Tuesday, September 29, 2009

Making Up Facts

I think I'll stop reading Willingham if he persists in making stuff up.

He writes, "Prior knowledge is vital to comprehension because writers omit information. For example, suppose you read “He just got a new puppy. His landlord is angry.” You easily understand the logical connection between those sentences because you know things about puppies."

It is no such thing. The two sentences have the form of an explanation. The landlord is angry [because] he got a new puppy. Nothing should be inferred regarding the cause of the landlord's anger: it may easily be allergies, dislike of yapping dogs, town bylaws about animals, or blind prejudice. And inference that involves the putative 'knowledge' about puppies would be an error.

In fact, the inference as to the structure of the explanation is based on a much more general inference, the observation that one statement is propositional (a 'fact') and the other expresses an attitude (an 'emotion'), and while facts explain emotions, emotions do not (generally) explain facts (though an _honest_ reader would leave open that he purchased the puppy in retaliation for the landlord's persistent and unreasonable anger).

Appealing to 'facts', especially when you're just making them up, leads to terrible reasoning, and even worse educational theory.

And let me add a point to this...

Willingham frequently makes this sort of appeal, where the reader is supposed to come to some sort of understanding based on a recollection of relevant (domain-specific) facts.

In logic, this sort of form is known as the 'implicit premise'. "Don't play in traffic. It's too dangerous," says the mother. You are supposed to know that traffic is dangerous, or that you should avoid dangerous things, or some such thing.

But - and this is the key point - implicit premises are filled not by domain-specific knowledge, but rather, by logical form. When reading the passage, the reader employs what is known as the 'principle of charity', which is, to infer that the writer means to state the premise or premises that would make the inference sound.

"Don't P. P is Q." Implicitly, "Don't Q". Because that is what makes the inference reasonable. What is "Don't Q"? Much as Willingham wants it to be, it is not domain knowledge. It is an inference, generated by the charitable reader, who has a good understanding of reasoning, and needs no particular understanding of the domain.

Of course, Willingham disguises this, by abducting, instead of the reasonable implicit premise, some domain-specific information that sometimes does and sometimes (as in this case) does not have anything to do with the inference. It's a sleight of hand. If the interpretations appear reasonable to you, it's not because you have domain knowledge (though he always keeps his examples simple, to foster the illusion), it's because you recognize the logical form of the argument.

It's a tired old trick and I wish Willingham would quit doing it and try some real psychology, or pedagogy, or whatever it is that he's doing.

Note: this post was submitted as a comment to the original article, where it did not survive author moderation

9 comments:

  1. I think it is more of an oversimplification than lies...don't you?

    I appreciate your detailed clarification, and questioning, always questioning. Thanks

    ReplyDelete
  2. Willingham's belief reminiscent of Schank's Script theory from the 70s, which was of some linguistic use -- for example in explaining how people tell stories about going to a restaurant, but don't explain the fact that they are skipping to the part where they get the food means they must have ordered first.

    But scripts kept multiplying, which is why people welcomed Relevance Theory.

    Relevance Theory provides a pragmatic way into parsing such statements without reference to a Schankian "script". In Relevance Theory it is assumed that your interlocutors will attempt to be maximally relevant -- they will not provide you with irrelevant information, or repeat things unnessarily, etc.

    This is easily demonstrable in that if we say:

    "He just got a new puppy. His landlord is pleased."

    we will invent reasons why the landlord would be pleased about a puppy.

    Now the reasons we invent *may* pull from domain knowledge. The implicatures. But the reasons are in essence backfilled from the understanding we have that a maximally relevant interlocutor would not put two statements together like that if one was irrelevant to the other, and that causality is the simplest linkage.

    I think this sounds similar to the principle of charity, but it has some neat demonstrations of broader application in language. It forms the basis of the most coherent explanation of irony, for example, and explains our ability to understand constructions like

    "Yes, but did you hit hit him?"

    Where we know that no one would unecessarily repeat a word ("hit hit") if there was no extra meaning added -- so we set about tryng to decide what a likely distinction between "hit" and "hit hit" might be -- with the assumption that "hit hit" will have a more precise meaning (not a less precise one, b/c if we have to process more, we demand more meaning).

    Long ramble, maybe, to get basically where you were. But if you haven't checked out Relevance Theory, you might try it -- it's neat, and I think you'd like it.

