Decision-making and the role of feedback in complex tasks

Taylor, Sheridan Anne (2018) Decision-making and the role of feedback in complex tasks. (PhD thesis), Kingston University, .

Abstract

This thesis investigates the process of decision-making in relation to complex tasks and considers the important role which dynamic information and real-time feedback play in shaping response behaviour and adaptation within such environments. Through empirical studies, the thesis explores the extent to which decision-makers can be said to act rationally when challenged by complex decision-making environments. Evidence relating to demand for information and the impact of feedback on behaviour is provided with two studies: The first uses a simulated auction platform to examine behaviour within overlapping auctions of short duration with close-to-identical items and minimal participation costs. Mouse tracking is used to capture data on relevant interactions of participants with the simulated online platform, including switching behaviour independent of bidding. The resulting data suggests that participants did behave in a manner consistent with utility maximisation, seeking to acquire the item at the lowest possible price and showing no bias in terms of auction preference. The impact of fixed-price offers in the form of a “Buy it Now” option is also examined with some evidence that participants again seek, and respond to, current information when deciding on their bidding strategy. The second study is a test of the impact of real-time feedback and demand for information within the context of financial markets. The study again uses a novel simulated environment which provides access to considerable amounts of relevant data which participants can choose to access. In addition, participants are exposed to regular feedback with regard to their own performance. Overall, demand for information is found to be dependent upon the type of feedback received and its context. Decision-makers then appear to behave objectively, apparently seeking the latest available information to support current decisions, although investor style is found to be important in determining overall trading propensity. The thesis starts by considering a number of the foundations and pathways which run through the judgment and decision-making literature. It is not a complete description, review or analysis of all of the prevailing lines of enquiry. Nevertheless, it seeks to achieve coherence in terms of bringing together some of the key themes dealing with risky choice under conditions of uncertainty and ambiguity. The field of judgment and decision-making is inevitably vast; its scope owing much to the fact that it transcends individual disciplines. The emergent behavioural sciences thus draw together important strands from various sources, notably Economics and Finance. In many areas, psychological traits can be applied to explain inconsistencies which are found in classical theory of rational behaviour. The recognition of behavioural traits has thus contributed greatly to the evolution of decisionmaking theories under conditions of uncertainty and ambiguity which are, in many cases, substantially more adaptable and robust than early normative theories of rational behaviour. The classical approach to rational decision making within Economics, together with some theoretical and empirical challenges to it, are considered in Chapter 1. It is here that we are introduced to the Rational Man. Like the mythical creatures found in Classical Antiquity, the Rational Man does not actually exist in the real world; he is nevertheless central to the concept of utility maximising rational choice which provided much of the foundation of Economics. Developments of expected utility theory (EUT) are considered, including its replacement of expected value, and the formalisation of rational behaviour within the context of axioms. When those logical axioms apply, decision-makers can be said to behave 'as if' they are utility maximises. The chapter ends with some empirical evidence, showing the types of approaches often used to explore rational decision-making. Some violations of EUT are explored, both in relation to notional gambles and consistency with regard to revealed preferences. Chapter 2 extends the narrative by considering rational decision-making in cases where there is no objective information about possible outcomes. Subjective utility theory (SEU) is then introduced, describing objective functions based upon preferences derived from combined utility and probability functions. The implications of the Allais’ and Ellsberg paradoxes are discussed, along with some possible solutions. It is here that we explore the concepts of uncertainty and ambiguity in more details and consider some theoretical formulations for addressing them. Chapter 3 covers the significant contribution to decision-making under conditions of uncertainty provided by Prospect Theory and, later, Cumulative Prospect Theory (CPT). Their evolution from the pioneering work of Markowitz is discussed within the context of reference points relative to which outcomes can be evaluated. The significance of stochastic dominance and rank dependence are explored. By this stage, we have examined numerous theories which have fundamentally transformed standard EUT into much more flexible and adaptable frameworks of rational choice. The core concepts of utility maximisation remain yet the initial, strictly concave utility function describing diminishing marginal utility is now substantially replaced by more complex weighted preference functions. From this theoretical base, the process of choice reduction and the application of heuristics in decision-making are considered. We again describe axiomatic behaviour compatible with rational choice. Therefore, decision-makers faced with multiple choices about which there may be little or no objective information about likely outcomes can nevertheless develop rational beliefs and expectations which can then be applied to reduce complex tasks to more manageable proportions. As well as considering these aspects from the point of view of actual choices, we also consider the processes by which decisions are taken. Thus, process tracing methods are introduced into the discussion. The chapter also explicitly considers the role of feedback in decision making. This includes a consideration of Bayesian inference as a process for updating probabilistic expectations subject to new information. From considering theoretical formulations form which we can judge rational behaviour, Chapter 5 looks at evidence for sub-optimal decision-making and bias. Bias with regard to probability assessments are considered along with empirical evidence of bias in relation to intertemporal discounting. Sunk cost bias is also considered as a clear example of irrational behaviour, leading in to a specific discussion about a number of persistent behavioural biases identified within financial markets. As an introduction to later chapters, this also covers the basic theoretical principles of market efficiency and evidence that real markets fail to adhere to those principles in important ways. Chapters 6 and 8 describe the empirical studies with Chapter 7 providing a more detailed introduction to the financial markets experiment, considering aspects of market efficiency, models of behaviour and other empirical evidence.

Actions (Repository Editors)

Item Control Page Item Control Page