Mutual information

…between spatial correlations and mutual information. In relativistic field theories the definition agrees with analytic continuation from spacelike to timelike regions. The presentation covers measurement protocols, calculations for the Ising chain and holographic theories, and connections to timelike pseudoentropy.

…decomposition approximates a normative information-theoretic objective where the

Maintaining physiological homeostasis is essential for survival. To achieve this, brains integrate various inputs, such as internal physiological and external environmental signals, to guide behavior. Inspired by brain-body literature, we consider three key axes along which the dynamics of these inputs can differ: their timescales, dependence on actions, and contribution to error signals that drive behavior. To understand how these axes impact physiological homeostasis, we consider an agent that can influence its internal physiology through interactions with the environment. We assume that env

Information theory provides a powerful theoretical framework for a quantitative understanding of the structure and function of neural circuits. Despite this, the application of these ideas to neuroscience and machine learning has been hampered by the fact that the computation and maximization of mutual information is only analytically tractable for simple coding problems, under the assumption of small network sizes and low dimensional stimuli. Using tools from the theory of spin-glasses we present a novel approach to estimating the entropy of large neural populations in response to arbitrary s

The natural world is full of noise, but the brain’s capacity for information transmission is severely limited. Therefore, discarding irrelevant information contained in sensory inputs while retaining salient features that are related to the input label, is key to survival. What are the salient features? And what are the underlying feature selection mechanisms? It is thought that the brain may implement information bottlenecks, which aim to optimise the trade-off between compression and preservation of salient information. However, information bottlenecks are notoriously difficult to implemen

Hippocampal granule cell (GC) models exhibit degeneracy with different ion channel parameters resulting in comparable functional behaviour [1,2]. However, it is unknown how the degenerate models are further constrained by energetic considerations and known computational functions of GCs. Here, we use Pareto optimality to investigate this question. The Pareto perspective is particularly useful for multi-objective optimisation with multiple trade-offs [3]. We assume that evolution constrains and simplifies the ion channel parameter space by favouring Pareto optimal neurons, which cannot improve

The role of feedback in sensory systems remains a major puzzle in neuroscience. Classical computational theories such as hierarchical predictive coding[1] or hierarchical Bayesian inference[2] postulate abstract normative goals for the role of feedback connections. However, predictions derived from these theories are typically specified at the level of individual neurons. Whether feedback shapes collective properties of neural populations and how such changes may impact encoding of sensory signals remain unknown. Here we tackle these questions directly by studying the population coding of na

…Although mutual information (MI) continues to be a useful tool when applying Information theory to neural systems, it does not suffice when trying to understand how informative individual stimuli are. In general such knowledge may be crucial for understanding the functional organization of sensory systems. To investigate this, we focus on ON-OFF Direction Selective Retinal Ganglion Cells (dsRGC). These cells respond robustly to motion in their preferred direction, and sparsely to motion in the op

Different stimuli elicit different transient neural responses in the brain. How is the information represented in the parallel neuronal activity of a neuronal population? We investigate this question in Neuropixel recordings of awake behaving mice by inferring the stimulus class (e.g., visual or tactile) from the activity. To quantify the separability, we use an optimally trained linear readout, which can be understood through the framework of Bayesian inference. We find that separability only depends on three geometric properties of the neuronal population vector, described by cones within th

…trade off energy consumption and information transmission. Specifically, neurons appear to reduce energy consumption by changing their integration properties through the adaptation of membrane resistance, resting potential, and synaptic efficacies, which comes at the cost of reduced information transmission. Furthermore, they found that individual neurons assume a broad range of values of these paramete

The vast majority of neurons in the vertebrate nervous systems transmits information in a digital manner via action potentials. Yet, a detailed understanding of the advantages of digital communication in the brain is still lacking. In our study, we investigate synaptic transmission in terms of an optimization problem, where the space of solutions comprises the digital and the analog regime. We analyze a binomial synapse model connecting two neurons, where the probability of transmitter release, modeled by a transfer function, depends on the axonal membrane potential. We optimize the transfer f

