London Judgment & Decision Making Group Autumn term 2013 – 2014 Organizers Emmanouil Konstantinidis University College London Contact details: Department of Cognitive, Perceptual & Brain Sciences Room 204c, 26 Bedford Way, London, WC1H 0AP UK Telephone: (+44) 020 7679 5364 E-mail: emmanouil.konstantinidis.09@ucl.ac.uk Neil Bramley University College London Contact details: Department of Cognitive, Perceptual & Brain Sciences Room 201, 26 Bedford Way, London, WC1H 0AP UK E-mail: neil.bramley.10@ucl.ac.uk LJDM website http://www.ljdm.info Web administrator: Dr Stian Reimers (Stian.Reimers.1@city.ac.uk) LJDM members’ (Risk & Decision) list Contact: Dr Marianne Promberger (marianne.promberger@kcl.ac.uk) Seminar Schedule October – December 2013 5:00 pm in Room 313, 26 Bedford Way, UCL Psychology 2nd October Value-integration, intuitive computation and preference-reversal Marius Usher Tel Aviv University / University of Oxford 9th October Cognitive modelling of experience-based decision-making: To generalise or not? Emmanouil Konstantinidis University College London 16th October The dark side of white lies Lily Jampol Cornell University 23rd October Adaptive gain control during human perceptual choice Sam Cheadle University of Oxford 30th October Modelling Interventions in Decision Analysis: Normative Foundations and a New Probability Revision Method Shweta Agarwal The London School of Economics 06th November NO SEMINAR – UCL READING WEEK 13th November Cognitive Training with Real-Time Strategy Gaming Brian Glass Queen Mary University of London 20th November tbc Piotr Winkielman University of California, San Diego 27th November When and why do graphs enhance risk comprehension? Considering the impact of graph literacy Yasmina Okan University of Leeds 4th December tbc tbc tbc 11th December Using less than we experience: The interplay between experience and memory in choice and valuation Nathan Ashby & Tim Rakow University of Essex Abstracts 02.10.2013 Marius Usher Tel Aviv University / University of Oxford Value-integration, intuitive computation and preference-reversal Value integration is a central process, in various types of decisions, such as choosing a flatmate or a selecting a stock on the basis of previous returns. While the classical approach to this process involves an analytic mode of serial application of digital operations, research in numerical cognition have indicated an intuitive/parallel stream of numerical processing, based on analogue representations. Here I will present a number of recent studies we carried out showing that: i) there are conditions that make intuitive computations of averages both better and faster than rule based ones, ii) unlike analytic computations, the intuitive averaging can performed at high speed and improves with the length of a sequence, iii) this computation is nevertheless subject to attentional biases that can explain paradoxical preference reversal effects. The results will be discussed in relation to the 2 system theory of decision processes. 09.10.2013 Emmanouil Konstantinidis University College London Cognitive modelling of experience-based decision-making: To generalise or not? The study of experience-based decision-making has benefited from the application of cognitive models which decompose individuals’ choice strategies and performance into latent cognitive processes. Model predictions are dependent on each individual’s previous choice history as well as the payoffs received from each option. The empirical validation of candidate models is usually assessed based on two methods: the “one-step-ahead” prediction method that uses past choices and payoffs to predict future choices and the simulation method which does not rely on any past choice information and only uses past payoffs. However, these two methods yield inconsistent results regarding which model should be preferred. A potential explanation of this discrepancy refers to the reliance of each model on past choices and payoffs: because the simulation method takes account of past payoffs but not past choices, a model which relies more on past choices will perform worse under this method. We sought to better understand the discrepancy between the two methods by employing the equal payoff series extraction (EPSE) technique (Yechiam & Ert, 2007) in a classical experience-based paradigm, the Iowa Gambling Task (IGT). Based on EPSE we were able to quantify the degree to which each model relies on past choices and thus predict which model will provide the best fit under the simulation method. In addition, we propose and evaluate a new reinforcement-learning model for the IGT. 16.10.2013 Lily Jampol Cornell University The dark side of white lies? The subtle effect of biased performance feedback on inequality in the workplace. How we give and receive performance feedback is complexly determined by myriad factors such as environment, social relationships, and individual characteristics. Yet little research has explored the how inconsistency in communication about performance can be a subtle yet important factor in the maintenance of inequality in the workplace. This project aims to expose a covert and endemic bias in how we give feedback and why this bias may impact the advancement of disadvantaged groups. Specifically, across several studies, we find that implicit gender stereotypes (e.g., that women need protection) may make women the targets of white lies (more positive but less accurate feedback) during performance evaluation. This may especially be the case for participants who hold traditional views about women, and despite research showing that women are traditionally evaluated as less competent then men. Collectively, the results of our studies suggest that although seemingly an act of benevolence, differential telling of white lies may actually undermine progress if particular groups of people are not receiving accurate feedback needed to improve performance. 