Seeing all Sides
An autonomous debating system
Noam Slonim et al.
Nature, 18 March 2021, Pages 379–384
Abstract:
Artificial intelligence (AI) is defined as the ability of machines to perform tasks that are usually associated with intelligent beings. Argument and debate are fundamental capabilities of human intelligence, essential for a wide range of human activities, and common to all human societies. The development of computational argumentation technologies is therefore an important emerging discipline in AI research1. Here we present Project Debater, an autonomous debating system that can engage in a competitive debate with humans. We provide a complete description of the system’s architecture, a thorough and systematic evaluation of its operation across a wide range of debate topics, and a detailed account of the system’s performance in its public debut against three expert human debaters. We also highlight the fundamental differences between debating with humans as opposed to challenging humans in game competitions, the latter being the focus of classical ‘grand challenges’ pursued by the AI research community over the past few decades. We suggest that such challenges lie in the ‘comfort zone’ of AI, whereas debating with humans lies in a different territory, in which humans still prevail, and for which novel paradigms are required to make substantial progress.
Interpolating causal mechanisms: The paradox of knowing more
Simon Stephan et al.
Journal of Experimental Psychology: General, forthcoming
Abstract:
Causal knowledge is not static; it is constantly modified based on new evidence. The present set of seven experiments explores 1 important case of causal belief revision that has been neglected in research so far: causal interpolations. A simple prototypic case of an interpolation is a situation in which we initially have knowledge about a causal relation or a positive covariation between 2 variables but later become interested in the mechanism linking these 2 variables. Our key finding is that the interpolation of mechanism variables tends to be misrepresented, which leads to the paradox of knowing more: The more people know about a mechanism, the weaker they tend to find the probabilistic relation between the 2 variables (i.e., weakening effect). Indeed, in all our experiments we found that, despite identical learning data about 2 variables, the probability linking the 2 variables was judged higher when follow-up research showed that the 2 variables were assumed to be directly causally linked (i.e., C→E) than when participants were instructed that the causal relation is in fact mediated by a variable representing a component of the mechanism (M; i.e., C→M→E). Our explanation of the weakening effect is that people often confuse discoveries of preexisting but unknown mechanisms with situations in which new variables are being added to a previously simpler causal model, thus violating causal stability assumptions in natural kind domains. The experiments test several implications of this hypothesis.
Harder Than You Think: How Outside Assistance Leads to Overconfidence
Matthew Fisher & Daniel Oppenheimer
Psychological Science, forthcoming
Abstract:
Cognitive ability consists not only of one’s internal competence but also of the augmentation offered by the outside world. How much of our cognitive success is due to our own abilities, and how much is due to external support? Can we accurately draw that distinction? Here, we explored when and why people are unaware of their reliance on outside assistance. Across eight experiments (N = 2,440 participants recruited from Amazon Mechanical Turk), people showed improved metacognitive calibration when assistance occurred after a delay or required active choice. Furthermore, these findings apply across a wide range of cognitive tasks, including semantic memory (Experiments 1a and 1b), episodic memory (Experiments 2a and 2b), and problem solving (Experiments 3a–3d). These experiments offer important insights into how we understand our own abilities when we rely on outside help.
Thinking beyond boundaries: A growth theory of interest enhances integrative thinking that bridges the arts and sciences
Paul O'Keefe et al.
Organizational Behavior and Human Decision Processes, January 2021, Pages 95-108
Abstract:
Innovations often arise when people bridge seemingly disparate areas of knowledge, such as the arts and sciences. What leads people to make connections that others might miss? We examined the role of implicit theories of interest -- the belief that interests are relatively fixed (a fixed theory of interest) or developed (a growth theory of interest) among people with established interests either in the area of arts or sciences. A stronger growth theory predicted that participants spontaneously noticed more stimuli from the area outside their interests (Studies 2 and 3) and generated better integrative ideas (Study 1). Furthermore, they were more likely to generate ideas that bridged the arts and sciences (Study 2), which was also found after inducing fixed or growth theories, establishing causality (Study 3). Finally, perceived utility of the outside area mediated this relation (Study 4). These results suggest that a growth theory may be important for integrative thinking and innovation across traditional disciplinary boundaries.
