BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Neural network models of free recall and spatial navigation - Stef
 ano Recanatesi (Weizmann Institute)
DTSTART:20170614T094500Z
DTEND:20170614T103000Z
UID:TALK73063@talks.cam.ac.uk
CONTACT:Daniel McNamee
DESCRIPTION:Two different models will be described during the seminar.\nA 
 Model of free recall addressing the capability human memory in retrieving 
 similar memories to a just retrieved one. This associative ability is at t
 he base of our everyday processing of information. Current models of memor
 y have not been able to underpin the mechanism that the brain could use in
  order to actively exploit similarities between memories. We introduce a n
 ovel mechanism capable to induce transitions between memories where simila
 rities between memories are actively exploited by the neural dynamics to r
 etrieve a new memory. The so generated spontaneous retrieval is compared t
 o experiments of free recall.\nA Model of spatial navigation motivated by 
 recent theoretical work trying to reconcile the declarative memory view an
 d the spatial navigation view of hippocampal functioning\, we investigate 
 a recurrent neural network model that shows how the hippocampus could inte
 grate episodic memories into generic "semantic relational networks". We pr
 opose an explicit computational mechanism for the learning of such relatio
 nal networks: predictive coding. The model learns to generate recurrent ne
 ural activations that are reminiscent of place cells and border cells in a
  simulated navigation environment\, and can naturally account for context-
 specific representations and "time cells”.
LOCATION:Cambridge University Engineering Department\, CBL\, BE-438 (http:
 //learning.eng.cam.ac.uk/Public/Directions)
END:VEVENT
END:VCALENDAR