BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Neuro-X Seminar: Rational Sensing from insects to rodents to human
 s to machines
DTSTART:20230302T130000
DTEND:20230302T140000
DTSTAMP:20260928T233033Z
UID:80dba8486953ce9d4c2a9a74bbe479e88df547260a90e43a99f725da
CATEGORIES:Conferences - Seminars
DESCRIPTION:Prof Rafael Polania\nNOTE: for logistical reasons\, please do 
 access the seminar room by the exterior of AI extension (level 0\, entranc
 e AI 0354)\n\n\nIs the role of our sensory systems to represent the physic
 al world as accurately as possible? If so\, are our preferences and action
 s—which are often labeled as irrational—decoupled from these “ground
 -truth” sensory experiences? We argue that the answer to both questions 
 is no. Perhaps counterintuitively\, we propose that accurate representatio
 ns of sensory signals do not necessarily maximize the organism’s chances
  of survival. To test this hypothesis\, we developed a unified normative f
 ramework for fitness-maximizing encoding by combining theoretical insights
  from neuroscience\, computer science\, and economics. Initially\, we appl
 ied predictions of this model to neural responses from large monopolar cel
 ls (LMCs) in the blowfly retina. We found that neural codes that maximize 
 reward expectation—and not accurate sensory representations—account fo
 r retinal LMC activity. Behavioral experiments in humans revealed that sen
 sory encoding strategies are flexibly adapted to promote fitness maximizat
 ion. Moreover\, human fMRI data confirmed that novel behavioral goals that
  rely on object perception induce efficient stimulus representations in ea
 rly sensory structures. Interestingly\, this result was confirmed by deep 
 neural networks with information capacity constraints trained to solve the
  same task in humans. Furthermore\, experiments in which rodents were trai
 ned to solve the same task in humans revealed that mice also adaptively al
 locate their sensory resources in a way that maximizes reward consumption 
 in novel stimulus-reward association environments. These experiments allow
 ed us to discover that arousal systems carry reward distribution informati
 on of sensory signals and that distributional reinforcement learning mecha
 nisms—a fundamental mechanism in state-of-the-art machine learning algor
 ithms—regulate sensory precision via top-down normalization. These findi
 ngs reveal how agents can efficiently perceive and adapt to environmental 
 contexts within the constraints imposed by neurobiology. Thus\, the often-
 observed irrationalities and biases attributed to downstream processing mi
 ght unavoidably originate from the way early sensory systems should adapt 
 to and process information in insects\, rodents\, humans\, and machines.\n
 \n 
LOCATION:AI 1153 https://plan.epfl.ch/?room==AI%201153 https://epfl.zoom.u
 s/j/62560106538?pwd=NHFDUG5mMUtQOUtib3RXNVVBZDVIUT09
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR