- Homepage
- Programme
- Registration
- Practical Guide
- About Neuronus
Wiktor Młynarski
Ludwig Maximilian University of Munich, Munich, Germany
The tremendous, almost impenetrable complexity of the nervous system is not just one, but a huge collection of mysteries we are all trying to solve. To develop understanding of neurobiological phenomena we need to seek simplicity. In the talk I will discuss one such way to search for simplicity and understanding - through building normative theories of neural computation. Normative theories attempt to identify goals and principles that may be shared by multiple, seemingly different neural systems. I will specifically focus on sensory systems which need to achieve a delicate balance between external and internal influences in order to accurately represent relevant information. Dynamic adjustments of the sensory code to these influences have been traditionally categorized depending on their origin and studied separately. Sensory adaptation is a response of a neuron to exogenouschanges in stimulus statistics, while internal modulation adjusts sensory representations to changes in the endogenous states of the brain such as behavioral goals, attention or uncertainty. I will present a theoretical framework which provides a unifying perspective on how sensory codes adapt to such changes regardless of their origin. Starting from the same set of basic principles grounded in information theory and Bayesian inference, our framework generates candidate normative explanations of the diversity of adaptive responses in the early visual system as well as the attentional modulation of neural populations in the primary visual cortex. I will conclude by presenting an experimental finding of spatio-temporal patterns of neural activity which dominate sensory responses in a brain region that has been thought to be predominantly a sensory relay - the superficial superior colliculus. These findings emphasize the need for new theories which will be required to understand the computational principles of dynamic sensory processing and to further tame the overwhelming complexity of the brain.
Sawicka Katarzyna, Włodkowska Urszula, Falińska Monika, Rafał Czajkowski
Laboratory of Spatial Memory, Nencki Institute of Experimental Biology, Polish Academy of Science, Warsaw, Poland
Episodic memories are crucial for our identity and heavily depend on the hippocampus. Despite this, understanding how these memories form, are maintained, and retrieved is still developing. The indexing theory suggests that hippocampal neurons are interconnected with sensory and associative cortical regions, yet it's unclear if cortical reinstating occurs neuron-to-neuron. To explore this, we induced cortical activity in the retrosplenial cortex (RSC) and observed increased c-Fos activity in hippocampal cells. We then replicated this hippocampal index and observed GCaMP signal activity in the RSC. Our goal is to compare the initially activated neurons with the reactivated population in the RSC. This research delves into cortical reinstatement's dynamics and its impact on episodic memory. By optogenetically stimulating cortical neurons, we aim to understand how the hippocampus and cortex interact during memory processes. These findings could enhance our understanding of memory formation and retrieval mechanisms, offering insights into conditions like amnesia and neurodegenerative diseases.
Emilia Kaczmarczyk, Maciej Kamiński
Biomedical Physics Division, Faculty of Physics, University of Warsaw, Warsaw, Poland
The aim of my research is to investigate the propagation of alpha rhythm in EEG signals using the PCMCI+ algorithm and to compare the results obtained with analyses using Granger causality-related algorithms. The alpha rhythm is a rhythmic activity of the brain cortex in the 8-12 Hz range, occurring during relaxation with eyes closed. This rhythm is generated in the visual cortex in the occipital lobe, and then is propagated to the front of the head by stimulating other structures. Previous studies aimed at investigating the propagation of alpha rhythm in terms of causality have been conducted using algorithms based on Granger causality, which are unable to detect causal relationships occurring at a rate faster than the sampling frequency. In this study we used PCMCI+ algorithm which is a derivative of the PC algorithm, based on the identification of a Bayesian network describing the given system. This method may help to better identify certain dependencies by examining contemporaneous causal relationships. Preliminary analyses indicate that the direction of alpha rhythm propagation may be correctly identified using only delay-free dependency analysis.
Magdalena Szponar1, Bartłomiej Gmaj2, Jan Kamiński1
1Laboratory of Neurophysiology of Mind, Nencki Institute of Experimental Biology, Polish Academy of Sciences, Warsaw, Poland
2Department of Psychiatry, Medical University of Warsaw, Warsaw, Poland
Currently, diagnosis of mental disorders is based predominantly on subjective and time-consuming selfreport techniques. Recent literature proposes applying machine learning classification on electroencephalography (EEG) data to distinguish psychiatric patients from healthy controls, with promising accuracies of 75-95%. However, these studies have substantial drawbacks, like small sample sizes (often less than 100 participants), training and testing the model with data from the same patients and lack of multi-categorical classification. Using the psychiatric hospital’s archival data, we prepared the database containing over 13000 restingstate EEG recordings of patients diagnosed with a wide range of disorders, providing a sufficient sample to train multi-categorical algorithms. We extracted over 20000 EEG measures for the five most common disorders in the database and trained a state-of-the-art shallow neural network on this data, using one-versus-rest scheme. We obtained the average accuracy of 67.5%, significantly higher than the chance level, for classification of independent patients. Furthermore, this accuracy increases with the model’s activation function value, allowing us to evaluate the model’s predictions. These results indicate that creating machine learning algorithms distinguishing several psychiatric disorders is possible. We will work further to increase our method’s accuracy and investigate features contributing to the correct classification. Hopefully, the final algorithm could aid psychiatrists as an objective and quick method for clinical diagnosis.
Funding: The research was conducted as part of the BRAINCITY project. The BRAINCITY project is carried out within the International Research Agenda Programme of the Foundation for Polish Science co-financed by the European Union under the European Regional Development Fund.
