalizations?

A provocative new approach to induction (Peikoff, 2007) argues that contrary to Hume, our first-discovered causal relationships are perceived directly, as when a child pushes a ball away from him. Even the phenomenon of gravity has a perceptual base, as when a child drops an object to the floor. More advanced causal generalizations require valid concepts. For example, Galileo, despite his great achievements, never grasped the concept of gravity, and this led him into errors (Harriman, 2007). Early psychologists never grasped what consciousness was, and this prevented them from being able to form and use valid psychological concepts and thus really understand human action (e.g., that goals affect how people act).

If this new approach is valid, it means that, like concepts, causal generalizations are based on inductions starting at the perceptual level. I noted earlier, however, that Bacon had said that induction is not a total enumeration of instances. We all know that demonstrating that x is often found with y is not evidence for causality. To show that x causes, or is a causal factor in y, based on the nature of x and y, is a critical step in making valid inductive generalizations.

[References:

Peikoff, L. 2007. Induction in physics and philosophy (Taped lecture). Available at www.aynrandbookstore.com

Harriman, D. 2007. Induction and experimental method. Objective Standard, 2(1):73-108.]

By that point, Locke is using the idea of "induction" in closer to its historic scientific use, but if one has to go through the procedure indicated in order to arrive at "valid [sic] inductive generalizations," then loop back to the notion that concepts are "formed inductively" in the first place: Seems to me that one is now revolving around a wheel on which there's no beginning!

I have the same objection to Rand's description in ITOE of how one arrives at the "essential" characteristic(s). She says (pg. 45, 1990) of "the process of determining an essential characteristic" that one proceeds by "the rule of fundamentality" (her emphasis), which she describes thus:

ITOE, 1990, pg. 45

When a given group of existents has more than one characteristic distinguishing it from other existents, man must observe the relationships among these various characteristics and discover the one on which all the others (or the greatest number of others) depend, i.e., the fundamental characteristic without which the others would not be possible. This fundamental characteristic is the essential distinguishing characteristic of the existents involved, and the proper defining characteristic of the concept.

Metaphysically, a fundamental characteristic is that distinctive characteristic which makes the greatest number of others possible; epistemologically, it is the one that explains the greatest number of others.

Thus, since she says that a definition is required for completing a concept, again it seems to me that we're presented with an endlessly circling wheel with no beginning: A causal theory is required to form a valid concept but valid concepts are required to from a causal theory......

Ellen

___

[Michael Stuart Kelly]

Ellen,

It will always go in circles so long as observation is treated as an unwanted bastard child of the mind, where pure logic and pure mathematics allegedly reside ("pure" meaning observation expunged from thought as much as humanly possible). This is allegedly the only place in the universe where certainty is possible and the bastard child keeps throwing a monkeywrench into those works.

But Objectivists include observation as part of reasoning. I see no reason to use the word "error" in talking about the Objectivist habit of using the word induction to mean "observe a bunch of similarities and differences and make a category" (which means form a concept).

You are trying to restrict induction to mean "prove a proposition" (like in a syllogism, which is deductive by definition), then claim that the Objectivist use of the word is an "error."

Without even getting into the merits of that, I learned a long time ago that dictionaries exist and that most words have more than one meaning without there being any "error."

A real error would be an Objectivist trying to use the inductive process of concept formation with the meaning you give to the word induction. But they don't. They define their terms and they are consistent. There is no error. There are only semantics and finger-pointing.

Anyway, I have no idea how a proposition can be formed without categories in the first place. That is another thing that will always go around in circles.

Michael

[Ellen Stuttle]

MSK,

It's often an exercise in amazement wondering where you get from a post the mis-readings you rail against. (You might try actually reading the post, carefully, before spouting off.)

Ellen

[tjohnson]

ITOE, 1990, pg. 45

When a given group of existents has more than one characteristic distinguishing it from other existents, man must observe the relationships among these various characteristics and discover the one on which all the others (or the greatest number of others) depend, i.e., the fundamental characteristic without which the others would not be possible. This fundamental characteristic is the essential distinguishing characteristic of the existents involved, and the proper defining characteristic of the concept.

Could someone give an example of this? How about 'table'? What is the fundamental characteristic of a table?

[merjet]

How about 'table'? What is the fundamental characteristic of a table?

"A man-made object consisting of a flat, level surface and support(s), intended to support other, smaller objects." - ITOE, 11.

