2pm, Tuesday, 13 Dec 2005<BR>
    LCV meeting room, Meyer Hall, rm 1024<BR><BR>

<CENTER><B>
A Bayesian View of Sensory Conflicts in Decision-Making
</B>

<P>
Angela Yu, with Peter Dayan and Jonathan Cohen
<BR>
Princeton University

</CENTER>

      In this work, we present a formal analysis of the computations  
underlying decision-making tasks in which the sensory inputs are in  
conflict.  These tasks have been used to study a wide variety of  
related cognitive functions, such as decision making, error  
detection, conflict monitoring, automaticity, selective attention,  
etc.  In particular, we concentrate on the Eriksen task, a classical  
paradigm for studying the integration of conflicting sensory  
information, and the role of selective attention in controlling this  
integration. We propose two distinct Bayesian models that explain  
existent behavioral data, one focusing on stimulus compatibility and  
the other on spatial uncertainty.  We will also discuss possible  
extensions to the classical Eriksen Task, for which the two models  
make distinct predictions that can be experimentally verified.