Cognitive Science Questionnaire

[xcthulhu]

Hey everyone,

I also posted this up in SCF, but I thought I might post this here for good measure.

For a cognitive science class I'm taking this semester we're running a questionnaire:

http://www.w-d.org/questionnaire/

If people are interested, I'd be happy to talk about my experiment after we've collected a bit of data.

You're feedback is greatly appreciated, and will help with our final analysis!

Thanks!

~XCT

----------------------------------


Edit: In what follows I'm happy to carry on the discussion of my research, but before looking please fill out the survey! Thanks!

(and now a little blank space...



























...there)

[Lincoln]

I looked over your questionaire. I found it difficult to answer. In the multiple types of questions, you had things like:

If A then B.
If C then B.

Some condition on B, question on A.

Since it was unclear whether the conditionals were logical ANDs or ORs, it was hard to get an unambiguous answer. Further, there were no words like always and so forth. Again, it made it hard to answer.

I think you'll get weird results due to the ambiguity of the question design. When you describe your experiment, I'll be curious as to the goals and expectations.

[CanadysPeak]

Yes, I too wondered if there is a difference in parallel construction due to language. I took them as ORs.

[xcthulhu]

Please fill out my survey before reading this! Thanks!

Our experiment is to test the effect of "you" versus other proper names in a reasoning task contrived by my BCS professor. The website above has a little PHP script that randomly assigns the viewer to one of two surveys.

The reasoning tasks test two forms of inference. "Modus Tollens" and "Modus Ponens". These have the form:

A --> B
A
-------- (Modus Ponens)
B

and

A --> B
~B
------- (Modus Tollens)
~A

The first 10 questions are like this. The next 10 questions are similar, but the hypothesis is that they are of the schematic form:


A --> B
C --> B
A
--------
?

and

A --> B
C --> B
~B
-------
?

...now my Prof has a complicated logic of his own contrivance (similar to the logic underlying the prolog programming language) to explain that in both of these cases the answer is maybe.

But this is why having feedback is critical. Since if I'm reading you correctly Canadyspeak, you are interpreting the second set of 10 questions like this:

(A --> B) v (C --> B)
~B
-------
?

Now we don't need special logic to see that no inference can be made (hence the "maybe" replies); classical logic will do.

I'm definitely going to mention this in my final write up.

Anyway, the second set of "psychology" questions is part of the Autism Quotient test, and my Prof's hypothesis is that autists will not answer "Maybe" in the second set of questions for crazy reasons I'm skeptical of. Hence the final question where we ask if you've been diagnosed with autism.

~XCT

[linford86]

I interpreted the two conditionals as independent. So, we have something like this:

A->B
C->B
A
------
B

since I interpreted the condition C to be independent of A. In that case, the premise C->B is irrelevant, and B is deduced by modus ponens.

Interesting study; I'd be interested in your results.

[32bitHermes]

I agree with Linford.

The second statement never negates the first so the answers don't change buy adding a the new sentence. For instance,

If my grandmother is sick, I will buy flowers. If my grandmother is not contagious, I will buy flowers. My grandmother is sick...

So you buy flowers, because there is no statement which says "If my grandmother is contagious, I will NOT buy flowers." Some people might infer the previous sentence, but it is never given in the problem. So "unsure" is always a wrong answer, and the answers don't change by adding C->B.

On another note. I had fun solving the logic questions but was a deeply offended by the personal questions in sections II, so I quit without submitting my results.

Answer the following question truthfully:

It is offensive (and perhaps irrational) to correlate logical agility with a mental impairment such as autism, which is an ill-defined condition but well-defined stigma.

A. Strongly agree
B. Mildly agree
C. Mildly disagree
D. Strongly disagree

My answer is A.

