September 29, 2009

The only difference between scientific reasoning and other kinds of reasoning is the content

From Professor Gross' article, which I discussed in the previous post, comes another gem on the nature of scientific reasoning.  What's so special about it?  In short nothing but the content.

What differentiates scientific from, say, historical reasoning?  Other than the content being reasoned about, I can't think of anything, so, I turn to the distinguished philosopher of science and epistemologist Susan Haack to discover that the notion of species of reasoning unique to science is unfounded.  Haack writes.

Scientific inquiry is continuous with the most ordinary of everyday inquiry.  There is no mode of inference, no "scientific method," exclusive to science and guaranteed to produce true, more nearly true, or more empirically adequate results ... And, as far as [science] is a method, it is what historians or detectives or investigative journalists or the rest of us do when we really want to find something out: make an informed conjecture about the possible explanationsof a puzzling phenomenon, check how it stands upto the best evidence we can get, and then use our judgment whether to accept it, more or less tentatively, or modify, refine, or replace it.
The practices of  good science are distinguished by that "informed conjecture"--by a special dependence upon technology (e.g., instruments that broaden the human range of perception), and by especially strong and well-enforced rules having to do with scrutiny and testing of claims and reproducibility of  results.  But they are not distinguished by an array of clearly identifiable, cognitively unique forms of reasoning.

What then, is to be understood by scientific reasoning?  The answer cannot be very deep because the question isn't.  Scientific reasoning is using, within a framework of scientific content, certain general cognitive abilities that develop over time or can be encouraged in most learners.  So, there is not much that is exclusively scientific about such reasoning other than the fact that one is thinking about scientific content.  Scientific reasoning is a sibling to, if not perfectly congruent with, historical reasoning, which is the use of similar cognitive basics in the context of records and commentary on the past.  Scientific reasoning is deployed with hypotheses and observations about nature.  It has other siblings as well: social, artistic, and literary reasoning for example.

The Bao et al. study showed that the Chinese and American cohorts had about the same level of general reasoning ability.  Yet, the American students  access this ability when attempting to reasonwhithin the Physics domain because they lacked the Phsyics content knowledge.

Also, if you want to believe that the typical Chinese education is painful rote and the typical American education is overly constructivist, then you're also led to the conclusion that a rote learning is every bit as capable of developing general reasoning skills as a constructivist education (or every bit as incapable depending on whether you think the skills demonstrated were sufficient).  I, however, think it's more a matter of degree between the more traditional education that the Chinese cohort received (which isn't rote, but mostly problem-solving) and the typical education of the American cohort which is a hodge podge of content-lite traditional education with an emphasis on constructivist learning activities.

So, to the extent that the Chinese cohort received more content instruction it aided their understanding within the confines of the content domain (i.e., physics) and it did not diminish their general reasoning abilities.  In contrast, the increased instructional time devoted to constructivist activities not only failed to improve general reasoning abilities (as the theory goes), but also hinered their ability to reason within a particular domain (physics) because the content of that domain was not learned.  And, whatever benefits that supposedly come from constructivist learning failed to compensate for teh lack of content analysis that was forfeited by reducing the amount of content knowledge taught while pursuing those constructivist activities.

There are opportunity costs in education, as there are in every other human endeavor.

6 comments:

Stephen Downes said...

Well, Susan Haack is definitely an authority; I have a well-worn copy of her excellent 'Philosophy of Logics' on my bookshelf.

But I'm not sure you track the implications of the assertion you cite. There are no domain-specific forms of reasoning. From which it follows, if you learn to reason in one domain, you can transfer those skills to another domain, without being required to learn the content of the domain in order to know how to reason within the domain.

Also, another error:

You write, "if you want to believe that the typical Chinese education is painful rote and the typical American education is overly constructivist, then you're also led to the conclusion that a rote learning is every bit as capable of developing general reasoning skills as a constructivist education"

Of course, the criticism is that American education is _not_ constructivist (and it certainly isn't when compared to, say, Canadian education) and is mostly based on drill and test. Which means that the conclusion that should be drawn is that American drill-and-test produces no more effective reasoning skills than Chinese drill-and-test.

The Chinese, at least, recognize that they need to change the way they do things.

Dick Schutz said...

The crux of Haack's point seems to be:

"make an informed conjecture about the possible explanations of a puzzling phenomenon, check how it stands up to the best evidence we can get, and then use our judgment whether to accept it, more or less tentatively, or modify, refine, or replace it."

