CSCI 3702
Why is it potentially surprising that people find modus ponens easier than modus tollens?
Because from the logical standpoint, they're essentially the same rule.
Which of the following best describes (on the basis of experimental evidence) people's use of logical reasoning?
People are capable of reasoning logically, but it takes effort and they do so inconsistently (depending on context) in everyday situations.
Which of the following modifications would eliminate the effectiveness of the Prisoner's Dilemma? (Choose all that apply)
Penalizing the Cooperate - Cooperate decision at the same loss as the Cooperate (you) - Defect (other) scenario Allowing the players to negotiate their choices with each other.
Self-driving cars are a particularly active (and controversial) area of current AI research. Among the following responses to the Turing test, which is most relevant (think of it as an obstacle to overcome) in creating self-driving cars?
he argument from informality of behavior
Why do we need to use a toroid (Links to an external site.) shape in the Schelling model of neighborhoods as discussed in the lecture?
A toroid allows all individuals to have exactly 8 neighbors, otherwise the corner location cannot possibly have 4 neighbors.
Which of the following assumptions / rules in the original Schelling model of neighborhoods as presented in lecture (not considering the adaptations on the original model)?
At each step, an unhappy family will move to an empty cell Each family needs a certain number of neighbors that are alike (red families want a certain number of red neighbors and blue families want a certain number of blue neighbors)
A dendrite serves what purpose in a (prototypical) brain cell?
Detecting inputs from other cells and contributing that information to the cell's decision.
To a good first approximation, our first conclusion about the relation between the distance of a point from us and its disparity is:
Disparity is inversely proportional to distance.
Using the Web as a research guide, read up a bit on Joseph Weizenbaum's famous "Eliza" (or "Doctor") program from the early days of AI. (In fact you can even interact with a running version of the program at: http://www.manifestation.com/neurotoys/eliza.php3 . You take the role of patient and let the program act as your therapist.) At the time of its creation, "Eliza" was considered an interesting AI system; today it is regarded as little more than a programming exercise. Here are some statements arguing for the unimportance of Eliza: which of these arguments is least substantial?
Doctor-patient conversations are not an interesting domain for the study of artificial intelligence or language understanding
Which of the following arguments follows (according to Searle) from John Searle's "Chinese Room" thought experiment?
Even if a computer did "pass" the Turing Test, this would not mean that the computer was exhibiting anything like human intelligence because of its lack of intentionality.
Based on our discussion of mental imagery, which of the following would you expect to be difficult / impossible?
Imagining a crowded bus and counting the number of passengers.
The inability to count the stripes on a mental image of a tiger is an example of what phenomenon of mental imagery?
Indistinctness
One limitation of our initial treatment of the disparity-distance formula is that:
It doesn't take into account the need for identifying individual points as seen from the left and right eye.
Look at the automata exhibited by the Cabaret Mechanical Theatre: https://cabaret.co.uk/ (Links to an external site.) Which of the following predecessors, do you think, is closest to the tradition in which these modern artists are working?
Jacques de Vaucanson
Which of the following can be said of "impossibility proofs" such as Turing's proof of the impossibility of solving the "halting problem"?
Knowing that something is impossible is, historically, often an important advance to theoretical knowledge.
Doing some Web research, you should find that one of the following artists did astonishing (and now vanished) work in the creation of automata. Which of these artists was an automaton-builder?
Leonardo da Vinci
Which of these sentences, referring to the distinction between judgment and problem-solving, is not true?
People are in general exceptionally good at both problem-solving and judgment tasks.
The Ponzo Illusion gives us evidence against a strict interpretation of which theory of mental imagery?
The symbolist interpretation
Look up the "hidden chair" illusion at: https://www.moillusions.com/the-hidden-chairs/ . How does this illusion illustrate the difficulty of the vision problem?
The very same two-dimensional scene can be produce by multiple arrangements of elements in three-dimensions.
Consider again the "$300 bonus" scenario. In what important sense are the two alternative choices (one with a $300 bonus, and one with a $500 bonus) identical?
They both involve a choice between a sure $400 and an even-money gamble between $300 and $500.
Which of the following is not true of the sorts of "abstract puzzle problems" discussed in this lecture?
They can be solved without the use of conscious effort.