"A piece of furniture consisting of a smooth flat slab fixed on legs." - Merriam-Webster Online Dictionary

[Michael Stuart Kelly]

MSK,

It's often an exercise in amazement wondering where you get from a post the mis-readings you rail against. (You might try actually reading the post, carefully, before spouting off.)

Ellen,

You are right up to a point. I was way too hasty and oversimplified a focus I perceived to the point of error. Sorry.

Here is what twitched my antenna:

This sort of vaguely broad usage can lead Objectivists down paths of error: For instance, into thinking that Hume, in denying the validity of inductive reasoning, was claiming that one can't learn anything from experience. Worse, into thinking that Rand's doctrine of "contextual certainty" answers Hume's "logical problem of induction." It doesn't.

This is a part that really needs fleshing out because of the omission of causality, which is a fundamental part of Rand's argument. You continue:

It accedes that Hume was right. Rand's "contextual certainty" doctrine holds that one could arrive at a generalization which is non-contradicted in one's current context of knowledge but which could turn out to be inadequate to an expanded context of knowledge. Thus her "contextual certainty" isn't in disagreement with the point in regard to the tentativeness of our generalizations which in other circles is called "scientific skepticism." In that respect it uses a different label for the same thesis (or one might say a new bottle for old wine).

There is something true in here and I have claimed it for a while now (especially in the discussions on Popper). I call it establishing a category. But it is more than just arriving at a generalization as you are stating. Actually the idea of making a category has very little to do with the following:

I've noticed, however, that often what Objectivists mean by "induction" -- or "inductively" -- is simply generalizing from experience.

A category is a very specific notion, even though it is open-ended to be able to include more details of the existent as they are discovered. It is arrived at through observation (experience), but it is not a vague lump of impressions. There is discernment, like noticing similarities and differences and figuring out which are the most important.

What I really object to is the lack of recognizing that this is what Objectivists do (form specific concepts, make specific categories), and instead claiming that they are being somehow over-general and this leads to their myriad errors.

Michael

[tjohnson]

"A man-made object consisting of a flat, level surface and support(s), intended to support other, smaller objects." - ITOE, 11.

"A piece of furniture consisting of a smooth flat slab fixed on legs." - Merriam-Webster Online Dictionary

OK, so which characteristic is the fundamental one and how are the others not possible without it? Is it the flatness, levelness, intended use?

[merjet]

OK, so which characteristic is the fundamental one and how are the others not possible without it? Is it the flatness, levelness, intended use?

All. Rand wasn't entirely consistent. Sometimes she used the singular, sometimes the plural. An excerpt that includes the latter:

"A definition is a statement that identifies the nature of a concept's units. A correct definition must specify the distinguishing characteristic(s) of the units (the differentia), and indicate the category of existents from which they were differentiated (the genus). The essential distinguishing characteristic(s) of the units and the proper defining characteristic(s) of the concept must be a fundamental characteristic(s)—i.e., that distinctive characteristic(s) which, metaphysically, makes the greatest number of other distinctive characteristics possible and which, epistemologically, explains the greatest number of others" (ITOE 84-5).

[Michael Stuart Kelly]

Merlin,

I have noticed the singular-plural issue with Rand's fundamental characteristic also.

In addition to that, I gave myself some exercises in concept formation to see if I could isolate a single fundamental characteristic. I got stumped right at the outset with giraffe. What is the one distinguishing characteristic of a giraffe? I see a lot of 'em.

I have a thought I want to leave somewhere, so here seems to be a good place. Rand seemed almost paranoid when the subject of man or human nature was involved in deep philosophical issues. She stressed individualism to the extent that issues like human species were fudged. The above case is another example. I cannot imagine her coming up with a single fundamental characteristic for giraffe (as opposed to many), but man has to have only one.

I speculate, but I think she was horribly afraid of what happens when collectivism runs wild due to growing up in Communist Russia. Thus when a subject looked like it could be used as intellectual grounds for collectivism, she came up with another theory to preempt that from happening. Like I said, that's only a speculation. But the more I read and mull this over, the more I see signs that point to it.

Nyquist accused Rand of imagining her ideal man, then building a philosophy to fit that vision. Although I do not buy that view for the entire philosophy, not even the majority of it, the charge does fit Rand's lapses and fits her radical anti-collectivism like a glove. It even explains that "only a man should be president" thing.

