To find a particular topic, scan the list below, and also search this document for specific keywords that interest you by using the FIND button on your web browser. (The FIND function is also located under the Edit menu.)
Ask new questions on the discussion web page.
From previous classes, but still relevant...
Questions regarding:
Wason card selection task
Heuristics and feelings in decisions
Aphasia, scripts, heuristics, speech errors,
grammar, additive info integration
Representational desiderata
ELIZA, CYC, Turing test, Spoonerisms
More on the Wason task
More on Spoonerisms
Too much! How can I learn it all?
Word integration (Carpenter & Daneman model).
Text integration (Kintsch model); and saccades.
Integration in text comprehension.
Pragmatic reasoning schemata.
Availability heuristic, famous names, and mood.
Miscellaneous! (Medical diagnosis, schemas in
language comprehension, logic of science, feelings in decision making,
means-end analysis and hill-climbing.)
Logic of science.
> Hi, I am very confused about the Wason experiment. I don't understand > why if a 7 did not have a D on the other side, why it would be wrong. > The rule says nothing about what a 7 should have on the side. I kind > of understand but I kind of don't. Thanks a lot.
To make sure that we're talking about the same rule, I'll re-state the version that I used in lecture:
|
|
|
|
Certainly the A must be checked, because if there is an odd number on the other side, then the rule is violated.
Many people think that the 4 must be checked, but this is not necessary. It doesn't matter what's on the other side of the 4. If there is a consonant on the other side, that's okay, because the rule only says what's required if there's a vowel on one side.
The other card that must be checked is the 3 (and many people find this difficult). To understand this, you need to realize that the rule would be violated if the backside of the 3 had a vowel. For example, suppose the backside of the 3 had a letter E (a vowel). Then, for the rule to be obeyed, the frontside must have an even number (not 3). So, to make sure that the rule isn't violated, you have to check the backside of the 3, to make sure that it doesn't have a vowel.
In general, for a rule of the form, If P then Q, you need to check any card with P and any card with NOT Q. People easily check cards with P, but find it difficult to check cards with NOT Q, when the rule is abstract.
When the rule is a "permission" or "cheating" situation, then it tends to be much easier to check. For example, consider the rule, If drinking, then must be at least 21. People find it trivially easy to check someone who is drinking and to check someone who is NOT at least 21.
For further reading, there are some related web pages linked into
the class schedule.
> I have a couple of questions about the test. First, I'm not sure if > I have all of the pertinent info about the Wason card selection > task. Could you outline the main ideas? In terms of problem > solving heuristic, I'm unsure about hill-climbing and working > backwards. Also, my notes are insufficient in regards to the role > of feelings in decision making. Thanks so much for your time!
Wason card selection task: See the Q and A above.
Hill climbing and working backwards: These are two problem solving heuristics. They are heuristics, not algorithms, because they are not guaranteed to work.
When you use the hill climbing heuristic, your next step is just whichever takes you closest to the goal, or the least far away from the goal. For example, when trying to solve a maze, when you encounter a fork in the maze, you take the route that appears to lead you in the direction of the goal. Of course, you need to know the direction of the goal to apply the hill-climbing heuristic. In pencil-and-paper mazes, you can see the direction of the goal. In a "rat's eye view" maze, you might not know the direction of the goal, and so you couldn't apply the hill-climbing heuristic. In the missionaries and cannibals problem (a.k.a. the hobbits and orcs problem, or the calvins and hobbes problem), using the hill-climbing heuristic implies that you always try to move a full boat load of beasties across the river when going from start-bank to finish-bank, because that gets you closest to your goal of everyone on the finish-bank, and you always try to move just one beastie back from finish-bank to start-bank, because that gets you least far away from your goal.
The working-backwards heuristic is simply starting at the goal and working backwards to the start. This can sometimes be useful for mazes (but not always), and it can sometimes be useful for other problems too. For example, when trying to write a term paper, it's sometimes helpful to look at a finished paper on the topic, and then work backwards to a thesis and argument of your own. This application of the heuristic solves the problem of getting the paper done, but doesn't always provide the best solution to the problem of learning the material well.
Feelings in decision making: Here's a repeat and expansion of an answer provided below. There were at least two main points, as follows:
First, it is not always easy to verbalize our feelings about a decision, yet some decisions depend crucially on feelings. For example, selecting a picture to hang on our wall depends on difficult-to-justify gut feelings. In fact, being forced to verbally justify a choice can negatively influence the choice.
