NeuroLunch: Max Heinrich (Bear Lab) & Lukas Vogelsang (Sinha Lab)
Description
Speaker: Maxwell John Heinrich (Bear Lab)
Title: The persistent sodium current underlies spontaneous, steady-state activity of layer 5 pyramidal neurons in the mouse visual cortical slice
Abstract: The mouse cortical slice is characteristically silent in the presence of traditional artificial cerebral spinal fluid (aCSF), exhibiting essentially no spontaneous activity across layers. Traditional aCSF contains significantly higher concentrations of calcium and magnesium ions than observed physiologically. Perfusion of mouse visual cortical slices with a more physiologic aCSF containing lower concentrations of these divalent cations drives action potentials in a subset of layer 5 pyramidal neurons in the absence of ionotropic synaptic transmission. Whole-cell experiments demonstrate that reducing extracellular divalence enhances the intrinsic excitability of layer 5 pyramidal neurons by hyperpolarizing the voltage-dependence of persistent sodium current (INaP) activation. Steady-state current-voltage (IV) curves used to measure INaP reveal a zero-slope peak, indicative of the point at which outward currents are overcome by INaP-mediated inward currents. In spontaneously active neurons, this zero-slope peak lies close to or below the zero-current line, indicating an unstable resting potential. Phase diagram analysis derived from steady-state IV curves predicts membrane potential bistability in the absence of action potential dynamics, explaining the experimentally observed ‘jump’ that brings the voltage of the neuron from stable potentials to the spike threshold during sparse spontaneous activity. Stability analysis further predicts the equivalence between the zero-slope peak current and the rheobase current, the minimum steady current required to generate an action potential, confirming the hypothesis that steady activation of layer 5 pyramidal neurons is governed by the competition between steady-state inward and outward currents.
Speaker: Lukas Vogelsang (Sinha Lab)
Title: Butterfly effects in perceptual development: On the functional significance of degraded sensory experience early in life
Abstract: Human perceptual development unfolds in a stereotypical temporal manner. Visual inputs, for instance, are initially strongly degraded and improve in quality over the first months or years of life. These initially degraded inputs have traditionally been considered to represent ‘limitations’ imposed by constraints of the developing neural system. We propose, however, that they may, in fact, be adaptive and facilitate the acquisition of important representations and processing strategies that subserve robust perception later in life. This proposal is supported by joint experimental and computational findings that we recently published – notably in the domains of visual acuity and color vision. Experimental evidence derives from studies of children born blind and treated for their blindness late in life. In stark contrast to neonates, these children commence their visual experience with a remarkably mature perceptual system and, therefore, effectively skip the initial period of degraded vision. Examinations of their perceptual profiles revealed marked recognition deficits in generalization to color-reduced or color-shifted images, as well as poor performance on tasks relying on extended spatial integration. Results of computational simulations with deep neural networks suggest that these deficits can be accounted for by the children’s lack of initially degraded inputs. Specifically, training with developmentally-inspired stimulus progressions from poor to rich led to strong generalization in the above tasks, whereas training on exclusively high-quality, low-quality, or high-to-low-quality inputs did not. These results have implications for understanding normal visual development, help account for some of the deficits reported in individuals who underwent atypical developmental trajectories, and provide inspiration for more robust training procedures for computational model systems. Together with results from the domain of prenatal hearing, they also point to a potentially broader developmental phenomenon transcending the visual domain.