Decision-Making PDF

7 Decision-making Introduction The heuristics and biases approach...

7 Decision-making Introduction The heuristics and biases approach to judgements Other factors that influence judgements Evaluation of the heuristics and biases approach Theories of decision-making Summary Review exercise Introduction Human lives tend to be complicated and we are faced with a multitude Copyright © 2003. Taylor & Francis Group. All rights reserved. of choices daily. Many of these are trivial, such as whether to have another cup of coffee or not; others can be life changing or life threatening, such as whether to end a relationship or where to cross a road. Sometimes decision-making requires a simple choice of one of two possible actions, such as whether to take a taxi or a bus. Even such a simple choice involves weighing the costs and benefits. However, making decisions often involves a more complex mix of weighing probabilities, personal desires and personal beliefs. Thus, given the same choices, one person may risk a great deal because of political or religious beliefs but another may be unwilling to take any risk. Sometimes the ‘logical’ decision may be resisted because of factors such as tradition or emotional ties. For example, it might make financial sense to trade in an old car that is costing a lot in repairs but 85 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  LANGUAGE AND THOUGHT a person may decide not to do so because they have become attached to it. Garnham and Oakhill (1994) point out that decisions involving probability judgements fall into two categories: decision-making under risk and decision-making under uncertainty. Decision-making under risk occurs when the probabilities of outcomes are known. For example, if you throw a dice there is a 1 in 6 chance of getting any number. Decision-making under uncertainty occurs when the probabilities of outcomes are not known. Most of the examples in this chapter are concerned with decision-making under risk and these depend on probability judgements of risk and rewards. We have to decide, given a possible reward, whether a risk is worth taking. This chapter starts by looking at some of the heuristics used to make these judgements, then moves on to discuss theories of decision-making. The heuristics and biases approach to judgements Making decisions can be difficult since we are required to weigh up various possibilities, often with critical pieces of information missing. The judgement we form in order to make decisions therefore often requires the use of short cuts or heuristics (see p.74). Heuristics can be invaluable since they enable decisions to be made quickly and, often, accurately. The study of heuristics in making judgements has been dominated by two researchers, Kahneman and Tversky. They have suggested that a limited number of heuristics are used in making judgements and, although these heuristics usually lead to accurate Copyright © 2003. Taylor & Francis Group. All rights reserved. decisions, they can also lead to errors and biases. Hence this approach has been called the heuristics and biases approach (Tversky and Kahneman, 1974; Kahneman and Tversky, 1996). Kahneman and Tversky tend to focus on how heuristics lead to errors in thinking about probability. Three of the heuristics – representativeness, availability, and anchoring and adjusting – are discussed below. The representativeness heuristic The representativeness heuristic is used when a person, object or event is judged to be a member of a class or group because there is a resemblance to a prototype or stereotype of that class or group. It is therefore based on similarity and often ignores other pertinent 86 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  DECISION-MAKING information such as relative size of the group or class. For example, if I described someone as competitive and very fit and then ask whether that person was more likely to be an Olympic 100-metre runner or a librarian you would probably choose the former. This is because the description of someone as competitive and fit seems representative of an athlete. However, this judgement ignores the fact that there are a lot more librarians than Olympic 100-metre runners. The use of the representativeness heuristic was illustrated in a study by Kahneman and Tversky (1973) when they asked participants to judge the occupation of someone based on a description. The participants were told the description was chosen at random from a set of a hundred descriptions of either lawyers or engineers. Half the participants were told there were descriptions of seventy lawyers and thirty engineers and the other half were told there were thirty lawyers and seventy engineers. The description given to the participants conformed to the stereotype of an engineer but not to a lawyer. The participants judged that there was a 90 per cent chance that the person was an engineer regardless of the other information they had been given (i.e. they ignored the 70:30 or 30:70 ratio of engineer to lawyer). Thus the decision was influenced by the similarity (representativeness) of the description to an engineer alone. This tendency to ignore information about how often something occurs in the population (or base rate) is called the base-rate fallacy. The representativeness heuristic has been used to explain the conjunction fallacy. This occurs when people believe that the conjunction of two events is more likely than either one of the two Copyright © 2003. Taylor & Francis Group. All rights reserved. events occurring alone. This was illustrated in a study by Tversky and Kahneman (1983, p.297), who gave participants the following description: Linda is 31 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. The participants were then given a number of options which they were asked to rank from most likely to least likely. These options included statements such as ‘Linda is a bank teller’, ‘Linda is a teacher’, ‘Linda is active in the feminist movement’ and ‘Linda is a bank teller 87 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  LANGUAGE AND THOUGHT and is active in the feminist movement’. Interestingly, most participants rated ‘Linda is a bank teller and is active in the feminist movement’ as being more likely than ‘Linda is a bank teller’. Kahneman and Tversky argue that logically this cannot be the case since not all bank tellers are active in the feminist movement. As there are more bank tellers than bank tellers active in the feminist movement it is more likely that Linda is a bank teller. However, the description contains elements that seem to make Linda representative of the feminist movement thus leading to the conclusion that she is both a bank teller and a feminist. The use of representativeness can also lead to misjudgements of small samples. A sample is often judged as more likely if it resembles the population as a whole. For example, if people are given the following sequences of coin tosses: THHTHT or TTTTTT and asked to judge which is more likely they tend to choose the first because it seems more representative. This is because in a large sample of coin tosses we would expect half heads and half tails. However, both sequences are equally likely since there are sixty-four possible sequences with six tosses and each sequence above represents one of those. When people are presented with a sequence such as TTTTTT and asked whether heads or tails is more likely next time they tend to judge heads more likely (Garnham and Oakhill, 1994). However, the odds of the coin falling as a head or a tail remains the same at 50:50 regardless of what has happened before. This is similar Copyright © 2003. Taylor & Francis Group. All rights reserved. to the gambler’s fallacy, which is the belief that a sequence of losses must be followed by a win. Although the representiveness heuristic can lead to misjudgements in many cases it is a sensible heuristic to use. If something shows many of the characteristics of a class of objects it probably belongs to that class (or as Kunda, 1999, puts it ‘anything that looks like a duck, walks like a duck, and quacks like a duck is most likely a duck’). 88 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  DECISION-MAKING Progress exercise Read through the following two lists quickly and then decide whether there are more women or men in each list. (Based on Kahneman and Tversky, 1973.) List A List B David Beckham Jennifer Lopez Jane Austen John Steinbeck Anna Ford David Niven Brad Pitt Kate Moss Katharine Hepburn Richard Briers Robbie Williams Victoria Beckham Cherie Booth Peter Mandelson Enid Blyton Lewis Carroll Tony Blair Margaret Thatcher Ewan McGregor Kylie Minogue Juliette Stevenson John Ford John Travolta Julia Roberts Iris Murdoch Joseph Conrad Angela Rippon Keith Moon Jonathan Ross Naomi Campbell Charlotte Rampling Tommy Cooper Barbara Stanwyck Robert Taylor George Bush Britney Spears Tom Cruise Kate Winslet Copyright © 2003. Taylor & Francis Group. All rights reserved. Now count the names in each list. Was your snap judgement correct? If not, think of a reason why each list seemed distorted (see p.90 for an explanation). The availability heuristic The second heuristic described by Tversky and Kahneman (1973) is the availability heuristic. They suggest that some judgements of the probability of an event, or that something will have certain characteristics, are based on the availability of information from memory. If people can think of a number of examples they tend to overestimate the probability, but if they cannot think of examples they 89 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  LANGUAGE AND THOUGHT tend to underestimate probability. The effect of availability was shown by Tversky and Kahneman (1973) when they asked participants whether letters such as K or R were more common as the first letter or third letter in English words. Most participants estimate that the letters are more common as the initial letter of words, but in fact both are much more common as the third letter. Tversky and Kahneman believe that the availability heuristic explains the error because it is easier to bring examples of words beginning with K or R to mind than examples when they are the third letters. The availability heuristic may explain why some highly publicised events are judged as more common than they actually are. People tend to overestimate the number of murders in Britain; perhaps this is because, although they are rare, they are headline news. Matlin (2002) describes two influences on the availability heuristic: recency and familiarity. Recency refers to the tendency for memories to fade; therefore more recent examples of an event are more readily available. For example, if people are asked to estimate the number of deaths on the railways following a major rail accident they tend to give a higher estimate than after a period of no accidents. This influence of recency may have applications in therapy. MacLeod and Campbell (1992) found that if people were asked to remember pleasant events they tended to believe that it was more probable that pleasant events would occur in the future (they became more optimistic). However, if they were asked to remember unpleasant events they tended to be more pessimistic and judged that unpleasant events were more likely. The availability heuristic can also