Second, people with specific brain damage (in medial, orbital
region of pre-frontal cortex), that insulates them from their
feelings, do not make very good decisions, even in "pure reasoning"
situations. Cases of this included Phineas Gage. It seems that these
people don't have the emotional links necessary for properly acting
upon a choice. Feelings are crucial even for the simplest everyday
decisions.
> Hi professor Kruschke I have a few questions about the test. > > Could you explain the Presidents speech and aphasic learning i don't > seem to have much about that, > > What should we know or focus on about the role of schemas and scripts? > > Means/end analysis and Hill climbing? > > Dell's model? > > Is there some relation between Chomsky's model and Mackay's model that > we should know about? > > The additive model based on our homework? > > If you could help me i would appreciate it. ARe these questions on > the web if so sorry for asking again thanks
Aphasia and the president's speech: This was a chapter from Oliver Sach's book, The Man Who Mistook His Wife for a Hat. When watching a televised speech of the president (presumably Ronald Reagan), many of the aphasics laughed at parts of the speech that the normal observers didn't find funny. These aphasics didn't understand the exact words, just his gesture and intonation. One aphasic had the opposite problem: A clear literal understanding of words, but no understanding of gesture and intonation. This patient thought that the speaker was either "brain damaged or lying". The point is that normals effortlessly and automatically integrate information from gesture, intonation, and words. The integration is broken in the aphasics. Another point is that politicians can take advantage of normal, automatic integration.
Schemas and scripts: The main thing to know about them in the present context is that they are one explanation for how we understand language. That is, language is full of ambiguity, and to understand even the simplest narratives, we bring to bear vast amounts of background knowledge. One possible form for this knowledge schemas and scripts. In artificial intelligence programs, schemas and scripts have specific forms, but the terms are also used more loosely to refer to just any kind of structured knowledge.
Means/end analysis and hill climbing: Hill climbing is described above. Means-ends analysis is just analyzing a problem in subproblems, and tackling the subproblem before tackling the whole problem. It's just a heuristic, because your particular subproblems might not have tractable solutions.
Dell's model of speech errors: The aspects of Dell's model that I emphasized in class were these: First, words are split into parts corresponding to the type of phoneme exchanges people make in speech errors - onset, middle, coda (end). Second, there is feedback of activation between layers, like the interactive activation model of letter perception. Third, there is residual activation from previous moments in time, so that phonemes might get activated enough for production at the wrong time.
Chomsky and Mackay: This is too big a topic for a quick summary. Nevertheless, here's an overview. Chomsky's early theory was just a formal sentence generator, without strong commitments to psychological processes. His goal was to specify the grammatical structure of language. His sentence generator started with phrase-structure rewrite rules to create a deep structure for a sentence, and then transformation rules were applied to generate a surface structure for the sentence. The hope was that this system would generate all grammatic sentences, and only grammatical sentences, so that it would constitute a complete specification of the structure of the language.
Psychologists then wondered whether Chomsky's formal sentence generator had anything to do with language processing in the mind. So they looked for correspondences between Chomsky's theory and human performance. Mackay's experiment is one such study.
The additive model based on our homework: You should know
what this model claims about information integration, in terms of
representation and process, and the (graphical) form of its
predictions.
> Hi, Professor Kruschke, I don't understand what representational > desiderata for problem solving is. I was wondering if you could > explain this concept further. thanks, > Hi, I was wondering if I could get a little help with something. > I'm having a hard time finding Representational desiderata for > problem solving in my notes. Could you help me on what this is? I > really appreciate it! Thank you,
In a representation and process for problem solving, there are
three desirable qualities: First, the representation of the start
state should be complete; i.e., include all the relevant information.
Second, the process that gets you from the starting representation to
the goal representation should be efficient. Third, the goal
representation must be explicit, otherwise the problem isn't solved
yet.
> I have a couple questions about the review sheet: > > 1. i can't find ELIZA, the Turing test and CYC project in my notes. > > 2. what is a spoonerism? > > 3. what do i need to know about aphasic comprehension of the > president's speech? > > thanks a lot!
ELIZA, CYC, and the Turing test: We had a computer demo in class of ELIZA in action, and we also saw it discussed in the video on the last day. ELIZA is the computer therapist, that uses a few lexical and grammatical tricks to (poorly) imitate conversation. CYC was discussed at length in the video. It's a massive program that tries to explicitly program common sense that's used for understanding language.
Spoonerism: Lots of examples were given in class, such as "The queer old dean" instead of the intended "The dear old queen". The Reverend Spooner was (in-)famous for his speech errors.
