Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session C: Friday, August 15, 2:00 – 5:00 pm, de Brug & E‑Hall

Predictive Perception via Simultaneous Learning and Inference

Mahdi Enan¹, Mario Senden¹, Yamil Dos Santos Vidal, Ryszard Auksztulewicz¹, Federico de Martino; ¹Maastricht University

Presenter: Mahdi Enan

The brain is continuously faced with noise in the soundscape and uncertainty in the input to the ears. This has led to the proposal that perception is not a direct consequence of sensory input, but rather an inferential process to determine the most probable state of the world. However, the algorithm through which the brain could realize this inferential process as well as its implementation are not yet well understood. In this work, we developed a novel framework for simultaneous learning and inference from first principles using exact inference and local Hebbian-like learning rules. We show that this framework allows unbiased inference and flexible model updating under noise and changing dynamics. Further, we show how this approach can reproduce local and global effects of prediction and prediction error in noisy environments. This model can be used to test (in silico) hypotheses related to predictive processing in noisy and non deterministic environments.

Topic Area: Predictive Processing & Cognitive Control

Extended Abstract: Full Text PDF