Decision-Making Lecture 1 PDF

Decision-Making Lecture 1 INTRO AND PROSPECT THEORY Acknowledgement of Country  Lecture style Was a student who preferred reading a story at my own pace instead of relying fully on hearing it. So, catering to my ‘past self’, the slides and notes (under slides) are w...

Decision-Making Lecture 1 INTRO AND PROSPECT THEORY Acknowledgement of Country  Lecture style Was a student who preferred reading a story at my own pace instead of relying fully on hearing it. So, catering to my ‘past self’, the slides and notes (under slides) are written to tell a story. All assessable content is in those slides and notes, as well as the readings. Because our stories can get technical, the slides and notes should be searchable through Ctrl+F. I will assess lecture content, but expect that the readings will assist with understanding by describing technical content ‘another way’. The University of Adelaide Slide 3 Module outline Practice quiz, Lecture content and summative quiz Exam readings and self-directed learning  Lecture outline Focus questions Lecture sections What is a decision? Introducing key concepts What is a model, and how have researchers used 'toy problems' to build models of decision- making? What are the basics of two early theories of decision-making: Prospect Theory Expected Utility Theory and Prospect Theory? What is the endowment effect predicted by Prospect Theory? The University of Adelaide Slide 5  Readings for this lecture The University of Adelaide Slide 6  Tutorial t u re 1 L ec Prospect Theory (1979) e3 tu re 4 tu r Lec Lec re 2 Fast and Lec tu Heuristics Bayesian models frugal and biases of cognition heuristics (1974) (1994) (1991) Tutorial in the week of Lectures 3 and 4 The University of Adelaide Slide 7 Introducing key concepts  Information processing Like the broader field of perception and cognition, the field of ‘judgement and decision-making’ conceives of the mind and brain as information processors. That is, the decision-making field assumes that: Information made available by the environment is processed by the mind/brain through a series of processing systems (e.g., attention, sensation, perception, short-term memory, and long-term memory). The mind’s processing systems transform or alter the information in systematic ways; that is, they engage in information processing. The goal of research is to define the mind’s processing systems and the steps they perform. The University of Adelaide Slide 9  Information processing Drawings by a patient with apperceptive agnosia following a stroke — copying the model (column 2), and drawing an apple, book and boat from memory (column 3). The University of Adelaide Slide 10  Mental representations The brain’s systems process internal representations of information. Internal representations – more commonly termed ‘mental representations’ – include: prior knowledge and expectations stored in memory firing rate of retinal cells in the eye to represent level of light brightness etc. The University of Adelaide Slide 11  The mind (vs. the brain) The concept of internal representations of the world is so central to psychology in general that it is common to hear researchers in psychology referring to the 'mind', a conscious internal world, rather than the 'brain', the biological entity housing the mind. The University of Adelaide Slide 12  Rationality The University of Adelaide Slide 13 How do you define ‘rationality’ to yourself – intuitively, before double-checking with Google or Siri? 14  Rationality: Classical view Heraclitus From ancient Greek times until the 1950s, rational behaviour (i.e., rationality) was defined as involving all of the following: complete, or at least clear and extensive, knowledge of the relevant aspects of one’s environment a well-organised and consistent system of preferences ability to calculate which action alternative from various available alternatives allows the highest attainable point on the preference scale to be reached Hegel The University of Adelaide Slide 15  Rationality: Classical view From ancient Greek times until the 1950s, For decision-making rational behaviour (i.e., rationality) was specifically, the implication of the classical view is that, defined as involving all of the following: to be considered rational, complete, or at least clear and extensive, human decisions must knowledge of the relevant aspects of one’s respect six principles: Ordering of alternatives environment Dominance a well-organised and consistent system of Cancellation preferences Transitivity ability to calculate which action alternative Continuity Invariance from various available alternatives allows the highest attainable point on the These are described in the preference scale to be reached reading by Plous (1993). The University of Adelaide Slide 16  Bounded rationality In 1955, Herbert Simon, an American economist, political scientist and cognition researcher, introduced the notion of 'bounded rationality’. Simon raised doubts as to whether the original conception of rationality was suitable as the basis for a theory of how people do— and should—behave. Instead, he proposed that definitions of rationality should consider an organism’s