Probabilistic computation in the cortex of the developing human brain

Duration: 
August, 2016 to July, 2018
Funding: 
US National Institutes of Health (Rochester University subgrant)
Research Area: 

The PI and consultant have been collaborating for studies of visual statistical learning since 2001 and have maintained a close and productive professional relationship over the ensuing 14 years. A key feature of that collaboration is a class of sophisticated Bayesian models that have been shown to capture many of the key properties of human performances. These computational models involve a close synergy between the designs of empirical studies and the simulations of putative learning mechanisms. That is, they are not ``off the shell`` simulations but rather custom tailored interactions that require on-going and substantial collaboration. Jozsef and his colleagues at the central European University (Berkes, Orban, Lengyel, Fiser, 2011) have implemented some of the most sophisticated probabilistic models of cortical development in the ferret brain. These modelling and data analysis skills are essential to our analyses of fNIRS recordings obtained from the infant cortex in this project.