Dr. Steven Glautier
Dr. Hedwig Eisenbarth
Dr. Anne Macaskill
A preference reversal is observed when a preference for a larger-later reward over a smaller-sooner reward reverses as both rewards come closer in time. Preference reversals are common in everyday life and in the laboratory, and are often claimed to support hyperbolic delay-discounting models which, in their simplest form, can model reversals with only one free parameter. However, it is not clear if the temporal location of preference reversals can be predicted a priori. Studies testing model predictions have not found support for them but they overlooked the well-documented effect of reinforcer magnitude on discounting rate. Therefore we directly tested hyperbolic and exponential model predictions in a pre-registered study by assessing individual discount rates for two reinforcer magnitudes. We then made individualised predictions about pairs of choices between which preference reversal should occur. With 107 participants we found 1) little evidence that hyperbolic and exponential models could predict the temporal location of preference reversals, 2) some evidence that hyperbolic models had better predictive performance than exponential models, and 3) in contrast to many previous studies, that exponential models generally produced superior fits to the observed data than hyperbolic models.
Throughout the day, many of our choices integrate information from multiple attributes about an item we are considering. How do people process information about multiple attributes and choose whether to select a presented option? In the simplest scenario, for one option and two attributes, the decision to either accept or reject the option is based on combinations of the two attributes.
Our model represents the evidence from each attribute towards accepting or rejecting the option as an accumulation process. We can model how the participant could combine this information into the final choice as combinations of these racing accumulators. For example, people may reject an option based on a single poor attribute but only accept the option if both attributes are highly valued. We constructed five different processing architectures and integrated them into a latent mixture modelling process to select between them. We use a hierarchical Bayesian approach to estimate individual participant processing architectures and overall group trends.
I will show an initial assessment of our modelling framework using data simulated from the five processing architectures. I will also discuss an experimental task where participants viewed a series of hotel options that differ on two attributes - price and hotel rating. In this task, participants received instructions on how to combine the attribute information for their decisions. The modelling framework recovered the expected processing architectures for the different instruction manipulations, demonstrating good selective influence. Understanding consumer attribute processing helps us present information in such a way as to keep consumers as informed as possible about the consequences of their choices.
Prior experience with options can impact our preferences and future choices. When strong preferences exist, context effects are hypothesized to diminish (Huber et al., 2014). In this study, we probed the effect of prior experience with options on the strength of attraction and compromise effects. Participants had the opportunity to choose from simple cognitive tasks (i.e., counting jobs) and complete the selected task. Using an ecologically valid and incentivized task, we found evidence for the formation of strong preferences with experience, yet reversals of context effects did not attenuate. The results were replicated in a pre-registered study and showed that our findings are robust to payment schemes and display format. These findings suggest that relative evaluation still plays a role in human decision-making, even when inherent preferences are accessible. We suspect what was learned from experience in our tasks is the weights for various attributes. As predicted by many models of multi-alternative, multi-attribute choice, context effects can emerge with unequal attribute weights formed through, for instance, prior experience with options.
Ms. Li Xin Lim
Ms. Madison Fansher
Performing an action often incurs a cost, such as exerting effort for a reward. Previous studies used the Effort Expenditure for Reward Task (EEfRT) to show devaluation of reward value with physical effort. However, it is unclear if a similarly structured attentional task would produce a similar reward devaluation with cognitive effort. In the present work, we propose a new task called the “shell game task” (SGT) as a cognitive effort-based decision-making paradigm. Participants performed both the EEfRT and SGT in a within-subject design. Using computational models of choice behavior, we showed that effort cost induced by the variability of task demands in the SGT is similar to the effort cost from the existing EEfRT in the devaluation of a given outcome in action choice selection. This result suggests that effort cost may be a stable idiosyncratic trait across the two tasks and shows how computational approaches can be used to estimate and compare measures of effort. In addition, the results suggest that the SGT can be used as an alternative to the EEfRT with subject populations with motor deficits.
Prof. Christoph Klauer
People rely on the choice context to guide their decisions, violating fundamental principles of rational choice theory and exhibiting phenomena called context effects. Recent research has uncovered that dominance relationships can both increase or decrease the choice share of the dominating option, marking the two ends of an attraction–repulsion continuum. However, empirical links between the two opposing effects are scarce and theoretical accounts are missing altogether. The present study used eye tracking alongside a within-subject design that contrasts a perceptual task and a preferential-choice analog in order to bridge this gap and uncover the underlying information-search processes. Although individuals differed in their perceptual and preferential choices, they generally engaged in alternative-wise comparisons and a repulsion effect was present in both conditions that became weaker the more predominant the attribute-wise comparisons were. To obtain a model-based characterization of individuals' behavior in terms of latent cognitive processes, we relied on the MLBA, a prominent cognitive model that is frequently used to model multi-attribute, multi-alternative choices in both the preferential and perceptual domain. Despite its past successes, the MLBA was unable to provide an accurate account of the data. Specifically, it was unable to jointly account for choices and their associated latencies, struggling the most with predicting the focal choice phenomenon, the repulsion effect. Altogether, our study corroborates the notion that repulsion effects are a robust and general phenomenon that current theoretical accounts cannot adequately account for and that need to take be taken seriously.
This talk presents a novel set of empirical data that aims to measure the influence of distributional entropy on the weight given to a particular probability. The experimental protocol extends a gamble-matching paradigm developed by Chechile and Barch (2013), which enables the assessment of probability weights without needing to postulate a choice rule or a utility function. The data is used to compare the performance of several existing probability weighting functions (Tversky-Kahneman, 1992; Prelec, 1998; Goldstein-Einhorn, 1987; Chechile-Barch, 2013) relative to a novel, information-theory-based probability weighting function (Valence-Weighted-Distance, Akrenius, 2020) and it's extension with the Sharma-Mittal family of entropies (Akrenius, 2021). The results of the study shed light on the influence of distributional context on probability weighting, individual differences in perceptions of uncertainty, the existence of probability distortion as a perceptual phenomenon, and the potential of using information entropy as a psychologically grounded construct in models of judgment and decision making.