The cortical microcircuitry of predictions and context – a multi-scale perspective

Conscious cognition depends on the ability of the neocortex to generate internal models of the outside world. During wakefulness, the neocortex maintains and updates knowledge of the world and uses this knowledge through top-down projections to make predictions, test hypotheses, and/or contextualise input from the senses. How are these information streams combined in cortical microcircuits? Is their computational function to test internal models on the basis of their predictions or to contextualise sensory signals, or both? In addition to their somatic integration zones, many pyramidal neurons have a site of top-down and other contextual  information integration near the top of the apical dendrite’s trunk. This architecture enables top-down contextualisation of bottom-up information, amplifying or attenuating sensory responses depending on prior knowledge and current context. However, current deep neural network models of sensory processing lack such a mechanism, and cognitive theories do not reach the explanatory level of intracellular two-compartment integration. In this interactive ‘live’ paper, we envision how a continued synthesis of multi-scale, multi-species experimental data and theoretical and data-driven models will drive further insights into the biophysics, microcircuitry and dynamics of context-sensitive two-compartment neurons, and their role in predictive cognition.