influence some people’s Copyright © 2003. Taylor & Francis Group. All rights reserved. perception of politicians. For example, Haddock (2002) asked par- ticipants to think of either positive or negative characteristics of Tony Blair and found that this influenced the perception of him in those participants with little interest in politics. However, there was less influence in the participants who did have an interest in politics, presumably because they were not influenced by the availability of recent memories. Familiarity refers to the amount of contact a person has with a subject. For example, a person who has met many people who are divorced will tend to give a greater estimate of divorce rates than a person who has little contact with divorcees (Matlin, 2002). The influence of familiarity was shown in a study by Kahneman and Tversky (1973) in which they read participants a list of thirty-nine 90 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  DECISION-MAKING names. The list either consisted of nineteen famous women and twenty less famous men or nineteen famous men and twenty less famous women. Participants were asked which sex was more common in the list and tended to give the incorrect judgement that the famous sex was more common than the less famous sex. Presumably the familiarity of the nineteen famous names misled participants into judging that there were more of that sex than the twenty less famous names (review your answers for the progress exercise on p.89). Kahneman and Tversky (1982) have suggested that there is a particular type of availability heuristic called the simulation heuristic. The availability heuristic is based on the ease of recall of actual examples, such as words or accidents. However, the simulation heuristic is based on how easy it is to imagine examples. If people find it easy to imagine that an event may happen they tend to judge it as more likely than an event they cannot imagine. For example, I have a friend who every year places a bet that Port Vale will win the FA Cup. As an avid supporter he imagines (or simulates) a successful cup run each year and this leads him to judge that Port Vale may win. Most other football supporters would find it difficult to imagine that Port Vale could win the cup and would not make the same judgement. The anchoring and adjustment heuristic The third heuristic proposed by Tversky and Kahneman (1974) is the anchoring and adjustment heuristic. They suggested that when making estimates or judging probability people often take an Copyright © 2003. Taylor & Francis Group. All rights reserved. initial value, an anchor, and then adjust it. For example, if you are negotiating the price of a second-hand car the price given by the vendor acts as an anchor. However, studies into judgement suggest there is a tendency to under-adjust if the anchor is inaccurate. Therefore if the anchor is inaccurate the final estimate or judgement tends to be inaccurate. If the anchor is very high people will adjust down, but only a little; conversely, if the anchor is very low people adjust up – but again it is an underestimate. For example, Tversky and Kahneman (1974) gave participants a random number between 0 and 100 and then asked them if that figure was higher or lower than the percentage of African countries in the United Nations. They then asked the participants to give an estimate of the percentage of African countries in the United Nations. When given an initial figure of ten 91 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  LANGUAGE AND THOUGHT participants gave an average estimate of 25 per cent but if the initial figure was sixty-five the average estimate was 45 per cent. They seemed to use the initial figure as an anchor even though they knew it was random. Matlin (2002) notes that the anchor and adjustment heuristic is not restricted to artificial estimates of numbers but has been applied to a variety of real situations such as risk assessment and estimates of property values. It has even been applied to aspects of social psy- chology such as judgements of people. For example, we may use stereotypes about particular groups (e.g. accountants) to anchor a person and even if the person does not fit the stereotype we are reluctant to adjust our view about them (Kunda, 1999). The heuristic can be difficult to avoid and many examiners are aware of it when they are the second person to mark a script or piece of coursework. There is a danger that the first mark acts as an anchor and influences the mark given by the second examiner. Many examiners re-marking or moderating marks avoid looking at mark sheets for this reason. There is some debate about whether the anchor and adjustment is a different type of heuristic from availability. The presence of an anchor obviously provides a value that is available and this may influence the estimates given by participants. However, the availability heuristic does not explain why the estimates are under-adjusted (Manktelow, 1999). Progress exercise Copyright © 2003. Taylor & Francis Group. All rights reserved. Explain how: 1. The availability heuristic may affect a doctor’s diagnosis of a rare disease. 