BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Extracting Universal Representations of Cognition from fMRI  mega-
 analyses - Bertrand Thirion\, Neurospin\, INRIA
DTSTART:20191118T123000Z
DTEND:20191118T133000Z
UID:TALK129298@talks.cam.ac.uk
CONTACT:Johan Carlin
DESCRIPTION:To map the neural substrate of mental function\, cognitive neu
 roimaging relies on controlled psychological manipulations that engage bra
 in systems associated with specific cognitive processes. In order to build
  comprehensive atlases of cognitive function in the brain\, it must assemb
 le maps for many different cognitive processes\, which often evoke overlap
 ping patterns of activation. Such data aggregation faces contrasting goals
 : on the one hand finding correspondences across vastly different cognitiv
 e experiments\, while on the other hand precisely describing the function 
 of any given brain region.\n\nIn this talk I will present two analysis fra
 meworks that tackle these difficulties and thereby enable the generation o
 f brain atlases for cognitive function.  The first one uses deep-learning 
 techniques to extract representations—task-optimized networks—that for
 m a set of basis cognitive dimensions relevant to the psychological manipu
 lations. This approach does not assume any prior knowledge of the commonal
 ities shared by the studies in the corpus\; those are inferred during mode
 l training.\n\nThe second one leverages ontologies of cognitive concepts a
 nd multi-label brain decoding to map the neural substrate of these concept
 s. Crucially\, it can accurately decode the cognitive concepts recruited i
 n new tasks. These results demonstrate that aggregating independent task-f
 MRI studies can provide a more precise global atlas of selective associati
 ons between brain and cognition.
LOCATION:Lecture Theatre\, MRC Cognition and Brain Sciences Unit\, 15 Chau
 cer Road\, Cambridge
END:VEVENT
END:VCALENDAR