Loading
Notes
Study Reminders
Support
Text Version

Reading and Decision Making

Set your study reminders

We will email you at these times to remind you to study.
  • Monday

    -

    7am

    +

    Tuesday

    -

    7am

    +

    Wednesday

    -

    7am

    +

    Thursday

    -

    7am

    +

    Friday

    -

    7am

    +

    Saturday

    -

    7am

    +

    Sunday

    -

    7am

    +

Reading and Decision Making

Welcome to the course introduction to cognitive processes I am Ark Verma from IIT Kanpur.And we have been talking about aspects of reasoning and decision making in this thein this lecture ah last lecture.I talked to you about kinds of reasoning; inductive reasoning, deductive reasoning andtoday I will talk about ah.So, I talked about deductive reasoning in the last lecture, today we talked about inductivereasoning and how do we reach two conclusions from particular kinds of evidences.We will also talk a little bit about how are ah making of conclusion from these evidencesare affected by various biases that we come to at some point.So, in inductive reasoning the premises are based upon the observations of one or morespecific cases and what we do is, we generalize from our observation from these cases to getto a more general conclusion.Now, one of the things is that in inductive reasoning these conclusions are ah that arejust suggestive they are not really very definitive.And these suggestions have various degrees of certainty, I can be 10 percent sure, 20percent sure, 30 percent sure or 70, 80 or 90 percent sure, but generally I am never100 percent sure.So, this is one of the classical differences major differences between inductive reasoningand deductive reasoning . Suppose I am reading out a couple of examplesfor you from goldsteins book, ah observation all the crows have seen in Pittsburgh areblack, when I visited my brother in Washington DC, the crows that I saw they were black too.A good conclusion would be there I think is a pretty good bet that all the all crows areblack.So, this is again I its a its a guess ah another observation could be here in Tucson the sunhas risen every morning, ah the conclusion could be the sun is going to rise ah you knowinduction tomorrow.So, different degrees of ah confidence in the last one the degree of confidence is muchmore.You do not really look for validity when we are trying to engage with detect inductivereasoning.ah What we are actually looking for is how strong the evidence is, you know how strongthe argument is.And strong arguments basically result in conclusion that are more likely than weaker arguments.We do not really talk about final you know conclusions or stuff.Now, the strength of these arguments can be basically determined by three things representativenessof observations.So, if you are looking at an observation, how well does the observation ah about a particularcategory resemble or represent all members of that category.Also a very important factor is number of observations how many observations have youmade?1, 2, 20, 200, 2000, 20000 adding more and more observations basically makes the ah argumentmuch stronger.If you remember the last example, here in Tucson the sun has risen every morning.So, the confidence in the conclusion is a bit more as compared to the confidence inthe earlier statement.And the third is the quality of evidence that you are looking at, the stronger evidenceleads to stronger conclusions and more definitive more ah you know let us say better conclusions.Now, let us talk a little bit about how our you know process of inductive reasoning isaffected by a variety of biases and variety of things that come in when you talk aboutavailability heuristic.So, when faced with a choice and somebody is asked to make a choice, what we are doingwhat we do is we are often guided by what we remember from our past you know and themost recent past is remembered most ah you know more.So, the idea is the availability heuristic states that events that are more easily rememberedmore and you know especially events that have just occurred, our choice as being more probablethan events that are less easily remembered ok.I will give an example when participants were asked to judge whether the ah you know whetherthere are more words with r in the first position versus more words there r in the third position,a lot of people basically responded that there are more words ah that begin with r 70 percentparticipants responded that there are more words that begin with.R compared to more words that have r in the third position.However it is actually ah the case that there are much more words that have r in the thirdcase in the third position as compared to words that begin with r.Because it was easier to remember words that begin with r as compared to be remember wordsthat have r in the third position, people thought that you know much more words mustbegin with r.So, this is one of the examples or demonstration of availability.I will show you table right now, which basically has likely causes of debts and people whoare asked to respond to them.You will see that as far as homicide is concerned, 20 percent you know homicides basically happen20 times more than you know ah then appendicitis.And 91 percent picked the more likely cause of death as compared to appendicitis.But if you look at some of the other things ah almost say ah 83 percent of the peoplewrongly chose ah pregnancy as you know causing more deaths as compared to appendicitis.So, in in this pair.So, the idea is that what is happening here is, ah