Michael

[tjohnson]

All. Rand wasn't entirely consistent. Sometimes she used the singular, sometimes the plural. An excerpt that includes the latter:

"A definition is a statement that identifies the nature of a concept's units. A correct definition must specify the distinguishing characteristic(s) of the units (the differentia), and indicate the category of existents from which they were differentiated (the genus). The essential distinguishing characteristic(s) of the units and the proper defining characteristic(s) of the concept must be a fundamental characteristic(s)—i.e., that distinctive characteristic(s) which, metaphysically, makes the greatest number of other distinctive characteristics possible and which, epistemologically, explains the greatest number of others" (ITOE 84-5).

This doesn't seem workable to me. It seems like there would be a great number of opinions about "which, metaphysically, makes the greatest number of other distinctive characteristics possible and which, epistemologically, explains the greatest number of others". I don't see any method for establishing this criteria.

[tjohnson]

I speculate, but I think she was horribly afraid of what happens when collectivism runs wild due to growing up in Communist Russia. Thus when a subject looked like it could be used as intellectual grounds for collectivism, she came up with another theory to preempt that from happening. Like I said, that's only a speculation. But the more I read and mull this over, the more I see signs that point to it.

I think you might be on to something there.

[Daniel Barnes]

(Ellen quoted Edwin Locke:)

A provocative new approach to induction (Peikoff, 2007) argues that contrary to Hume, our first-discovered causal relationships are perceived directly, as when a child pushes a ball away from him.

LOL! So this is Peikoff's brilliant theory at last. Sadly, after decades of massive philosophical exertion he doesn't seem to have noticed it doesn't reply to Hume at all.

(Edwin Locke:)

Even the phenomenon of gravity has a perceptual base, as when a child drops an object to the floor.

"Perceptual base" indeed! Hume has a "perceptual base" too! The point is the inference from the "perceptual base."

(Edwin Locke:)

Early psychologists never grasped what consciousness was, and this prevented them from being able to form and use valid psychological concepts and thus really understand human action (e.g., that goals affect how people act).

What outstanding drivel this Locke fellow writes.

(Edwin Locke:)

If this new approach is valid, it means that, like concepts, causal generalizations are based on inductions starting at the perceptual level. I noted earlier, however, that Bacon had said that induction is not a total enumeration of instances. We all know that demonstrating that x is often found with y is not evidence for causality. To show that x causes, or is a causal factor in y, based on the nature of x and y, is a critical step in making valid inductive generalizations.

Even more drivelicious, with extra points for stealing yet again the concept "valid."

Worse, into thinking that Rand's doctrine of "contextual certainty" answers Hume's "logical problem of induction." It doesn't. It accedes that Hume was right.

Absolutely. Let's just say that again, loud and clear: Rand's "contextual certainty" in fact accedes that Hume was right.

[merjet]

I have noticed the singular-plural issue with Rand's fundamental characteristic also.

In addition to that, I gave myself some exercises in concept formation to see if I could isolate a single fundamental characteristic. I got stumped right at the outset with giraffe. What is the one distinguishing characteristic of a giraffe? I see a lot of 'em.

Now observe, on the above example, the process of determining an essential characteristic: the rule of fundamentality. When a given group of existents has more than one characteristic distinguishing it from other existents, man must observe the relationships among these various characteristics and discover the one on which all the others (or the greatest number of others) depend, i.e., the fundamental characteristic without which the others would not be possible. This fundamental characteristic is the essential distinguishing characteristic of the existents involved, and the proper defining characteristic of the concept.

Metaphysically, a fundamental characteristic is that distinctive characteristic which makes the greatest number of others possible; epistemologically, it is the one that explains the greatest number of others. (ITOE, 35)

I think Rand lacked rigor and confounded two key terms -- fundamental and distinguishing.

Consider the following definition of giraffe: a large fleet African ruminant mammal (Giraffa camelopardalis) that is the tallest of living quadrupeds and has a very long neck and a short coat with dark blotches separated by pale lines - Merriam-Webster Online Dictionary

The most distinguishing characteristics of a giraffe are its height and very long neck. However, these don't explain much else, arguably only its ability to feed on trees. On the other hand, the giraffe being a ruminant mammal says a lot about it -- most of its body parts and more.

[Ellen Stuttle]

You are right up to a point. I was way too hasty and oversimplified a focus I perceived to the point of error. Sorry.