> Hello again. I forgot to ask somthing when i emailed the first > time. For the essay questions do we have to know details of any > specific experiments? For instance, the Wason card selection task, > do we need to know representation, process, task, IV, DV, etc? If > we do, could you help me with e the details of the Wason task? I'm > confused on therepresentation and process (i usually always am)? > Thanks again!
You need to know such details only to the extent that they were actually specified in lecture or in the book. Such details were not emphasized in the case of the Wason card selection task.
The I.V.'s of the Wason experiment include the type of card (for the rule "If P then Q", the cards could be P, Q, Not-P, or Not-Q) and the contextual situation (e.g., abstract situation vs. checking drinkers). The D.V. is accuracy in selection, and perhaps also speed of selectin. The task and results are specified in a previous Q and A.
The representations and processes in the various theories were only
vaguely specified. We discussed four possible types of reasoning:
First, abstract logical rules (check P and check Not-Q). Second,
exemplar-memory, in which you don't really "reason out" an answer, you
just remember the right thing to do from previous cases. Third,
pragmatic reasoning schemas, such as how to check for rules of
permission. Fourth, cheater detection mechanisms, which are activated
by appropriate social situations.
> What do spoonerisms imply about the structure of speech?
Spoonerisms have a very specific form: They are always exchanges of
initial phonemes. This implies (perhaps) that words have certain
types of psychological parts, namely, an onset part that is separate
from other parts.
> I'm a student in your P335 class and I'm beginning to > study for the upcoming exam. The thing is I'm having great > difficulty trying to piece together information to fully > answer your example topics. I don't really know how to > tackle them. When I look at the readings my mind goes > bonkers because I feel that the readings are fairly difficult > to understand and very technical. Should I be focusing on > the notes more than the book or both. I need help!!!! > Thanks. > Sincerely, Confused
Keeping your knowledge organized is crucial for understanding and for remembering. The time you spend figuring out how it all fits together will pay off in terms of performance on exams and in terms of your enjoyment of the class!
To help digest the readings, it's crucial always to keep in mind the "big picture" --- why is the author discussing this particular thing in this particular place? What point is it supposed to be illustrating or demonstrating? Be sure you can answer this question; if you can't, then you don't really understand!
The knowledge organization diagram from the syllabus is
intended to help you keep the big picture. Whenever you read
about an experiment in the book, or hear about one in lecture,
you should be able to fill in all the leaves of the knowledge
organization tree. Sometimes an experment is discussed in
terms of its results only, without a theory thoroughly
described. In these cases, you should at least be aware that
if a theory were proposed, it would have to specify
representation and process.
> ON THE REVIEW SHEET, YOU WROTE THAT WE SHOULD READ PAGES > 309-311 FOR CARPENTER AND DANEMNAN'S (1981) EXPERIMENT. > I READ THOSE PAGES, BUT ALL THAT I FOUND THAT RELATED TO THE > TWO RESEARCHERS WAS THE CHART. IS THIS WHAT WE SHOULD GET FROM > THOSE PAGES-AN EXPLANATION OF THE CHART. THANK YOU VERY MUCH,
> Professor Kruschke- > > Can you briefly describe the difference between the Carpenter and > Danemans model of reading and Kintsch's model of text comprehension. > The explanation in the book is confusing to me. > > In my notes on perceptual span and reading it mentions something about > Scades, but I'm not sure of the correct definition. > > Thank you for your help, > Prof. Kruschke, > What exactly do you want us to know about Integration in Text > Comprehension? Is the main idea that there are three variables? > Thanks,
The Carpenter & Daneman model addresses word-by-word integration of meaning. The Kintsch model addresses phrase-by-phrase or sentence-by-sentence integration of meaning. The two models have a lot in common, in that they both address how meaning is integrated from one unit (word or sentence) to the next. But they also differ in some details, of course. Does the word-integration model do reinstatement searches? Does the text-comprehension model have error recovery heuristics?
Saccades are just rapid movements of the eyes between fixations. When people read, their eyes move in a series of rapid jumps (saccades) and stops (fixations), not in a smooth flow.
Regarding integration in text comprehension again... You should
know how Kintsch's model addresses the three factors that affect
integration. You should also know the basic processes in the model,
including, but not necessarily only, the construction-integration
cycle described in lecture, and the various ways that the model tries
to achieve integration using STM, inference, reinstatement search,
etc.
> Professor Kruschke, > While looking up information about the Wason card-selection > experiment, the book lists permission and obligation as the two types of > pragmatic reasoning schemata. However, I find permission and > authorization in my notes. I was wondering if you could help clear up > this confusion. Thank you.