23.10.2013 Sam Cheadle University of Oxford Adaptive gain control during human perceptual choice Neural systems adapt to background levels of stimulation. Adaptive gain control has been extensively studied in sensory systems, but overlooked in decision-theoretic models. I will describe evidence for adaptive gain control during the serial integration of decision-relevant information. Human observers judged the average information provided by a rapid stream of visual events (samples). The impact that each sample wielded over choices depended on its consistency with the previous sample, with more consistent or expected samples carrying greater weight. This bias was also visible in the encoding of decision information in pupillometric signals, and in cortical responses measured with functional neuroimaging. These data can be accounted for with a new serial sampling model in which the gain of information processing adapts rapidly to reflect the average of the available evidence. 30.10.2013 Shweta Agarwal The London School of Economics Modelling Interventions in Decision Analysis: Normative Foundations and a New Probability Revision Method One of the key challenges in decision analysis is modelling decision makers’ beliefs about the world in a way which is both manageable in terms of judgmental burden and natural for the decision makers themselves. The problem context explored in this project is one where actions can affect the probability of one or more uncertain events (for instance advertisers designing marketing campaigns to improve the chance of success of a new product or operational risk mitigations). A convenient modelling tool of choice for such decisions is the influence diagram. We address the normative foundation of influence diagrams where probabilities of a chance variable depend on a decision and argue that a general foundation for influence diagram modelling is provided by Causal Decision Theory (CDT). Then we show that generalizations of existing probability revision methods to model interventions in decision analysis (generic controller of Matheson and Matheson), coincides with a class of linear probability revision rules—‘imaging’—and expose the theoretic advantages of the linearity property. We also propose a convenient method to operationalize the concept of linear probability revision rules and illustrate its application using a real world case study conducted for a leading insurance company. 06.11.2013 UCL READING WEEK – NO SEMINAR 13.11.2013 Brian Glass Queen Mary University of London Cognitive Training with Real-Time Strategy Gaming Training in action video games can increase the speed of perceptual processing. Additionally, we find that video-game training can lead to broad-based changes in higher-level competencies such as cognitive flexibility, a core and neurally distributed component of cognition. Two versions of a real-time strategy (RTS) game are compared to determine which video gaming characteristics can enhance cognitive flexibility and why these changes occur. Using a meta-analytic Bayes factor approach, we found that the gaming condition that emphasized maintenance and rapid switching between multiple information and action sources led to a large increase in cognitive flexibility as measured by a wide array of non-video gaming tasks. Theoretically, the results suggest that the distributed brain networks supporting cognitive flexibility can be tuned by engrossing video game experience that stresses maintenance and rapid manipulation of multiple information sources. Practically, these results suggest avenues for increasing cognitive function. 20.11.2013 Piotr Winkielman University of California San Diego tbc tbc 27.11.2013 Yasmina Okan University of Leeds When and why do graphs enhance risk comprehension? Considering the impact of graph literacy In order to make informed medical decisions people need to understand information about risks, benefits, and drawbacks of different treatments. Graphical displays—including bar charts, icon arrays or line plots—are powerful tools that can facilitate the communication and comprehension of such information. However, inadequately designed graphs may not only be unhelpful, but can also lead to errors in comprehension and mislead decision makers. Individuals with low graph literacy (i.e., the ability to understand graphically presented information) can be at a higher risk of misinterpreting the data depicted. I will present results of a series of experiments designed to (1) identify the errors that are most prominent among individuals with low graph literacy, (2) uncover the cognitive processes underlying such errors, and (3) test graphical design features (i.e., dynamic displays) that can reduce errors and foster informed medical decision making. Experiments included web questionnaires as well as laboratory-based studies involving process tracing methodologies (i.e., eye-tracking). Results revealed that graph literacy affected allocation of attention to regions of graphs containing essential information for accurate interpretations. I will discuss implications of our results for the design of graphical risk communications, including the customization of health-related decision-support systems. 04.12.2013 tbc tbc tbc tbc 11.12.2013 Nathan Ashby & Tim Rakow University of Essex Using less than we experience: The interplay between experience and memory in choice and valuation Recent research investigating decisions from experience suggests that not all information is treated equally with more recently encountered information being weighted heavier in the decision process (Hertwig, Barron, Weber, & Erev, 2004). We report three studies investigating how this differential treatment of information affects subjective valuations of and choices between risky prospects and what role individual differences in working memory play. In Study 1 we find that a model averaging only a subset of the most recently encountered outcomes fits the data best. In Study 2 and 3 we replicate and expand on this finding by showing that the amount of information used to form valuations and choices varies between individuals and that digit span explains a significant portion of this variation. Combined, these results indicate a direct link between cognitive capacity and information usage, providing further insight into the processes involved in the construction of value.