Dr. Artur Luczak – Lethbridge Brain Dynamics

Artur Luczak

Director of TrainBioData NSERC CREATE program
Professor of Neuroscience, Canadian Centre for Behavioural Neuroscience

Department of Neuroscience, University of Lethbridge

4401 University Drive, Lethbridge, AB, T1K 3M4, Canada
e-mail: Luczak@uleth.ca , phone: (403) 394-3974, office: SA8240, Lab: SA8118

Research Interests

We are using electrophysiological and machine learning methods to study information processing in the brain. One of our main contributions is a development of ‘neuronal packet’ concept, which describes basic building blocks of neuronal code (Nature Rev Neurosci 2015, Neuron 2013, J Neurosci 2013, Neuron 2009, PNAS 2007). Moreover, we derived a predictive learning algorithm from basic cellular principles, i.e. from maximizing metabolic energy of a neuron, which may offer a step toward a general theory of neuronal learning (Nature Machine Intelligence; 2022). We are also studying changes in neuronal activity caused by neurological disorders, especially epilepsy (Brain 2017). To facilitate it, our lab developed Deep Neural Networks for detecting neurological deficits (PLOS Biology 2019).

My talk about our research on Brain Learning Mechanisms and Consciousness given at the Royal Society of Canada meeting in 2022.

Research highlights from Luczak lab

Single neuron predictive learning. Neurons have biochemical mechanisms capable of performing complex computations. Thus, the simple models of neurons used in machine learning may be missing the essential computational elements of the brain. We showed that to maximize metabolic energy, individual neurons need to predict their own expected future activity. This results in a new learning algorithm which could be a crucial component of the brain’s learning mechanism. Nature Machine Intelligence paper

Here is my talk about our epilepsy research and at 17:40, I also describe single neuron predictive learning.

Neuronal packet theory. We described that in the sensory cortex, information is not processed continuously, but rather is divided in 50~500ms long “packets”, which have specific sequential structure of neuronal activity. Each packet can be conceived of as a discrete ‘message’, with neurons active at the beginning of a packet providing general information (e.g. it is a face), while neurons active in the latter phase encode more precise information (e.g. this is face of my friend John). This packet-like organization of neuronal activity may provide an explanation for multiple puzzling observations about neuronal coding. Nature Rev Neurosci paper.

Short video about our research on neuronal packets, and here is more in depth talk about packets (+ slides)

Job opportunities for PhD student or Postdoc

Our lab seeks highly motivated individuals with strong computational backgrounds to work at the interface of neuroscience and machine learning. In our lab we are also recording activity of hundreds of neurons in normal and epileptic animals, and successful candidate is welcome to participate in those projects. Preferred candidate should have strong background in modeling neurons and networks, and should be very familiar with models of synaptic plasticity (e.g. BCM). Our lab at the Canadian Centre for Behavioural Neuroscience in Lethbridge is located in the sunniest area of Canada and next to scenic Rocky Mountains.

Undergraduate students interested in any of the above topics may also apply for Independent study in our lab.

Teaching

Introductory workshop on computational methods in neuroscience
Statistics and Programming in Matlab
Advanced Application of Computational Methods (online class)
Brain and Behavior

Short bio

2018 prof. – CCBN, University of Lethbridge
2016 elected member of the College of the Royal Society of Canada
2016/17 visiting assoc. prof. – Stanford University (I. Soltesz lab)
2014 assoc. prof. – CCBN, University of Lethbridge
2009 assistant prof. – CCBN, University of Lethbridge
2004 postdoc – Rutgers University (K.D. Harris lab)
2002 postdoc – Yale University (R. Coifman lab in Comp. Sci. Dept. and M. Laubach lab in Neurosci. Dept.)
2002 Ph.D. – Jagiellonian University, Medical College, Poland. (thesis: The application of fractal geometry in neuroanatomy; J. Trabka lab)
2001/02 Marie Curie Fellowship – International School for Advanced Studies , Italy (A. Treves lab)
1997 M.Sc in Biomedical Engineering – Wroclaw University of Technology.

Publications (Google Scholar profile)

An orexigenic subnetwork within the human hippocampus.

Reinforcement Learning with Brain-Inspired Modulation can Improve Adaptation to Environmental Changes.

Editorial: Deciphering population neuronal dynamics: from theories to experiments.

Hippocluster: an efficient, hippocampus-inspired algorithm for graph clustering.

Biologically-inspired neuronal adaptation improves learning in neural networks.

Neurons learn by predicting future activity.

Predictive neuronal adaptation as a basis for consciousness.

Combining Backpropagation with Equilibrium Propagation to improve an Actor-Critic Reinforcement Learning framework.

Epileptic seizures and link to memory processes.

Sensory experience selectively reorganizes the late component of evoked responses.

Spatiotemporal structure of sensory-evoked and spontaneous activity revealed by mesoscale imaging in anesthetized and awake mice.

A Neural Network Reveals Motoric Effects of Maternal Preconception Exposure to Nicotine on Rat Pup Behavior: A New Approach for Movement Disorders Diagnosis.

Spatiotemporal patterns of neocortical activity around hippocampal sharp-wave ripples.

Diverse Perspectives on Interdisciplinarity from the Members of the College of The Royal Society of Canada.