Don't worry about the difference in terminology. They're really
just different labels for the same thing.
> Professor Kruschke- > I have a question regarding the second essay topic on the review > sheet. In the book it talks about a direct and indirect way of > keeping information active in STM. I understand the direct examples, > but the indirect example is a little confusing. Is it important to be > able to explain them both?
The direct and "less direct" ways of keeing information active in
STM are really not that different. The direct way simply has the very
same words or concepts repeated in consecutive sentences. The
indirect way has the same schema or mental model maintained
in consecutive sentences, without explicit repetition of the concepts.
Both cases lead to the same conclusion: Comprehension is better when
relevant preceding information is still active in STM.
> I also have a question regarding the last essay topic on the review > sheet. Can you clarify the explanation for why people make the > mistake of thinking there are more women than men in the Famous Names > Experiment. And are you refering to the Wright and Bower > experiment(1992) when you mention the influence of mood and > probability estimates? > could you please review whatyou want us to know about the famous names > experiment from lecture? thanks for your time,
The availability heuristic states that people try to generate examples of an event in order to estimate the probability of an event. The more examples they can come up with, the higher the probability estimate.
For the famous names experiment, the famous names are easier to remember. So when estimating the probability (or proportion) of women in the list, the people actually are remembering the famous names, and the list happened to have more famous women than famous men, even though the list had more men overall.
Yes, the "mood" topic does refer to Wright & Bower (p. 424).
What should you know about the famous names experiment? Know the
four leaves of the knowledge
organization diagram (although in less detail than usual since we
did not dwell on this in class). Know the basic procedure and task,
and the result. We did not discuss the independent variables in class,
although you can deduce that one I.V. was the proportion of women's
names and another was the proportion of famous women's names.
Then know the basic explanation in terms of the availability
heuristic. This explanation emphasizes process, and does not specify
much about representation.
> professor kruschke, i went through the study guide and found some > questions i needed help with. what should we know as far as medical > diagnosis, the role of schemas and scripts in lang comp, > propositional reasoning and the logic of science, and the role of > feelings in decsion making. i found there was either too much or too > little material on these topics and wanted to make sure i studied > the right things. also, for mean-end and hill climbing, all i have > is that means end is a subgoal to achieve and that hill climibing is > a particular way to achieve a subgoal. do i need to no more than > that? thanks for your help.
It's difficult for me to write a long discussion of every one of these topics; I'll just provide some pointers. Please e-mail me again if you have more specific questions.
Medical diagnosis. (pp. 433-435) What was "one of the major findings" of this study? What was "a central issue"? (The phrases are in quotes because the book uses them exactly.)
Schemas in language comprehension. This is a very broad topic and was discussed in several chapters of the textbook. It was first brought up on pp. 278-285 of Chapter 9, and then similar issues were discusses throughout Chapters 10 and 11. For example, see Figure 11.1, p.327.
Logic of science. Science can only deny the consequent; i.e., it can only prove that theories are wrong in some way. Science cannot prove that a theory is correct (that is, affirming the consequent - confirming a prediction - doesn't prove that the theory is correct).
Feelings in decision making. This was discussed in lecture. There were at least two main points. First, it is not always easy to verbalize our feelings about a decision, yet some decisions depend crucially on feelings. For example, selecting a picture to hang on our wall depends on difficult-to-justify gut feelings. Second, people with specific brain damage, that insulates them from their feelings, do not make very good decisions, even in "pure reasoning" situations. It seems that these people don't have the emotional links necessary for properly acting upon a choice.
Means-end analysis and hill climbing. Both of these are
heuristics for problem solving. Means-end analysis tries to solve
problems by analyzing sub-goals and solving the sub-goals. Means-end
analysis can get stuck if a sub-problem can't be solved with the
available operators. Hill-climbing is another heuristic, which tries
to solve problems by getting closer to the goal. Hill-climbing can
get stuck if the true solution path requires a detour farther away
from the goal.
> I was going over the exam four review sheet for P335 recently and was > confounded by the propositional reasoning and logic of science question > topic under the multiple choice (#15). Did we discuss this in class? I have > attended every class and can't find it in my notes. > Thanks,
This was discussed in class, along with the topic of conditional
reasoning, after the Wason card selection task. Science can only deny
the consequent; i.e., it can only prove that theories are wrong in
some way. Science cannot prove that a theory is correct (that is,
affirming the consequent - confirming a prediction - doesn't prove
that the theory is correct).
>