Thanks.

Here is what twitched my antenna:

This sort of vaguely broad usage can lead Objectivists down paths of error: For instance, into thinking that Hume, in denying the validity of inductive reasoning, was claiming that one can't learn anything from experience. Worse, into thinking that Rand's doctrine of "contextual certainty" answers Hume's "logical problem of induction." It doesn't.

I thought that that was probably what triggered your response.

Let's recap the background of what I was talking about there. I'd been pointing out that in the only place in the text of ITOE where Rand mentions "induction" and "deduction" -- the passage on pg. 28 -- she isn't using the terms in the way they're used in formal logic. She was using a broader meaning, whereby "induction" is referring to forming conclusions arrived at from experience and "deduction" to applying previous conclusions to new examples.

This broad meaning of "induction" (and of "inductively") -- I went on to say in the post to which you replied -- is often what Objectivists are talking about when they speak of arriving at conclusions by "induction" (or "inductively").

But they can then be thrown off, with resultant error in interpreting thinkers other than Rand, because they import their usage into what those other thinkers are talking about even when those other thinkers are using the formal meanings. They thus end up believing, for example, that David Hume denied that you can get knowledge from experience ("induction" broadly defined) because he denied the validity of inductive inference as a formal method of reasoning from particulars to certainty about general laws. The error I'm talking about is their not realizing that they aren't meaning the same procedure by the same term.

I'll continue in a second post so as to lessen the clutter of nested quotes.

Ellen

___

[Ellen Stuttle]

This sort of vaguely broad usage can lead Objectivists down paths of error: For instance, into thinking that Hume, in denying the validity of inductive reasoning, was claiming that one can't learn anything from experience. Worse, into thinking that Rand's doctrine of "contextual certainty" answers Hume's "logical problem of induction." It doesn't.

This is a part that really needs fleshing out because of the omission of causality, which is a fundamental part of Rand's argument.

I don't understand the comment.

You continue:

It accedes that Hume was right. Rand's "contextual certainty" doctrine holds that one could arrive at a generalization which is non-contradicted in one's current context of knowledge but which could turn out to be inadequate to an expanded context of knowledge. Thus her "contextual certainty" isn't in disagreement with the point in regard to the tentativeness of our generalizations which in other circles is called "scientific skepticism." In that respect it uses a different label for the same thesis (or one might say a new bottle for old wine).

There is something true in here and I have claimed it for a while now (especially in the discussions on Popper).

Yes, I'm aware that you did see, during the discussions on Popper, that the difference between "contextual certainty" and "scientific skepticism" was more that of a different way of describing the situation than of substance.

However, I don't understand the next comment:

I call it establishing a category.

You continue directly:

But it is more than just arriving at a generalization as you are stating.

But I'm NOT stating that establishing a category is just arriving at a generalization. I even agree with your next comments:

Actually the idea of making a category has very little to do with the following:

I've noticed, however, that often what Objectivists mean by "induction" -- or "inductively" -- is simply generalizing from experience.

A category is a very specific notion, even though it is open-ended to be able to include more details of the existent as they are discovered. It is arrived at through observation (experience), but it is not a vague lump of impressions. There is discernment, like noticing similarities and differences and figuring out which are the most important.

I would therefore say that Rand's throwing in the reference to "induction" and "deduction" in that brief comment on pg. 28 was unfortunate because the result was to confuse the issue of the process she was describing rather than to clarify.

What I really object to is the lack of recognizing that this is what Objectivists do (form specific concepts, make specific categories), and instead claiming that they are being somehow over-general and this leads to their myriad errors.

That isn't what I was claiming, as I hope is clear now from the previous post.

Ellen

___

[Ellen Stuttle]

Daniel,

The Locke article is a "mixed bag." The parts which are pretty straight O'ism I mostly think are poor, although he does do a good job of briefly describing Rand's theory of concept-formation. He's simplistic or outright wrong in his interpretation of Popper.

HOWEVER, there are features of his suggestions re methodology in psychology that I could very much wish would have been written and paid heed to back in the '60s. They'd have been thrown out of court without any consideration by most research psychologists back then, but they provide leads to sound method in psychology.

(The paper is amongst the subjects slated for discussion at our traditional Thanksgiving Seminar.)

