Para Limes

Seminar: Modeling and Data Analysis in Primary Visual Cortex

Seminar: Modeling and Data Analysis in Primary Visual Cortex

Dr Malte J. Rasch

Post-doc at Beijing Normal University, China

Dr Malte J. Rasch was born in Northern Germany. He studied (theoretical) biophysics at the Humboldt University in Berlin. His study and research topics were mainly mathematical modelling of biological processes and computational neuroscience. For instance, he built a model of the (possible) function of adult neurogenesis in the hippocampus (together with L. Wiskott). After the master degree, he pursued PhD with W. Maass and N. Logothetis, analysing LFP/spikes using machine learning techniques and building spiking network models of the visual system. After PhD graduation, he went to China, learnt to speak Mandarin and worked as postdoc in Shanghai (ION) and Beijing Normal University in the neural information processing lab of Si WU. He continued to combine neural data analysis with mathematical modelling in the visual system (optical imaging, with data from Lu Haidong and Anna Roe) and further researched on the neural codes of perceptual learning (together with Wu Li).

Date: 16 August 2012

Time: 3pm – 4pm

Venue: Function Room 2, Level 4, Nanyang Executive Centre, Nanyang Technological University, Singapore

Address: 60 Nanyang View, Singapore 639673


The primary visual cortex (V1) is relatively well understood in principle but the correspondence of model predictions and multidimensional experimental data are often not tested directly in quantitative manner. In this talk, the speaker will give an overview over some recent studies in which experimental data from V1 were analyzed and attempts were made to relate neural recordings to theoretical models of V1.

First, the speaker will present analysis of electrophysiological recordings from monkeys, and show how the knowledge of local field potentials (LFPs) predict spiking activity in V1 using machine learning techniques, a question relevant for interpreting fMRI measurements. He will further discuss briefly the information content about a movie coded in spiking activity relative to the phase of slow LFP oscillations.

Then the speaker will ask how the current state-of-the-art network models of V1 relate to spike recordings in terms of their statistical structure. For that the speaker built a biophysical detailed V1 neural network model and compared the statistical structure of recordings and model responses using similar movie stimuli in experiment and model.

Finally, the speaker will show that at least in the case of average population responses to rigid random dot motion measured with optical imaging in V1 of anesthetized macaques a simple model of V1 is sufficient to predict measurements in a quantitative accurate way.