a substantial proportion of parchmentsare misjudging the relative likelihood of these causes of death.ah Large number of errors are basically associated with causes that has been publicized by media.For example, 58 percent of people thought that they were more deaths caused by tornadosthan asthma, while in reality 20 times more people died of asthma than from tornados,but because tornados in the in the United States are more on the news they are publicizethey are talked more and they are you know there are so many pictures and visuals available,that a lot of people misjudge the fact that asthma basically leads to more deaths as comparedto tornadoes.You can look at the ah fourth figure four figure here, now 42 percent people are onlyare making their correct choice.Another example is that 41 percent and people thought 40 percent ah people thought thatparticipants ah thought that botulism causes more deaths than asthma.So, you see here asthma in botulism.So, 41 percent people are thinking I am making the wrong judgment and selecting botulismover asthma.Now, an experiment was done by Stuart McKelvie ah in 1997 and that demonstrates the factthat ah availability heuristic ah you know it kind of demonstrates different way.So, what did they did was McKelvie presented lists of 26 names to participants.There were two conditions in the famous men condition ah 12 of the names were famous menand 14 were non famous women.While in the famous women condition 12 of the names were famous women and 124 were ahand 14 were non famous ah were non famous men ok.So, 12 famous men, 14 non famous women, 12 famous women, 14 non famous men.When participants were asked to estimate whether they were more males or more females in thelist they had heard their answer was in influenced by the fact that whether they had ah hearda famous male or the famous female list . 77 percent of the participants who had heardthe famous male list ah suggested that there were more males in the list, even though yousee there were 14 non famous ah women and 81 of the participants who had heard the famousfemale list suggested that there were more females in the list.So, the result is consistent with the availability heuristic because you hear more ah famousah you know you hear more of you know famous people, the idea is that that is influencingyour judgment of you know how many people were there in all .The next heuristic that affects our decision making is referred to as the representativenessheuristic.Now the representativeness heuristic basically is based on the idea that people often makejudgments based on how much one event resembles the other event you know how suppose for example,one of the you have observed one event, and you kind of try and liken this event to allthe other events you know this has happened, this one did it resembles what I saw lastyear or this one resembles what I saw 2 days ago or 5 years ago or 5 years ago and youkind of make this likeness.So, in other words what is happening is that the, probability that for example, A is amember of class B can be determined by how will the properties of A resembles ah youknow to what we associate with being the general properties of being I show you this with anexample.So, randomly you know ah people picked one male from the population of the United States,the male Robert and it speaks and quietly and he wears glasses and he reads a lot.Now then people were asked whether Robert could be a librarian or a farmer.So, you can guess it while I am going to the next slide ah what do you think Robert basicallydoes?When Tversky and Kahneman ah produced you know they present in this question in an experiment,more people judged that Robert was a librarian why?Because this is more associated to how we perceive librarians they were influenced bythe representativeness heuristic into basing their judgment on how closely, they thinktheir characteristics of this male Robert resemble that of a typical librarians.What you are doing is you are likening these characteristics to what the stereotype ofa librarian is.However, in doing so, they were ignoring a very important source of information thatthe base rate of farmers and librarians.There are far more farmers in the country as compared to librarians.So, if somebody is randomly picking some male, there is a much higher chance of that personbeing a farmer as compared to a librarian I am sure this kind of example would workvery well in our country as well.So, this aspect of base rate, what is the base rate?Base rate is basically ah the relative proportion of different classes in the population.So, by that you know ah maybe Robert was more likely a farmer because in 1972 at that time,you guys had much more farmers than librarians.If I talked of India; obviously, the thing applies here ah much more that you know weprobably have much more farmers than librarians in the country.So, what happens suppose if you tell the base rates to people if you inform them of thebase rate while you are asking them to make these decisions?So, participants ah given this problem correctly guessed ah that there would be a 30 chanceof picking up an engineer; however, for some participants another description was added.So, the similar task was there and they were basically given that you know the base rateis 30 percent of the chances that you will select an engineer.Some other for a different group of artisans other information was also added.Now adding this description ah cause participants to greatly increase their estimates that randomlyah picked person was