The Quantum Decision Theory, developed recently by the authors, is applied to clarify the role of risk and uncertainty in decision making and in particular in relation to the phenomenon of dynamic inconsistency. By formulating this notion in precise mathematical terms, we distinguish three types of inconsistency: time inconsistency, planning paradox, and inconsistency occurring in some discounting effects. While time inconsistency is well accounted for in classical decision theory, the planning paradox is in contradiction with classical utility theory. It finds a natural explanation in the frame of the Quantum Decision Theory. Different types of discounting effects are analyzed and shown to enjoy a straightforward explanation within the suggested theory. We also introduce a general methodology based on self-similar approximation theory for deriving the evolution equations for the probabilities of future prospects. This provides a novel classification of possible discount factors, which include the previously known cases (exponential or hyperbolic discounting), but also predicts a novel class of discount factors that decay to a strictly positive constant for very large future time horizons. This class may be useful to deal with very long-term discounting situations associated with intergenerational public policy choices, encompassing issues such as global warming and nuclear waste disposal.
Deep Dive into Physics of risk and uncertainty in quantum decision making.
The Quantum Decision Theory, developed recently by the authors, is applied to clarify the role of risk and uncertainty in decision making and in particular in relation to the phenomenon of dynamic inconsistency. By formulating this notion in precise mathematical terms, we distinguish three types of inconsistency: time inconsistency, planning paradox, and inconsistency occurring in some discounting effects. While time inconsistency is well accounted for in classical decision theory, the planning paradox is in contradiction with classical utility theory. It finds a natural explanation in the frame of the Quantum Decision Theory. Different types of discounting effects are analyzed and shown to enjoy a straightforward explanation within the suggested theory. We also introduce a general methodology based on self-similar approximation theory for deriving the evolution equations for the probabilities of future prospects. This provides a novel classification of possible discount factors, which
The concept of risk is widely used in economics, finance, psychology, as well as in everyday life. Respectively, there exist several definitions of risk and different ways of evaluating it. In any application, the notion of risk is always related to the necessity of taking decisions under uncertainty. It is impossible to achieve optimal results in any science without correct decisions, leading to optimal consequences following from the taken decision. This is why the notion of risk and the problem of its evaluation has, first of all, to be understood in the frame of decision theory. It is precisely the aim of the present paper to formulate a novel approach for taking into account the risk in decision making and to demonstrate in concrete examples, related to temporal effects in making decisions, that this new approach is free of defects and paradoxes plaguing the application of standard decision theory.
Classical decision theory is based on expected utility theory, which was advanced by Bernoulli [1] and was shaped into a rigorous mathematical theory by von Neumann and Morgenstern [2]. In this theory, a decision maker chooses between several lotteries, or gambles, each being composed of a set of outcomes, equipped with a probability measure. Initially [2], the probabilities were assumed to be objective. Savage [3] extended utility theory to the case of subjective probabilities. Savage’s generalization has been demonstrated to be tremendously flexible in representing the attitude of decision makers towards risk and uncertainty. Starting with Pratt [4] and Arrow [5], different measures of risk have been proposed. Extensions and modern developments are covered, e.g., in [6][7][8].
Notwithstanding a remarkable breadth of successful applications, classical decision theory, when applied to real humans, leads to a variety of paradoxes that remain unsolved in its framework. The first such anomaly was described by Allais [9], which is now known as the Allais paradox. Other well known paradoxes are Ellsberg’s paradox [10], Kahneman-Tversky’s paradox [11], the conjunction fallacy [12,13], the disjunction effect [14], and Rabin’s paradox [15]. These and other paradoxes are reviewed in Refs. [16,17].
There has been many attempts to modify expected utility theory in order to get rid of the paradoxes that plague its application to the processes involving decision making of real human beings. One of these approaches is the cumulative-prospect theory or reference-point theory [18], which assumes that decision making is not based on the absolute evaluation of payoffs but depends on a reference point that is specific to the present state of the decision maker. Because the reference point is shifted as a result of the consequences emerging from a first decision, the subsequent decision performed, according to the reference-point theory, is therefore sensitive to the difference between subsequent payoffs rather than to the absolute payoff deriving solely from the second decision.
One of the main problems encountered when using reference-point theory is that the reference point of a decision maker is not uniquely defined: for a similar payoff history, each decision maker can possess (and actually does possess) his/her own specific reference point, which is generally unobservable. Moreover, reference-point theory is more suited to address those anomalies that arise in gambles involving at least two-steps, in which the reference point can be expected to be shifted after each outcome. But, the majority of paradoxes appear in single-step gambles, where reference-point theory is not applicable. In the hope of explaining the paradoxes mentioned above, many other variants of the so-called non-expected utility theories have been suggested. A review of a variety of such non-expected utility theories can be found in Machina [19][20][21]. A rigorous analysis of these theories has been recently performed by Safra and Segal [22], who concluded that the non-expected utility theories cannot explain all paradoxes. Though it is possible to invent a modification of utility theory that will fit one or a few paradoxes, the problem is that many others will remain unexplained at best, or new inconsistencies will arise at worst.
The basic difficulty in taking into account and evaluating risk, when deciding under uncertainty, is that the usual approaches assume that decision makers are rational. However, real human beings are only partially rational [23], as is well documented by numerous empirical data in behavioral economics and neuroeconomics [24][25][26][27]. Risk is always related to emotions. But how could one describe emotions within a quantitative framework suitable for decision making?
A new approach to decision making, called Quantum Decision Theory (QDT), has been advanced in Refs. [16,17,28]. The main idea of this approach is to take into account that realistic decision-making problems are composite, consisting of several parts intima
…(Full text truncated)…
This content is AI-processed based on ArXiv data.
