David Harriman's Book

[Robert Campbell]

Since the thread on David Harriman's book The Logical Leap has long since quit being about his book, over the next few days I will be posting some quotations from that opus that are I think are indicative of key points in Harriman's argument.

Comments are welcome.

Robert Campbell

[Robert Campbell]

It should not be difficult to identify which parts of the book came from which brain. In essence, the original philosophic ideas belong to Dr. Peikoff, while I provided their illustration in the history of science. In particular, the philosophic foundation presented in Chapter 1 is taken nearly verbatim from Dr. Peikoff's lectures. Also, I have incorporated into Chapter 2 his discussion of concepts as "green lights to induction." Finally, many of the essential points in Chapter 7, including the explanation for the role of mathematics in physical science, are taken from his lectures.

In addition, every chapter of the book has benefited greatly from his line-by-line scrutiny. […] Dr. Peikoff has been a very generous editor and teacher. Of course, any errors in the science and its history are entirely my responsibility.

While Dr. Peikoff taught me how to write this book, the Ayn Rand Institute gave me time to write it. (Preface, p. 2)

[Robert Campbell]

One widely held but false view is that induction is based simply on enumeration. According to this superficial approach, the inducer's method is merely to collect instances of a generalization; the greater the number of instances, the greater the probability of the generalization (the proponents of this view deny that we can ever reach certainty).

[…]

Enumeration is not the method of induction, and it provides no basis to infer from "some" to "all," not even with a degree of probability. This is why all attempts to ground inductive reasoning on statistics have failed. A generalization reached merely from enumeration is necessarily arbitrary, and must therefore be dismissed without discussion from the field of rational consideration. As we shall see, there are valid inductions based on a single case; and there are generalizations with millions of instances, which are yet utterly illegitimate (e.g., "All men seek pleasure"). Why such a difference? Simple enumerators have no answer. (Chapter 1, pp. 8-9; bolding added by me).

[BaalChatzaf]

One widely held but false view is that induction is based simply on enumeration. According to this superficial approach, the inducer's method is merely to collect instances of a generalization; the greater the number of instances, the greater the probability of the generalization (the proponents of this view deny that we can ever reach certainty).

[…]

Enumeration is not the method of induction, and it provides no basis to infer from "some" to "all," not even with a degree of probability. This is why all attempts to ground inductive reasoning on statistics have failed. A generalization reached merely from enumeration is necessarily arbitrary, and must therefore be dismissed without consideration from the field of rational consideration. As we shall see, there are valid inductions based on a single case; and there are generalizations with millions of instances, which are yet utterly illegitimate (e.g., "All men seek pleasure"). Why such a difference? Simple enumerators have no answer. (Chapter 1, pp. 8-9; bolding added by me).

Very well. Then what grounds induction if not a replication of instances?

Ba'al Chatzaf

[Daniel Barnes]

Most theories of induction are basically Underpants Gnomes-type theories. That is:

Step 1: Observation(s)

Step 2: ???????

Step 3: Induction!

From what I've seen so far Harriman doesn't seem to be an exception.

[George H. Smith]

Most theories of induction are basically Underpants Gnomes-type theories. That is:

Step 1: Observation(s)

Step 2: ???????

Step 3: Induction!

From what I've seen so far Harriman doesn't seem to be an exception.

You can do a better parody if you actually know something about a position. I discussed the approach defended by Harriman (though I didn't have him in mind -- it's a standard approach) in some of those posts on Popper than you didn't want to look for. Popper's arguments against induction are some of the silliest things he ever wrote.

Ghs

[Robert Campbell]

Bob K and Daniel B,

Harriman thinks induction proceeds from a foundation of first-level generalizations (what in Peikoff-land is called the "gen theory"). I'll put up some quotes about the gen theory tomorrow.

George S,

A generalization reached merely from enumeration is necessarily arbitrary, and must therefore be dismissed without discussion from the field of rational consideration.

You've said the approach presented by Harriman is a standard one. Do you agree with Harriman's claim that any generalization based only on instances already observed is arbitrary?

Harriman is Leonard Peikoff's protégé, so it's wisest to assume that he's attributing everything to "the arbitrary" that Peikoff does. (In fact, nearly every aspect of the Peikovian doctrine is presented somewhere in The Logical Leap.)

Robert C

[George H. Smith]

Bob K and Daniel B,

Harriman thinks induction proceeds from a foundation of first-level generalizations (what in Peikoff-land is called the "gen theory"). I'll put up some quotes about the gen theory tomorrow.

George S,

A generalization reached merely from enumeration is necessarily arbitrary, and must therefore be dismissed without discussion from the field of rational consideration.