2. The anchoring and adjustment heuristic may affect the decision of what price to offer for a house. Other factors that influence judgements Framing effect One strong influence on judgements and risk-taking behaviour is the way the problem is presented to participants; this is known as the 92 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  DECISION-MAKING framing effect. The framing effect has been shown in numerous studies, but its influence tends to be caused by one of two factors: the wording of the problem or the background context of the problem (Matlin, 2002). A lot of research has shown the influence of the wording of a problem on decision-making. For example, Tversky and Kahneman (1981) gave different groups of participants two versions of the ‘Asian disease problem’. These are shown as problems 1 and 2 below: Problem 1 In problem 1 participants were told that there was going to be an outbreak of an unusual Asian disease in the United States which was expected to kill 600 people. There were two possible programmes to deal with the disease: A and B. If programme A were used 200 people would be saved. If programme B were used there would be a one-third probability that 600 people would be saved and a two-thirds probability that no one would be saved. Problem 2 In problem 2 participants were given the same background information as for problem 1 (an Asian disease was expected to kill 600 people), but the two programmes (A or B) were phrased differently. If programme A were used 400 people would die. If programme B were used there would be a one-third probability that nobody would die and a two-thirds probability that everybody would die. In both problems the deep structure is the same – only the surface Copyright © 2003. Taylor & Francis Group. All rights reserved. structure differs (see p.78). Problem 1 is positively framed (it emphasises the number of lives saved), whereas problem 2 is negatively framed (it emphasises the number of lives lost). Statistically all four answers produce the same average outcome: 200 people are saved and 400 die. However, the framing of the question produced very different decisions. When presented with problem 1, 72 per cent of the participants chose programme A. However, when presented with problem 2, 78 per cent of the participants chose programme B. It seems that if programme A is presented as a sure gain participants avoid taking a risk (programme B) but if it is presented as a sure loss participants are willing to gamble on saving more lives. In this instance the wording of the problem had a strong influence on risk-taking behaviour. 93 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  LANGUAGE AND THOUGHT Another factor that influences the framing effect is the background information that is provided. For example, Tversky and Kahneman (1981) asked participants to imagine that they were buying goods in a store that cost $140 in total (one item for $125 and another for $15). They were then asked to imagine that another store sold the same items for $5 less and asked whether they would be willing to travel to that store to save the $5. Tversky and Kahneman (1981) found that the decision depended upon the item that was $5 cheaper. If the $15 item was $5 cheaper participants were more likely to decide to go to the other store than if it was the $125 item. However, in both cases the saving and the inconvenience are the same! It seems that participants regard a saving of $5 on a $15 item as being more than a saving on a $125 item. In a similar example Tversky and Kahneman (1981) asked participants whether they would buy another ticket costing $10 if they arrived at the theatre and found they had lost their original ticket. Alternatively, participants were asked whether they would buy a ticket costing $10 if they arrived at the theatre and found they had lost $10. Participants were more likely to buy a ticket if they had lost $10 than if they had lost the original ticket, even though the financial loss was the same. In both these examples the decision taken is influenced by the background to the problem rather than the financial consequences. Although these studies might be criticised as being artificial and lacking in ecological validity a number of studies have shown that the framing effect can influence consumer behaviour. For example, Johnson (1987) found that people are more likely to buy minced beef Copyright © 2003. Taylor & Francis Group. All rights reserved. if it is described as ‘80 per cent lean’ than if it is described as ‘20 per cent fat’. Advertising is essentially framing the information about the product to try to influence the consumer to buy it. Overconfidence It seems that people tend to be overconfident about their decisions (i.e. they are more confident about the accuracy of their decision than their actual performance warrants). This is typically studied by asking participants to choose one of two answers to a question and then asking them to rate their confidence in the answer. A rating of 50 per cent would be chance level (i.e. the participant has no idea which answer is correct) and a rating of 100 per cent would indicate that the 94 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  DECISION-MAKING participant was absolutely sure that they had chosen the correct answer. Most studies show that when participants report 100 per cent confidence in their answer they are actually correct about 80 per cent of the time (e.g. Lichtenstein et al., 1982). The level of overconfidence is influenced by a number of factors such as the type of question (being lowest on easy questions and highest on ‘trick’ questions) and gender (males tend to be more overconfident in their decisions than females). As with the studies of the framing effect these studies could be criticised as being artificial. However, the concept of overconfidence in decisions has been found to apply to real-life situations, including military decisions about whether to launch missiles and students’ decisions about when to begin assignments (Matlin, 2002). Hindsight bias Most people are aware that answers seem obvious when they are told the answer; this tendency leads to the hindsight bias. This is