*cognitive dissonence =
a state of holding two or more conflicting cognitions simultaneously
*cotards syndrome =
a mental disorder in which people hold a delusional belief that they are dead
intensity of belief has little to do with the ...
ability to express more then one interpertation in info/data
subjective cultural or emoitonal association that a word carries
relation between obj in which one obj designates or acts as a means to connect to the other
*operational definition =
def in terms of how something is measured
*surrogate measure =
intended to stand in for quantites that cant eaisly be measured directly
3 problems regaurding terms in research and arguments are ...
- (i) *Ambiguity - variations of a term affect one's willingness to accept arg
- (ii)*connotation - terms have emotional impact
- (iii)*reference - ook at how something is defined
reasons + conclusion
facts or general principals
what IS the case
*prescriptive argument =
what SHOULD be the case
unstated or unsupported reasons necessary for an arg to work
reasons necessary for arg to work but common to both sides
some words that identify reasons are ...
- - as a result of
- - because of the fact that
- - is supported by -studies show
- - for the reason that
- - in view of
- - because the evidence is
- - first ... second...third
*implicit reasons =
unstated assumptions; taken for granted
*linkaged assumptions =
*definitional assumptions =
assuming definition that affects the argument
*General laws =
reason that is not tied to single time or place (principal)
*explicit reasons =
process of deriveing a conclusion from premises know or assumed to be true
*value assumption =
implicit preference for one value over another
*descriptive assumption =
unstated belief abut how the world was, is, or will become
Implicit assumptions include ...
- - unreliable specifics
- - false linkaged assumptions
- - definitional assumptions that affect the arg
red flags w/in assumptions are...
- - untrue linkage assumptions
- - definitional assumptions that affect the arg
- - untrue specifics(facts)
- - poor use of inference rules
- - unexamined presupposition
thinking in a linear step-by-step manner about how a problem can be solved
experience-based technique for problem solving
*availability heuristic =
judgement about probability of events by the ease which example comes to mind
*representiveness heuristic =
gaug prob. that person belongs to group by comparing characteristics to 'typical' memeber
*system one =
pattern completion - rules of thumb
*system two =
a reasoning 'trick' used to persuade you to accept a conclusion
*ad hominem =
attack or insult on the person rather than directly addressing person's reason
making the assumption that a proposed step will set off uncontrollable chain of undesirable events, when procedures exist to prevent such a chain of events
*searching for perfect solution =
falsely assuming that because part ofa problem remains after a solution is tried, the solution should not be adopted
a key word or phrase is used with two or more meanings in an argument such that the argument fails to make sense once shifts in reasoning is recognized
*Ad populum =
justifying claim by appealing to sentiments that large groups of ppl have in common falsely assumes that anything favored by group is desirable
*appeal to questionable Authority =
supporting conclusion by citing an athority who lacks expertise
*appeals to emotions =
use of emotionally charged language to distract readers and listeners from relevant reasons and evidence.
*straw person =
distorting our opponents point of view so that it is easy to attack; thus we attack point of view that does not exist
*false dilemma =
assuming only two alternatives when there are more then two
*explaining by nameing =
falsely assuming that because you have provided name for some event or behaviour, you have also adequately explained the event
*glittering generality =
the use of vague, emotionally appealing virtue words that disponse us to approve something without closely examining the reasons
*Red Harring =
in irrelevant topic is presented to divert attention from the origional issue and help to win an argument by shifting attention away from the arg and to another issue.
*begging the question =
arg in which conclusion is assumed in the reasoning
*impossible certainty =
assuming that a research conclusion should be rejected if it is not absolutely certain
*faulty analogy =
occures when analogy is proposed in which there are important relevant dissimilarities
everything used to determine the truth of an assertion
*simpson's paradox =
paradox in which a correlation present in different groups is reversed when groups are combined
*vivdness heuristic =
putting undue weight on concrete examples
*selection bias =
error in choosing individuals or groups to take part in study
what are some factors that can affect results when using statistics as evidence?
- - sociological factors (why believe this person or group)
- - cognitive factors (human observation is imperfect - we are very poor at handling uncertainty)
- - statistical factors (meaningless w/o context or point of comparison
set of expectations about world
*confirmation bias =
evidence that supports current mindset is not questioned
*null hypothesis =
generally corresponds to default position
problems with observation are ...
- - is based on knowledge to interpert input
- - precieve what we expect to percieve
- - always based on theories -> null hypothesis
how do we avoid the "you see what you know" problem in observation?
