Exam 3 Flashcards

Question

In AI, what is a general problem solver

Answer 1

A computer program used to perform non-systematic searches.

Answer 2

1. Productive Thinking: Thinking based on a grasp of the general principles that apply to the situation at hand 2. Structurally Blind Thinking: The tendency to reproduce thinking appropriate for other situations but not the one at hand

Answer 3

1. Artificial Narrow Intelligence (ANI), also known as Weak AI or Narrow AI, refers to a type of artificial intelligence that is designed to perform a specific task or set of tasks. 2. Unlike Artificial General Intelligence (AGI), which aims to replicate human-like intelligence and reasoning across a wide range of domains, ANI systems are focused on excelling at a particular task or set of tasks

Answer 4

1. Speech recognition: These recognize and transcribe human speech, including Siri, Alexa, Google Assistant 2. Image recognition: These recognize and classify images, including facial recognition systems and object recognition systems 3. Recommendation engines: These provide personalized recommendations to users, including Netflix’s movie suggestions 4. Language translation: These translate text from one language to another, including Google Translate and Microsoft Translator

Answer 5

1. Data availability: They rely on large amounts of high-quality data 2. Algorithmic bias: They can exhibit bias if the data is biased 3. Explainability: It can be difficult to understand how ANI systems make their decisions, particularly for complex systems 4. Adaptability: They are designed to perform specific tasks, and may not be able to adapt to new or changing situations

Answer 6

1. How to define and measure intelligence in a general way 2. How to ensure AGI systems are aligned with human values 3. How to model and simulate human cognition, emotions, and consciousness 4. How to enable AGI systems to interact with humans and other agents in a natural and cooperative way 5. How to balance the trade-offs between generality and efficiency, exploration and exploitation, and learning and reasoning in AGI systems

Answer 7

Artificial intelligence can teach us about human cognition by revealing the limitations and biases of human thinking, such as heuristics, fallacies, errors and illusions Artificial intelligence can teach us about human cognition by inspiring new questions and hypotheses about the nature and origin of intelligence, such as the role of evolution, culture, emotion and consciousness

Answer 8

As long as computers do not directly interact with the world, they cannot achieve human intelligence

Answer 9

The textbook (wrongly) describes this as a quasi-need to finish tasks The Zeigarnik effect states that people can remember the details of unfinished tasks better than those of finished tasks E.g. Study in more than one session, interrupting them with unrelated work or For a class with midterms and a cumulative exam, it’s best not to breathe too big a sigh of relief after the midterms – keep the task open!

Answer 10

One of our biggest limitations is how we generally treat unexpected findings. Our natural tendency is to resist, and ignore, information that disconfirms our pre-existing hypotheses At least an AI probably doesn’t have pre-existing hypotheses (unless it learns them from us) The scientific method recognizes this limitation and requires us to instead adapt our hypotheses

Answer 11

In the case of scientific problem-solving, in vivo research involves the observation of ongoing scientific investigations, while in vitro research involves laboratory studies of scientific problem solving.

Answer 12

Reasoning done by more than one person (e.g. weekly team meetings). Distributed reasoning is particularly effective in changing problem representations because different people reach different conclusions

Answer 13

Biased evaluation of evidence, in accordance with one’s prior views and beliefs. Recall from Chapter 10 that we often will ignore information that disconfirms our pre-existing hypotheses. A strongly related issue, motivated reasoning is different from simply ignoring information. It’s choosing to evaluate information in a way that is consistent with our beliefs. People are more likely to read the conclusion first. If they disagree with the conclusion, they try to find a flaw in the premises

Answer 14

A syllogism consists of two premises and a conclusion. Each of the premises specifies a relationship between two categories

Answer 15

1. Universal affirmative: All A are B. All cows are animals. All right angles are 90-degree angles. 2. Universal negative: No A are B. No tomatoes are animals. No acute angles are 90-degree angles. 3. Particular affirmative: Some A are B. Some animals are dangerous. Some pigeons are clever. 4. Particular negative: Some A are not B. Some animals are not cows. Some pigeons are not clever.

Answer 16

The belief that logical reasoning is an essential part of human nature.

Answer 17

One in which two premises point to a conclusion that calls for action. Premise 1: I need to understand psychology as a whole. Premise 2: The only way to understand psychology as a whole is through the study of cognition. Conclusion: Therefore I need to study cognition.

Answer 18

People often find it difficult to separate the validity of a syllogism from the issue of whether the syllogism is consistent with their experience or beliefs. Thus they may accept an invalid syllogism if they believe that the conclusion is true in the real world. However, the effect of believability is greater if the syllogism is invalid. Believable but invalid syllogisms were accepted as valid by 71 per cent of participants, while unbelievable and invalid syllogisms were accepted by only 10 per cent.

Answer 19

If it’s unbelievable, they then try to find some flaw in the premises that renders the conclusion invalid. However, if the conclusion is believable, they don’t try to establish that the syllogism is invalid: instead, they try to find some way of thinking about the premises that renders the conclusion valid. In other words, they set themselves the goal of discovering a syllogism to be invalid only if the conclusion is unbelievable.

Answer 20

Reasoning involving premises that express the relations between items (e.g., A is taller than B). Also known are transitive relations.

Answer 21

Linear syllogisms consisting of two comparative sentences from which a conclusion must be drawn

Answer 22

the relations between the parts of the model correspond to the relations between the parts of the situation it represents. It also shows that you can get more out of a mental model than you put into it.

Answer 23

Once you have constructed a mental model, you can see relationships that were not evident before you constructed it.

Answer 24

A process that refers to itself is called recursive. Recursion can have interesting effects on reasoning. The first thing to recognize about recursion is that it can sometimes lead to awkward forms of thought. The most famous example is the liar paradox : There is a very old story about Epimenides the Cretan, who is supposed to have said that All Cretans are liars. Because Epimenides is a Cretan, he is including himself when he says that All Cretans are liars.

Answer 25

A natural deduction system makes use of propositions stored in working memory. Propositions are statements structured around connectives such as if . . . then, and, or, and not. The system uses deduction rules to draw conclusions from these propositions.

Answer 26

Participants are told that the three numbers 2, 4, 6 conform to a simple relational rule that the experimenter has in mind, and that their task is to discover the rule by generating sequences of three numbers. The experimenter tells them each time whether the rule has been followed.

Answer 27

A strategy based on attempting to falsify your hypotheses, in order to eliminate incorrect beliefs.

Answer 28

A way of presenting the various combinations of the constituents of logical statements. p q If p then q T T T T F F F T T F F T

Answer 29

There is little evidence to support the hypothesis that people reason in accordance with the rules of a single logical system. If we did use a single system, our reasoning processes should not be affected by the content of the problem. What constitutes adaptive reasoning may differ from one type of problem to another. Consequently, different types of problem may require the use of different, domain-specific inference procedures.

Answer 30

The theory that inference procedures have evolved to deal with social contracts in which people give something up in order to gain something else.

Answer 31

A heuristic, as we noted in Chapter 10, is a rule of thumb or shortcut that works in some situations but can mislead us in others, while a bias is a predisposition to see a particular type of situation in a particular way.

Answer 32

The larger the sample size the closer the statistic will be to the true value of the population

Answer 33

To adequately assess a conclusion, we need to consider all possible interpretations of the premises and conclusions. But people don’t always work out all models (e.g. mental model of the word 'some')

Answer 34

Anchoring Illusory Correlations Availability Law of Averages Gambler’s Fallacy

Answer 35

Anchoring is when the first value someone is exposed to biases the decision that we would make about subsequent values. This is most easily seen numerically, but ‘value’ does not necessarily mean numbers. Consider the two calculations below. Without doing the math, which produces the larger number? 8x7x6x5x4x3x2x1 1x2x3x4x5x6x7x8 Research showed the first was estimated at 2250 on average, while the second was only 512 (both are way off)

Answer 36

The mistaken belief that events go together when in fact they do not. Findings correlations to confirm our hypothesis instead of looking at all available correlations to draw proper conclusions (relationship between symptoms and disease)

Answer 37

The belief that an answer which comes to mind easily is more likely to be correct. Retrospective decision making is based on our incomplete memory. We confuse the frequency with which we can remember something with how frequently it actually occurs Peak-end rule: Retrospective judgments of the total painfulness of an event are formed by averaging the pain experienced during the most painful moment of the event and that felt at the end of the event. Duration neglect: The finding that retrospective judgments of the total painfulness of an event are unrelated to the event’s duration.

Answer 38

The incorrect belief that events of one kind are always balanced by events of another kind .

Answer 39

The gambler’s fallacy is the belief that when some event doesn’t occur, this increases the probability of it occurring. This is a fallacy specifically in the context of independent trials Whenever there is no dependency in the probability of the outcomes, the probability will not increase. If there actually is a dependency, e.g., from a biased coin, then the probability would indeed increase (potentially by a small amount) “Due to win” when you haven’t won in a long time

Answer 40

The representativeness heuristic is the belief that a sample of data “looks right”, given our (incorrect) expectations for randomness. This is responsible for a lot of logical errors, even on important tests Given a properly balanced coin, tossed 10 times, which outcome is more likely? HHHHHHHHHH HHTHTTHTHT None, they are as likely

Answer 41

The mistaken belief that a small sample should be representative of the population from which it is drawn. May lead to representativeness heuristic

Answer 42

When choosing between two objects (according to some criterion), if one is recognized and the other is not, then select the former.

Answer 43

A heuristic is ecologically rational if it produces useful inferences by exploiting the structure of information in the environment.

Answer 44

Sometimes the person who knows less is able to make a better judgment than the person who knows more but is unable to use that knowledge in the situation at hand.

Answer 45

For example, procedural memory can be described as anoetic (non-knowing) because it does not go beyond itself; it stays in the here and now.

Answer 46

By contrast, semantic memory is noetic (knowing) because when we use it we are aware not only of our immediate surroundings, but also of things that lie beyond it.

Answer 47

Finally, episodic memory is autonoetic (self-knowing) because it involves remembering personal experiences.

Answer 48

A surgical procedure, now abandoned, in which the connections between the prefrontal lobes and other parts of the brain were severed; also known as prefrontal lobotomy. One goal of the procedure was to calm patients who ruminated excessively about them selves and their problems:

Answer 49

He has called our sense of subjective time chronesthesia, and considers it a cognitive capacity as important in its own way as the capacity to see or hear.

Answer 50

The level of consciousness that operates without our attention, continuously monitoring and changing the contents of thought, and tracking and changing behaviour to address goals.

Answer 51

The basic level of awareness.

Answer 52

The level of consciousness at play when you direct your attention to your own state of mind.

Answer 53

The process of transforming information into one or more forms of representation is known as encoding.

Answer 54

Thus a single word might be encoded in terms of multiple dimensions: frequency of occurrence, how we feel about it, as well as its physical characteristics, such as its length and sound. This process is called multidimensional encoding.

Answer 55

Limen: A threshold Subliminal perception (or unconscious perception) : Perception without awareness; occurs when the stimulus is too weak to be consciously recognized, but still has an impact on your behaviour.

Answer 56

Backward masking involves presenting a stimulus, called the target, to the participant and then covering, or masking, the target with another stimulus Because the mask is the last stimulus presented, subsequent feedback and feedforward processing brings the mask into consciousness. The end result is conscious experience of the mask and failure to consciously perceive the initial stimulus. However, the initial feedforward sweep did carry information about the stimulus to the inferior temporal cortex, creating the opportunity for perception without awareness

Answer 57

To determine whether or not a stimulus has affected cognitive processes, then, it might not be necessary to rely on direct measures of cognitive processes, such as participants’ verbal reports that they have not seen the stimulus; indirect measures of perceptual processes could be used instead (i.e. the effects of an undetected stimulus on a subsequent task)

Answer 58

An experimental strategy designed to show that it is possible to perceive stimuli in the absence of any conscious awareness of them.

Answer 59

The point at which participants can detect a stimulus at a chance level versus the point at which they say they did not perceive it.

Answer 60

An experimental technique that requires participants not to respond with items they have observed previously

Answer 61

The effect on a person’s experience, thought, or action of an object in the current stimulus environment in the absence of, or independent of, conscious perception of that event.

Answer 62

Failure to consciously detect an obvious change in a scene. One of the main points raised by change blindness is that our internal representation of the world is not as rich as we think.

Answer 63

The illusion that what we see in our visual field is a clear and detailed picture of the world.

Answer 64

The blurring of information on the retina that occurs during fast eye movements.

Answer 65

The halting of visual processing during an eye movement

Answer 66

The halting of visual processing during an eye blink.

Answer 67

The incorrect impression that a stimulus occupies a section of the visual scene when in fact it occupies only the region around it.

Answer 68

Conscious awareness of what is occurring in one’s consciousness.

Answer 69

self-caught method A technique used to catch episodes of mind-wandering in which participants are asked to monitor their consciousness and report anytime their mind has wandered. probe-caught method A technique used to catch episodes of mind wandering by presenting participants with a probe asking them whether they were mind-wandering just before the probe was presented. Even though participants were consciously trying to monitor their mental activity, the probes still caught a substantial number of mind-wandering episodes. This indicated that meta-consciousness was suspended at times, allowing consciousness to drift to task-unrelated thoughts without the participants’ noticing it.

Answer 70

The general technique of asking people to comment on the contents of their consciousness at specific moments.

Answer 71

The temporary disengagement of meta consciousness, resulting in lack of awareness of the contents of consciousness. It’s even possible that they are the norm and meta-consciousness is the exception

Answer 72

electrooculogram (EOG) A record of the changes in electrical potentials generated by the movements of the eye. electromyogram (EMG) A record of the changes in muscle activity of the body.