[linford86]

Ugh. As I understand what xct said, the study does not equate logical ability with a mental impairment. It equates one particular logic with autism and another logic with those who do not suffer from autism and furthermore is designed to see who uses what logic. Neither one of these logics is the true logic*, and as I understand what xct is studying, the question is how do different people interpret language and then make inferences based on those interpretations. Of course, it makes no presumption that any of the factors of interest are necessarily linked. It merely tests to see if there is a correlation. And I shouldn't have to remind you that correlation does not imply causation. If you are correct, and there is no correlation, then there will correspondingly be no problem. If, however, there is a correlation, there is a lot more that needs to be answered before a causal link can be inferred. As far as the definition of autism, the autism spectrum (and associated disorders) are defined in the DSM-IV-TR, ICD-10, and in several other places. Xct's definition, as I understand what he wrote, comes from the Autism Spectrum Quotient. The AQ is a standard psychological test used in diagnosing autism spectrum disorders. So, sorry, 32bit, I think you just completely misunderstood the study. At any rate, if we grab beers tonight (as we rightfully should, after being bombarded by "offensive" surveys :) ) then we can talk more about it then.

I should also point out that even if the autism spectrum disorders cause people to more easily make certain kinds of inferences, then it could well be that having certain forms of autism would be advantageous instead of crippling. If so, the "stigma" would be ill founded.

------------------------------------------------------------------------------

*As I understand xct's point of view, and I would agree with him here, there is no one true logic. Despite the colloquial use of the term, "logic" is merely a group of formal languages which are useful for making inferences. Just like natural languages, there are many different logics, with different vocabularies, and which are useful for making inferences about different kinds of things.

[32bitHermes]

Well, I would say that one can only answer the questions with the information that is presented. Then there really is a correct answer, because by definition A->B, so if A is given, then I know the answer is B. The only way a person would ever give a different answer is if they made outside assumptions. I'll talk about this in detail below in my pre-planned reply:

I've thought of some comments that are more constructive than my previous post. Instead of speculating what psychological disorder people who do well suffer from, why not explore why so-called "rational" people do poorly? I have some theories.

It has to do with persistent illusions, and I am borrowing these ideas from the behavioral economist, Dan Ariely. Consider these two tables

http://dencemond.files.wordpress.com/20 ... usion2.jpg

Even though you can easily check that distance from end-to-end is the same for both, you can't help but continue to see one as being longer than the other. This is the result of a bias in your sensory input which interprets perspective. You can over-ride this bias with abstract logic, "I have measured the tables and know they are the same, and I choose to believe measurement over perception." But knowing the answer does not destroy the illusion. This is called a persistent illusion, and if you have a lapse in judgment you will once again assume the left table is longer. You cannot help it. It is a bias built into your system which is on average useful and correct, but which can also be wrong, as is the case with the two tables.

Ariely's great insight is to realize that these biases are not limited to your visual field of input, but extend to all regions of the mind. His work has shown how people consistently answer certain questions differently based on the context of the question, even when the underlying logic has not changed. The idea works this way:

Consider the following question.

C_N
YO_
RE_D
THI_
_ENT_NCE ?

A. yes
B. no

The fact that you even think you can answer A or B is a remarkable testament to the power of the mind. It can fill in the empty spaces so that the above gibberish actually IS a sentence. The answer to the question is no, however, because the sentence actually reads, "Can Yon read this sentence?" where Yon is an illiterate sheep-herder from the Himalayas. You probably answered yes because you thought it asked, "Can YOU read this sentence?" Your mind filled in the blanks based on previous experience. You have seen these words so often you do not need every letter to understand it. If you forget that the name Yon, and someone asks you to read the sentence again, you will again answer "yes". This is an example of a consistent bias, or persistent illusion.

I wager that this study is another example of consistent bias. Even though it was never stated in the question that "if my grandmother is contagious I WON'T buy her flowers" by stating the inverse "if my grandmother is not contagious, I will buy her flowers" most socially aware people will fill-in-the-blank and assume the latter sentence implies the former. It is not true logically, but it can be inferred socially because if someone told you the latter sentence, you would *assume* that they meant it as a condition on whether they will buy there grandmother flowers if she is sick. Socially, it doesn't make sense to add a second sentence which does not qualify the first. So socially, the second sentence that you will also buy flowers if she is not contagious does not seem to follow from the first. This is why more sociable people will fill-in-the-blanks and assume the sentence means something different from what it says. It is a persistent social bias.