Optical illusions are useful tools for studying the computational view of vision because:
They highlight "gaps" in our vision algorithms - situations where the algorithms give the wrong answers.
In Axelrod's "Prisoner's Dilemma World", how many different options are there for defining a "creature"?
2^70
Which of the following statements is the best description of Game Theory?
A model of making decisions in the presence of other decision-making agents.
As a very first step in treating vision as a computational problem, we can think of a retinal image as:
An array of pixels, where each pixel denotes a light intensity value.
Which of the following analogies has not played an important role in the history of science?
Cells are like corporate organizations.
In responding to the behaviorist tradition of psychology, which of the following was not an element of the computational metaphor of mind?
Computers can be linked together to communicate with each other directly, providing an analogy to conversation (or, perhaps, to telepathy).
What can we say about geons and facial recognition?
Geons can identify that we are looking at some human face, but they aren't especially helpful for individual face recognition.
Which of the following are not key defining aspects of the prisoner's dilemma, that separate it from other game theory scenarios?
I can always see what my opponent is going to choose before I make my decision. I personally know my opponent, and can guess their strategy. I can never remember what happened in the last round.
What is a common problem with breadth-first search as a general method of searching a problem space?
If the nearest solution is many moves away (and the graph is large), the number of states we need to search might be practically impossible.
Scientific analogies, in general, are not exact (nor are they supposed to be). Which of these observations does not reveal a flaw in a common scientific analogy?
If the solar system is like a clockwork, then we should be able to tell time by the movement of the sun, moon, and planets in the sky.
What is the likely conclusion to draw from the "bartender" example shown in lecture?
It is easier for people to reason about meaningful situations (in human experience) than it is to reason about abstract symbols.
The "decouple-the-metal-rings" problem is difficult to approach via the same methods as (say) Rubik's Cube. Which of the following reasons is most relevant to this difficulty?
It is hard to know, on inspection, what constitute the distinct "states" of the problem space.
The "$300 bonus" scenario described in lecture illustrates which of the following properties of human judgment?
People are more reluctant to risk a loss than to gamble for a reward, even when the situations are objectively identical.
Which of these facts is not especially relevant to making the conceptual connection between "She shook her piggy bank" and "It made no sound"?
Piggy banks are shaped like pigs.
The primary visual cue that we use to identify classes of common objects (e.g., cup, bucket, suitcase) is:
Shape.
As discussed in lecture (both this week and earlier in the course), one of the early tenets of cognitive science is that:
Software : Hardware :: Mind : Brain
The "mutilated chessboard" example from lecture illustrates which of the following ideas about problem solving?
Sometimes the best way to solve a problem is to experiment with a novel representation of the problem (other than the obvious representation).
The sum of the two distances between the projection of a point on the left and right retinas and the centers of those respective retinas is called:
The disparity of the point.
Consider the statement "If the battery is out, the radio will not work." Which of the following is an example of the "Affirmation of the Consequent" logical fallacy?
The radio is not working, and thus the battery is out.
Doing some Web research, you should find that one of the following thinkers experienced a spectacular failure trying to market a "talking doll" (most people find it scary rather than appealing). Which of these thinkers ventured into the business of automata?
Thomas Edison
Examine following photograph form Wikipedia: https://en.wikipedia.org/wiki/Penrose_triangle#/media/File:LargeTribarGotschuchenAustria.JPG
Vision, because it involves all sorts of guesswork and heuristics in interpreting three dimensions from a two-dimensional projection, is capable of being confused or misled by certain images
Examine following photograph form Wikipedia: https://en.wikipedia.org/wiki/Penrose_triangle#/media/File:LargeTribarGotschuchenAustria.JPG What does this photograph suggest about the difficulty of the "vision" problem?
Vision, because it involves all sorts of guesswork and heuristics in interpreting three dimensions from a two-dimensional projection, is capable of being confused or misled by certain images
"Type 1" problems, as discussed in lecture, are problems for which we don't know where to start to answer them, or even (sometimes) whether they could be answered. Which of the following is an example of such a problem?
What's the purpose (if any) of our existence as human beings?
When considering an image as a 2D array of pixels, what denotes an edge?
A line of pixels with adjacent pixels with very different intensity levels (high and low values).
Which of these situations most resembles the "$300 bonus" scenario?