You've said the approach presented by Harriman is a standard one. Do you agree with Harriman's claim that any generalization based only on instances already observed is arbitrary?

Harriman is Leonard Peikoff's protégé, so it's wisest to assume that he's attributing everything to "the arbitrary" that Peikoff does. (In fact, nearly every aspect of the Peikovian doctrine is presented somewhere in The Logical Leap.)

Robert C

I don't know if I agree with all this. I would have to read Harriman's book. I was thinking of the claim that we don't need complete enumeration to justify an inductive inference. This was discussed in another ongoing thread about Harriman's book -- the really long and polemical one. Here is one of my posts where I address this issue:

http://www.objectivistliving.com/forums/index.php?showtopic=8903&view=findpost&p=105498

I don't know why these damned links don't work. This takes you to #758, whereas I copied the link to #751. But this will get you into the ballpark.

I posted some things on induction both before and after this one, but this looks like a representative sample.

Ghs

[Ellen Stuttle]

Popper's arguments against induction are some of the silliest things he ever wrote.

Here we go again.....

Ellen

[George H. Smith]

Popper's arguments against induction are some of the silliest things he ever wrote.

Here we go again.....

Ellen

Why didn't you post this comment when you read Daniel's ignorant caricature? If Daniel posts an intelligent comment on induction, even if I disagree with him, then you will find that I am quite civil. And if hell freezes over, then I will go ice skating....

I am not generally a fan of Peikoff or Peikovians. But neither am I willing to ridicule an author I haven't read just because he is a Peikovian.

I think I will order Harriman's book from Amazon tomorrow.

Ghs

[kiaer.ts]

My opinion for years, without knowledge of Peikoff's or anyone else's, has been that concept formation is induction, pure and simple. Where else do they come from? One cannot deduce from any concept that hasn't already been induced. Induction is implicitly valid, since any disproof of its validity would have to be induced.

What beyond the obvious would a theory of induction provide? One cannot say much ahead of time about how induction is done, other than the methods of similarity and difference, just as there is no algorithm of algorithms, other than making an educated guess, testing the results, and improving the hypothesis.

All sorts of propositions, especially those dealing with a concepts definition and essential attributes are implicit in concepts. Darwin induced the concept evolution (by natural selection). Newton induced the concept of force (as mass times acceleration). I think this would have been Rand's position, had she been more scientifically literate. If essential characteristics are those that best explain others, then will not essence be tied up with cause? Is not the reason why the black swan example silly because there is nothing causal about the whiteness of the swan to justify inducing it as a necessary attribute of swans? Am I missing something?

[George H. Smith]

Enumeration is not the method of induction, and it provides no basis to infer from "some" to "all," not even with a degree of probability. This is why all attempts to ground inductive reasoning on statistics have failed. A generalization reached merely from enumeration is necessarily arbitrary, and must therefore be dismissed without discussion from the field of rational consideration. As we shall see, there are valid inductions based on a single case; and there are generalizations with millions of instances, which are yet utterly illegitimate (e.g., "All men seek pleasure"). Why such a difference? Simple enumerators have no answer. (Chapter 1, pp. 8-9; bolding added by me).

Robert,

This is the passage I had in mind. Essentially the same point has been made many times by logicians, and Harriman is right -- even if he is a Peikovian.

Ghs

[Later edit]

Robert,

Sorry, I didn't quite get your earlier point. I've been doing "real" writing tonight and writing these posts at breakneck speed, and I got careless.

Is inference based merely on enumeration arbitrary? Well...no...I wouldn't call that. It depends on the conclusion that is drawn. It may just be a case of fallacious reasoning. I agree that Peikovians tend to overdo use of the word "arbitrary."

Nevertheless, I agree with the other points in the quoted passage by Harriman.

Ghs

[Selene]

Question gentlemen:

If I "observe" a number of particulars are similar and distinct from all other particulars and I generalize a statement about those particular set of particulars and that generalization deductively leads to verification of another particular which was unknown when I started observing those original particulars, is the first part of this totality not induction?

Adam

[Ellen Stuttle]

Here we go again.....

Why didn't you post this comment when you read Daniel's ignorant caricature?

Oh, for heaven's sakes. I didn't see his caricature until you'd quoted it and replied to it. I copied only the last sentence of your comment in order to save space, instead of copying the whole sequence. Sorry I said anything. Actually, I only did say anything because I was chuckling at the thought of how short the lull in battling had been.

< sigh >

Ellen

[George H. Smith]

Here we go again.....

Why didn't you post this comment when you read Daniel's ignorant caricature?