Therefore, I believe this study measures biases related to how socialized a person is. It tests if they are willing to fill-in-blanks and make inferences outside the construct of the problem. A logical person would never do this. A socialized person has biases which compel them otherwise. That somehow doing well on this test should correlate with autsim only proves that "autism" is a term invented by psychologists meant to malign the socially awkward. The term itself represents the biases of the psychological community, who value social interaction above all else (hence their career choice), who seek to reinforce social norms and punish people who don't adhere to them with stigmatizing labels. If you include an analysis on whether or not autistic people do well on this test then you will be complicit in this modern-day witch trial.

[CanadysPeak]

32bitHermes,

I have some doubt about the "Gestalt" sentence. If we look at the bipolar cells in the retina, they're always being excited or inhibited by what goes on next door. In the end, the visual cortex must interpret; there's no such thing as a "u" photoreceptor or an "n". So, how do we say which it is? What else do we have as a measuring device? What is the gold standard? I think the "trick" too glib.

[32bitHermes]

your eye doesn't fill in the "u" or "n". Your mind does when it tries to interpret the sentence. Just like when you read the second sentence on the logic test your mind wants to interpret it as a qualifier to the first sentence (because that is the way people talk), but in reality the sentence is just given and the assumptions about a qualifier are your own personal bias.

[Lincoln]

Right...the assumptions about the qualifier are a bias. The text is insufficiently precise to define itself. Thus, I infer that this is actually not at all about logic, but rather about the psychology of the reader. If it were a logic test, it would be stated in a more solid manner. Or, if it was a logic test, it was a dreadfully poor and imprecise one.

I found myself not being all that rigid about imposing an "AND" or an "OR" qualifier, but rather picking the one that fit best with the specific text of the specific language. I certainly leaned one way, but the specific grammar would occasionally sway me.

[CanadysPeak]

Right...the assumptions about the qualifier are a bias. The text is insufficiently precise to define itself. Thus, I infer that this is actually not at all about logic, but rather about the psychology of the reader. If it were a logic test, it would be stated in a more solid manner. Or, if it was a logic test, it was a dreadfully poor and imprecise one.

I found myself not being all that rigid about imposing an "AND" or an "OR" qualifier, but rather picking the one that fit best with the specific text of the specific language. I certainly leaned one way, but the specific grammar would occasionally sway me.

I hadn't thought about the grammar part but, yes, that also introduces another bias. Depending on your background and your native tongue, you are likely to use "will" and "shall", for example, in different ways. In 2008, I had to translate a drawing package from Japanese into English. Manufacturing drawings, of course, have to be exceedingly precise and I would sometimes wind up with a very detailed, expensive procedure; the odd visiting Japanese engineer would sometimes look at that and say, "Oh, that was just a aside; we don't mean you should have to do that."

[xcthulhu]

It is offensive (and perhaps irrational) to correlate logical agility with a mental impairment such as autism, which is an ill-defined condition but well-defined stigma.

A. Strongly agree
B. Mildly agree
C. Mildly disagree
D. Strongly disagree

My answer is A.

This is not my hypothesis, this is my Profs. I am very skeptical of it. However, I have to object to what you say here. I have known a few high functioning autists (people with Asperger's) and I don't consider it a mental impairment. As far as I can tell, people with Asperger's are basically like everyone-else, maybe a bit nerdier, with a tendency to have higher mathematical ability in my experience.

Moreover, as Lincoln remarks, the test was ambiguously worded; there's no "logically correct" answer. It was intentionally worded. Our research group was consciously imitating this paper authored by my professor:

ftp://ftp.fcdonders.nl/pub/pdfs/Neurops ... nacker.pdf

Although there are a number of difference between our study and his (our's is shorter, to start).

This isn't mathematics or physics - I don't consider one answering style "more logically agile" than another.