the tendency to be more confident about the predicting accuracy of an answer once the answer is known. For example, Fischhoff (1977) gave one group of participants a number of general knowledge questions and asked them to choose the correct answers and then assign a probability of the answers being correct. Another group was given the correct answers and was then asked to assign a probability that they would have given those answers. The second group gave higher confidence levels than the first. In hindsight they were confident that Copyright © 2003. Taylor & Francis Group. All rights reserved. they would have given the right answers. The hindsight bias explains the tendency of people to say ‘I knew that’ when you tell them the answer to a question or problem. Evaluation of the heuristics and biases approach The heuristics and biases approach introduced by Kahneman and Tversky has dominated the study of judgement and decision-making during the past three decades. The research methods they used and the concept that heuristics form the basis of most human judgements and decisions permeate throughout the literature in this area. Their theory has been used to explain a number of different types of research findings and, as Manktelow (1999, p.175) notes, ‘one of the strengths 95 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  LANGUAGE AND THOUGHT of Kahneman and Tversky’s heuristic theory is that it can predict results in a wide range of tasks’. Eysenck and Keane (2000) point out that the theory has generated much research and is relevant to a number of important practical applications of psychology. For example, the approach has been applied to decision-making in business, and Maule and Hodgkinson (2002) describe how heuristics are used in strategic decisions in corporations and how biases such as the framing bias can distort business decisions. This explains why Kahneman was awarded a Nobel Prize in 2002, not for Psychology, but for Economic Science. The concepts of heuristics and biases have even been used to explain belief in the paranormal (Blackmore and Troscianko, 1985). People who believe in paranormal occurrences tend to be more prone to errors of judgement about the probability of events happening and therefore do not believe that events happen by chance. However, there is not total agreement with this approach and an alternative view of decision-making has emerged which is largely associated with Gigerenzer and his colleagues (e.g. Gigerenzer, 1991, 1996). Matlin (2002) characterises the two different positions as the pessimistic and optimistic views of decision-making. Kahneman and Tversky’s approach is the pessimistic view since it tends to empha- sise the errors and seeming irrationality of human decision-making. Gigerenzer’s approach is the optimistic view since it emphasises the accuracy of decision-making when people are given better information about factors such as base rate. One of the main differences in the two approaches is that Kahneman and Tversky tend to present people with information expressed as probabilities whereas Gigerenzer tends Copyright © 2003. Taylor & Francis Group. All rights reserved. to present information expressed as frequencies. Gigerenzer (1996) believes that people are better able to deal with information expressed as frequencies. He also argues that Kahneman and Tversky tend to focus on a statistical analysis of problems without considering the influence of real-life experiences. Furthermore, he suggests that Kahneman and Tversky do not explain how or when the various heuristics are used. In other words they provide a description of heuristics which do not explain a full understanding of the processes involved. The debate between these two views is an intense one and is well illustrated by articles in which the two positions are argued (see Research Articles 3 and 4 – Kahneman and Tversky, 1996 and Gigerenzer, 1996 on pp.114–117). The outcome of the debate may well 96 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  DECISION-MAKING influence the future direction of research into human judgement and decision-making. Progress exercise Briefly describe the following three heuristics: 1. Representativeness. 2. Availability. 3. Anchoring and adjustment. Indicate how each can lead to biases in judgement. Theories of decision-making The discussion in this chapter so far has concentrated on the judgements of probability people use in order to make decisions. Some decisions are made solely on the basis of such judgements (e.g. the decision to take an umbrella because it is judged that rain is probable). However, many decisions are not made on the basis of judgement of probability alone because a lot of decisions are also influenced by needs and desires. As Manktelow (1999, p.191) notes ‘decision making... fundamentally concerns combining information about probability with information about desires and interests’. So, for example, the decision of whether to revise or go out with friends is based both on judgements about how likely it is you will pass the exam and on how Copyright © 2003. Taylor & Francis Group. All rights reserved. much you want to pass (and how much you want to go out!). There are a number of theories of how people use various types of information to make decisions but the basis for most is the subjective expected utility theory. Subjective expected utility (SEU) theory The subjective expected utility (SEU) theory suggests that when making decisions people weigh up the expected utility of an outcome against the probability that the outcome will happen. In this context expected utility is something that is useful or desirable, such as more money or