Journal of Mathematical Psychology最新文献

We introduce a context-dependent theory for choice under risk, called range utility theory. It builds on Parducci’s range principle from psychophysics and modifies expected utility by positing that risky prospects are evaluated relative to the range of consequences of all prospects in the decision context. When the context is fixed, choices typically exhibit the four-fold pattern of risk preferences, yet are fully consistent with expected utility (linear in probabilities) without invoking rank-principles. We illustrate this advantage in game theory contexts. As the same time, when the context varies, the relative value of an alternative also does, yielding different forms or preference reversals, some of which have been robustly documented.

In this paper, we propose a dynamical model for the best–worst choice task. The proposed model is a modification of the multi-episode Cube model proposed and studied in so-called (Mallahi-Karai and Diederich, 2019, 2021). This model postulates that best–worst choice (or more generally, ranking) task is the outcome of sequential choices made in a number of episodes. The underlying model is a multivariate Wiener process with drift issued from a point in the unit cube, where episodes are defined in terms of a sequence of stopping times. This model can also be extended to an attention-switching framework in a standard way.

Psychological theories are often formulated at the level of latent, not directly observable, variables. Empirical measurement of latent variables ought to be valid. Classical psychometric validity indices can be difficult to apply in experimental contexts. A complementary validity index, termed retrodictive validity, is the correlation of theory-derived predicted scores with actually measured scores, in specifically designed calibration experiments. In the current note, I analyse how calibration experiments can be designed to maximise the information garnered and specifically, how to minimise the sample variance of retrodictive validity estimators. First, I harness asymptotic limits to analytically derive different distribution features that impact on estimator variance. Then, I numerically simulate various distributions with combinations of feature values. This allows deriving recommendations for the distribution of predicted values, and for resource investment, in calibration experiments. Finally, I highlight cases in which a misspecified theory is particularly problematic.

Forward-graded and backward-graded structures of knowledge are two important classes of knowledge structures. Spoto and Stefanutti (2020) establish necessary and sufficient conditions for skill maps to delineate these structures. We introduce fuzzy skills to describe varying levels of proficiency in skills and extend the theoretical results of Spoto and Stefanutti (2020) for delineating forward- and backward-graded knowledge structures using fuzzy skill maps. The paper establishes necessary and sufficient conditions for fuzzy skill maps to delineate a backward-graded simple closure space, a forward-graded knowledge space, and a forward-graded simple closure space. Furthermore, the competence-based local independence model (CBLIM) with fuzzy skills is introduced and its unidentifiability is discussed.

We introduce a new approach to modelling decision confidence, with the aim of enabling computationally cheap predictions while taking into account, and thereby exploiting, trial-by-trial variability in stochastically fluctuating stimuli. Using the framework of the drift diffusion model of decision making, along with time-dependent thresholds and the idea of a Bayesian confidence readout, we derive expressions for the probability distribution over confidence reports. In line with current models of confidence, the derivations allow for the accumulation of “pipeline” evidence that has been received but not processed by the time of response, the effect of drift rate variability, and metacognitive noise. The expressions are valid for stimuli that change over the course of a trial with normally-distributed fluctuations in the evidence they provide. A number of approximations are made to arrive at the final expressions, and we test all approximations via simulation. The derived expressions contain only a small number of standard functions, and require evaluating only once per trial, making trial-by-trial modelling of confidence data in stochastically fluctuating stimuli tasks more feasible. We conclude by using the expressions to gain insight into the confidence of optimal observers, and empirically observed patterns.

This paper demonstrates how averaging over individual learning and forgetting curves gives rise to transformed averaged curves. In an earlier paper (Murre and Chessa, 2011), we already showed that averaging over exponential functions tends to give a power function. The present paper expands on the analyses with exponential functions. Also, it is shown that averaging over power functions tends to give a log power function. Moreover, a general proof is given how averaging over logarithmic functions retains that shape in a specific manner. The analyses assume that the learning rate has a specific statistical distribution, such as a beta, gamma, uniform, or half-normal distribution. Shifting these distributions to the right, so that there are no low learning rates (censoring), is analyzed as well and some general results are given. Finally, geometric averaging is analyzed, and its limits are discussed in remedying averaging artefacts.

