News

May 29th, 2012 Dr. Schiller gave a talk at Ulm University, Germany

"Developing Cognitive Models of Gambling and Problem Gambling Behaviour" (Marvin Schiller)

Gambling is an attractive laboratory for examining human information processing and decision making. Problem gambling, i.e, excessive gambling with harmful consequences, raises important questions about the underlying psychological mechanisms. Current research suggests that gambling is sustained by a complex interplay between different factors relating to the player and her environment. These factors include individual predispositions and motivations, the processing of cues from the gaming environment, reinforcement, the role of emotions, and biases and beliefs held by the player. In the presented work, (computational) cognitive models for gambling behaviour are developed, using the cognitive architecture CHREST. CHREST has previously been used to accurately model human processes of learning and expertise, and puts a particular emphasis on the modelling of perception and memory. In the current work, a mechanism for modelling affective memory and reinforcement learning - which are linked to the development and maintenance of (problematic) gambling behaviour - has been introduced to CHREST. Models for the Iowa Gambling Task, aspects of fruit machine gambling, and the acquisition of strategies in blackjack will be presented.
Mar 13th, 2012 Dr. Schiller gave a talk as part of the Brain Awareness Week" campaign at Brunel University

"Simulating the Gambler's Mind" (Marvin Schiller and Fernand Gobet)

The study of gambling behaviour is a fascinating topic for researchers in psychology, decision making, cognitive science and economics. An intriguing question is not only how people play, but also why they play, sometimes continuing beyond reasonable limits. Recent research provides evidence for mechanisms that are thought to underlie excessive gambling (also called problem gambling or pathological gambling). These mechanisms relate not only to the individual (e.g. impulsivity and cognitive biases and misconceptions) but also to the environment (e.g. misleading cues provided by fruit machines). In this work, a snapshot of the work at the Centre for the Study of Expertise at Brunel University is provided, which aims to deepen the understanding of gambling and problem gambling behaviour using computer models and simulations. The cognitive architecture CHREST is used to simulate processes of perception, memory, emotion and decision-making in the gambling context. This includes work on modelling an artificial gambling task (the Iowa Gambling Task), the effect of near wins on the persistence of fruit machine players, and the acquisition of strategies in blackjack.
Mar 5th, 2012 Dr. Schiller gave a talk at Birmingham University:

Cognitive Models of Gambling and Problem Gambling Behaviour: Current Developments (Marvin Schiller and Fernand Gobet)

Gambling activity provides a comprehensive and stimulating laboratory for studying human information processing and decision making. Excessive gambling with harmful consequences - so-called problem gambling - raises important questions about the underlying psychological mechanisms. Current research suggests a complex interplay between different factors relating to the player and her environment, including individual predispositions and motivations, the processing of cues from the gaming environment, reinforcement, the role of emotions, and biases and beliefs held by the player. In the presented work, (computational) cognitive models for gambling behaviour are developed. A particular focus is to study the role of affective memory and reinforcement learning on gambling behaviour, which are linked to the development and maintenance of (problematic) gambling behaviour. This work uses the cognitive architecture CHREST, which was developed as a computational model for human perception, attention, learning, memory, and problem solving. Models for the Iowa Gambling Task, aspects of fruit machine gambling, and the acquisition of strategies in blackjack will be presented.
Jan 18th, 2012 Dr. Schiller gave a talk in the Computer Science Research Colloquium at Hertfordshire University

"Developing Cognitive Models of Problem Gambling: Mechanisms for Memory, Reinforcement, and Emotions" (Marvin Schiller and Fernand Gobet)

Cognitive Modelling is a promising tool for investigating human information processing and decision making. In this talk I will present work on modelling gambling activity (and in particular problem gambling), since gambling provides a comprehensive and stimulating laboratory for the study of decision making. Problem gambling is defined as gambling for money with harmful consequences. Problem gambling research has identified various aspects of typical behaviour, pathways towards problem gambling, and particular risk factors. In this talk, I will argue that the development of a gambling problem may involve a multitude of various individual and external factors, and that their interactions and their underlying mechanisms are still poorly understood. The aim of the presented work is to investigate the behaviour under question by implementing and analysing it using the cognitive architecture CHREST. A particular focus of the current work is on affective memory and reinforcement learning, which is assumed to have a pivotal role in problem gambling. A mechanism to model this type of affective learning has recently been introduced to CHREST, and was used to model the Iowa Gambling Task, aspects of fruit machine gambling and blackjack.
Sep. 13th and 14th, 2011 The 2011 London Workshop on Problem Gambling and the CHREST Tutorial: An Introduction to Cognitive Modelling took place 13th and 14th September at Brunel University, London.
June 2011 Dr. Schiller presented at the 8th Nordic Conference on Prevalence, Prevention, Treatment and Responsible Gaming, June 13-15th 2011, Reykjavik, Iceland

"Development of Problem Gambling: Towards a cognitive Model" (Marvin Schiller and Fernand Gobet)

The rigorous study of the aetiology of problem gambling is key for understanding, preventing and treating the disorder. Recent theories suggest a complex interplay of factors in the development of gambling problems. These factors include individual predispositions and motivations, the processing of cues from the gaming environment, reinforcement, the role of emotions, and biases and beliefs held by the player. Computer modelling is an excellent tool for developing well-specified and testable theories of such complex interactions, but surprisingly, this has hardly been exploited in the case of problem gambling. Using the cognitive architecture CHREST, we develop models for investigating the interplay of cognitive and affective processing during play, drawing on recent theories of reinforcement, implicit learning and affective memory. The CHREST framework was developed as a computational model for human perception, learning, memory, and problem solving, with a particular focus on human perception and attention. It has proved to accurately model aspects of human cognition and expertise in various domains. In this talk, we present our first results pertaining to the modelling of slot machine play.
May 2011 Dr. Schiller presented a paper at EuroCogSci 2011, May 21st-24th 2011, Sofia, Bulgaria

"A Manifesto for Cognitive Models of Problem Gambling" (Fernand Gobet and Marvin Schiller)

Research on problem gambling has identified a wealth of phenomena related to the aetiology and maintenance of pathological gambling, its prevalence and risk factors, and has suggested programs and treatments. Remarkably, techniques of systematic formal modelling have rarely been used in the investigation of problem gambling despite their potential. We summarise and assess the current state of the field and highlight the opportunities and the potential impact of cognitive modelling.
April 2011 Dr. Schiller gave a presentation at the "International Conference on Gambling Studies" in Nottingham, UK, April 3rd-5th, 2011.

"The Making-Of the Problem Gambler: Towards a Cognitive Model" (Marvin Schiller and Fernand Gobet)

Abstract: Those forms of gambling that are most addictive feed on a dangerous amalgamation of contributing factors. As current research highlights, these pertain to the reinforcing mechanisms and cues, designed in purpose, of the gaming environment and the processing of this information by the player, which is modulated by specific motivations, predispositions, biases and rationalisations. In spite of the gambling industry's expertise in enhancing the attraction of their games, the understanding of the underlying psychological processes of gambling, and in particular the onset of problem gambling, requires further scientific investigation. While links with psychological theories of learning and affective processing have been proposed previously, these theoretical accounts ideally need to be refined into precise, testable models that reveal how excessive gambling develops from a complex interaction of processes. We propose an approach based on computational cognitive modelling using the CHREST framework. CHREST has previously been used to accurately model human processes of learning and expertise, and puts a particular emphasis on the modelling of perception and memory. Our modelling has initially focused on slot machine playing and aims to bring together results from a growing body of theory and empirical research, with a special focus on attention, (implicit) learning and emotion.

Click here to see a photo taken by the organiser