cognitive capacity (e.g. memory limitations), and the structure of the environment the organism inhabits. In a later commentary piece, he concluded: Human rational behavior is shaped by scissors whose two blades are the structure of the task environment and the computational capabilities of the actor. Simon, 1990, p. 7 The University of Adelaide Slide 17 Calinescu, 2021 Slide 18  Models Models in decision-making draw heavily on Simon’s approach. But what are ’models’? The University of Adelaide Slide 19  Models Models in decision-making and perception and cognition more generally are theories that: recognise recognisean ananalogy analogybetween the brain/mind and a computer between the brain/mind and a computer The University of Adelaide Slide 20  Models Models in decision-making and perception and cognition more generally are theories that: recognise an analogy recognise the critical role of between the brain/mind the physical environments and a computer (since minds inhabit computers are information processors) The University of Adelaide Slide 21 The University of Adelaide Slide 22  Models Models in decision-making and perception and cognition more generally are theories that: recognise an analogy recognise the critical role of between the brain/mind the physical environments and a computer minds inhabit tend to answer one of three questions concerning the nature of the task and the nature of brain machinery; the questions can be considered separate levels of analysis The University of Adelaide Slide 23  Models: Three levels of analysis This level of analysis… Answers this question… Computational (strategic) level What problem does the process solve? Algorithmic level Through what series of steps is input transformed into output? Implementational level How are the transformational steps physically realised in the brain? Models at the computational and implementational levels are more common and inform models at the algorithmic level. The University of Adelaide Slide 24  Models: Three levels of analysis A snippet of data relevant to an algorithmic model – brain areas activated during trials with heightened levels of error corrections; error correction is calculated based on degree to which choices in a computerised task changed following corrective feedback Love, 2015 The University of Adelaide Slide 25  Models: Three levels of analysis All the decision-making models we will discuss are computational. Most are ‘descriptive’, but some are ‘mathematical’. This level of analysis… Answers this question… Computational (strategic) level What problem does the process solve? Algorithmic level Through what series of steps is input transformed into output? Implementational level How are the transformational steps physically realised in the brain? The University of Adelaide Slide 26  Decisions: The phenomena being modelled A judgement involves discerning a pattern in some cues to form a conclusion—an estimate— about an unobserved state of the world. For example, an oncologist might need to bring together information from a number of diagnostic tests to make a judgement about whether a patient’s prostate tumour is life- threatening or benign. A decision is a choice about several possible action alternatives based on a judgement. However, many judgements presuppose courses of action, so judgements and decisions are often equivalent. There is often no right answer in a decision- making situation. The University of Adelaide Slide 27 Decisions: Different types Significance Medical diagnoses (i.e., what is at stake) Everyday decisions Most decisions Slide 28  Decision-making in toy problems Given that there are few objective metrics of decision-making accuracy in the real world, theories about how people make decisions in day-to-day contexts are often based on studies of decision- making with toy problems. A toy problem has a set of relevant cues that are well-understood by the researchers. Toy problems also have a correct answer that is known to the researchers. For example, several researchers have developed theories to explain patterns of responding to the question: ‘Which city has a larger population—Frankfurt or Leipzig?’ Participants are asked this question for all possible pairings of 83 German cities and, for each city in the question set, the researchers know: the city’s population (the basis for the correct answer) where the city stands in terms of nine features (cues) that people might bring to mind as they make their judgement—the researchers know, for example, whether the city: o has hosted an Olympic Games o has a team in the national top-tier soccer league o has an international airport. The University of Adelaide Slide 29  Decision-making in toy problems Computational-level descriptive and mathematical models of decision-making in the German cities problem have sought to answer questions such as: Do people pay attention to all cues, or only the ones most predictive of the correct answer? In what order do people consider the cues? The University of Adelaide Slide 30 Decisions: Different types Significance Medical diagnoses (i.e., what is at stake) Everyday decisions Decisions in toy problems Most decisions Correct answers are known