Session 4: Thursday 11 February, 1pm-2pm
Most studies of ambiguity aversion — the preference for ‘risk’ (a mathematical specification of probability) over ‘ambiguity’ (immeasurable lack of certainty) when holding utility constant — rely on experimental paradigms involving contrived monetary bets. Thus, the extent to which ambiguity aversion is evident outside of such contexts is largely unknown, particularly in those contexts which cannot easily be reduced to monetary terms. The present work seeks to understand whether ambiguity aversion occurs in a variety of different qualitative domains, such as work, family, love, friendship, exercise, study and health. We presented participants with 24 qualitative vignettes and measured the degree to which participants preferred risk to ambiguity, as well as asking for prior probability estimates for the ambiguous events presented. Ambiguity aversion was observed in response to the vast majority of vignettes, but at different magnitudes. These magnitudinal differences among the vignettes were not predicted by participants’ prior probability estimates of the ambiguous events presented, except in rare circumstances. Our results suggest that ambiguity aversion occurs in a wide variety of qualitative contexts, but to different degrees, and is not generally driven by unfavourable prior probability estimates of ambiguous events.
Prof. Ben Newell
Most theoretical accounts of non-instrumental information seeking suggest that the nature of future rewards has a direct impact on the attractiveness of the information in two specific ways. First, positively valenced outcomes (e.g., monetary gains) are predicted to result in preference for information, while negatively valenced outcomes (e.g., monetary losses) are predicted to result in information avoidance. Second, the magnitude of rewards is assumed to be proportional to the strength of information seeking (or avoidant) behaviour. In a series of experiments using both primary and secondary reinforcers, we explore the extent to which observed information seeking behaviour tracks these predictions. Our findings indicate a robust independence of information seeking from outcome valence and magnitude with preferences for information largely remaining constant across different reward valence and magnitudes. We discuss these results in the context of current computational models with suggestions for future theoretical and empirical work.
Dr. Timothy Ballard
Many decisions we face daily entail deliberation about how to coordinate resources shared between multiple, competing goals. When time permits, people appear to approach multiple-goal pursuit problems rationally, integrating information analytically to arrive at a prioritisation decision. However, it is not yet clear if this normative strategy extends to situations characterised by limited deliberation time. We evaluated the question of how limited deliberation time affects goal prioritisation decisions using a gamified experimental task, which required participants to make a series of interdependent goal prioritisation decisions. We fit several candidate models to experimental data in order to identify decision strategy adaptations at the individual subject-level. Results indicated that participants tended to opt for a simple heuristic strategy that was reliant on goal deadlines when deliberation time was low. This suggests that deadlines became particularly salient for most participants at the expense of other relevant information when deliberation time was limited.
Hypothetical choice scenarios provide insight into a consumer’s decision-making process when considering products or services. Perceptual decision-making studies have shown how important decision time and the impact that time pressure has on choice processes. Yet the impact of time on choice is rarely explicitly tested in the consumer choice literature. To address this, we combined methods from experimental psychology and consumer choice research, implementing a speed-accuracy tradeoff (SAT) manipulation in a preferential choice scenario. We extend the use of the response-signal (RS) task to a multi-dimensional stimulus in the form of a discrete choice experiment (DCE) for consumer-like products. This work is a modified version of our first RS experiment, with the addition of a stimulus mask. We used SAT functions to understand preferential choices by examining when different sources of information are 'cognitively online' and influence choices. We use these results to infer which product attributes influence the choices people make, and when.
According to the derived attention theory, organisms attend to cues with strong associations (Le Pelley, Mitchell, Beesley, George, & Wills, 2016). Combined with a Rescorla-Wagner style learning mechanism, derived attention explains phenomena such as learned predictiveness (Lochmann & Wills, 2003), inattention to blocked cues (Beesley & Le Pelley, 2011) and value-based salience (Le Pelley, Mitchell, & Johnson, 2013). However, existing derived attention models cannot explain the inverse base rate effect (Medin & Edelson, 1988) or retrospective revaluation (Shanks, 1985). We have developed a Bayesian derived attention model that explains a wider array of results and gives further insight into the principle of derived attention. Our approach is Bayesian linear regression combined with the assumption that the associations of any cue with various outcomes share the same prior variance. The new model simultaneously estimates cue-outcome associations and prior variance through approximate Bayesian learning. A significant cue will develop large associations, leading the model to estimate a high prior variance and hence develop larger associations from that cue to novel outcome: this provides a normative, statistical explanation for derived attention.Through simulation, we show that this Bayesian derived attention model not only explains the same phenomena as existing derived attention models, but also retrospective revaluation and the inverse base rate effect. We hope that further development of the Bayesian derived attention model will shed light on the complex relationship between uncertainty and predictiveness effects on attention (Pearce & Mackintosh, 2010).
Prof. Ramesh Srinivasan
Fitting drift-diffusion models (DDMs) to multiple participants’ choices and response times during perceptual decision making tasks result in parameter estimates that have cognitive interpretations such as individual differences in speed-accuracy tradeoffs and the average rates of evidence accumulation. The cognitive interpretations of DDM parameters can then be verified with experimental conditions and manipulations. Fitting neural drift-diffusion models (NDDMs) to participants’ scalp-recorded EEG as well as choices and response times can reveal additional individual differences in cognition. In particular it is thought that the collection of EEG data can reveal individual differences in visual attention, visual encoding time (VET), and evidence accumulation paths. We discuss our recent efforts to verify cognitive interpretations of EEG potentials and NDDM parameters in a preregistered study. In particular, we show evidence that individual differences in the onset of evidence accumulation can be measured, but show mixed evidence for understanding individual differences in evidence accumulation paths. Through this work we have discovered best practices for joint computational modeling of human EEG and behavior and make suggestions for the future of similar work.
Dr. David Sewell
Dr. Jamal Amani Rad
Levy Flights models have obtained good fits to experimental data. They differ from conventional evidence accumulation models in assuming a power-law distribution for accumulation noise, which causes some jumps during the information accumulation process. In this study, we plan to examine whether jumping during the accumulation process has real psychological meaning or only causes overfitting of the data. To this end, we have examined the effect of practice and feedback on within-trial variability, (i.e. the parameter that defines jump size) against the between trial variability parameters. Four versions of the Levy Flights model, with and without parameter variability, were fitted to behavioral data from a study by Evans and Brown (2017), and a five-layer deep inference neural network is utilized to fit the models. The results show the jump size has a systematic decreasing trend but the other variability parameters do not have any specific pattern during the experiment, speaking against the possibility of this parameter trading off with existing variability parameters. The results suggest that the Levy Flights model is not simply improving fits due to increased model complexity, and that further investigation of a potential psychological interpretation of jumps in evidence accumulation are warranted.
Dr. Scott Brown
How do the instructions of a random dot motion task affect the participants' response behaviour? How many thresholds are possible, and are they possible using any type of instruction? Participants were either given clear written instruction, totally ambiguous instruction using colour as a theme, or ambiguous-but-ordered instruction related to dots on a scale. We used five levels of speed-accuracy trade off; extreme speed, speed, neutral, accuracy, and extreme accuracy. Participant accuracy and response time formed two clusters, where extreme speed and speed were significantly different from neutral, accuracy, and extreme accuracy, but differences were small inside these groups. Early modelling suggests that participants adopted two response thresholds, one for the speeded conditions and the other for the neutral and accurate conditions. The way instructions were presented seemed to have little impact on accuracy, response time, or thresholds.