Computational Model of Surprise in Neuroscience
Event details
| Date | 25.06.2019 |
| Hour | 14:00 › 16:00 |
| Speaker | Alireza Modirshanechi |
| Location | |
| Category | Conferences - Seminars |
EDIC candidacy exam
Exam president: Prof. Michael Gastpar
Thesis advisor: Prof. Wulfram Gerstner
Co-examiner: Prof. Martin Jaggi
Abstract
Shaping perception and making decisions in the absence of complete knowledge requires estimation of unknown variables. To survive despite its natural lack of knowledge, our brain is evolved in a way to continuously estimate the environmental unobserved variables and predict the consequences of its actions. A common hypothesis is that the brain makes a probabilistic model of the world, and learns its parameters during time. Then it uses this probabilistic model for estimation, inference, and prediction. In this proposal, we first introduce a unifying notation for such probabilistic models, and then review three impactful articles which proposed such models for human perception and behavior.
Background papers
Mathys, Christoph, et al. A Bayesian foundation for individual learning under uncertainty. Frontiers in human neuroscience5 (2011): 39.
Friston, Karl, et al.Active inference: a process theory. Neural computation 29.1 (2017): 1-49. (only first 19 pages, until the end of section 2)
Maheu, Maxime, Stanislas Dehaene, and Florent Meyniel.Brain signatures of a multiscale process of sequence learning in humans. Elife.
Exam president: Prof. Michael Gastpar
Thesis advisor: Prof. Wulfram Gerstner
Co-examiner: Prof. Martin Jaggi
Abstract
Shaping perception and making decisions in the absence of complete knowledge requires estimation of unknown variables. To survive despite its natural lack of knowledge, our brain is evolved in a way to continuously estimate the environmental unobserved variables and predict the consequences of its actions. A common hypothesis is that the brain makes a probabilistic model of the world, and learns its parameters during time. Then it uses this probabilistic model for estimation, inference, and prediction. In this proposal, we first introduce a unifying notation for such probabilistic models, and then review three impactful articles which proposed such models for human perception and behavior.
Background papers
Mathys, Christoph, et al. A Bayesian foundation for individual learning under uncertainty. Frontiers in human neuroscience5 (2011): 39.
Friston, Karl, et al.Active inference: a process theory. Neural computation 29.1 (2017): 1-49. (only first 19 pages, until the end of section 2)
Maheu, Maxime, Stanislas Dehaene, and Florent Meyniel.Brain signatures of a multiscale process of sequence learning in humans. Elife.
Practical information
- General public
- Free