Slide 31  Four models of decision-making Researchers have used toy problems to develop models/theories of everyday decision-making. In lectures 1-4, you will be introduced to four theories of decision-making in toy problems: Prospect Theory, first described in a paper in 1979 Heuristics and Biases, first described in a paper in 1974 Fast-and-Frugal Heuristics, which gained prominence in the early 1990s Bayesian approaches, which rely on advanced computing methods available since the 2000s. These theories have built on each other. The Bayesian approach is slowly becoming dominant, but it will not be covered in much depth because understanding it requires high-level statistical and computer science skills. The University of Adelaide Slide 32  t u re 1 L ec Prospect Theory (1979) e3 tu re 4 tu r Lec Lec re 2 Fast and Lec tu Heuristics Bayesian models frugal and biases of cognition heuristics (1974) (1994) (1991) Prospect Theory  Overview Prospect Theory is a highly influential theory within decision-making research. Its precursors trace back to the 18th century when mathematician Daniel Bernoulli noted the curious fact that even though humans are supposedly financially rational… The University of Adelaide Slide 35  Overview (cont.) In effect, Bernoulli noticed that the 'utility' (or desirability) of money declines with the amount already possessed. Picking up on this point, Expected Utility Theory—a family of theories popular in the late 1940s—proposed that the utility of a sum of money can be different from its dollar value, without this meaning that people are irrational in the classical sense. As the upcoming slides will show, Prospect Theory, proposed by Tversky and Kahneman in the 1970s, expands on Expected Utility Theory. The University of Adelaide Slide 36  Expected utility (perceived usefulness) Expected Utility Theory (1940s) Own wealth Hardman, 2009  Expected value Loss amount Gain amount Reference: current psychological state Expected harm Hardman, 2009  Expected value Loss amount Gain amount Expected harm Hardman, 2009  Expected value Loss amount Gain amount Expected harm Hardman, 2009  Expected value Loss amount Gain amount Loss aversion Expected harm Hardman, 2009 Which would you prefer? A. A certain gain of $3,000 B. An 80% chance of gaining $4,000, otherwise nothing 42  Expected value Loss amount Gain amount Which would you prefer? A. A certain gain of $3,000 Loss aversion B. An 80% chance of gaining $4,000, otherwise nothing Expected harm Hardman, 2009 Image source: https://order.hpa.org.nz/products/what-s- your-look-don-t-make-it-dumbburn-poster- female The framing effect  The framing effect In addition to whatever you own, you have been given $1,000. Which would you prefer next? A. A 50% chance of gaining $1,000 B. A sure gain of $500  The framing effect In addition to whatever you own, In addition to whatever you own, you have been given $1,000. you have been given $2,000. Which would you prefer next? Which would you prefer? A. A 50% chance of gaining A. A 50% chance of losing $1,000 $1,000 B. A sure loss of $500 B. A sure gain of $500  The framing effect In addition to whatever you own, In addition to whatever you own, you have been given $1,000. you have been given $2,000. Which would you prefer next? Which would you prefer? A. A 50% chance of gaining A. A 50% chance of losing $1,000 $1,000 B. A sure loss of $500 B. A sure gain of $500  The framing effect Reverses for small probabilities  The framing effect Reverses for small probabilities Which would you prefer? Which would you prefer? A. A 1% chance of gaining $5,000 A. A 1% chance of losing $5,000 B. A sure gain of $5 B. A sure loss of $5 Risk seeking with a gain frame. Risk averse with a loss frame.  Recap Let’s recap the three basic tenets of Prospect Theory. Firstly, the theory predicts loss aversion reflected in a steeper expected value curve for losses. Secondly, it predicts framing effects reflected in a concave expected value curve for gains and a convex one for losses. A third and final proposal is that framing effects are reversed for small probabilities. The University of Adelaide Slide 52  Prospect Theory and rationality As discussed by Levy (1997, p. 101), Kahneman and Tversky stated the following about the extent to which Prospect Theory considers human behavior rational: The curves proposed by Prospect Theory follow from Expected Utility Theory, which claims to demonstrate that human decisions are rational in a classical sense Loss aversion reflects the fact that pain is more urgent to attend to than pleasure. Risk aversion reflects the fact that organisms settle into habits. The University of Adelaide Slide 53  Evidence of individual and situational differences (i.e., some deviations from the theory’s predictions) In any decision-making textbook or economics journal, you can find reviews of many studies that have observed survey answers consistent with the key proposals of Prospect Theory. However, even though Prospect Theory states that its proposed risk preference patterns are observable in all people, recent studies have demonstrated that risk preferences vary across individuals and situations. The University of Adelaide Slide 54 Individual