better health. The notion of expected utility was first proposed by von Neumann and Morgenstern (1944), who suggested 97 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  LANGUAGE AND THOUGHT that decision-making is like gambling and we bet on the best probable outcome. This idea was modified by Savage (1954), who argues that the concept of utility is subjective since what is desirable for one person might not be for another. Thus a first prize of a case of wine in a raffle might have high expected utility for one person but low for another whose religion prohibits alcohol. For this reason Savage suggested that the idea of subjective expected utility was preferable. In its simplest form SEU can be calculated by multiplying the probability of an outcome by its subjective utility: SEU = (probability of an outcome) × (subjective utility) Thus an outcome may have a low probability of happening but have a high subjective utility and therefore a high SEU. This may explain why millions of people buy lottery tickets each week despite knowing that the probability of winning is very low – the subjective utility of a win is very high. The SEU theory is built on the idea that if people follow a number of principles (or axioms) they will maximise their expected utility. Some of the more important principles are: 1. People’s preferences are well ordered. So if there are two outcomes, A and B, then either A is preferred to B, B is preferred to A or neither is preferred (indifference). 2. Following on from the first principle the second suggests that people’s preferences are transitive. This means that if someone prefers A to B and prefers B to C then A is preferred to C. Copyright © 2003. Taylor & Francis Group. All rights reserved. 3. The third principle is the independence principle. This is that if there is a factor that remains the same no matter what the decision is then it should not influence the decision. For example, if there were two alternative routes to a friend’s house and both took 20 minutes then time should not be a factor in deciding the route. 4. Finally there is the sure-thing principle. This suggests that if A is preferred to B in every situation then A should be preferred even when the situation is not known. For example, if you prefer route A to your friend’s house whether it is sunny or raining you should choose route A even when you do not know the weather. 98 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  DECISION-MAKING Evaluation These principles seem very straightforward and may even seem like common sense. However, the SEU has been challenged because a variety of studies have shown that people do not follow these principles. For example, there are a number of ‘paradoxes’, such as the Allias and Ellsberg paradoxes, which show that people do not always use the principle of independence. There are even studies that show people do not always follow the sure-thing principle. Tversky and Shafir (1992) asked participants to imagine they had taken a qualifying exam and now had the opportunity of a desirable five-day holiday in Hawaii at a very low price. However, this was a special offer which ended the next day. Participants were asked whether they would buy the holiday, put down a deposit to secure the holiday at the discount price, or not take the holiday. Participants were placed in one of three conditions: (a) they knew they had passed the exam, (b) they knew they had failed the exam, or (c) they would not know the result until after the offer had expired. The decision made by the first two groups were approximately the same, and about 55 per cent opted for buying the holiday and 30 per cent opted to put down a deposit. Thus the majority opted to buy the holiday whether they had passed or failed. According to the sure-thing principle the group that did not know the result should be similar. However, this group were different as only 32 per cent opted to buy the holiday and 61 per cent chose to place a deposit (i.e. they did not follow the sure-thing principle). This study suggests that when people do not know what state they are in they Copyright © 2003. Taylor & Francis Group. All rights reserved. tend to avoid making decisions. Tversky and Shafir labelled this the disjunction effect. Another problem with the SEU is that it does not explain the various studies of the framing effect (see p.92). In these studies the context of the problem, or the way the problem is phrased, affects the decision that participants make. However, analysis of the various outcomes shows them to be the same utility. Thus the subjective expected utility is the same but the decisions are not. Prospect theory The prospect theory, proposed by Kahneman and Tversky (1979), is essentially a modification of the SEU that attempts to explain the framing effect and apparent paradoxes in decision-making. It starts 99 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  LANGUAGE AND THOUGHT with the same central assumptions as the SEU, such as regarding decisions as if they are gambles and that people gamble on achieving the best outcome. It also shares the assumption that decisions are made on the basis of judgements about probability and subjective utility. However, the prospect theory differs in a number of important respects. Firstly, it suggests that people do not make decisions based on absolute gains and losses but rather assess expected utility relative to reference points. A reference point is usually the state a person is in prior to the decision and gains and losses are assessed relative to this point. This idea seems to explain the framing effect. A common aspect of the framing effect is risk-aversion if there is