That's pretty close to and an improvement on "Doing the damndest with one's mind, no holds barred."

As she says, that's not very deep. There are routines/conventions that are applicable to different reasoning domains. But the depth comes in the background information (declarative knowledge and procedural knowledge) and this does not transfer widely from domain to domain.

If you're going to delve into "reasoning" Ken, the standard cognitive sci reference is "The Cambridge Handbook of Science and Reasoning." The handbook includes chapters on "Legal Reasoning," "Scientific Thinking and Reasoning," and "Thinking and Reasoning in Medicine."

(Apparently lawyers reason but don't think. If they stopped to think, they probably wouldn't be lawyers. My reasoning, not that of the author of the chapter on legal reasoning).

The chapters identify more differences than Haack does, but they pursue matters at a micro level. At the macro level, my reading is that the chapters support your interpretation rather than Stephen's.

KDeRosa said...

So how does one "make an informed conjecture about the possible explanations of a puzzling phenomenon" in the absence of knowledge about that phenomenon?

How does one "check how it stands up to the best evidence we can get" without knowing the best evidence or understanding why it is "best"?

And, how does one "then use our judgment whether to accept it, more or less tentatively, or modify, refine, or replace it" without an understanding of the alternatives and the nature of the mofications and refinements.

All of this is highly content dependent.

It's like saying that there is a general ability to decode written text. You can decode words across all domains which is true. But you can't understand the decoded text unless you know the content (meaning) of the underlying text. Until you know the content, you're general decoding ability is of little value to you at least in the domain for which you don't know thw content.

Of course, the criticism is that American education is _not_ constructivist (and it certainly isn't when compared to, say, Canadian education) and is mostly based on drill and test.

Now we've hit No True Scotsman territory. Maybe you should stick to conclusory arguments.

Tracy W said...

From which it follows, if you learn to reason in one domain, you can transfer those skills to another domain, without being required to learn the content of the domain in order to know how to reason within the domain.

So why don't people reason well outside their domain? It seems to be a reasonably well-supported theses in cognitive science that experts are only experts in their area of expertise.

If reality differs from the logical result, doesn't that imply that the logic is missing something? Or is the summary congitive scientists make of the empricial evidence wrong, and in fact an expert in, say, physics with no specific training in biology can reason well about biological problems?

Eric said...

if you learn to reason in one domain, you can transfer those skills to another domain, without being required to learn the content of the domain in order to know how to reason within the domain.

What percentage of ed school faculty believe this? What eduction is so deficient as to allow this statement to be put forward as a credible proposition? There's more to reasoning than modus ponens and modus tollens (which, of course, transfer effortlessly). A good bit of reasoning is expended on getting the models right against which propositional and predicate calculus can be applied.

Try reasoning about the Kolb Learning Cycle and Howard Gardner's Disciplined Mind. Presumably educators will do a better job with this than non-educators; but in this case, the exception proves the rule.

Eric said...

Here's an example of getting the model wrong. Yes, reasoning (i.e. arithmetic) transfers across knowledge domains. But modelling is the hard part. Sure, little more than common sense is required to correct the modelling mistake in this riddle (that's the point-it's a riddle). But it nicely illustrates the ease with which a novice can get a model wrong in an actual discipline.

The Riddle of the Missing Dollar:
Three men were on a business trip out of town,so they decided they would share the cost of a hotel room. They went to the desk and the clerk told them the room would cost 30 dollars,so each man handed the clerk 10 dollars each. The clerk took the $30 back to the manager and the manager advised the clerk the room was on special for $25, so the manager gave the clerk 5 one dollar bills. The clerk was unsure how to divide 5 one dollar bills 3 ways, so he gave each man a dollar back and stuck 2 dollars in his pocket. That means each man paid $9 each that's ($27 total) plus the $2 the clerk put in his pocket makes $29--where did the other dollar go?

E. D. Hirsch uses force problems from physics to illustrate the complexities of modelling. Allingham contrasts an abstract problem with an identical problem in a concrete setting (trains, tickets and immunizations).

In contrast, Howard Gardner tries to make a similar point with a counter-intuitive physics simulation and a gimmicky math word problem.

So among the three:
Allingham best makes the point that contextualizing a problem can aid reasoining about the problem;
Hirsch best makes the point that expertise in a domain is important for getting the model right so the math yields the right answer;
Gardner best embarrasses cocktail party guests with misleading Algebra I word problems.

Riddle: Which one is on Harvard's faculty?