Richter, C. 1 & Vezoli, J. 2
1 BCBL
2 Ernst Strüngmann Institute (ESI)
The environment is rich in spatio-temporal structure and statistical regularity. This symposium will explore evidence that the brain actively exploits environmental regularity forming predictions of the incoming sensory stream, and will discuss putative neural mechanisms that account for predictive processing. In the first talk, Dr. Caspar Schwiedrzik will discuss evidence based on electrophysiological recordings and functional magnetic resonance imaging (fMRI) in monkeys and humans, that argues that perceptual inference occurs within hierarchical cortical networks that may support predictive coding. Remarkably, these hierarchical networks endow lower level cortical areas with discriminative abilities beyond their basic receptive field properties generating prediction error signals containing content usually ascribed to higher order association cortices. Dr. Matthias Grabenhorst will then explain, in our second talk, recent findings on temporal predictions that generalize across the visual, auditory and somatosensory modalities. He finds that reaction times are best explained by a non-linear transformation of the probability density function of events, rather than a mechanism based on the hazard rate. Based on these findings he argues that temporal evidence accumulation is a central, domain general mechanism. Dr. Lars Muckli will give the final talk discussing experiments conducted with ultra-high field fMRI. He will give evidence demonstrating laminar-specificity of visual input and top-down predictive processing streams in area V1. This segregation provides a hierarchical substrate that may support mechanisms of predictive coding and sheds light on how top-down visual predictions may interact with incoming sensory evidence.
S 3.1: Making and testing abstract predictions in neural circuits. Caspar Schwiedrzik
S 3.2: Temporal-probabilistic inference is independent of sensory input modality. Matthias Grabenhorst
S 3.3: Visual predictions in different layers of visual cortex. Lars Muckli