Ellen

___

[Ellen Stuttle]

I speculate, but I think she was horribly afraid of what happens when collectivism runs wild due to growing up in Communist Russia. Thus when a subject looked like it could be used as intellectual grounds for collectivism, she came up with another theory to preempt that from happening. Like I said, that's only a speculation. But the more I read and mull this over, the more I see signs that point to it.

I think you might be on to something there.

I think so, too.

Ellen

___

[tjohnson]

Here is a connection between induction and probability theory that goes back to Laplace's time.

The point is that many of the most important problems of current science and engineering are just problems of inductive reasoning, in which no "random experiment" is involved in any way.

....the relations of probability theory are perfectly valid when used in the Laplace sense of the "calculus of inductive reasoning", whether or not there is any connection between probability and frequency.

....I hope to show that we can not only obtain important, useful, and non-trivial results; we can also clear up some of the paradoxes surrounding communication theory, statistical mechanics, and quantum mechanics.

[Daniel Barnes]

I speculate, but I think she was horribly afraid of what happens when collectivism runs wild due to growing up in Communist Russia. Thus when a subject looked like it could be used as intellectual grounds for collectivism, she came up with another theory to preempt that from happening. Like I said, that's only a speculation. But the more I read and mull this over, the more I see signs that point to it.

I think you might be on to something there.

I think so, too.

Ellen

___

Ditto that.

[Michael Stuart Kelly]

Ellen,

Here is where I am confused about with your position. Here is your first statement (and I put the part in bold that jumped out at me).

I'd been pointing out that in the only place in the text of ITOE where Rand mentions "induction" and "deduction" -- the passage on pg. 28 -- she isn't using the terms in the way they're used in formal logic. She was using a broader meaning, whereby "induction" is referring to forming conclusions arrived at from experience and "deduction" to applying previous conclusions to new examples.

That is not precise. Here are Rand's words (borrowing the quote from my post above so I don't have to look it up again). I am putting the pertinent part in bold:

ITOE (2nd), Chapter 3, Abstraction from Abstractions, p. 28:

Thus the process of forming and applying concepts contains the essential pattern of two fundamental methods of cognition: induction and deduction.

The process of observing the facts of reality and of integrating them into concepts is, in essence, a process of induction. The process of subsuming new instances under a known concept is, in essence, a process of deduction.

Integrating facts of reality into a concept is not the same as "forming conclusions arrived at from experience." There might be a time lapse here that is causing the confusion. Concept formation happens at the beginning, with the first integration, not after the concept is made. Forming a concept is essentially identifying a category.

For instance, a baby forms a concept of man after he has encountered a few. The file folder gets opened. He will no longer in his life form a new concept of man. Instead, he will add to the already formed concept. Over time he will form new concepts about different aspects of man and add those to his original concept, but never again will he do the concept of man from scratch. Once done, its done.

That is not the same thing as forming a conclusion, which is more in the universe of propositions, not primary cognitive identification. I see you understand the difference with the following phrase:

But I'm NOT stating that establishing a category is just arriving at a generalization.

I don't understand how you reconcile your first statement with this one in light of Rand's words, unless maybe you are thinking about the time after a concept has been formed and imagining that she is also talking about that. Then I agree that this would be forming conclusions (dealing with propositions), but that is not what I understand Rand to be doing.

btw - I do agree that many Objectivists I have read (especially snarky posters) do not understand concept formation or induction or Rand's attribution of induction as a process of concept formation. They trash Hume et al., but I do not attribute this to their use of induction as meaning a generalization, then transposing this meaning to Hume. I understand these people to be aping Rand and Peikoff and that's about as far as it goes.

I will agree that the more intelligent Objectivists I have read (including Rand) mistakenly attribute the concept formation idea I gave above (i.e., establishing categories) to Hume's meaning of induction, which instead deals with deductive propositions after a concept has been established, not primary identification. (His concept of swans already exists. He is only interested in whether he will find an oddball member or not. He is not interested in looking at a bunch of birds he has never seen before and figuring out which ones are similar enough to each other and different enough from all others to form a new category, i.e. swan.)

And when you remember that Rand considered entities to be (1) a source of causality, and (2) the primary referent for the lowest level concepts, these Objectivists arrive at the conclusion that Hume was saying you can't really form a concept of an entity that includes all members of that category, thus you cannot really identify a category except for some loose verbal convenience, thus you cannot identify the causality arising from the entity (or type of entity) since not all members will have it. When that premise is taken to a logical conclusion, they continue that Hume was denying that causality even exists.