Oh, for heaven's sakes. I didn't see his caricature until you'd quoted it and replied to it. I copied only the last sentence of your comment in order to save space, instead of copying the whole sequence. Sorry I said anything. Actually, I only did say anything because I was chuckling at the thought of how short the lull in battling had been.

< sigh >

Ellen

Okey-dokey.

As much as I like Popper in many ways, his objections to inductive reasoning have never made much sense to me. (I reread some of his stuff just a few weeks ago. ) I would therefore like to engage an intelligent Popperian on this subject. If you know where I can find one, please let me know.

Ghs

[George H. Smith]

I just ordered Harriman's book from Amazon. Since I am no longer an Amazon Prime customer (I don't order nearly as many books as I used to), I specified standard shipping, so I expect it to arrive on Friday or Monday. It will be good to get an overall perspective on his approach. Snippets can be tricky.

Ghs

[Brant Gaede]

I've yet to learn why "The Problem of Induction" is a problem.

--Brant

[anthony]

I've yet to learn why "The Problem of Induction" is a problem.

--Brant

I'd like to know too. Apart from it not including logic, the inferences drawn from induction are as valid a means towards knowledge and rationality as deduction, aren't they? If not more so. And speedier.

With the proviso perhaps that one checks the conclusions (concepts, and so on), deductively.

There would appear to be some tie-up with intuitive, pattern-seeking, right-brain 'hard-wiring' involved; but that's just a (inductive) guess.

Tony

[Daniel Barnes]

As much as I like Popper in many ways, his objections to inductive reasoning have never made much sense to me. (I reread some of his stuff just a few weeks ago. )

I thought you said a minute ago that they were "some of the silliest things he ever wrote." Now it seems you're saying you didn't quite get them, and wouldn't mind talking them over with someone just in case they might not be so silly. That seems a bit more reasonable.

I would therefore like to engage an intelligent Popperian on this subject. If you know where I can find one, please let me know.

I recommend my good friend Rafe Champion, who posts over at Matt Dioguardi's most excellent CriticalRationalism blog. If there's a relevant thread, perhaps you could drop in with any questions you might have.

Alternately I recommend the Critical Rationalism Yahoo group, which Matt also moderates. Though I haven't posted over there for a long time, Ken Hopf is a particularly good commenter and I see he's posting a bit right now, so you could fire a few queries his way. Incidentally, he's a former Objectivist, so he knows that side of the street too.

Happy to be of service...;-)

Edited September 7, 2010 by Daniel Barnes

[BaalChatzaf]

I've yet to learn why "The Problem of Induction" is a problem.

--Brant

I'd like to know too. Apart from it not including logic, the inferences drawn from induction are as valid a means towards knowledge and rationality as deduction, aren't they? If not more so. And speedier.

With the proviso perhaps that one checks the conclusions (concepts, and so on), deductively.

There would appear to be some tie-up with intuitive, pattern-seeking, right-brain 'hard-wiring' involved; but that's just a (inductive) guess.

Tony

The problem with induction is that it is possible to infer a false conclusion from true premises. Think about the black swan and the albino crow. Deduction guarantees the conclusion that follows logically from true premises is true. Induction does not have that guarantee. A less simple minded example is how the aether hypothesis was arrived at from the wave aspects of light (electromagnetic radiation). Or how caloric was inducted/abducted from some of the effects of heat.

Even so, we are bound to use induction and abduction (inferring a cause from effects)* for if we don't, we will never be able to get beyond a limited finite set of facts to anything new. No induction/abduction = no physical science.

We are never going to comprehend physical reality by purely a priori reasoning.

Ba'al Chatzaf

*most of the induction that physical scientists do is abduction (a term invented by C.S. Peirce) which is a way of getting to the most likely or plausible cause from a collection of effects. This kind of reasoning was extensively worked on by J.S.Mill in the 19th century and is still used today. We cannot get to the real reality by a priori reasoning so we are bound to use induction/abduction.

Edited September 7, 2010 by BaalChatzaf

[Ellen Stuttle]

I've yet to learn why "The Problem of Induction" is a problem.

--Brant

Because there is no way you can be certain that a generalization you arrive at will be true for all cases. NO WAY. And Harriman has not provided such a way. If there were such a way, who would need scientific investigation?

Unlike George, I find Popper's views on induction exceedingly intelligent.

I think that what maybe confuses people of O'ist orientation is the term "valid." Sure, we can't exist without proceeding on assumptions that tomorrow will be, in significant respects, like today. But inductive reasoning can't ever provide necessarily true conclusions. Our inductive reasoning could turn out to be wrong. It's happened over and over and over in the history of science.

