Have you ever wondered how to fit a model to your behavioral data? The Amsterdam Center for Brain & Cognition (ABC) is organizing a 1½ day workshop with the aim to introduce cognitive mathematical models of decision making and their application. International and national speakers will discuss why the model-based approach is useful and attractive in both the cognitive and neuroscientific field. Topics will range from the theoretical background of several sequential sampling models to their application in neuroimaging studies. The course is particularly designed for PhD students, postdoctoral researchers, and anyone how takes an interest in getting more out of their behavioral data.
|Date||Start 27 May 2014||End 28 April 2014|
Prof. Roger Ratcliff, Ohio State University, Columbus, Ohio, USA
Dr. Leendert van Maanen, University of Amsterdam, Amsterdam, the Netherlands
Dr. Martijn Mulder, University of Amsterdam, Amsterdam, the Netherlands
Dr. Brandon Turner, Stanford University, Stanford, California, USA
(Leendert van Maanen / Martijn Mulder)
Many cognitive and neuroscientific studies involve tasks in which participants have to make responses to sensory stimuli in different experimental conditions. The effect of the experimental manipulation is often summarized by reporting the difference in the proportion correct responses (accuracy) and/or response times (RT) between conditions. Although, at a descriptive level, this might be useful, such a summary does not allow answering the question why the speed and accuracy of responses change with the task manipulation.
To answer this question, mathematical models have been developed that describe and predict the mechanisms that might drive changes in accuracy and RT. Typically, these models conceptualize the different response conditions as a perceptual decision, in which sensory information accumulates over time toward a decision threshold. Such a process of accumulating evidence is not only behaviorally intuitive but has proven to be neurobiologically plausible as well. Gradually, the model-based approach has been adopted by both the clinical and basic neuroscientific community, where it has shown to be useful as an intermediate level between the neurological and behavioral observations (see for a review Forstmann et al., 2011, TICS). On this first day of our workshop, we will introduce the model-based approach and show that such an approach is useful and fairly easy to apply to most perceptual/cognitive paradigms.
(Roger Ratcliff / Brandon Turner)
On the second day of our workshop, Prof. Roger Ratcliff will discuss the popular drift-diffusion model (DDM), also known as the Ratcliff diffusion model. The DDM lent its popularity by the fact that it fits extremely well to a large variety of perceptual and cognitive two-choice data (see for a review Ratcliff and McKoon, 2008, Neural Computation). In this lecture, Prof. Ratcliff will first discuss and illustrate the main features and advantages of the model. Next, a practical handson session will take place where workshop participants perform model-fits to RT and accuracy data (note, that familiarity with either R or Matlab might be advantageous, but is not a requirement).
Brandon Turner will discus future directions such as the “Joint modeling” framework. The "joint modeling" framework provides a principled way to constrain behavioral models in neurologically plausible ways, and allows for a straightforward interpretation of how the abstractions assumed by cognitive models are related to actual neural processes.
Tuesday 27th of May (3pm - 6pm)
Wednesday 28th of May (9.30am - 5pm)
Graduate and PhD students in the field of experimental/mathematical Psychology, Neuroscience, AI, Computer Science, Neuroeconomics or similar subject.
Some programming skills (R, Matlab) are advantageous.
Attendance is free but limited to 30 participants!
The workshop is full.