This time lapse is critically important to understanding what each side is talking about, i.e., whether one is discussing before the concept is formed or after. This is what I meant when I said time should not be brushed aside if predicting the future is what Hume (and scientists in general) was after with the problem of induction.

Michael

[Ellen Stuttle]

Ellen,

Here is where I am confused about with your position. Here is your first statement (and I put the part in bold that jumped out at me).

I'd been pointing out that in the only place in the text of ITOE where Rand mentions "induction" and "deduction" -- the passage on pg. 28 -- she isn't using the terms in the way they're used in formal logic. She was using a broader meaning, whereby "induction" is referring to forming conclusions arrived at from experience and "deduction" to applying previous conclusions to new examples.

That is not precise. Here are Rand's words (borrowing the quote from my post above so I don't have to look it up again). I am putting the pertinent part in bold:

ITOE (2nd), Chapter 3, Abstraction from Abstractions, p. 28:

Thus the process of forming and applying concepts contains the essential pattern of two fundamental methods of cognition: induction and deduction.

The process of observing the facts of reality and of integrating them into concepts is, in essence, a process of induction. The process of subsuming new instances under a known concept is, in essence, a process of deduction.

Michael, the source of confusion here was Rand because of her using the terms "induction" and "deduction" in that context at all. I'm pointing out that she was using the broader sense of the terms; however, she was using even that sense of the terms loosely, not precisely; as I said in an earlier post, she was making an analogy. Her "in essence" (and "the essential pattern") is fudge language in the context.

Integrating facts of reality into a concept is not the same as "forming conclusions arrived at from experience."

I agree with you. It isn't the same. Which is why she shouldn't have made that statement.

That is not the same thing as forming a conclusion, which is more in the universe of propositions, not primary cognitive identification.

Again, I agree. The terms "induction" and "deduction" refer to "the universe of propositions." The process of concept formation Rand describes isn't one of propositional reasoning.

I see you understand the difference with the following phrase:

But I'm NOT stating that establishing a category is just arriving at a generalization.

I don't understand how you reconcile your first statement with this one in light of Rand's words [...].

Very easily, since I think that Rand's words in that passage are erroneous and misleading and only loosely applicable to the process she herself described. I think the passage is a goof. In saying that I think she had the broad meanings of "induction" and "deduction" in mind in writing the passage, I'm not therefore saying that I think that even these broad meanings (let alone the technical meanings) are correct in the context. I think that they're incorrect in the context.

I'm afraid I couldn't follow the rest of your post well enough to comment on it.

Ellen

___

[Mindy Newton]

I don't understand how the terms ["deduction" and "induction"] are being used loosely, vaguely, or "non-technically" (unless that was supposed to be "vaguely non-technical." Do you limit induction to enumeration?

I mean by "induction" reasoning from a group of non-contradicted particular observations "that X (obtains)" to the universalized proposition "that X (obtains)" (whether by virtue of hypothesized intervening causal mechanisms or not).

I've noticed, however, that often what Objectivists mean by "induction" -- or "inductively" -- is simply generalizing from experience. I'm in process of reading (I've almost finished it) a recent (2007) paper by a prominent Objectivist, industrial psychologist Edwin A. Locke, in which "induction" and "inductively" are used just that way. (See * below.)

This sort of vaguely broad usage can lead Objectivists down paths of error: For instance, into thinking that Hume, in denying the validity of inductive reasoning, was claiming that one can't learn anything from experience. Worse, into thinking that Rand's doctrine of "contextual certainty" answers Hume's "logical problem of induction." It doesn't. It accedes that Hume was right. Rand's "contextual certainty" doctrine holds that one could arrive at a generalization which is non-contradicted in one's current context of knowledge but which could turn out to be inadequate to an expanded context of knowledge. Thus her "contextual certainty" isn't in disagreement with the point in regard to the tentativeness of our generalizations which in other circles is called "scientific skepticism." In that respect it uses a different label for the same thesis (or one might say a new bottle for old wine).

--

* For instance, Locke writes:

"The Case for Inductive Theory Building"

Journal of Management, Vol. 33 No. 6, December 2007 867-890

DOI: 10.1177/0149206307307636

© 2007 Southern Management Association. All rights reserved.

pg. 882

Concepts are formed inductively, from observing reality. The facts discovered about members of the class are generalized to all members, including those not yet seen.

