BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Compositional Generalization and Learning - Rui Xia\; Daniel Korna
 i
DTSTART:20260212T090000Z
DTEND:20260212T110000Z
UID:TALK244627@talks.cam.ac.uk
CONTACT:124819
DESCRIPTION:Cognition is highly flexible—we perform many different tasks
  and continually adapt our behaviour to changing demands. One way to flexi
 bly switch between tasks and rapidly learn new tasks is through reuse of n
 eural representations and computational components. In the first part of t
 he journal club\, Rui will present a paper [1] where it is shown that such
  compositionally is found in the monkey brain when performing three compos
 itionally related tasks. In neural recordings\, the authors found that tas
 k-relevant information about stimulus features and motor actions were repr
 esented in subspaces of neural activity that were shared across tasks. Mon
 keys adapted to changes in the task by iteratively updating their internal
  belief about the current task and then\, based on this belief\, flexibly 
 engaging the shared sensory and motor subspaces relevant to the task.\n\nH
 owever\, the paper contains no attempts to model subject behaviour\, and m
 akes no suggestions for how such modelling could be done. Therefore\, in t
 he second half of the talk\, Daniel will introduce “infinite composition
 al contextual bandits”\, a class of Bayesian models which are currently 
 being developed to investigate the effects of compositional generalisation
  on decision-making during continual learning in animals.\n\n[1] Tafazoli\
 , S.\, Bouchacourt\, F. M.\, Ardalan\, A.\, Markov\, N. T.\, Uchimura\, M.
 \, Mattar\, M. G.\, ... & Buschman\, T. J. (2025). Building compositional 
 tasks with shared neural subspaces. Nature\, 1-9. \n
LOCATION:CBL Seminar Room\, Engineering Department\, 4th floor Baker build
 ing
END:VEVENT
END:VCALENDAR