KogWis 2016 Tutorials

Tutorial T1: Detecting and Discouraging Non-Cooperative Behavior in Online Rating Tasks
Jana Häussler, University of Wuppertal; Tom Juzek, Nuance Communications

More and more researchers use crowdsourced rating tasks for data collection. In a rating task, participants are asked to evaluate some stimulus with respect to a given scale (e.g. they evaluate the similarity of two stimuli on a 7-point scale). However, previous studies demonstrated that crowdsourcing is quite susceptible to non­-cooperative behaviour (NCB), i.e. some participants are not complying with the task. Critically, NCB has a significant impact on the quality of the results that goes beyond mere noise.
This workshop presents response-time based strategies for detecting and discouraging NCB. In Session 1, we motivate their relevance, outline their functioning, and walk through the statistical part. We will show why a median-based criterion is more effective than a mean-based or absolute one and we will justify a response-time-based warning mechanism that discourages NCB effectively. Common platforms used for crowdsourced ratings tasks, e.g., Amazon Mechanical Turk or Prolific Academic, do not offer response times, let alone real-time access to them. Session 2 therefore provides the hands-on knowledge necessary for setting up an external rating website that allows the researcher to collect response times, to fully randomise items (with a Fisher-Yates shuffle), to implement the on-line/real-time warning mechanism, to intersperse booby trap items, and to collect personal data from the participants (using JavaScript and PHP; the code will be provided and explained, no prior technical knowledge is required).

Tutorial T2: Kant and Cognitive Science
Tobias Schlicht, Ruhr-Universität Bochum

Theoretical positions from historical figures in philosophy are not only interesting in their own right but can sometimes be especially helpful in teaching us systematic ways of inquiry that are ignored or simply unknown in contemporary debates. It has been claimed that many of Kant’s ideas make him the intellectual godfather of cognitive science (e.g. his distinction of percepts and concepts, his method of transcendental argument). In several recent publications, authors have suggested that various claims from Kant’s tentative Philosophy of Mind not only have counterparts in the contemporary cognitive science of the mind but can guide cognitive science in its quest to discover the function and nature of consciousness, perception and other phenomena.
This tutorial has two purposes: First, to (a) outline central claims of Kant’s philosophy of Mind. This is no easy task since Kant has not fully developed a full-fledged theory of consciousness or mental phenomena; rather, everything he has to say about the structure and function of mental phenomena is in the service of his epistemological project of developing a theory of knowledge. The second purpose is to (b) situate Kant’s claims in contemporary debates on consciousness, (c) to evaluate which of his claims are still of use for a thoroughly naturalist approach to the mind and, more specifically (d) to evaluate whether recent claims that recent developments in cognitive neuroscience suggest a “Kantian brain” are justified.

Tutorial T3: Workshop on Creativity
Bipin Indurkhya, Jagiellonian University, Krakow, Poland 

Human creativity has always fascinated psychologists and cognitive scientists. In the last fifty years or so, many cognitive aspects of creativity have been studied, and based on them many techniques for stimulating creativity have been developed. In this workshop, you will participate in a creativity-stimulating exercise that is based on one such technique. There are no prerequisites for participating, except to bring a fresh and open mind. 

This workshop is related to my talk in the KogWis 2016 symposium PROSOCRATES: Problem Solving, Creativity and Spatial Reasoning in Cognitive Systems.

Tutorial T4: Introduction to Cognitive Modeling with ACT-R
Nele Rußwinkel, Sabine Prezenski, Marc Halbrügge, Stefan Lindner Technische Universität Berlin 

ACT-R is the implementation of a unified theory of human cognition. It has a very active and diverse community that uses the architecture to model laboratory tasks as well as applied scenarios. The structure of ACT-R is oriented on the organization of the brain. This cognitive architectures states to be hybrid since it holds symbolic and subsymbolic components. The aim of working on cognitive models with a cognitive architecture is to understand how bottlenecks and errors occur in human behaviour occur. 

In this tutorial the cognitive architecture ACT-R is introduced (J. R. Anderson, 2007: How can the human mind occur in the physical universe? New York: Oxford University Press). The focus of the tutorial is on the symbolic parts. In the beginning a short overview about recent work and ACT-R's benefit for applied cognitive science is given. Then a short introduction of the background, structure and scope of ACT-R is provided. Two hands-on examples of how to write ACT-R models are the core part of the tutorial. The first short example introduces important mechanisms of ACT-R (productions and chunks). This is followed by an in-depth introduction on mechanisms such as visual and manual processing. For the second example, the participants work on their own model version of a letter-selection task. Assistance and advice will be given during the exercises. Different solutions for the second example will be discussed. In the end information on further mechanisms of ACT-R such as subsymbolic components for learning processes are given.
No prior experience or programming knowledge is required. Please bring a laptop and preferably download the ACT-R software (standalone version) prior to the event.

Tutorial T5: Bayesian Data Analysis: Main Ideas, Practices, and Tools
Michael Franke, Fabian Dablander, University of Tübingen

Bayesian approaches to statistical inference are often portrayed as the new cool kid in town and heralded as superior to classical techniques. Naturally, the hype is also perceived critically. This course is meant to critically introduce the Bayesian approach in a nutshell. Participants who are as of yet unfamiliar with it will receive enough information to form an opinion and to know where to obtain more information that suits their needs. Those who are familiar with the main ideas can benefit from a concise rundown of the most important recent developments. In particular, this course will do two things: (i) on the conceptual level, we provide an overview of the main ideas, advantages, and challenges of Bayesian data analysis, in direct comparison to classical approaches; (ii) on a practical level, we give an executive summary of some of the most recent and convenient tools for hands-on Bayesian data analysis.