Ben R. Newell’s research while affiliated with University of South Wales and other places

The goal of the Paris Agreement is to keep global temperature rise well below 2°C. In this agreement—and its antecedents negotiated in Copenhagen and Cancun—the fear of crossing a dangerous climate threshold is supposed to serve as the catalyst for cooperation among countries. However, there are deep uncertainties about the location of the threshold for dangerous climate change, and recent evidence indicates this threshold uncertainty is a major impediment to collective action. Early warning signals of approaching climate thresholds are a potential remedy to this threshold uncertainty problem, and initial experimental evidence suggests such early detection systems may improve the prospects of cooperation. Here, we provide a direct experimental assessment of this early warning signal hypothesis. Using a catastrophe avoidance game, we show that large initial—and subsequently unreduced—threshold uncertainty undermines cooperation, consistent with earlier studies. An early warning signal that reduced uncertainty to within 10% (but not 30%) of the threshold value catalysed cooperation and reduced the probability of catastrophe occurring, albeit not reliably so. Our findings suggest early warning signals can trigger action to avoid a dangerous threshold, but additional mechanisms may be required to foster the cooperation needed to ensure the threshold is not breached.

The ability to discover patterns or rules from our experiences is critical to science, engineering, and art. In this article, we examine how much people’s discovery of patterns can be incentivized by financial rewards. In particular, we investigate a classic category learning task for which the effect of financial incentives is unknown (Shepard et al., 1961). Across five experiments, we find no effect of incentive on rule discovery performance. However, in a sixth experiment requiring category recognition but not learning, we find a large effect of incentives on response time and a small effect on task performance. Participants appear to apply more effort in valuable contexts, but the effort is disproportionate with the performance improvement. Taken together, the results suggest that performance in tasks that require novel inductive insights is relatively immune to financial incentives, while tasks that require rote perseverance of a fixed strategy are more malleable.

The naiveté of the dominant ‘cognitive-miser’ metaphor of human thinking hampers theoretical progress in understanding how and why subtle behavioural interventions—‘nudges’—could work. We propose a reconceptualization that places the balance in agency between, and the alignment of representations held by, people and choice architects as central to determining the prospect of observing behaviour change. We argue that two aspects of representational (mis)alignment are relevant: cognitive (how people construe the factual structure of a decision environment) and motivational (the importance of a choice to an individual). Nudging thinkers via the alignment of representations provides a framework that offers theoretical and practical advances and avoids disparaging people’s cognitive capacities.

An unexamined assumption in many studies of learning and decision making is that people learn underlying probability distributions. However, the acquisition of distributional knowledge is rarely the focus of investigations. We report five experiments (N = 580 adults) that provide this focus and highlight the factors that impact people’s ability to accurately learn and reproduce underlying distributions. We find that people accurately reproduced the distribution only when either the environmental signal is strong (e.g., discrete bimodal distributions) or sufficient cues are provided to aid construction of mental representations (e.g., items from the modes in a noisy bimodal distribution are presented in different colors). We interpret these results in terms of participants testing and learning discrete rules corresponding to salient features of the environment rather than spontaneously representing entire distributions. As such, the findings challenge strong assumptions about the role of probability distribution knowledge in explanations of learning and decision making.

Do people change their preferences when they are offered the same risky lotteries at different times (now vs. the future)? Construal level theory (CLT) suggests that people do because our mental representation of events is moderated by how near or distant such events are in time. According to CLT, in the domain of risk preferences, psychological distance causes payoffs and probabilities to be differentially weighted or attended between present and future timepoints: Temporal distance increases the influence of payoffs and decreases the influence of probabilities. Specifically, CLT predicts that high probability/low amount lotteries (i.e., %-lotteries) are preferred in the present, whereas low probability/high amount lotteries (i.e., $-lotteries) are preferred in the future, even when the expected value of these lotteries is identical. We present a functional characterization and systematic investigation of this putative pattern of risk preferences and develop a formal model that incorporates CLT’s predictions. In five experiments, we examined several factors that could moderate the effect (e.g., outcome and probability magnitude, lottery presentation format, incentivization procedures). Both our behavioral observations and modeling results suggest the effect is labile, and if it does occur, it is not fully consistent with our formal model of CLT.

Public trust in scientists may be essential for widespread acceptance of science-based solutions to societal problems, including climate change. Across 68 countries (N = 69,534), individuals expressed less trust in climate scientists than scientists in general. In most countries and overall, conservative political orientation was more strongly associated with lower trust in climate scientists than with trust in scientists in general.