The first sentence illustrates the vague use of "inductively." The second I find a rather remarkable statement, given the obvious discrepancy with Locke's own brief description in the same paper of Rand's theory of concept-formation, in which description he says:

pg. 881

[his emphasis]

But how do you form the concept chair when in reality every chair is in some way different from every other (e.g., even "identical" chairs will differ somewhere by a millimeter [also of course in location -- my add][?] Her [Rand's] original insight was that, although the shape and dimensions of chairs entail a range of measurements, in forming the concept, the measurements are omitted. They are assumed to exist but are not stated.

How then could "facts discovered about members of the class" be "generalized to all members, including those not yet seen," given that, in particular measurements, each member of the class is said to differ from all others? From what he goes on to say, I have to assume he's talking narrowly about causal "facts discovered about members of the class."

He writes (repeating the above two sentences and continuing):

pg. 881

[his emphasis]

But how do you form the concept chair when in reality every chair is in some way different from every other (e.g., even "identical" chairs will differ somewhere by a millimeter [also of course in location -- my add][?] Her [Rand's] original insight was that, although the shape and dimensions of chairs entail a range of measurements, in forming the concept, the measurements are omitted. They are assumed to exist but are not stated. How then do we discover causal generalizations?

A provocative new approach to induction (Peikoff, 2007) argues that contrary to Hume, our first-discovered causal relationships are perceived directly, as when a child pushes a ball away from him. Even the phenomenon of gravity has a perceptual base, as when a child drops an object to the floor. More advanced causal generalizations require valid concepts. For example, Galileo, despite his great achievements, never grasped the concept of gravity, and this led him into errors (Harriman, 2007). Early psychologists never grasped what consciousness was, and this prevented them from being able to form and use valid psychological concepts and thus really understand human action (e.g., that goals affect how people act).

If this new approach is valid, it means that, like concepts, causal generalizations are based on inductions starting at the perceptual level. I noted earlier, however, that Bacon had said that induction is not a total enumeration of instances. We all know that demonstrating that x is often found with y is not evidence for causality. To show that x causes, or is a causal factor in y, based on the nature of x and y, is a critical step in making valid inductive generalizations.

[References:

Peikoff, L. 2007. Induction in physics and philosophy (Taped lecture). Available at www.aynrandbookstore.com

Harriman, D. 2007. Induction and experimental method. Objective Standard, 2(1):73-108.]

By that point, Locke is using the idea of "induction" in closer to its historic scientific use, but if one has to go through the procedure indicated in order to arrive at "valid [sic] inductive generalizations," then loop back to the notion that concepts are "formed inductively" in the first place: Seems to me that one is now revolving around a wheel on which there's no beginning!

I have the same objection to Rand's description in ITOE of how one arrives at the "essential" characteristic(s). She says (pg. 45, 1990) of "the process of determining an essential characteristic" that one proceeds by "the rule of fundamentality" (her emphasis), which she describes thus:

ITOE, 1990, pg. 45

When a given group of existents has more than one characteristic distinguishing it from other existents, man must observe the relationships among these various characteristics and discover the one on which all the others (or the greatest number of others) depend, i.e., the fundamental characteristic without which the others would not be possible. This fundamental characteristic is the essential distinguishing characteristic of the existents involved, and the proper defining characteristic of the concept.

Metaphysically, a fundamental characteristic is that distinctive characteristic which makes the greatest number of others possible; epistemologically, it is the one that explains the greatest number of others.

Thus, since she says that a definition is required for completing a concept, again it seems to me that we're presented with an endlessly circling wheel with no beginning: A causal theory is required to form a valid concept but valid concepts are required to from a causal theory......

Ellen

___

Thanks for that explanation, Ellen. I understand better. I think the Objectivist view is that induction is whatever is bottom-up in terms of abstraction.

I believe you'll find Rand's positions are right in line with Aristotle on this. Her "metaphysical" account of universals is his account of "essence" and her "epistemological" account is his account of definition. Here's how it's put in the article on Aristotle in the Oxford Companion to Philosophy: "...he outlined an account of what each thing's essence is (the feature which provides the fundamental account of its other genuine properties), [and] of how things should be defined (in terms of their basic explanatory features)..." Aristotle also had a sort of syllogism called a demonstration which aimed at explanation, and worked to organize characteristics hierarchically. It isn't enumerative. (This is part of his Analytics.)