Answer 73

According to Chalmers (1995), the easy problems of consciousness are those that can be answered using the standard methods of cognitive neuroscience: for example, which brain areas are active when one is conscious of a visual object. By contrast, the hard problems of consciousness involve understanding the specific subjective qualities of an individual’s consciousness.

Answer 74

When a different image is presented to each eye, the viewer becomes conscious of only one of the images at a time.

Answer 75

Chalmers now postulates that material processes in the brain cannot explain consciousness because consciousness is a fundamental property of matter itself. In fact, he has suggested that all matter has some form of consciousness: this view is called panpsychism.

Answer 76

When different images are presented to each eye and one of the images is replaced, the new image enters consciousness and the image presented to the other eye is suppressed from consciousness.

Answer 77

The view that each individual has “many different visual consciousnesses that are distributed over space and time.” In Chapter 3 we saw that different areas along the visual pathways are responsible for different aspects of visual perception. For instance, if an area known as vision area 4 (V4) is damaged, then colour perception will be compromised. According to Zeki (2009), the various areas involved in visual perception are responsible for distinct microconsciousnesses, which combine to give us the sense of a single unified consciousness.

Answer 78

Spikes in the electroencephalogram that occur at roughly 40 cycles per second

Answer 79

These findings indicate that neither hemisphere is conscious of the meaning of the word presented to the other hemisphere. Rather, each hemisphere is drawing an image of its word and integrating them on the same piece of paper. The only thing that is shared is control of the drawing hand

Answer 80

Phantom limbs occur as a consequence of the way the body schema represents the parts of our bodies and their relationships.

Answer 81

It’s a selecting agency that chooses one thing from many others The chosen information is emphasized and accentuated, while others are suppressed

Answer 82

When the behaviourists were in control of the field, they would have described consciousness something like this (paraphrased) Consciousness? No such thing If there is consciousness, it’s irrelevant (epiphenomenal) for understanding behaviour, so stop worrying about it

Answer 83

The most recent perspective rejects both James’ view and that of the behaviourists Consciousness is an awareness of our environment, ourselves, and the operations of our own minds - whether it’s memories, smells, thoughts, or bodily sensations In his book Mind and Body, Mandler (1975) proposed an interesting model of consciousness where it is useful and even necessary for the proper functioning of several basic cognitive processes

Answer 84

It’s a complex and dynamic system of mental processes that interact with each other and with the environment, involving three main components (and, of course, attention) 1. Perception: largely automatic, conscious if surprising or ambiguous 2. Memory: Schemas are used to guide perception and action, and work with emotional evaluation to determine importance 3. Emotional Evaluation: Comparison of current state with schema expectations

Answer 85

Aside from just enjoying the awareness of our own existence, Mandler suggested three main functions for consciousness 1. Learning: Consciousness enhances learning and is often necessary for speeding up the development of new skills 2. Making Judgments: Using Kahneman’s System 2 to consider alternatives 3. Troubleshooting: Handling unexpected situations that violate well-learned routines (Kahneman’s System 1 decisions)

Answer 86

Change blindness illustrates how easy it is to create a breakdown in the overall system of conscious information processing Change Blindness: is the inability to detect changes to the object(s) of a scene It is easy to reproduce, you just need to have a brief interruption to the perceptual processing of the scene (a few hundred ms is sufficient)

Answer 87

Getting a sense that something is changing, people are aware that there is something going on without being able to pinpoint it

Answer 88

Autonoetic consciousness representing the highest form and anoetic is lowest

Answer 89

Gap between the eyes They never perceive the exact same information due to the separation of the eyes in space, and that’s how we get most of our depth perception

Answer 90

FFA: Fusiform Face Area

Answer 91

Parahippocampal Place Area

Answer 92

When experiencing binocular rivalry, they showed a switch in the relative strength of activity in these two areas, corresponding to the time of the conscious alternation

Answer 93

Damage to the brainstem, typically via TBI or stroke, can make it impossible for you to make any voluntary movement Eye movements are the only behaviour typically spared, as they are connected above the brainstem, but not always Higher order brain functions are relatively intact

Answer 94

At least 1 in 5, or 20%

Answer 95

Severe head trauma is very likely involved, so what if they are experiencing anterograde amnesia Anterograde amnesia would mean that the patient also does not remember how to play tennis, thus cannot engage in the prompt. It does not necessarily mean they are not conscious

Answer 96

When different images are presented to each eye and one of the images is replaced, the new image enters consciousness and the image presented to the other eye is suppressed from consciousness.