BEGIN:VCALENDAR
VERSION:2.0
PRODID:talks.ox.ac.uk
BEGIN:VEVENT
SUMMARY:Modelling human planning with recurrent neural networks - Professo
 r Marcelo Mattar (New York University\, USA)
DTSTART;VALUE=DATE-TIME:20250609T150000
DTEND;VALUE=DATE-TIME:20250609T160000
UID:https://talks.ox.ac.uk/talks/id/a6c5789e-294d-40ca-b656-5069c837d02c/
DESCRIPTION:ABSTRACT:\n\nWhen interacting with complex environments\, huma
 ns can rapidly adapt their behavior in response to changes in task or cont
 ext. To facilitate this adaptation\, people often spend substantial period
 s of time contemplating possible futures before acting. In this talk\, I w
 ill present empirical and modeling work exploring the critical balance bet
 ween thinking and acting\, and the factors affecting the content of our th
 oughts when we are making a decision. First\, I'll introduce a recurrent n
 eural network model that learns when planning is beneficial\, explaining v
 ariations in human thinking times and patterns observed in rodent hippocam
 pal activity during navigation. Second\, I'll discuss how meta-learning en
 ables neural networks to discover human-like planning strategies that blen
 d elements of tree search and rollout algorithms. Finally\, I'll examine h
 ow working memory constraints shape reward encoding during sequential plan
 ning\, revealing how humans strategically allocate cognitive resources bas
 ed on decision relevance. Across these studies\, recurrent neural networks
  emerge as powerful models for capturing the dynamic\, iterative nature of
  human planning processes and their implementation in brain circuits\n\nAB
 OUT THE SPEAKER:\n\nProf Marcelo Mattar is an Assistant Professor at New Y
 ork University's Department of Psychology and the Neuroscience Institute. 
 He holds a PhD in Psychology from the University of Pennsylvania\, where h
 e also earned a Master's in Statistics. His academic journey includes a Ba
 chelor's in Electronics Engineering from the Aeronautics Institute of Tech
 nology in Brazil. Prior to his current role\, Prof. Mattar was a postdocto
 ral researcher at Princeton University under the mentorship of Nathaniel D
 aw and at the University of Cambridge's Department of Engineering\, workin
 g with Mate Lengyel.\nHis research integrates neuroscience\, psychology\, 
 and computational modelling to explore how the brain supports flexible dec
 ision-making. Focusing on the interplay between learning\, memory\, and pl
 anning\, Prof Mattar's work aims to understand how the brain learns intern
 al models from experience and simulates future scenarios to guide decision
 -making.\n\nSpeakers:\nProfessor Marcelo Mattar (New York University\, USA
 )
LOCATION:New Radcliffe House (2nd floor seminar room)\, Walton Street OX2 
 6NW
TZID:Europe/London
URL:https://talks.ox.ac.uk/talks/id/a6c5789e-294d-40ca-b656-5069c837d02c/
BEGIN:VALARM
ACTION:display
DESCRIPTION:Talk:Modelling human planning with recurrent neural networks -
  Professor Marcelo Mattar (New York University\, USA)
TRIGGER:-PT1H
END:VALARM
END:VEVENT
END:VCALENDAR