The demonstration syllogism may play the role of induction in your detection of circularity in Rand/Locke's account. So it is an intermediate step in forming--fully forming--a concept. A rudimentary concept is a grouping of similar objects (simplifying). But noticing that things resemble one another doesn't immediately separate out all common and not-in-common characteristics. So there is a period of examination and discovery that sorts that out, including the hierarchical organization of facts about and features of the objects.

I'm sure this won't be fully satisfactory, but it offers another perspective on what you see as circularity.

--Mindy

[Michael Stuart Kelly]

Integrating facts of reality into a concept is not the same as "forming conclusions arrived at from experience."

I agree with you. It isn't the same. Which is why she shouldn't have made that statement.

Ellen,

Maybe to you she shouldn't have made the statement, but she did and that's what most of (what I call) the intelligent Objectivists understand induction and deduction to be. You keep insisting that Rand's usage should have been only proposition-based (including for those who adhere to her system), but it isn't, except the primitive level I mention below.

Also, I made that statement because you keep stating and/or insinuating that Rand's usage of induction was really forming conclusions and her using it to mean concept formation was a mistake or somehow was muddying the waters. But waters can't be muddied when meanings are clear. And Rand made her meaning very clear in this instance.

Voila. A clear meaning emerges and you don't like the meaning. All I can say is get used to it.

... since I think that Rand's words in that passage are erroneous and misleading and only loosely applicable to the process she herself described. I think the passage is a goof.

This is where I go back to my dictionary statement. I don't think it was a goof. And I don't think it was misleading. Rand was very clear that she was using the term induction for concept formation (establishing categories), not for propositional reasoning. You can either accept that or not, but it is a huge mistake to call that a mistake. Any dictionary on earth gives more than one meaning for a word. Only on online forums have I seen this practice disparaged.

btw - Making a category actually is, on a very primitive level, a proposition. It can be verbalized as "Those things I observed have enough similarities that they can be grouped into a category." Only after that proposition is satisfied do you go about enumerating some of the similarities (and differences with other stuff). That proposition is pure induction. The problem is you can't falsify it with deduction because you can only verify it with observation.

If you need a definition, on that level you get what Rand called an ostensive definition, essentially pointing and saying "I mean that."

I personally think Rand was brilliant for noticing induction as the essential reasoning process in concept formation. It was a first-class piece of original thinking.

Michael

[Mindy Newton]

Integrating facts of reality into a concept is not the same as "forming conclusions arrived at from experience."

I agree with you. It isn't the same. Which is why she shouldn't have made that statement.

Ellen,

Maybe to you she shouldn't have made the statement, but she did and that's what most of (what I call) the intelligent Objectivists understand induction and deduction to be. You keep insisting that Rand's usage should have been only proposition-based (including for those who adhere to her system), but it isn't, except the primitive level I mention below.

Also, I made that statement because you keep stating and/or insinuating that Rand's usage of induction was really forming conclusions and her using it to mean concept formation was a mistake or somehow was muddying the waters. But waters can't be muddied when meanings are clear. And Rand made her meaning very clear in this instance.

Voila. A clear meaning emerges and you don't like the meaning. All I can say is get used to it.

... since I think that Rand's words in that passage are erroneous and misleading and only loosely applicable to the process she herself described. I think the passage is a goof.

This is where I go back to my dictionary statement. I don't think it was a goof. And I don't think it was misleading. Rand was very clear that she was using the term induction for concept formation (establishing categories), not for propositional reasoning. You can either accept that or not, but it is a huge mistake to call that a mistake. Any dictionary on earth gives more than one meaning for a word. Only on online forums have I seen this practice disparaged.

btw - Making a category actually is, on a very primitive level, a proposition. It can be verbalized as "Those things I observed have enough similarities that they can be grouped into a category." Only after that proposition is satisfied do you go about enumerating some of the similarities (and differences with other stuff). That proposition is pure induction. The problem is you can't falsify it with deduction because you can only verify it with observation.

If you need a definition, on that level you get what Rand called an ostensive definition, essentially pointing and saying "I mean that."

I personally think Rand was brilliant for noticing induction as the essential reasoning process in concept formation. It was a first-class piece of original thinking.

Michael

Ellen,

"Forming conclusions based on experience" would include forming a definition of a group of similar things.

--Mindy