Right now, I will say already that either people gave responses like Canadyspeak's or Lincoln's, with lots of "maybes", or they answered like Linford with few or no maybes. I will report, when we have finished collecting data and performed our analysis, whether one's AQ was any sort of indicator which group one would fall in.

[32bitHermes]

I think there is a most logical solution. Look at this problem for example

1. If my grandmother is sick, I buy flowers.
2. If my grandmother is not contagious, I buy flowers.

My grandmother is sick.

Let's make a truth table.

.......................grandmother is sick
......................|....yes....|....no.....|
.................yes|.............|.............|
contagious...no|.............|.............|

Now proposition one tells us that if there is an x in the first column then we buy flowers. Proposition 2 tells us if there is an x in the bottom row then we buy flowers. The second condition never negates (or qualifies) the first condition, so the only reason we wouldn't buy flowers is potentially if there was an x in the right column, top row. But we know the x is in the first column, so we must buy flowers.

There is only ambiguity if you *interpret* the above sentences as:

1. If my grandmother is sick, I buy flowers.
2. If my grandmother is not contagious, I buy flowers, OTHERWISE I DO NOT.

But notice that we have to *add* that line by hand. It is not stated in the problem. Everyone who answers "unsure" read those invisible lines when the problem itself never provided them. It is illogical to infer that the sentence meant something it did not say. It is logical to assume that the sentence means only what it says. If you allow for people to insert arbitrary words into sentences then language loses all meaning and becomes relative, and logic becomes impossible to communicate.

In conclusion, if this test was a test of logic, then there really is only one unique answer which does not require outside assumptions. And any good feelings one might have had for finding the unique solution didn't last long because that person was then immediately categorized as autistic.

[Bill Davis]

I find a problem with comparing common interpretation of those who do not have Autism and those who do. The problem is: the assumption that the common interpretation is more accurate than the one the Autistic has. There are functional biases depending on how well someone is able to have empathy with another. This affects all sorts of interaction and learning aspects. On the other hand, there are a wide range of areas of awareness that someone who is Autistic is far more adept at than the non-autistic person (as alluded to in the question on noticing changes). The trick is, an Autistic may not notice your face has shifted from pleasure to pain but notice that the third book on the shelf is pushed in a little bit more or any other clue Sherlock Holmes would find.

[jshort]

xct...

Since you asked the question regarding whether or not one has formal training in logic/math, it would be interesting to note how a "mathematician" answers this questionnaire (hopefully giacomo will take it too).

I didn't click any "maybe's", but perhaps this is because I treated this as a purely logical test. Logically, there is only correct answer to each question for the reasons already mentioned by Linford and 32bit.

[Giacomo]

I used the Truth-table as shown on this link

http://www.philosophypages.com/lg/e10b.htm

[linford86]

xct,

I think it would help both myself and 32bit to understand your study if you showed us why the arguments were ambiguous. I thought this was a valid syllogism (albeit with one extraneous premise):

1. A->B
2. C->B
3. A
4. Therefore, B (by MP, from (1) and (3))

My thoughts on the matter: Now I suppose that if C implies not-A this would be contradictory and not ambiguous. However, it isn't clear to me in any of the examples that the language should be interpreted in that way. I could also imagine that if C implied possibly not-A (i.e. ) then the argument would only possibly be contradictory (which is funny, and I don't know enough about modal logic to really resolve it.)

[Lincoln]

It's because this is a >>cognitive science<< questionnaire and not a >>logic<< one.

The question is (for me) one of language. The way it is written, it was unclear to me if the sentences should be combined to the logical

If (A and C) -> B
or
If (A or C) -> B.

Language is not as precise as symbolics. The questions were not expressed symbolically and, in doing so, you are imposing your interpretation. It well could be that the point of the exercise was to simply see what fraction of the population imposed which interpretation.

For me, some of the nuances of the language made me lean one way and some the other. If I treated each problem separately, I would have answered in a way that seems erratic to someone who imposed a strict rule.