Data-driven analyses of motor impairments in animal models of neurological disorders.

Using neuron spiking activity to trigger closed-loop stimuli in neurophysiological experiments.

Direct Current Stimulation Improves Limb Use After Stroke by Enhancing Inter-hemispheric Coherence.

Effect of body position on relieve of foreign body from the airway.

Phase of EEG theta oscillation during stimulus encoding affects accuracy of memory recall.

Deep Convolutional Auto-Encoder with Pooling – Unpooling Layers in Caffe.

Predict Consequences as a Method of Knowledge Transfer in Reinforcement Learning.

Involvement of fast-spiking cells in ictal sequences during spontaneous seizures in rats with chronic temporal lobe epilepsy.

UP-DOWN cortical dynamics reflect state transitions in a bistable balanced network.

Chronic Mild Stress Exacerbates Severity of Experimental Autoimmune Encephalomyelitis in Association with Non-coding RNA and Metabolic Biomarkers.

Creation of a Deep Convolutional Auto-Encoder in Caffe.

Head-down self-treatment of choking.

Computational Properties of the Hippocampus Increase the Efficiency of Goal-Directed Foraging through Hierarchical Reinforcement Learning.

Context-Switching and Adaptation: Brain-Inspired Mechanisms for Handling Environmental Changes

Packet-based communication in the cortex.

Beyond the Silence: Bilateral Somatosensory Stimulation Enhances Skilled Movement Quality and Neural Density in Intact Behaving Rats.

Formation and reverberation of sequential neural activity patterns evoked by sensory stimulation is enhanced during cortical desynchronization.

Transcranial Direct Current Stimulation in Stroke Rehabilitation – A Review of Recent Advancements.

Gating of sensory input by spontaneous cortical activity.

Consistent sequential activity across diverse forms of UP states under ketamine anesthesia.

Temporal variability of the N2pc during efficient and inefficient visual search.

Default activity patterns at the neocortical microcircuit level.

Neural correlates of auditory distraction revealed in theta-band EEG.

Recording Large-scale Neuronal Ensembles with Silicon Probes in the Anesthetized Rat.

Measuring neuronal branching patterns using model-based approach.

How do neurons work together? Lessons from auditory cortex.

Spontaneous events outline the realm of possible sensory responses in the auditory cortex.

Population coding of tone stimuli in auditory cortex: dynamic rate vector analysis.

Sequential structure of neocortical spontaneous activity in vivo.

Spatial embedding of neuronal trees modeled by diffusive growth.

Spectral representation – analyzing single-unit activity in extracellularly recorded neuronal data without spike sorting.

Multivariate receptive field mapping in marmoset auditory cortex.

Modeling stimulus-response functions in the auditory system.

A cluster of workstations for on-line analyses of neurophysiological data.

Estimating neuronal variable importance with Random Forest.

Model of neuronal distribution during development in rat cortex based on cellular automata

Fractal modelling of dendritic structures as a paradigm for morphogenetic structure of neurons

Simulation and modelling as the cognitive procedures.

Modeling of growth and shape of neurons by the application of fractal geometry

Parametric description of neuron shape on the basis of a generator of artificial neurons

Book chapters

Packets of sequential neural activity in sensory cortex; In "Analysis and modeling of coordinated multi-neuronal activity – Sequence phenomena and memory-trace replay". Editor: Tatsuno M.

Shaping of neurons by environmental interaction; In "Dendritic computations through morphology and connectivity". Editors: Torben-Nielsen B, Remme M, Cuntz H.

Patent:

Brain state dependent therapy for improved neural training and rehabilitation (patent pending in USA and Canada; filed in June 2015). full text
Description: This invention provides means to assess how receptive to learning (plastic) the brain is at any given time. This invention has several commercial applications: (1) Rehabilitation centers could use devices based on this technology to measure brain responsiveness to therapy, which could improve rehabilitation after e.g. brain injury. (2) In addition, a consumer version for the general public could be used to focus learning/training to times of maximal brain receptivity, and as a biofeedback device for self-training to produce plastic brain states. Considering that this idea could result in significant health benefits, my colleagues and I started a company DeepBrain Analytics Inc., to facilitate bringing this invention to the market.

Outreach activities:

I am past President of the Lethbridge Chapter of the Society for Neuroscience (SfN), where I was responsible for organizing multiple events to promote brain research, which engaged over 500 people annually. The main annual events included:

I co-organize with M. Mohajerani Satellite Symposium at the Canadian Neuroscience Meeting: Neural Signal and Image Processing: Quantitative Analysis of Neural Activity; in Montreal (2017); and also with T. Murphy in Vancouver, (2018), and with S. Prescott in Toronto (2019). Each year we have ~50 participants from across Canada, and in 2019 we expanded format of this workshop to two days to accommodate student demand.

Funding:

NSERC CREATE Grant (2024-2030) Training Biology Students for Jobs in Data Science (TrainBioData)
CIHR Project Grant (2019-2024)
CIHR Priority funding: Data science (2019-2021)
NSERC Discovery Grant (2010-2025)
NSERC DG Accelerator (2015-2018, awarded to the top ~4%)
University of Lethbridge Research Fund (2018)