>> constantly reasses your mindset, look at as much info as possible, select between alternatives
belief that explains incomplete data
*situational analysis =
analysis focused on one instance of an event or situation
analysis of event or situation across multiple occurences
process of moving meaning from one target to another target
decision-making strategy that attempts to meet an acceptability threshold
method of working by adding to a project using many small steps instead of large jumps
law or rule that has to be, or is usally followed
a hypothesis is useful way to explain ...
gaps or inconsistencies in data
*situational analysis has ...
low generalty (few instances) and high articulation (many issues)
a theory has...
high generality (many instances) and low articulation (few issues for each instance)
*analogy (comparison) has ....
qualities of both a situational analysis and a theory
what is the function of the imagination?
- i) creation of mental picture - seeing something in your mind
- ii) rearrangment - used for exploring processes e.g. thought experiment
what are some limitations to the imagination?
- - interactions with facts
- - imagery is perception run in reverse, it inherets many of its limitations
what is a warrent?
a warrent is a principal or law which is ment to connect ones reasons to claims
Statistical significance is important to determine if a difference is statistically meaningful. Usally 'meaningfulness' has to do with the size of an effect in statistics, what are some other ways we could assess meaninfulness?
meaningfulness can also be attributed to the impact to which the observed difference will make. We could observe a statistically significant difference within the affects of an allergy medication, yet the meaninfulness might not be that great because the medication barely reduces symptons.
Are abstract words more likely to be ambiguous?
yes, the more abstract the word gets the less concrete it is. Abstracting a term is the process of abstracting general meaning from a particular class of obj. e.g in cs a linked list is a very specific data structure with specific properties, but it is a memeber of an abstract class of collections. Collections encompass many different types of data structures.
'Anchoring' to a number is the reason for why ppl do not react to their total accumulated wealth, but to differences of wealth from whtever number number they are currently anchored to. would we want an intellegent system to behave this way?
No, we would want the system to act on long term net gain and be able to accept some short term loss..
according to beverage, what is the main purpose of hypothesis?
The main function is to suggest new experiments. It also helps one see the significance of an object or event that outherwise would mean nothing. e.g. being prepared for evolution might help one make more significant observations
how can imagination help a researcher?
imagination helps the researcher generate new patterns of thought. Once we have contemplated a set of data, the mind tends to follow same line of thought even if line is unprofitable. Imagination can help generate new lines of thought to follow.
Whether your subject is a rat or a computer program, the task of science is the same to, to provide an answer to three basic research questions:
1) How will a change in the agent's structure affect its __(a)__?
2) how will a change in an agent's task affect its __(b)__?
3) how will a change in an agent's environment affect __(c)__?
- (a) behavior given a task and an eviornment
- (b) behavior in a particular environment
- (c) its behavior on a particular task
__(a)__ data analysis finds things in a haystack, where as __(b)__ puts them under a microscope and tells us whether they are needles and whether they are sharp.
- (a) Exploratory
- (b) statistical hypothesis testing
What are six components common between studying AI systems and studying moderately intelligent animals like rats?
- 1) Agent
- 2) Task
- 3) Protocal
- 4) Environment
- 5) Data collection
- 6) analysis
According to Kuhn, what is a scientific revolution?
when a whole paradigm of thought and methods shifts. Usally these shifts are also correlated with progress in a specific field that had been static for a long time.
Nonconscious mind plays a large role in doing research. Different aspects of this often lumpedtogether under the term ___.
What are the 3 different aspects that are often lumped under the term "intuition"?
- Immediate reaction - reflex, “instinct”, “know-how” (“just do it”)
- Sudden appearance - to conscious mind [cf. Beveridge] —“illumination”, “insight”
- Feeling - that appears out of nowhere [cf Claxton] —“gut feeling”, “taste”
Nonconscious intelligence is better then the conscious one at...
- pattern detection / creation
- can often make a call even when consciously the situation seems incomplete
- good for integrating different sources (non-analytical – cf Schooler example)
conscious mind is good for...
- lift problem out of embedding context (isolate important part of pattern)
- decompose problem into parts (handle combinatorics/systematization/rearrangements)
- carrying out mental simulation / thought experiments (visualization)
- communicating with others
- evaluating via logic (rational)
what type of reasoning is best for simple problems (formalized, less than 4 variables)? i.e. What type of reasoning ought you use if you need percise control (symbolic systems) ?
conscious reasoning (System 2)
what type of reasoning is best for less well-defined problems? i.e. What type of reasoning ought you use if you need to rely on patterns (neural networks)?
nonconscious reasoning (System 1)
what type of reasoning is best for novel problems? i.e. What type of reasoning ought you use if you need to break out of familiar patterns and pick out what might be relevant?
conscious reasoning (System 2)
what type of reasoning is best for uncertainly?
combination of nonconscious reasoning (System 1) and conscious reasoning (System 2). use conscious mind to consider several alternatives. use nonconscious mind to get a “feeling” about each
What is the Clustering illusion? And what causes it?