    ReplyDelete
  3. You're right - I like it, and it does take us to the same place.

    ReplyDelete
  4. Here are four questions that are domain specific. Each is readily answerable with minimal knowledge in the specific domain and some general reasoning skills. However, let's assume you lack the required content knowledge and a means for acquiring that knowledge. Use your (superior) general reasoning skills to derive the same answer an expert would give for each.

    The total enthalpy of any non-isolated thermodynamic system tends to decrease over time, approaching a minimum value. Why?

    As the location of the subatomic particle becomes more precise, what would you infer about its momentum?

    Jones sacrificed and knocked in a run. Where is Jones and the runner?

    When John walked out onto the street, he nictitated rapidly. Where might John have just come from?

    ReplyDelete
  5. I don't see what you expect to prove by this. yes, you need domain knowledge to answer requests for facts within that domain. So what? This has nothing to do with Willingham's argument.

    Indeed, the form of *your* presentation is that of circular reasoning. You are trying to demonstrate that facts are required to reason within a domain. To prove this, you pose as an instance of reasoning in a domain a demaind for some fact specific in the domain.

    I've said on numerous ocasions, simply responding to a demand for some fact does not constitute reasoning. These questions may resemble high school tests - but they do not resemble inference or reasoning.

    ReplyDelete
  6. As willingham wites, "Prior knowledge is vital to comprehension because writers omit information." In the example I gave, the writer has assumed you know the meaning of certain words and other domain knowledge and has omitted those explanations. In Willingham's example, your general reasoning skills can still lead to a correct answer and domain knowledge did not constrain your ability to reason your way to an answer. In the examples I gave, lack of knowledge does constrain your ability to reason your way to an answer.

    Also, contrary to your assertion, answering all these questions requires some reasoning ability. For example, to answer "When John walked out onto the street, he [blinked] rapidly." You can reason your way to an answer using general reasoning skills and general knowledge as follows. Humans blink when coming from a dark place to a bright place. John is a human. Streets are often bright due to daylight. Movie theaters are one example of a dark place. Therefore, John could have come from a movie theafter. Many would be able to reason their to the right answer with the blink in the sentence. Fewer would with the word nictitate--perhaps some vets.

    That these answers can be provided in the form of declarative knowledge does not detract from the premise that the answered must understand the underlying knowledge to activate the declarative knowledge needed to answer the question.

    ReplyDelete
  7. *sigh*

    You (and Willingham) are writing as though 'understand' were an on-off switch, with the options being 'zero comprehension' and 'total comprehension'. That's what forces both of you into making your example subjects jump into wild flights of unjustified infrence.

    Let's take a sentence, any sentence in the world. Prior knowledge is needed in order to understand that sentence. You need to be a speaker of the language, you need to know its grammar and syntax, and you need to know what the words mean.

    Grammar and syntax can be more or less complex, and a person's vocabulary may be more or less large, and so already you have a range of possible understandings of different sentences.

    But now: none of this yet involves declarative knowledge - that is, none of this involves the remembering of specific propositions as propositions in the mind. A person's understanding of language, and concordantly, of a vocabulary, is a skill. It's not a matter of absorbing and reciting facts, it's a matter of executing complex linguistic tasks.

    Take what you purport to be a piece of specific knowledge: "Humans blink when coming from a dark place to a bright place." mostly, nobody ever learned that as a fact. They may behave, and perform language games, as though it were true, and if you asked them, they would probably agree that it is true, but the knowledge of it is implicit in the tangled web that is the person's large set of observations and experiences.

    This kind of 'knowledge as skill' -- which includes and is part and parcel of critical thinking - is sufficient for the vast bulk of a person's understandings. It is often enough to give a person at least a partial understanding of a sentence, enough to go on.

    ReplyDelete
  8. (continued from previous due to Google's new 4096 rule)


    Needless to say, it does not include delatative (propositional) knowledge from domains outside the person's experience. Put a question about baseball to a person who has never watched a game, much less grown up with the game, and they will not be able to respond with an answer.

    Now - if you provided the person with the putative fact, "people who sacrifice do so from home plate" then the person will be able to answer the questions thta demanded such facts. But, crucially, that is not how people learn baseball. For most people, nobody ever told thenm the 'fact', "people sacrifice from home plate". It is not a 'fact' that ever exists in a person's mind until the question is asked.