Do Animated Line Graphs Increase Risk Inferences?
Junghan Kim & Arun Lakshmanan
Journal of Marketing Research, forthcoming
Abstract:
This article shows that animated display of time-varying data (e.g., stock or commodity prices) enhances risk judgments. We outline a process whereby animated display enhances the visual salience of transitions in a trajectory (i.e., successive changes in data values), which leads to transitions being utilized more to form cognitive inferences about risk. In turn, this leads to inflated risk judgments. The studies reported in this article provide converging evidence via eye-tracking (Study 1), serial mediation analyses (Studies 2 and 3), and experimental manipulations of the process factors: transition salience (graph type; Study 3) and utilization of transitions (global trend; Study 4 and investment goals; Study 5) and in the process, outline boundary conditions. We also demonstrate the effect of animated display upon consequential investment decisions and behavior. This paper adds to the literature on salience effects by disambiguating the role of inference-making in how salience of stimuli causes biases in judgments. Broader implications for visual information processing, data visualization, financial decision-making, and public policy are also discussed.
Interpersonal closeness impairs decision memory
Pinar Uğurlar, Ann-Christin Posten & Michael Zürn
Social Psychology, March 2021, Pages 125-129
Abstract:
We hypothesized that self-other confusion as a result of interpersonal closeness impairs people’s memory of their own decisions. Four studies (min N = 352) tested whether closeness affects memory in cooperative decisions. Participants played trust games in which they entrusted resources to another person and then had to recall their own decisions. Study 1 showed that people with an independent self-construal recalled their decisions more accurately, suggesting that less self-other overlap results in higher accuracy. Studies 2–4 showed that people made more recall errors when they played the trust game with a close in comparison with a distant partner. The findings suggest that interpersonal closeness impairs people’s memory of cooperative decisions.
Deliberation, Single-Peakedness, and Coherent Aggregation
Soroush Rafiee Rad & Olivier Roy
American Political Science Review, forthcoming
Abstract:
Rational deliberation helps to avoid cyclic or intransitive group preferences by fostering meta-agreements, which in turn ensures single-peaked profiles. This is the received view, but this paper argues that it should be qualified. On one hand we provide evidence from computational simulations that rational deliberation tends to increase proximity to so-called single-plateaued preferences. This evidence is important to the extent that, as we argue, the idea that rational deliberation fosters the creation of meta-agreement and, in turn, single-peaked profiles does not carry over to single-plateaued ones, and the latter but not the former makes coherent aggregation possible when the participants are allowed to express indifference between options. On the other hand, however, our computational results show, against the received view, that when the participants are strongly biased towards their own opinions, rational deliberation tends to create irrational group preferences, instead of eliminating them. These results are independent of whether the participants reach meta-agreements in the process, and as such they highlight the importance of rational preference change and biases towards one’s own opinion in understanding the effects of rational deliberation.
Man-Bites-Dog Contagion: Disproportionate Diffusion of Information about Rare Categories of Events
Alice Jayoung Jang & Jesse Shore
Boston University Working Paper, January 2021
Abstract:
How do social networks affect the diffusion of information? While previous research has mainly focused on the spread of specific messages, we study how the overall mix of information that diffuses through multiple intermediaries can become distorted or biased based on what categories of information people pass on versus filter out. We conducted randomized online laboratory experiments of diffusion through multi-step social networks. We find support for our pre-registered hypotheses that (1) the further someone is down a diffusion chain, the more the mix of information that they receive is biased toward rare categories of events because (2) information about rare categories of events is passed on disproportionately frequently. Our data is consistent with a preference for variety in what is shared, as well as a perceptual bias in favor of rare events. We name the disproportionate diffusion of rare categories of events "Man-Bites-Dog Contagion." Even when people intend to be accurate and informative, multi-step diffusion risks de-emphasizing the importance of empirically common categories of events and over-emphasizing the importance of empirically rare categories of events.