[linford86]

Lincoln, in xct's post, where he explains how this is supposed to work, he notes that there is an ambiguity involved in classic propositional logic. I therefore wanted to see how that ambiguity came about. Of course, natural language is full of ambiguity -- that's why we make the jump to symbolics in the first place. But the way I interpreted what xct had said was that there is, in fact, an ambiguity even in the formal logic case.

[Lincoln]

That's all well and good, but he also said "take the quiz before you read any further." Whether that was wise or not is, of course, open to debate. But in doing so, this necessarily introduces the language question. Had the terms and definitions been carefully laid out before the quiz was taken, that would be one thing. But since they weren't, that's something else. Because of this, the test was not rigorously controlled and language ambiguity plays a role in people's answers.

Whether this was due to poor design or intentional to address sneaky and hidden things is a separate question.

[linford86]

Well, I did actually take the quiz *before* I read that. I agree with you that how people interpret language, especially in the face of ambiguity, is essential to xct's study. Nonetheless, we can imagine study participants who already know elementary propositional logic (like myself) and consider how this group would choose to answer the questions.

[xcthulhu]

Because of this, the test was not rigorously controlled and language ambiguity plays a role in people's answers.

Whether this was due to poor design or intentional to address sneaky and hidden things is a separate question.

The wording of the test was completely intentional; the ambiguity is part of the design of these tests.

The hypothesis of my Cog-Sci prof is that people who have autistic tendencies like people with ASD or Asperger's tend to fail to recognize that the sentences require some interpretation to either (A /\ B -> C) or (A \/ B -> C) depending on context.

People with autistic tendencies always seem to interpret to (A \/ B -> C) (which in classical logic is exactly the same as (A -> C) /\ (B -> C)). So far, what we have found in our test is that there are two types of people - people who reason to an interpretation and people who don't (ie, people who either read the sentences as you read them or people who read the sentences as linford reads them). The question now is whether a high-autism quotient is a good indicator for being in linford's group of people, or something else such as education in logic.

I know linford has taken courses in analytic philosophy where he was more or less trained to read this sort of sequence of sentence in the manner which he is reporting. Another person I chatted with who answered like linford is a philosophy graduate student who's a moderator on these forums. I sort of suspect that having a background in logic is really the deciding factor but we haven't finished our analysis.

Further, we did run a control and an experimental group. The control has a variety of scenarios involving Julie and Josh and so on according to our Prof's research. The experimental group has all of the scenarios in terms of "you". The questionnaire link above is actually a script that randomly assigns the visitor to one of these two surveys. Our hypothesis is that autists have a hard time reasoning about other minds, so it could be that this simple phrasing plays a role in the reasoning behaviours you guys have been arguing about. So far it would appear that the answer is negative...

[magnificentlenny]

jesus. what a waste of time. Social science- the deification of classic liberalism and the Utopian attempt to make man fit for society, not make society fit for man.

don't worry though, your education will come in handy when proctor and gamble needs someone on their team teaching children how to tell their parents that they "need" some junk on tv, not they "want" it. Or maybe you'll hit the big-time and work in a team of researchers devoted to determining the best method to make parents feel guilt over lack of fundamental development and language acquisition in their child so that they can be convinced to medicate them. the future is wide open. lot of people to decieve.

[Lomax]

For some reason this has sat in my "stuff to read" folder for an age; I finally got 'round to it.

I interpreted the questions the same way as Linford did. My line of thinking was basically the same as this:

There is only ambiguity if you *interpret* the above sentences as:

1. If my grandmother is sick, I buy flowers.
2. If my grandmother is not contagious, I buy flowers, OTHERWISE I DO NOT.

But notice that we have to *add* that line by hand. It is not stated in the problem. Everyone who answers "unsure" read those invisible lines when the problem itself never provided them. It is illogical to infer that the sentence meant something it did not say.

It doesn't, to me, seem preferable to insert an additional clause which not only isn't stated, but actually allows cases which contradict the first premise. For that reason I treated the second premise as irrelevant.

jesus. what a waste of time. Social science- the deification of classic liberalism and the Utopian attempt to make man fit for society, not make society fit for man.

Prft, post-structuralism.

Lomax