engineer.So, what they are doing is, they are including the base rate information in their judgment. Let us let us say this example in a groupof people there are seventy lawyers and thirty engineers what is the chance that we pickone person from the group at random and that person will be an engineer so; obviously,30 percent will be an engineer.Now, let us take a different example, ah jack is a 45 year old man he is married and hasfour children, he is generally conservative careful and ambitious, he shows no interestin political and social issues and spends most of his free time on on his many hobbies,which include home carpentry, sailing and solving mathematical puzzles.Now, apparently when only base rate information is available, what people do is people usethe base rate information, but when other descriptions are available, people disregardthe base rate information and this also can lead to potential errors in judgment as youwill see in the last lecture.So, what is happening is another characteristic of this representativeness heuristic is theah conjunction rule, that is the probability of a conjunction of two events is much lessthan the probability of you know it cannot be higher than the probability of a singularevent.So, if I am talking about two scenarios, let us say ah you know I have a description here,Linda is 31 year 31 years old single and outspoken very bright, she is majored in philosophyas a student she was deeply concerned with issues of discrimination and social justicealso ah you know at the end she participated in anti nuclear demonstrations now what isthe likelihood of the following two alternatives and there are two examples Linda is a bankteller, and second is Linda is a bank teller whose active in a feminist movement.Now, the fact is a lot of people basically mistook this thing.A lot the technical thing is hear that because feminist bank tellers are basically a subgroupof all bank tellers the probability of feminist bank tellers should be much lesser as comparedto the probability of Linda being a bank teller ok.So, a lot of people would make this kind of a mistake, but the second you know a differentkind of heuristic that also operates while people are doing inductive reasoning is thelaw of large numbers.What is the law of large numbers?ah The idea is that is the largest a number of individuals that are randomly drawn froma population, the more representative ah the resulting group b group will be of the entirepublic.Suppose I am drawing a sample of 5 people, how likely is it that these 5 people representthe whole population?Suppose I am drawing a sample of 1000 people, how likely is that this 100 people will representthe more population?So obviously, you will assume that the second picking we will have more representative youknow we will be more representative of the entire population.So, ah you know let us have this example again borrowed from Goldstein.A certain town is served by two hospitals in the larger hospital about 40 babies areborn each day and in the smaller hospital around 50 babies are born each day.As you know about 50 of all babies are boys, how the exact percentage of boys and girlsmust be varying from a you know a day to day?Sometimes it might be that it might be the fact that higher than 50 percent ah of themare boys or sometimes lower than 50 percent are boys.For a period of one year both of these hospitals are large hospital where 45 babies are bornevery day, and a small hospital at 55 15 babies are born each day; ah recorded the day isin which the more than 60 percent of the babies were boys which hospital do you think wouldhave recorded more such days?So, the larger hospital or the smaller hospital or both would have done about the same.Now, what happens here is when Kahneman and Tversky basically presented this 22 percentof people picked larger hospital, and 56 percent of the people ah picked there they were therates will be about the same.Now, what is happening with the second group is, ah that they they are assuming that therewill be no difference because in the larger and in the long run and the birth rate forboth the hospitals would be almost identical which should you know will not be really verydifferent because its being observed for up to one year.But the correct answer is that there would be more number of days with 60 percent male,but more male births in this smaller hospital ok why is that because the fact is again thereare there is a sample sizes are very small there, but people what they are doing is becausethey are looking at the global sample size, they are looking at one year and they aresaying in one year time both ah the birth rates in both of are really should be identical. Another bias that I can talk about is referredto as the confirmation bias that is our tendency to selectively selectively look for informationthat conforms to our hypothesis and overlooked information that r use against it.You look around yourself you see how you read you news or how you listen to news or howyou you know evaluate what people are telling you, you will find that a lot of times weare much more likely to believe we are much more ready to believe, what is consistentwith our points of view as compared to what is not consistent with our points of view.Let us say ah you know Wason basically did this a very interesting task and he presentedthe participant should following instructions.I am going to give you a bit of an experimental demonstration.So, what happened was, the instructions were like you will be given three numbers ah whichconform