A man refuses to pay more than $200 for a ticket to a Broadway show. He acquires the ticket for $200, and is then offered $250 for the ticket. He turns down the offer.
If our convolution function is centered on a low-intensity (dark) pixel along a dark-to-light transition, what kind of value will the function output?
A negative number
The human eye, to a first approximation, works like which of the following?
A pinhole camera.
Suppose we have a problem space representation in which the (lone) goal state can be reached via a sequence of edges from some states but not others. Which of the following statements is true of this problem (and its graph)?
As long as our start state is among the states with paths to the goal state, then we can consider this problem solvable.
In fuzzy logic, it is possible to speak of "degrees of truth" using real numbers between 0 and 1. (Here, 0 corresponds to "false" and 1 to "true".) Thus, we might say that the statement "The 2018 Boston Red Sox are a good baseball team" is "0.98 true" (i.e., close to certain) and "The 2018 Baltimore Orioles are a good baseball team" is "0.02 true" (i.e., very close to certainly false). Which of these statements - none of which is entirely true or false -- is most true?
February is a cold month in Boulder.
Which of these does not represent a potential complicating factor for our "pixel-array" portrait of the retina?
There are many wavelengths that the eye is not responsive to.
Consider the following propositional logic statements: i. IF (P AND Q) THEN R ii. P AND S iii. Q AND V Which of the following statements cannot be derived from these three statements?
(NOT P) OR (NOT Q)
Consider the "10 coins in three cups" problem given at: https://curiosity.com/topics/can-you-solve-the-3-cups-10-coins-logic-puzzle-curiosity/ What might make this difficult to solve with a computational system?
A "standard" search program assumes a particular representation of the problem, while this particular problem involves finding a creative reconsideration of the problem statement itself.
Many more people die annually of stroke than as a result of tornadoes; yet many people might find this surprising, since they are more afraid of tornadoes (from watching the nightly news) than stroke. Of which judgment phenomenon is this an example?
Availability
What did Hubel & Weisel's Nobel Prize-winning study discussed in lecture (the one involving cats) discover?
Cats have a particular neuron that responds to a bar of light in a particular orientation moving in a particular direction.
Which of the following tasks did Turing mention as a promising beginning candidate for exploring (what would come to be called) artificial intelligence?
Chess
Why do the vision and language problems seem easy to us?
Everyone seems to learn how to see and how to speak by a relatively early age, so the problems seem easy, though they prove to be hard to represent in program form.
The "rotating-quarters" problem is difficult to approach via the same methods as (say) Rubik's Cube. Which of the following reasons is most relevant to this difficulty?
Finding a solution appears to involve elements of mental imagery or mental simulation.
The crucial information for "parsing" a shape into constituent geons is:
Finding the boundaries or junctions where geons meet.
Suppose we have a "brute-force" technique for checking a 3-digit combination (like "048" or "311"). It would take 1000 guesses (at worst) to go through all the possibilities. Now we add three more digits to our combination, which means that it will take a million guesses to go through all the possibilities. Now we add three more digits, which means it will take a billion guesses. Which of the following expresses the time complexity of this technique, where N is the number of digits in the combination?
For an N-digit combination we need 10^N guesses.
Under which circumstances would it be sensible to use (or consider) means-ends analysis in solving a problem?
Given a state of the problem, we can compare it to the goal state and decide what is the most important difference between our state and the goal.
How did Kosslyn help to settle (some of) the debate between pictoralists/visualists and symbolists?
He used MRI to compare the differences in brain activity between a mental imaging task and a picture viewing task.
The "Economist subscription" example described in lecture illustrates which of the following properties of human judgment?
Human judgment can be manipulated by the way in which choices are presented.
In the chapter by Pretz et al., there is a discussion of "well-defined" versus "ill-defined" problems. Which of the following is true of ill-defined problems?
Ill-defined problems generally don't lend themselves to "problem-space" representation.
The "piano-in-the-mirror" example illustrates which of the following ideas?
It is impossible in principle to recover a unique three-dimensional structure from a two-dimensional projection.
The "vision problem" as discussed in lecture is impossible to solve because
It is impossible to recover 3D objects from 2D information (of the sort on the retina).
roblem" as discussed in lecture is impossible to solve because
It is impossible to recover 3D objects from 2D information (of the sort on the retina).