already a gain, but risk-seeking if there is already a loss (Garnham and Oakhill, 1994). For instance, in the Asian disease example on p.93 in problem 1 participants typically chose programme A because from this reference point there are 200 people saved so why risk killing them all? However, in problem 2, participants typically chose option B because from this reference point 400 people will die so why not taking a chance on saving everyone. The absolute values of how many survive or die in all options do not vary but the outcome relative to the reference point does. The second major difference between the SEU and the prospect theory is that the SEU assumes there is a linear relationship between value and utility but the prospect theory does not. The prospect theory suggests there is a non-linear S-shaped relationship between value and utility. This suggests that it is the gains and losses nearest to the reference point that are the most important. So, for example, most people would enjoy winning £10 yet the subjective utility of winning Copyright © 2003. Taylor & Francis Group. All rights reserved. £1,000 would be even higher. However, the subjective utility of winning £1,010 would not be noticeably higher than the subjective utility of winning £1,000. Evaluation The prospect theory is much better at describing how people make decisions than the SEU. It can also explain some of the paradoxes and effects of framing that could not be explained by the SEU. The prospect theory has been described as ‘the most influential contemporary descriptive theory of decision making’ (Manktelow, 1999, p.204). However, it does not deal with all aspects of decision-making – the role of emotions, for example. Some theories, such as the regret theory 100 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  DECISION-MAKING (Loomes and Sugden, 1982), focus on the influence of anticipated emo- tions in decision-making (e.g. would the regret of making a decision that was wrong outweigh the rejoicing if the decision was right?). Summary Kahneman and Tversky argue that there are a number of heuristics involved in making judgements and that, although these heuristics often work, they can lead to biases in judgement; this is known as the heuristics and biases approach. The representativeness heuristic is the tendency to judge that something belongs to a group or class based on its similarity to a prototype of the group or class. This can lead to the base-rate fallacy and the conjunction fallacy. The availability heuristic is the tendency to make judgements based on examples available from memory rather than reality. It seems to be influenced by both familiarity and recency. Kahneman and Tversky also described the anchoring and adjustment heuristic, which is the inclination to use given value as a starting point (anchor) and then to base judgements around it (adjustment). Other influences on decision-making include the framing effect. This is the influence of presentation of a problem on decisions and this can be affected by the wording of the question or the background information. Other studies have shown that people are inclined to be overconfident about the accuracy of their decisions and that this can have important consequences for decisions in real life. Another form of overconfidence is the hindsight bias. This is the tendency to be more confident about the ability to give an accurate Copyright © 2003. Taylor & Francis Group. All rights reserved. answer when the correct answer is known. Kahneman and Tversky’s approach has been very influential in the study of judgement and decision-making but has been criticised by Gigerenzer who has proposed an alternative, more ‘optimistic’ approach. There are a number of theories of decision-making but most are based to some extent on the subjective expected utility theory (SEU). The SEU suggests that decision-making is based on judgements of probability of an outcome and the subjective utility of the outcome. However, this theory fails to explain the framing effect or various studies that show paradoxes in the decisions participants made. The prospect theory suggests that decisions are made using reference points, not absolute values of gains and losses. This theory seems to explain some paradoxes and the framing effect more effectively. 101 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.  LANGUAGE AND THOUGHT Review exercise 1. What is meant by the ‘heuristics and biases approach’ to judgements? 2. Explain why understanding how people make judgements is central to understanding decisions. 3. What other major factor influences decision-making? 4. Explain why the prospect theory accounts for the framing effect better than the SEU. Further reading Manktelow, K. (1999) Reasoning and Thinking. Hove: Psychology Press. This is an undergraduate textbook but is written in a clear and engaging style that is accessible at A-level. Chapters 8 and 9 cover the material of this chapter in more depth. Matlin, M.W. (2002) Cognition (5th edn). Orlando, Fla.: Harcourt, Inc. This is a general cognitive psychology textbook that has a good chapter on decision-making. A recent issue of The Psychologist (2002, vol. 15, no. 2) focuses on judgement and decision-making. The articles cover a wide range of issues but the main focus is on how the ideas and theories discussed in this chapter have an impact on real-life decisions in business, on juries, etc. Copyright © 2003. Taylor & Francis Group. All rights reserved. 102 Lund, Nick. Language and Thought, Taylor & Francis Group, 2003. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/cityuhk/detail.action?docID=199320. Created from cityuhk on 2024-11-22 07:41:17.