What Harriman argues in fact does not give one any firmer conclusion about necessarily true than any scientist would claim for current scientific "law." Harriman merely says, in keeping with the convoluted "contextual certainty," that you need to add the proviso "as best we know now." Adding this proviso makes your conclusion ok. But it doesn't solve the historic "problem of induction," which is the question of how to BE SURE your generalization is timelessly correct. No way to be sure of that.

Ellen

[anthony]

I've yet to learn why "The Problem of Induction" is a problem.

--Brant

I'd like to know too. Apart from it not including logic, the inferences drawn from induction are as valid a means towards knowledge and rationality as deduction, aren't they? If not more so. And speedier.

With the proviso perhaps that one checks the conclusions (concepts, and so on), deductively.

There would appear to be some tie-up with intuitive, pattern-seeking, right-brain 'hard-wiring' involved; but that's just a (inductive) guess.

Tony

The problem with induction is that it is possible to infer a false conclusion from true premises. Think about the black swan and the albino crow. Deduction guarantees the conclusion that follows logically from true premises is true. Induction does not have that guarantee. A less simple minded example is how the aether hypothesis was arrived at from the wave aspects of light (electromagnetic radiation). Or how caloric was inducted/abducted from some of the effects of heat.

Even so, we are bound to use induction and abduction (inferring a cause from effects)* for if we don't, we will never be able to get beyond a limited finite set of facts to anything new. No induction/abduction = no physical science.

We are never going to comprehend physical reality by purely a priori reasoning.

Ba'al Chatzaf

*most of the induction that physical scientists do is abduction (a term invented by C.S. Peirce) which is a way of getting to the most likely or plausible cause from a collection of effects. This kind of reasoning was extensively worked on by J.S.Mill in the 19th century and is still used today. We cannot get to the real reality by a priori reasoning so we are bound to use induction/abduction.

Thanks Ba'al (and Ellen)

I don't think that anyone is suggesting ditching deduction.

However, the next question is although induction can be flawed, how often does it get it right?

We all know that we have to sort through many uncertainties to discover any certainty - and deduction alone will make that painfully slow. (I'm speaking more for the layman, rather than the scientist.)

Even a physicist raises a 'working hypothesis', and then empirically shoots it down - or otherwise.

A great example is of Ayn Rand. If I understand correctly, she constructed an entire philosophy via induction. Roughly, she got 90% exactly,incontrovertibly, right; 5%, a bit 'iffy'; and 5%, dead wrong. (My humble estimates.)

She did it all without leaving her desk.

Tony

[Daniel Barnes]

I've yet to learn why "The Problem of Induction" is a problem.

I will try to give you a snapshot version. As I'm currently on a tropical island without my books and somewhat limited internet access, I will call this the "pool bar" explanation.

The problem as you will know ad tedium was made famous by Hume, even though pre-dates him and he never called it such. We may summarise it as the problem of obtaining universal laws from particular instances - in effect, obtaining knowledge of future occurrences (for universal laws are predictive) from observations of past ones.

We might call "induction" the belief that this process is reasonable, and even possible. After all, it seems to track with our everyday experience: we see the sun rise every morning and have done all our lives, it seems unthinkable that it will not do so tomorrow. In fact, the more often we've observed something occur in a certain fashion, the more confident we feel that it will occur in this fashion in the future. Once we've seen something happen often enough, why, we just know that it will happen the same way again.

Hume, being a particularly rigorous fellow and disinclined to take anything for granted, especially the obvious, looked at this seemingly undeniable process and said "Hey...wait just a second. What exactly is the justification for this seemingly undeniable belief?" There's no logically valid way of getting from 1, or 1,000, or 1,000,000 observations to anything like a universal law, for any such generalisation must always outstrip the evidence for it. Further, if we try justify induction a different way, 'inductively' - by saying that hey, induction has worked in the past, so it must work in the future, right? - this too is a fallacy, because it's simply another appeal to experience to justify the first appeal to experience. Hume pointed out that this question could be renewed each time this justification was offered, so effectively led to an infinite regress. So no such luck there either. Even appeals to probability - that, ok, if not certain, then perhaps a conclusion justified by prior observations would be more probable - are, sadly, fatally destroyed in the same stroke.

So it turned out this perfectly natural and seemingly indispensable assumption had no rational justification whatsoever. Even more inconveniently, so secure had this natural assumption seemed that the basic foundations of science had been erected around it. By the slow, painstaking accumulation of millions of observations, it was believed, a rock-solid theory would eventually form, like a stalagmite of truth from a vast accumulation of tiny limestone drips. Yet Hume was pointing out this was an entirely false hope - in reality it was more like making observations of millions of rising house prices and concluding therefore they would always rise in future...