differences in the framing effect Abedellaoui et al., 2008 Individual differences in the framing effect Abedellaoui et al., 2008  Individual differences in the framing effect In addition to whatever you own, you have been given $2,000. Which would you prefer? A. A 50% chance of losing $1,000 B. A sure loss of $500 Abedellaoui et al., 2008 Situational effects on loss aversion Walasek & Stewart, 2014 Situational effects on loss aversion Walasek & Stewart, 2014  Additional criticisms of Prospect Theory In the reading, Levy (1997) points out, in the ‘Theoretical Limitations’ section that: Prospect Theory was developed to account for experimental findings – not describe all of human decision-making; thus, the theory overall says little regarding whether human judgements in the two-choice tasks are rational Prospect Theory does not explain how people might arrive at the two choices they have in judgements of the kind Prospect Theory is concerned with Overall, as a computational model of an aspect of human decision-making (of choice under conditions of risk), Prospect Theory defines some processes (risk aversion, framing effect, etc.) but does not specify the purpose these processes serve in maintaining effective (i.e., rational) functioning in the environment. It also does not explain why, in some circumstances, people deviate from the proposed processes. The University of Adelaide Slide 60  More criticisms: The ‘endowment effect’ Finding across multiple experiments (Box 1 in reading): People who were given a mug and told ”you own this mug” were more reluctant to trade the mug for a bar of chocolate than people who were not told anything about being owners of the mug. The reverse was the case when participants were given a chocolate bar that they could exchange for a mug (i.e., participants were more reluctant to trade the chocolate bar). Initial explanation based on loss aversion: The state of owning something makes exchanging it feel more like a loss, and, according to Prospect Theory, people are averse to losses more than they value gains. One alternative explanation draws on the notion of ‘psychological ownership’. According to this explanation, merely owning a product results in a more positive perception of it because ownership makes the product part of a person’s self-concept, and: - self-concepts tend to be positive. - products tend to have more positive than negative features, and people recall information (including product features) better when those are relevant to their self-concept; meaning that owners can think of more positive features overall. Morewedge & Giblin, 2015 Supporting evidence for the alternative ‘psychological ownership’ explanation (Box 4 in reading) Morewedge & Giblin, 2015  Summary Prospect Theory is one model of decision-making covered in this module. There are three to go, and they will be covered in the next three lectures. Prospect Theory concerns decision-making in tasks with two choices. These are highly simplified toy problems in which the correct answers are known. This is generally not the case in real-world decision-making problems (whether they are high-stakes or low-stakes). The ‘German cities’ problem is another example of a toy problem, and we will return to it in later lecture. Prospect Theory extends on Expected Utility Theory, which sought to demonstrate human rationality. Prospect Theory says little about rationality, but rationality and bounded rationality are key concepts in the decision-making literature that will appear in later lectures. Beyond rationality, bounded rationality, and toy problems, the following key concepts were introduced for you to be aware of in future lectures and readings: information processing (of mental representations) models (at different levels of analysis – including the computational) There are multiple criticisms of Prospect Theory, and one concerns the fact that the ‘endowment effect’ typically attributed to loss aversion has multiple other evidence-based explanations. The University of Adelaide Slide 63 64  References Abdellaoui, M., Bleichrodt, H., & L’Haridon, O. (2008). A tractable method to measure utility and loss aversion under prospect theory. Journal of Risk and Uncertainty, 36, 245. Hardman, D. (2009). Judgment and Decision Making: Psychological Perspectives. John Wiley & Sons. Chapter 7: Decision-making under risk and uncertainty. Love, B. C. (2015). The algorithmic level is the bridge between computation and brain. Topics in Cognitive Science, 7, 230–242. Morewedge, C. K., & Giblin, C. E. (2015). Explanations of the endowment effect: An integrative review. Trends in Cognitive Sciences, 19(6), 339–348. Plous, S. (1993). The Psychology of Judgment and Decision Making. McGraw Hill. Chapter 10: The representativeness heuristic. Simon, H. A. (1955). A behavioral model of rational choice.The Quarterly Journal of Economics, 69, 99– 118. Simon, H.A. (1990). Invariants of human behaviour. Annual Review of Psychology, 41, 1–19. Walasek, L., & Stewart, N. (2015). How to make loss aversion disappear and reverse: tests of the decision by sampling origin of loss aversion. Journal of Experimental Psychology: General, 144, 7–11. The University of Adelaide Slide 65