to a simple rule that I have in mind your aim is to discover this rule by writingdown sets of three numbers together with your reasons for your choice of them.So, the ideas Wason will give you three numbers and then people will basically be writingthree numbers which should test the rule which Wason has in mind.After you have written down each set you shall tell you whether your numbers confirm to therule or not.When you feel highly confident that you have discovered the rule you you have to writedown the rule and tell me about it.So, this is what the instruction that Wason gave.After the instructions of presented Wason give the first set of numbers, which were2,4, in 6 the participants then began creating their own sets of three numbers and they startreceiving feedback from us.Wason told participles only whether the numbers proposed fit their rule the participants didnot find what their rule was correct until they felt confident ah enough actually toannounce the rule.The most common initial hypothesis was increasing intervals of 22 plus 2, 4 plus 2, 6.Because the actual rule was three numbers increase increasing order of magnitude therule increasing intervals of two is incorrect even though it kind of fits the overall youknow rule even though it kind of creates the you know sequences that satisfy Wasons rule.So, they were not being able to test it.And the secret to determining the correct rule is to try and create a sequence thatdoes not satisfy the rule.So, if somebody comes over this sequence 2, 1 and 5 or something like that that is whatwe test the rule.So, thus determining that the sequence 2, 4; 5 is correct allows us to reject our initialhypothesis and it allows us to make a new hypothesis ok.So, there were few participants whose rule was correct in their first guess followedby a strategy ah and basically what they did was they followed a strategy by testing anumber of hypotheses themselves before announcing their who by creating sequences that weredesigned disconfirm their current hypothesis, the idea was what these people were doingis and they saw this three were and three numbers that Wason was given and they cameup with an hypothesis then they started coming with numbers which dis confirm their hypothesisand its you know kept doing it again and again till they ah were very confident that theyhave figured out the correct hypothesis.In part in contrast participants who did not who could not guess the correct rule on theirfirst try they tended to keep creating hypotheses that confirm their current you know sequencesthat confirm their correct with hypothesis though they kept creating a numbers whichkind of fitted with what they were thinking.Now, this confirmation bias you know this is really what confirmation bias actuallyworks like.So, the confirmation bias kind of acts like a pair of blinders, you know its just likeyou put two things on your eyes and what we do is you seeing the world according to therules you think are correct or we think are correct and.So, that we are never dissuade it from this view because we seek out only evidence thatconfirms this rule.If you have a particular kind of a opinion about somebody if you have a particular kindof you know opinion about a particular you know ah class, what you tend to do is youcan you only look at those aspects of the behavior of this person, that fit what youinitially thought.You kind of tend to very conveniently ignore everything else that does not fit your hypothesis,and this in sense is the problem with your reasoning and this is basically what confirmationbias is all about.And this confirmation bias is so, strong that it can actually affect peoples reasoning bycausing them to ignore sometimes very relevant information as well.So, lord and coworkers they wanted to test this out and they basically demonstrated thisin an experiment that tested how peoples attitudes are affected by exposure to evidence, thatcontradicts those attitudes.So, sometimes people are not able to change their attitudes even in presence of you knowparticular kinds of evidence.Let us see by means of a questionnaire lord identified one group of participants, whowere in favor of capital punishment, and another group of participants that were not in favorof capital punishment.Each participant was then presented with descriptions of research studies on capital punishmentshowing either that capital punishment was acting as a deterrent or not.So, they were presenting each participant with the studies as some studies showed thatit was acting as a deterrent; some study showed that it was not acting as a deterrent.When participants were reacting to these studies their responses reflected the attitudes theyinitially had they had at the beginning of the experiment.For example A 1 an article representing the evidence that supported the deterrence ofthe punishment was rated as convincing by proponents of capital punishment and unconvincingby those that were against it.Here you see the working of how the confirmation bias is going.So, people are kind of you know finding things that are you know, they think that anythingthat ah confirms with my perspective is more convincing evidence and anything that doesnot confirm with my perspective is less convincing evidence and you will see this operating allthe time all around you, ah and in the way people talk and create arguments.So, this is all about inductive reasoning from my side, today ah we have one more lectureto go ah from reasoning and decision making and we will talk about that in the next lecture.Thank you.