    Put the question in the form of a demand for a fact, and the person will, if able, resppond appropriately, with a fact.Doing, in other words, what was requested. But what happens is not that the person had the fact, but rather, that the person created the fact out of the complex mass of previous observations and experiences, and presented it to you. The 'fact' never existed in the mind until you asked the question.

    And - what's more - the person will (as Caulfield suggested, appropriately) to use whatever tool they can in order to create this fact. This includes knowledge inside and outside the domain, and especially, critical reasoning, language learning, analogies form other domains, related knowledge from other domains, unrelated stuff from a movie he saw yesterday, and the rest.

    the result is, people will answer the questions very differently from each other, and most will not engage in wild flights of reasoning. You say, of the example, "When John walked out onto the street, he [blinked] rapidly," that "Humans blink when coming from a dark place to a bright place. John is a human. Streets are often bright due to daylight. Movie theaters are one example of a dark place. Therefore, John could have come from a movie theafter. Many would be able to reason their to the right answer with the blink in the sentence."

    But there is no RIGHT answer here. You mad eit up., You created a case in which a person would (putatively) respond only in one fact-specific way, and then use it as an argument to show (?) that the person had those facts in their brain.

    Not only is it an instance of really poor reasoning (seriously, who would reason that way?) it does not in any way support your contention.

    ReplyDelete
  9. You (and Willingham) are writing as though 'understand' were an on-off switch, with the options being 'zero comprehension' and 'total comprehension'.

    Understanding is not merely an on-off switch, I've merely used extreme examples. I could have just as easily referred to the vast body of cognitive science research in which the understanding and reasoning abilities of those with expertise is shown to be superior than those with little expertise within that domain, but not outside the domain.

    I do like your little rhetorical trick characterizing the unknowns as mere facts or mere declarative knowledge and then arguing from that limioted perspective. I purposesly used the term knowledge, not facts, to encompass all those points you try to distinguish over in the middle paragraphs of your comment.

    As you point out, certain knowledge must be known for the reader to understand the question (and even more to answer the question with a declarative fact). That knowledge does not have to be in fact form. Moreover, that understanding might be less than perfect--there is a continuum of understanding.

    So, as usual, you choose to argue against a strawman, rather than confront the actual argument I made.

    So, when you write, "the knowledge of [the dark place/blinking statement] is implicit in the tangled web that is the person's large set of observations and experiences" I'll point out that this is also my underlying assumption. The representation of that knowledge determines, at least in part, how well the knowledge is understood. And, of course, the reader will use whatever "tool" at their disposal to "create" the "fact" from their represented knowledge base.

    And, of course I never suggested that the answer I gave to the nictitate question was the only "right" answer. I explicitly wrote "You can reason your way to an answer" which should suggest to any reasonably skilled English language reader, that the answer I provided was one of possibly many suitable answers, each with varying degree of correctness. Again, you attack a strawman.

    Needless to say, it does not include [declarative] (propositional) knowledge from domains outside the person's experience. Put a question about baseball to a person who has never watched a game, much less grown up with the game, and they will not be able to respond with an answer.

    And, yet that person retains their general reasoning abilities, but is unable to use them in a domain outside their expertise. Put a question about [physics, biology, chemistry, history, etc., etc] to a person who has [no domain knowledge relevant to physics, biology, chemistry, history, etc., etc], and they will not be able to respond with an answer. Which is why an important aim of formal education is to develop domain knowledge in these areas becausee we would like people to respond with coherent answers in these domains.

    Now - if you provided the person with the putative fact, "people who sacrifice do so from home plate" then the person will be able to answer the questions thta demanded such facts. But, crucially, that is not how people learn baseball. For most people, nobody ever told thenm the 'fact', "people sacrifice from home plate". It is not a 'fact' that ever exists in a person's mind until the question is asked.

    The are many ways to acquire the needed knowledge base. Being provided the facts is on eof many ways. And, the famous chicken sexing experiment tells us that it is a particularly effective way.

    Dismantle the strawmen and address the argument I actually made, not the argument you would have liked me to make.

    ReplyDelete

I welcome your comments - I'm really sorry about the moderation, but Google's filters are basically ineffective.