- Seeing clusters as structure.
- Cause: representative heuristic - like goes with like
What is the Causality illusion? And what causes it?
- Attributing correlation to some cause.
- Cause: confabulation - need to invoke an agency of some kind
What is the Regression fallacy? And what causes it?
- Expecting extremes to stay extreme.
- Cause: prediction should resemble the predictor
- if an event is an extreme, the subsequent event should stay extreme,rather than have (on average) the mean value.
How do you test systematically whether chance is the responsible a factor?
Look at whether chance can explain given data – null hypothesis
What is a null hypothesis?
The null hypothesis typically corresponds to a general or default position. For example, the null hypothesis might be that there is no relationship between two measured phenomena or that a potential treatment has no effect.
What is are the 2 dimensions of science?
- basic: degree to which work affects theories - academic science
- applied: degree to which work affects applications - engineering
When choosing among alternatives with regaurds to a hypothesis, one need to...
give all possibilities fair consideration.
“The only really valuable thing is ___ .”— Albert Einstein
“The human understanding supposes a greater degree of __(a)__ in things than it really finds; and although many things in nature be sui generis and most irregular, will yet invest parallels and conjugates and relatives where no such thing exists.” — Francis Bacon
(a) order and equality
“The test of a first-rate intelligence is the ability ____ and still retain the ability to function.”- F. Scott Fitzgerald
to hold two opposing ideas in mind at the sametime
If you believe that you are correct in a hypothesis, you should...
persevere if you’re convinced. But this should be done on groundsother than emotional ones—you should have some basis for this, and know what thealternatives are.
What are the 8 steps for Analysis of Competing Hypotheses ?
- (1) Generate as wide a range of hypotheses as possible
- (2) For each hypothesis, list evidence for and against
- (3) Determine the diagnosticity of each piece of evidence - Identify evidence that is most helpful (critical data)
- (4)Refine set of hypotheses; Delete evidence that is not diagnostic
- (5)Assess relative likelihoods of each hypothesis
- (6)Sensitivity analysis - Cf perception – generic viewing conditions
- (7) Pick max likelihood
- (8) Prepare for new evidence when it arrives / look for new evidence - keep evolving!
“Science is built up with facts, as a house is with stones. But a collection of facts is no more ___…”—Henri Poincaré
a science than a heap of stones is a house
What is science?
Science (from Latin scientia, meaning "knowledge") is a systematic enterprise that builds and organizes knowledge in the form of testable explanations and predictions about the universe.
Science is not just a matter of data it involves __(a)__to provide understanding. Facts are just a way to get to the __(b)__ of the world
- (a) theories
- (b) underlying structure
Why might there often exist major failures to accept advances in science (Beveridge)?
There can be a failure to fit data into prevailing mindset.
What should one always consider when confronted with any theory?
you might be constrained by existing assumptions (conventional wisdom)
What is a theory? And, what are 3 important ideas behind it?
- A theory is an explanation of something:
- (1) set of basic concepts/ideas, along with relationships between them
- (2) tries to generalize— goes beyond specifics of a single case
- (3) helpful for understanding - prediction
What are concepts or ideas? and how do they relate to the real world?
abstract entities, linked to things in the real world. only some aspects of real world are captured.
A set of necessary conditions on relations that always hold =
set of sufficient conditions to capture some aspect of the world =
What is a necessary condition?
A necessary condition of a statement must be satisfied for the statement to be true.
What is a sufficient condition?
A sufficient condition is one that, if satisfied, assures the statement's truth.
what is it that mindsets and theories have in common?
interpret sensory input/data, determine what is / isn’t important. Additionally, they are both biased to confirm expect result (invoked by particular features/null hypothesis)
What is a Paradigm ?
a set of distinct concepts or thought patterns. According to Kuhn, "A paradigm is what members of a scientific community, and they alone, share."
How do paradigms allow for efficiency?
They allow group work. effectiveness and efficiency increased (cf. Gauch), even if wrong or less complete; benefits outweigh the costs.
A paradigm defines what is considered a legitimate __(a)__, and a legitimate __(b)__. Essentially they, define a __(c)__.
- (a) problem
- (b) solution
- (c) science
What are are the 2 possible outcomes when there are problems / phenomena that don’t fit into the paradigm?
- (1) leave it unaffected (“bounce”)
- (2) replace it (“paradigm shift”) – revolutionary science. There is a complete change in our understanding of something.
A paradigm shift is a ...
change in the basic assumptions, or paradigms, within the ruling theory of science.
“What, then, is a good experiment? It is that which teaches us something more than an isolated fact. It is that which enables us to ___.”—Henri Poincaré
predict, and to generalise.
What is an Experiment?
A study of causality: how an independent variable affects a dependent variable.
What are the 3 ideal stages of research investigation?
- (1) Find an effect. i.e. show that it really exists. e.g., determine structure, correlation; description, prediction.
- (2) Find an explanation for it. i.e. show that the proposed explanation is the best available choice. e.g. determine relationships or prediction.
- (3) If something is incomplete, to go back to 1 (repeat) and try again. i.e. repeat until everything is understood (as much as possible)
What is the goal of exploratory studies?
to discover patterns and discover hypotheses
What is the goal of confirmatory studies?
show that the structure / cause really exists
What are the 2 guidlines of an experiment design?
- (1) Find the right factors (variables)
- (2) Control the right factors
what is a operational definition?
An operational definition defines something (e.g. a variable, term, orobject) in terms of the specific process or set of validation tests used to determine its presence and quantity. The operational definition of a peanut butter sandwich might be simply "the result of putting peanut butter on a slice of bread with a butter knife and laying a second equally sized slice of bread on top"
To find the right factors (variables), what 2 aspects does someone need?
- (1) aspect of behavior and structure that can be measured
- (2) aspect of behavior and structure that is relevant
What are the 3 guidlines of an Analysis Design?
- (1) Take interactions into account
- (2) Independence of measurements is central
- (3) Be careful about context
“____ in biology has probably produced more insights than all experiments combined.”—Ernst Mayr
What are the 3 Different types of study? i.e.ways to investigate phenomena systematically.
- 1. Controlled (manipulation) experiment: e.g. behaviour of an agent or algorithm
- 2. Natural (observation) experiment: association between variables
- 3. Comparative study – look at structure: groupings, correlations of properties
what is a controlled experiement? And what can it investigate?
Manipulation: an experiement that uses use controlled variation (e.g drug dosage).
What is a Natural experiment? What is the control in such experiements? what is the advantage/disadvantage to them?
: association between variables. i.e. use natural variation.
: selection of conditions
- Advantage: doesn’t disturb system (cf. anthropology)
- Disadvantage: can’t determine causality – look at likelihood that correl is causal
What is a Comparative study? What is the control in such experiements? what is the advantage/disadvantage to them?
- Comparative study – look at structure: groupings, correlations of properties. i.e. use natural variation.
- Control: control is in selection of what to focus on (observe)
- Advantage: more widely applicable than natural experiment
- Disadvantage: can’t determine causality — only structure
A Comparative Analysis is ...
a powerful method - used to determine structure and relationships in the world
A comparative analysis is based on __(a)__ instead of causation. There is no __(b)__ or __(c)__ variables.
- (a) correlation
- (b) independent
- (c) dependent
When is a comparative analysis useful?
- useful: when experiments are impossible/impractical.
- e.g. biology – compare structure of animals. e.g. given three species, which two are more closely related. e.g. theory of evolution.
“The implication here is that everything in nature, everything in the universe, is composed of ____ of two elements, or two parts in functional relationship to each other…”—Jonas Salk
what is meant by the idea that there are "emergent aspects" to systems?
more than the sum of the parts
What is a hierarchical system?
A system where sub-systems are built out of components that are themselves sub-systems (holons). Additionally, each level has structure above, and structure below.
One of the reasons that hierarchies are so common, is that they are easier to __(a)__ and fairly __(b)__.
for a system to be a hierarchy, what must it's design control be? What is an issue with hierarchies desgin?
- design of control hierarchies: each unit (holon; sub-system) should function at least somewhat autonomously
- important design issue: what is being controlled at what level. This applies to biological organisms as much as companies (or other organizations).
What are the two main kinds of hierarchy? Explain what they are.
- (1) constitutive hierarchy: higher-level structures made out of lower-level ones. e.g. cells built from proteins
- (2) functional (control) hierarchy: higher-level structures control lower-level ones. e.g. army.
What is a network?
a generalized hierarchie with arbitrary kinds of connections.
What are 2 counterintuitive properties that networks have? and explain them.
- property I: a few nodes have most of the connections. one might expect each node to have roughly the same connections.
- property II: any node is just a few links away from any other. one would expect average separation to be quite high. However, there is usally six degrees of separation.
