Ede RANZ, INMED, INSERM, Equipe Circuits Corticaux
"A biological model system for studying predictive processing"
Abstract : Despite the increasing recognition of predictive processing in circuit neuroscience, little is known about how it may be implemented in cortical circuits. We aim to gain access to predictive processes by understanding and controlling everything else: the sensory environment, feed-forward and feed-back inputs, integrative properties, spiking activity and functional output.
To this end, we defined and characterised a biological model system. At the centre are layer 5 pyramidal cells (L5PN) of the mouse secondary visual cortex (V2). We show that L5PNs in V2 receive a wide variety of long-range inputs, which can be classified as either feed-forward or feed-back. Moreover, these inputs are diversely distributed over the dendritic tree, often contradicting accepted connectivity rules. We also found that the integrative properties of L5PNs differ fundamentally from canonical L5PNs of the primary sensory cortices.
Furthermore, we show that V2 L5PNs project to eye movement-related areas and can directly influence eye movements in awake, head-fixed mice. Lastly, I will introduce a novel rotation platform we developed to re-engage the vestibular system in head-fixed mice to control both the vestibular and visual sensory environment and introduce prediction errors.
In summary, we introduce a holistic model system to study the biological substrate of predictive processing. We show that the connectivity and integrative properties of L5PNs in V2 disagree with received wisdom, questioning the existence of canonical rules for cortical organisation.