The synchronization of the phase of neuronal oscillations and the correlated fluctuations in the envelopes of neuronal signals are two prominent coupling modes that govern the brain dynamics. They mediate in an efficient and selective manner the communication in networks of cortical and subcortical areas to facilitate sensory processing, behavior, and perception. However, it is not clear yet if and how these two types of intrinsic coupling modes (ICMs) interact with each other. Here we investigate the relation between both coupling modes across different brain states. In accordance with our pr

…of activity that maximises the mutual information (MI) between the transformation and the relevant task variables, since MI is the most general meas

Perception is molded by sensory input, prior knowledge, and contextual cues, collectively giving rise to perceptual biases whose neural mechanisms remain poorly understood. This study aimed to elucidate these mechanisms by examining prefrontal cortex (PFC) neural recordings from monkeys performing a vibrotactile frequency discrimination task (Fig. 1a). Monkey’s behavior exhibited the contraction bias (Fig. 1b), a contraction of the perceived first frequency towards its mean. This bias increased with longer delay periods, suggesting a Bayesian explanation of this effect. Examination of PFC’s ne

Several studies highlight motivation's crucial role in motor decision-making for maximizing potential rewards. However, how explicit rewards impact the neural dynamics of decision-making processes and their effects on neural organization at the network level, is still unexplored. To address this gap, we analyzed neuronal activity recorded from the premotor cortex (PMd) of three monkeys while performing a modified version of the stop signal task. In this task, the motivation to either act or inhibit action was manipulated using Cues that informed about different reward contexts. We conducted an

Variations in brain states and behavioral profiles clearly influence neuronal activity in the visual cortex[1,2]. However, the mechanisms behind these influences and their consequent effect on sensory encoding remain to be determined. Here, we address this issue by analyzing simultaneous recordings of spiking activity and local field potentials (LFPs) from visual and subcortical regions of (N = 25) awake mice viewing a naturalistic movie. Applying a hidden Markov model (HMM) to cortical LFPs, we consistently identified three stable states in each mouse: a high-frequency state (dominated by gam

Planar shape, i.e., the silhouette contour of a solid body, carries rich information important for object recognition, including both local (curvature) and global shape cues. While curvature-selective neurons have been identified in area V4 of primate, it remains unclear whether a) curvature is the best way to characterize the shape selectivity of these neurons and b) whether selectivity is limited to local shape. Here we employ a unique array of shape stimuli to dissociate tuning for local and global shape properties. These stimuli have been used previously to identify an intriguing congruenc

…of past neural data that have maximal mutual informa

…protocols that may not yield enough information about these parameters. Here, we propose using Bayesian

…brain areas can interact to encode information. When two brain areas interact with a third, what fraction of the mutual information between the former and the latter can be attributed uniquely to each of the former? What fraction is redundant between them, and what fraction is synergistic? These “partial information” components quantify how different brain areas interact, which can help us understand the efficiency of various neural coding or communication schemes. Information-theoretic measures called partial i

…two worlds, showing that prominent information processing tr

Efficient coding theory, which stipulates that the nervous system attempts to minimize redundancy in stimulus coding, has proven widely successful in predicting observed features in early sensory systems, including center-surround receptive fields (RFs), ON and OFF parallel pathways, and the relative arrangement of ON and OFF RF mosaics. However, relatively limited work has addressed the encoding of spatiotemporal stimuli. Ocko et al. (2018) examined spatiotemporal encoding in the retina using an approach based on a convolutional autoencoder, replicating known features of retinal ganglion cell

…of probability and relative frequency information? We propose a Bounded Log-Odds Model (BLO) of probability and relative frequency distortion based on three assumptions: (1) log-odds: probability and relative frequen

…perception, we have to understand how information enters the brain, how it is transformed and how it is transferred. But, how do we measure information transfer in the brain? This presentation will start with a general introduction of what mutual information is and how to measure it in an experimental setup. Next, the talk will focus on how this can be used to develop brain models at different (spatial) levels, from the microscopic single neuron level to the macroscopic network and behavioural level. How can we incorporate the knowl