“The ___ Principle asserts that a minority of causes, inputs, or effort usually lead to a majorityof the results, outputs, or rewards.”—Richard Koch
What do power laws say about webstites?
most websites have a few links; a few have a whole lot
What do power laws say about metabolic networks?
just a few molecules participate in most reactions
What was the origianl formulation of the power law by pareto supposed to address?
original formulation by Pareto:a small percentage of people have most of the wealth. order individuals in terms of wealth; top 20% (or so) have 80% of total wealth. n.b. not always exactly 80/20; could be 80/10, or 90/20.
What is the power law distribution?
f(a) = a y for some y
Why do power-law distributions exist? 3 reasons.
- - 1: every entity (person, item) the same (no linkage)
- - 2: every cause the same -> gaussian curve (linkage to interior)
- - 3: every opportunity the same -> power law (linkage to exterior)
“If you are to do important work then you must work on ____. Without any one of the three, you may do good work but you will almost certainly miss realgreatness.”—R.W. Hamming
the right problem at the right time and in theright way
What 2 methods might one take to finding a particular question/problem? (generation phase)
- Evolutionary Design 1: Start with a particular problem, and modify it
- Evolutionary Design 2: Start with general issue, and progressively articulate it (narrow it down)
What is the danger in trying to formuate a particular research question or problem? how might it be avoided/
- danger: denaturation – lose the essence of the issue
- try to remove irrelevant factors (variables), while keeping something of interest
What is the ultimate goal when formulating a research design? 3 things.
- (1) question - external aspect (computational level)
- (2) problem - internal aspect (representational level)
- (3) solution
What is the MAGIC Criteria? define each part.
- 1. Magnitude: how much of a difference is found
- 2. Articulation: how focused the results might be— number of useful distinctions / details (bits).
- 3.Generalizability: how widely applicable are the conclusions
- 4.Interestingness: ability to change people’s minds
- 5. Credibility (Compatibility): results should be believable - - methodology should be believable
within the MAGIC criteria, when addressing Magnitude, what is needed in terms of data and the hypotheses?
- in terms of data (behaviour) — how big the effect is (effect size)
- in terms of hypotheses (explanation) – how many possbilities are affected. Additionally, how many alternatives can be ruled out (or suggested).
within the MAGIC criteria, when addressing Articulation, what is needed in terms of nature and applicability?
- nature - clearly defined (vs. vague)
- applicability - detailed? (vs. vague)
within the MAGIC criteria, when addressing interestingness, what is meant in terms of surprise, importance, & elegance?
- surprise = number of existing beliefs that are (unexpectedly) changed
- importance = extent of theories or practices that need to be changed
- elegance = parsimony; number of things described by a single idea/law
within the MAGIC criteria, when addressing Credibility (Compatibility), what is needed in terms of results & methodology?
- results: should be believable. i.e. should fit in with knowledge about behavior of world (phenomenal). should fit in most existing theories about the world (theoretical).
- methodology: should be believable. i.e. compatible with existing analysis techniques. methodology should be as sound as possible; minimal assumptions. logical development should be as airtight as possible; assumptions believable
According to Meltzoff, what are tow ways that a hypothesis can orginate?
- (1) a hypothesis can spring from experiential observation; it is not offspring of any formal theory. issue with this is that they lead to a collection of isolated facts with no theory to attach them to.
- (2) a hypothesis can be deduced from theories. These benifit when their broader and more general origins are explained to the reader.
According to Bevridge, how can imagination help a researcher?
- Leads us to new and orginal facts (new combinations of information)
- Pushes our efforts because it allows us to "see" possible outcomes to imagined facts
Are abstract words more likely to be ambiguous, or less likely? Why?
- More likely to be ambiguous.
- the more abstract the word gets the less concrete it is -- meaning gets further away from unambiguous concrete example
According to Heuer, does previous experience with poor information help or hinder subsequent analysis? Why?
- Hinders it
- =>poor info leads to an incorrect mental model thats resistant to change.
According to Beveridge, what is the main purpose of hypothesis?
- suggest new experiments and observations
- => leads to discoveries even when not correct its self
What does the power law mean in terms of distribution?
- there is a 80:20 distribution between events with low magnitude to those with high magnitude.
- e.g. on the internet 20% of the hubs account for most of the connections
According to Kuhn, what is a scientific revolution?
when a paradigm of thought and methods shifts
Could Polya have used a better catagoriy of words in his comparative analysis of European languages? Why or why not?
No, meaning is universal
Does Gilovich think that the tendency to see patterns in random noise is ever useful? why or why not?
- If noise is truely random then no, because it will drive us to come up with a false theory about the underlying order in the noise.
- However, tendency to pick up on patterns is an evolutionary adaptataion. that helps if there are patterns in the environment that could aid in our survival