Despite the simplicity of our first model of vision - interpreting black and white photos - it is not entirely unfair because:
It is, after all, a task that we as human beings are capable of.
Which of the following best expresses the meaning of a symbol like "P" or "Q" in propositional calculus?
It represents a true/false statement (e.g., "Eleven is a prime number.")
The idea of functionalism holds a certain natural appeal for computer scientists. Which of these statements comes closest to explaining why that's the case?
Just as we can study algorithms without worrying about the fine structure of the computers on which they run, we can study cognitive models independent of the physical substrate (neural or mechanical) on which they run.
Which of the following is not true of the concept of probability?
Probability is an easy concept to master.
Let's see if you can recognize another (famous) example of the "conjunction fallacy". Here's a description of Linda: Linda is thirty-one years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear demonstrations. Here are your choices - your job is to rank order them in terms of probability: Linda is a teacher in an elementary school. Linda works in a bookstore and takes yoga classes. Linda is active in the feminist movement. Linda is a psychiatric social worker. Linda is a member of the League of Women Voters. Linda is a bank teller. Linda is an insurance salesperson. Linda is a bank teller and is active in the feminist movement. Which of these statements istrue?
Linda is more likely to be a bank teller than a bank teller who is active in the feminist movement., Linda is most likely to be a teacher.
Besides binocular vision, which of the following is not a cue to the distance between you and objects that you are looking at?
Living objects are generally closer than are inanimate objects, so if you see an animal or plant you have reason to believe it's close.
Poke around the Web to see some descriptions of the Hamiltonian cycle problem. As it happens, the only sure-fire algorithm for solving the Hamiltonian cycle problem for a graph with n vertices is to list all possible arrangements of the n vertices (there are n! of those), and see whether any one of them corresponds to an actual cycle of edges present in the graph. So the problem requires exponential time (in the number of vertices) to solve. Which of the following statements is also true of this problem?
Making a (perhaps very lucky) guess of a correct cycle, and checking that this guess is correct by examining the graph, is relatively easy.
The perceived parallels between machines and people inspire different emotional themes in literature. Which of the following themes is least prominent in this literature?
Military glory
In the past year, an interesting new variation of the Turing test has emerged. Consider this article, published just this past summer in the British newspaper "The Guardian": https://www.theguardian.com/technology/2018/jul/06/artificial-intelligence-ai-humans-bots-tech-companies It describes a strange new variation of the Turing test (although they don't put it that way), in which people hired by software companies, are forced into pretending to be AI systems. In other words, in this case, the person's job is to successfully imitate a computer. (Or maybe: the person's job is to imitate an imagined computer program imitating a person. Are you lost yet?) Not unsurprisingly, this could be interpreted as a sort of troubling job to have — what if you, the customer, think you're talking to a software system, but are actually talking to another person?... In any event, which of these strategies would be least plausible as a technique for imitating a computer in this situation?
Misspell the occasional long or difficult word in answering a question. Sometimes a given word will be spelled correctly, sometimes incorrectly.
De la Mettrie's ideas about mechanical interpretations of the human animal are part of a larger movement in scientific history. Which of these thinkers do you think is most representative of the tradition in which de la Mettrie was working?
Newton, who visualized the solar system as a sort of "celestial clockwork"
When we say of a particular problem, "This is a hard problem", we might mean all sorts of things. Which of these very likely isn't what we really are saying?
No one has ever stated this problem before.
Consider the bowling-ball-and-ping-pong-ball problem shown in lecture. What makes this problem interesting for our purposes?
One can get an exact solution by solving the physics equations, but it's much easier to visualize a rough solution by making a reasonable approximation first.
Consider the famous "farmer crossing the river with a chicken, a fox, and bag of grain" problem (a brief statement of the problem can be found at https://riddlesbrainteasers.com/fox-chicken-sack-grain/ ). Which of the following is not true of this problem?
One cannot represent this problem in a problem space format.
When we speak of the issue of "optimality" in lecture, to what idea are we referring?
Our problem-solving method may fine one particular solution, but that solution may not be (for our purposes) the best solution.
Consider the Kosslyn, Ball, & Reiser experiment (visualizing the island). The naive computational model of shifting attention (e.g. spiraling outward) does not capture the appropriate understanding of what aspect of human attention?