The response to this radical finding of Hume's was...nothing at first. Even Hume himself - the most hard-headed of men - couldn't quite bring himself to believe it. Pretty much everyone felt the same way except a certain middle-aged scholar over in Prussia name of Immanuel Kant. His reaction on reading Hume was: Holy fucking shit...This Is BAD!! For he realised that if Hume was right, human knowledge, including that last, best hope for certainty, science itself - or at least as it was being constructed at the time - was fatally undermined. Nothing like a rock-solid law could ever in fact be formed by such a process. So he went back to the drawing board to try and figure a way round the problem. (Whether he succeeded or not is a matter for debate - I think not.)

So was set what Bertrand Russell called the ticking "time bomb" under philosophy. For the problem was not just a logical one, but a psychological one too. For as Hume observed, we fundamentally believe in this idea of induction even though it has no rational justification - we cannot, apparently, do without it. Thus humans are fundamentally irrational - all hopes for a rational philosophy are doomed. Despairingly, Russell said that our only hope against this inexorable conclusion was that Hume's problem might one day be solved - however, all attempts so far had failed.

And that is the pool-bar explanation why the problem of induction is a problem.

Make mine a Pina Colada.

Edited September 7, 2010 by Daniel Barnes

[Daniel Barnes]

I think that what maybe confuses people of O'ist orientation is the term "valid."

It's a stolen concept..;-)

[BaalChatzaf]

I've yet to learn why "The Problem of Induction" is a problem.

--Brant

I'd like to know too. Apart from it not including logic, the inferences drawn from induction are as valid a means towards knowledge and rationality as deduction, aren't they? If not more so. And speedier.

With the proviso perhaps that one checks the conclusions (concepts, and so on), deductively.

There would appear to be some tie-up with intuitive, pattern-seeking, right-brain 'hard-wiring' involved; but that's just a (inductive) guess.

Tony

The problem with induction is that it is possible to infer a false conclusion from true premises. Think about the black swan and the albino crow. Deduction guarantees the conclusion that follows logically from true premises is true. Induction does not have that guarantee. A less simple minded example is how the aether hypothesis was arrived at from the wave aspects of light (electromagnetic radiation). Or how caloric was inducted/abducted from some of the effects of heat.

Even so, we are bound to use induction and abduction (inferring a cause from effects)* for if we don't, we will never be able to get beyond a limited finite set of facts to anything new. No induction/abduction = no physical science.

We are never going to comprehend physical reality by purely a priori reasoning.

Ba'al Chatzaf

*most of the induction that physical scientists do is abduction (a term invented by C.S. Peirce) which is a way of getting to the most likely or plausible cause from a collection of effects. This kind of reasoning was extensively worked on by J.S.Mill in the 19th century and is still used today. We cannot get to the real reality by a priori reasoning so we are bound to use induction/abduction.

Thanks Ba'al (and Ellen)

I don't think that anyone is suggesting ditching deduction.

However, the next question is although induction can be flawed, how often does it get it right?

We all know that we have to sort through many uncertainties to discover any certainty - and deduction alone will make that painfully slow. (I'm speaking more for the layman, rather than the scientist.)

Even a physicist raises a 'working hypothesis', and then empirically shoots it down - or otherwise.

A great example is of Ayn Rand. If I understand correctly, she constructed an entire philosophy via induction. Roughly, she got 90% exactly,incontrovertibly, right; 5%, a bit 'iffy'; and 5%, dead wrong. (My humble estimates.)

She did it all without leaving her desk.

Tony

Deduction from true premises yields true conclusions 100 percent. No iffy, No dead wrong. The problem is knowing when the premises are true. That is not a problem in logic, that is a problem in observation and measurement, for those premises that assert something about the world.

It is the certainty of true conclusion (given that the premises are true) absolutely and necessarily that has made deductive systems of mathematics and logic the Golden Calf before which intellectuals through the millenia have bowed and prayed. The Greeks who invented deduction or were the first to clearly understand its nature were totally smitten by it.

Just another note. It is impossible to learn something true about the world (even as a starting point) without leaving one's desk. There are no innate apodictic ideas nor are there any specific statements about the world (non-tautologous statements) which can be gotten a priori. In order to find out something true about the world one must look, measure, touch and sometimes even taste and smell. We cannot pull the world out of our heads nor can we pull out of Plato's gizzards.

Nothing is in the mind that is not first in the senses.

Ba'al Chatzaf