Paolo Avesani

Today at Rovereto takes place the third edition of the Workshop on Functional and Neural Perspectives of Concepts, Actions and Objects (CAOS).

In the afternoon Susanne Greiner, Emanuele Olivetti and Paolo Avesani will present a poster on “MVPA: Face Perception and Free Design”. The contribution is summarized by the following abstract:

The use of pattern recognition methods and machine learning techniques to discover distributed patterns of brain activity is a recent trend in fMRI data analysis. This approach, called Multi Voxel Pattern Analysis (MVPA) or, investigates the relation between the joint timecourses of multiple brain voxels with respect to the stimulation protocol, reversing the traditional analysis paradigm introduced by Statistical Parametric Maps (GLM/SPM). Often MVPA methods consist in predicting the stimulus (or action) the subject experiences. The prediction is based just on inspection of the distributed patterns of BOLD activations recorded by the MR scanner, concurrently. The aim is twofold: first to find evidence of the presence of information encoding the stimulus, second to localize the brain areas carrying that information (mapping). Cognitive neuroscience investigations may take advantage of MVPA techniques since they could be less restrictive with respect to the constraints imposed by block and event-related designs. The focus of our empirical investigation is the robustness of these methods of analysis once we shift from an event-related stimulation protocol towards a more free-design protocol (e.g. shorter ISI). In this work we illustrate an experiment designed for this purpose based on face perception/recognition and we provide preliminary results of the empirical investigation.

You may have a look at the poster as pdf file.