Our visual/mental search incorporates some memory of the spatial relationship between objects
In the reading on "prospect theory" from Kahneman's book, which feature of human valuation is most relevant to the $300 bonus scenario?
People's valuation function is asymmetric, placing a greater relative weight on loss than on gain.
Which of the following is not a typical example of the sort of problem discussed in this lecture?
Poker
Consider the hammerhead shark. Why do you think the animal evolved this sort of head shape?
Presumably the distance between the two eyes helps with binocular vision (particularly for points in the region between the two eyes).
As we have seen, many puzzle-like problems can be represented in a graph ("problem space") format. Which of the following is not (typically) true of this sort of format?
Problem space graphs always have astronomical-size (or infinite) numbers of vertices (states) among which to search.
Among Polya's problem-solving heuristics is the suggestion "Look for a related problem that you know." Why might this be an interesting or challenging suggestion for a computational problem solving system?
Pursuing this heuristic would involve tackling the notion of similarities or analogies between various (superficially distinct) problems.
If software is to hardware as mind is to brain, then that implies
Studying neuroscience is analogous to studying computer architecture - interesting, perhaps, but not necessary to the understanding of algorithms.
Broca's area was discovered to be important in speech through which methodology?
Studying the speech response patterns of brain injury patients with damage in that area.
Which of the following examples from earlier in the class is an example of a convolution step?
The "sombrero" function in edge detection
Read a bit about the "uncanny valley" on the Web. Which of the following works (look them up on the Web) does not play on ideas about the "uncanny valley" to achieve a scary or eerie effect?
The Exorcist
Which of the following is not a typical example of the sort of problem discussed in this lecture?Under which circumstances would it be sensible to use generic breadth-first search in solving a problem?
The closest possible solution is sure to be at most a few moves away, and number of available moves at each step is not exceptionally large.
Hilary Putnam's article "The Nature of Mental States" (1967) presents an argument for a functionalist theory of mind. Which of the following statements is not representative of the functionalist view?
The functionalist account is less susceptible to charges of "mind-body dualism" (the idea that mind and body are entirely different types of entities, and that an immaterial "soul" can direct the body) than the brain-state account.
Before William Harvey promoted the analogy that the heart is like a pump, most physicians followed the classical analogy (due to Galen). Using the Web a source of research, answer: which of the following does not characterize Galen's model of the heart?
The heart is soft tissue.
Consider the "monkey climbing a rope" problem given at: https://activityworkshop.net/puzzlesgames/monkey/index.html What makes this problem difficult?
The problem involves elements of physics knowledge and (most likely) mental simulation and imagery.
Which statement best represents the relationship between human vision and the convolution function approach to edge detection?
The retinal ganglion cells perform a similar function to the convolution function, looking for differences in signal from an area of photoreceptor cells.
Look up some videos or animations of people solving the Tower of Hanoi problem (e.g., on YouTube). These examples show an initial puzzle of (say) 6, or perhaps 7 or even 8 disks. Why doesn't anyone show the solution for a tower of 100 disks?
The solution requires about 2^100 steps, and even if each of those steps takes only a microsecond, the resulting video would be impossibly long in human terms.
The "Teddy Roosevelt" problem is difficult to approach via the same methods as Rubik's Cube. Which of the following reasons is most relevant to this difficulty?
There is a potentially (extremely) wide range of real-world or common-sense knowledge involved in answering the question; so the problem is not self-contained as many puzzles are.
One reason that geons are plausible candidates as "shape primitives" is that:
They retain their qualitative properties regardless of small changes in viewing angle or lighting.
Which of the following is not a usual tenet of functionalism?
Thinking can take place only in human or animal brains.
Consider the "sand timers" problem (Problem 4) at the following website: https://www.analyticsvidhya.com/blog/2016/07/20-challenging-job-interview-puzzles-which-every-analyst-solve-atleast/ What can you say about this problem?
This problem seems, in fact, amenable to a "standard" problem-space representation, and to solution via search.
Which of the following best summarizes Searle's response to the "Robot Reply"?
You could incorporate sensors and actuators (like the input from TV cameras, and motors to control arms and legs) into the Chinese characters that are used to communicate with the room.
Which of the following describes the general class of problems for which neural nets are well suited?
pattern recognition problems