NeuroImage paper out
In this paper, we tested how well human groupings of naturalistic images was reflected in patterns of fMRI activity in visual cortex. Contrary to what is commonly assumed, there was not a clear-cut mapping of the brain responses in ventral-temporal cortex to behavior, and comparisons with deep network models suggested that activity in these regions reflected lower level processing than what guided behavior. This suggests there's still a lot we don't understand about the representations in higher-order visual regions, especially not for naturalistic images! Note that the stimuli and data are freely available so others can test more models of visual representation against these behavioral and fMRI data!
NYU FAS Travel Award
I received the NYU Faculty of Arts and Sciences Postdoctoral Travel Award to sponsor my attendance at OHBM this year! I will be presenting our ECoG and fMRI work on temporal dynamics in visual cortex, in both a poster and a (selected) talk session.
Talk in Japan
On February 21st 2019, I will be giving an invited talk at the 5th CiNet Conference themed ‘Computation and Representation in Brains and Machines’, Center for Information and Neural Networks (CiNet) in Osaka, Japan.
I'm very excited to be among this line-up of esteemed speakers! It promises to be a very interesting meeting.
New paper in PLoS Comp Biol
It took a while, but I finally published the last chapter of my PhD thesis! The wonderful Sara Jahfari and I led this project together, looking at the influence of scene complexity (as determined by low-level image statistics) on detection of animal in natural scenes. It turns out that scene complexity affects the degree of evidence accumulation (as formalized in a drift diffusion) which for highly complex scenes appears to be driven by feedback activity.
New paper in eLife!
In this paper, we tested how well scene categorization behavior and patterns of fMRI activity in scene-selective cortex could be explained by three different models of scene information, quantifying the unique contributions of each. We found that the relative contributions of each model differ depending on whether you're looking at behavior or the brain - suggesting we have more work to do in terms of developing computational models that can explain both behavior and neural responses in the human brain.