We study a sufficiently general regret criterion for choosing between two probabilistic lotteries. For independent lotteries, the criterion is consistent with stochastic dominance and can be made transitive by a unique choice of the regret function. Together with additional (and intuitively meaningful) super-additivity property, the regret criterion resolves the Allais’ paradox including the cases were the paradox disappears, and the choices agree with the expected utility. This super-additivity property is also employed for establishing consistency between regret and stochastic dominance for dependent lotteries. Furthermore, we demonstrate how the regret criterion can be used in Savage’s omelet, a classical decision problem in which the lottery outcomes are not fully resolved. The expected utility cannot be used in such situations, as it discards important aspects of lotteries.

In many decision tasks, we have a set of alternative choices and are faced with the problem of how to use our latent beliefs and preferences about each alternative to make a single choice. Cognitive and decision models typically presume that beliefs and preferences are distilled to a scalar latent strength for each alternative, but it is also critical to model how people use these latent strengths to choose a single alternative. Most models follow one of two traditions to establish this link. Modern psychophysics and memory researchers make use of signal detection theory, assuming that latent strengths are perturbed by noise, and the highest resulting signal is selected. By contrast, many modern decision theoretic modeling and machine learning approaches use the softmax function (which is based on Luce’s choice axiom; Luce, 1959) to give some weight to non-maximal-strength alternatives. Despite the prominence of these two theories of choice, current approaches rarely address the connection between them, and the choice of one or the other appears more motivated by the tradition in the relevant literature than by theoretical or empirical reasons to prefer one theory to the other. The goal of the current work is to revisit this topic by elucidating which of these two models provides a better characterization of latent processes in $m$-alternative decision tasks, with a particular focus on memory tasks. In a set of visual memory experiments, we show that, within the same experimental design, the softmax parameter $\beta$ varies across $m$-alternatives, whereas the parameter $d^{\prime }$ of the signal-detection model is stable. Together, our findings indicate that replacing softmax with signal-detection link models would yield more generalizable predictions across changes in task structure. More ambitiously, the invariance of signal detection model parameters across different tasks suggests that the parametric assumptions of these models may be more than just a mathematical convenience, but reflect something real about human decision-making.

A better method named MAGL (Multi-Attribute Gain Loss) is proposed to predict choices made by consumers in a multi-attribute setting. The MAGL method uses the tenets of prospect theory, Kauffman’s complexity theory, norm theory, and context-dependent choice theory. Since the choice processes are often found to be affected by the context or the choice set, the proposed MAGL method is able to model and predict the context-dependent choice behavior of consumers. The predictions of the MAGL method are useful to marketing/product managers in designing new products. The output of the MAGL method can be analyzed to determine which combination of attribute values is outperforming in a specific competitive market condition. A decision support system can be designed and developed for marketing/product managers where they can experiment by introducing, redesigning, or removing products and simulate the market share of various products for a similar consumer population.

Stopping rules are criteria for determining when data collection can or should be terminated, allowing for inferences to be made. While traditionally discussed in the context of classical statistics, Bayesian statisticians have also begun exploring stopping rules. Kruschke proposed a Bayesian stopping rule utilizing the concept of Highest Density Interval, where data collection can cease once enough probability mass (or density) accumulates in a sufficiently small region of parameter space. This paper presents an alternative to Kruschke’s approach, introducing the novel concept of Relative Importance Interval and considering the distribution of probability mass within parameter space. Using computer simulations, we compare these proposals to each other and to the widely-used Bayes factor-based stopping method. Our results do not indicate a single superior proposal but instead suggest that different stopping rules may be appropriate under different circumstances.