Biology subjects

Lenninger, M.

Publications and source records attributed to Lenninger, M..

3 recordsLinked to original sources

bioRxiv2026

NMDAR-mediated shift of neuronal gain across the cortical hierarchy

Neuronal gain is a critical feature of neuronal information processing, reflecting a cells excitability in response to presynaptic inputs. Evidence from spine counts in non-human primates suggests that the number of presynaptic inputs increases across the cortical hierarchy. Therefore, understanding the regulation of the neuronal gain can yield important insights into different integration strategies across the cortex. Along with the increasing spine counts, there is further evidence for a similar increase in NMDA-receptor-mediated currents. Here, we show, using detailed simulations of a thick-tufted layer 5 pyramidal neuron, that the amount of current mediated through NMDA receptors (NMDARs) strongly affects the neuronal gain function with balanced presynaptic inputs. Without NMDA, postsynaptic activity quickly saturates and declines with increasing inputs. However, with NMDARs, the gain of postsynaptic activity shifts into higher input regimes, enabling integration of dense inputs. We show that, with NMDARs, a significant fraction of postsynaptic activity arises from feedback between the voltage dependence of NMDARs and voltage-gated sodium channels. Lastly, we show, using a network model, that NMDAR knockout in single cells has differential effects on postsynaptic activity depending on network size (a proxy for cortical hierarchy). Thus, we propose that an NMDA gradient across the cortical hierarchy may be necessary to maintain excitability amid increasing presynaptic input.

neuroscience↗

bioRxiv2023

Predictive Motor Control Based on a Generative Adversarial Network

Predictive processing models suggest that a brain decides actions through inference using its internal generative model over the worlds states and their transitions. Most of the predictive processing models have been formalized using explicit representations of the probability distributions, with explicit structures and parameters. They are difficult to learn in general and needs explanation about representation of structure and parameters of distributions, and the method for statistical arithmetic on them, each of which are questions not easy to answer. In this study, we explore an alternative representation for predictive processing which is based on an implicit model known as generative adversarial networks, which has been widely explored recently in machine learning studies as they can learn a distribution directly from data. We demonstrate how a generative adversarial network can be trained to learn an implicit generative model of motor dynamics. And then, we show that such a model can perform approximate inference using the trained model, providing the necessary computations for both the forward and inverse model of motor control. Our framework may provide another formalization for brains inference model, especially for learning process. Additionally, we suggest that the functional architecture of the cortical-basal ganglia circuit may modeled as the generator and discriminator in the generative adversarial network model.

neuroscience↗

bioRxiv2022

How short decoding times, stimulus dimensionality and spontaneous activity constrain the shape of tuning curves: A speed-accuracy trade-off

According to the efficient coding hypothesis, sensory neurons are adapted to provide maximal information about the environment given some biophysical constraints. Early sensory neurons modulate their average firing rates in response to some features of the external stimulus, creating tuned responses. In early visual areas, these modulations (or tunings) are predominantly single-peaked. However, periodic tuning, as exhibited by grid cells, has been linked to a significant increase in decoding performance. Does this imply that the tuning curves in early visual areas are sub-optimal? We argue that the time scale at which neurons encode information is imperative to understanding the relative advantages of single-peaked and periodic tuning curves. Because, if decoding ability scales differently with time for the different shapes of tuning curves, the time scale at which the neurons operate becomes critical. Here, we show that the possibility of catastrophic (large) errors due to overlapping neural responses for distinct stimulus conditions creates a trade-off between decoding time and decoding ability. Unfortunately, standard theoretical measures such as Fisher information do not capture these errors. We investigate how (very) short decoding times and stimulus dimensionality affect the optimal shape of tuning curves for stimuli with finite domains. In particular, we focus on the spatial periods of the tuning curves (or the number of "peaks") for a class of circular tuning curves. We show a general trend for minimal decoding time, i.e., the shortest decoding time required to produce a statistically reliable signal, to increase with increasing Fisher information implying a trade-off between accuracy and speed. This trade-off is reinforced whenever the stimulus dimensionality is high or there is ongoing activity. Thus, given constraints on processing speed, we present normative arguments for the existence of single-peaked, rather than a periodic, tuning organization observed in early visual areas.

neuroscience↗