link to homepage

Institute of Neuroscience and Medicine

Navigation and service


Morning Session

8:55 - 9:00introduction
9:00 - 9:35Xiao-Jing Wang, New York University
9:35 - 10:10Martin Giese, University Clinic Tübingen
10:10 - 10:40 morning coffee
10:40 - 11:15Stefan Mihalas, Allen Institute for Brain Science
11:15 - 11:50Steven Bressler, Florida Atlantic University
11:50 - 13:30 lunch

Afternoon Session

13:30 - 14:05Joana Cabral, University of Oxford
14:05 - 14:40Paula Sanz-Leon, The University of Sydney
14:40 - 15:10afternoon coffee
15:10 - 15:45Maximilian Schmidt, Jülich Research Centre
15:45 - 16:20Matthieu Gilson, Pompeu Fabra University
16:20 - 16:55André M Bastos, MIT


What does it mean to build a  large-scale brain circuit model?

Xiao-Jing Wang

Center for Neural Science, New York University, New York and New York University East China Normal University Joint Institute of Brain and Cognitive Science, New York University, Shanghai, China


Neural model for visual action recognition and its interaction with motor representations

Martin Giese

Section Computational Sensomotorics, CIN & HIH, Department of Cognitive Neurology, University Clinic Tübingen, Germany

Action perception and execution are intrinsically linked in the human brain. Consequently, visual action representation involves a whole spectrum of cortical functions, ranging from the visual processing from shape and motion, the processing of spatial relationships and action semantics, and the interaction of visual representation with neural representations of motor programs. We present a neural theory that has been developed in close connection to neural and behavioral data, and which unifies a number of observations in visual action recognition, and its interaction with motor representations. The core of the model is a physiologically-inspired neural hierarchy (deep architecture) that mimics the properties of the visual pathway, which is combined with a neural mass / field model for the representation of sequences. This model is extended in different ways to account for experimental data on action recognition and its interaction with motor execution. For the processing of goal-directed actions, the hierarchy has to be extended by special mechanisms that process the spatial relationship between effectors and objects. In order to account for the experimentally observed interaction between action perception and motor execution the model has to be extended by neural representations for motor programs and their dynamic interaction with the visual representation. It is shown that such extended models can account in a unifying manner for a number of experimental results in visual action processing.

Supported by EC Fp7-PEOPLE-2011-ITN PITN-GA-011-290011 (ABC), FP7-ICT-2013-FET-F/604102 (HBP), FP7-ICT-2013-10/611909 (Koroibot), BMBF, FKZ: 01GQ1002A, DFG GI 305/4-1 + KA 1258/15-1.


Cortical circuits implementing optimal context integration

Stefan Mihalas

Allen Institute for Brain Science, Seattle

Neurons in the primary visual cortex (V1) predominantly respond to a patch of the visual input, their classical receptive field. These responses are modulated by the visual input in the surround. This reflects the fact that features in natural scenes do not occur in isolation: lines, surfaces are generally continuous. There is information about a visual patch in its surround. This information is assumed to be passed to a neuron in V1 by neighboring neurons via lateral connections. The relation between visual evoked responses and lateral connectivity has been recently measured in mouse V1.

In this study we combine these three topics: natural scene statistics, mouse V1 neuron responses and their connectivity. We are interested in addressing the question: Given a set of natural scene statistics, what lateral connections would optimally integrate the cues from the classical receptive field with those from the surround?

First, we assumed a neural code: the firing rate of the neuron maps bijectively to the probability of the feature the neuron is representing to be in the presented image. We generate a database of features these neurons represent by constructing a parameterized set of models from V1 electrophysiological responses. We used the Berkeley Segmentation Dataset to compute the probabilities of co-occurrences of these features. We computed the relation between probabilities of feature co-occurrences and the synaptic weight which optimally integrates these features. The relation between evoked responses and connectivity which leads to optimal context integration is qualitatively similar to the measured one, but several additional predictions are made. We integrate this connectivity in existing models of cortical columns and we make additional physiological predictions. We hypothesize that this computation: optimal cue integration is a general property of cortical circuits, and the rules constructed for mouse V1 generalize for other areas and species.


The wave packet in multi-area cortical modeling

Steven Bressler

Cognitive Neurodynamics Laboratory Center for Complex Systems & Brain Sciences Florida Atlantic University

Extensive experimental evidence on the olfactory spatial amplitude modulation (SAM) pattern led Freeman (2003) to propose the wave packet as a fundamental unit of cortical function at the level of neuronal populations. Based on Freeman’s wave packet theory, I proposed (Bressler 2007) a model in which global neurocognitive state emerges in the cerebral cortex as a result of reentrant interactions among cortical areas (Seth et al. 2004). In the model, each area generates wave packets of high-­‐frequency oscillatory activity, with SAMs that represent the area’s local assessment of its own current state. Sets of cortical areas mutually constrain their constituent wave packets through their dynamic interaction. These interactions lead to the emergence of distributed wave packets that have cognitively consistent SAM patterns. This process creates in toto a neurocognitive state that is globally unified across the cortex.

Bressler, S.L. (2007) The formation of global neurocognitive state. In: LI Perlovsky, R Kozma (Eds.) Neurodynamics of Higher-­‐Level Cognition and Consciousness, Springer, New York, 2007, pp. 61-­‐72.
Freeman, W.J. (2003) The wave packet: An action potential for the 21st century. J. Int. Neurosci 2:3-­‐30.
Seth, A.K., McKinstry, J.L., Edelman, G.M., Krichmar, J.L. (2004) Visual binding through reentrant connectivity and dynamic synchronization in a brain-­‐based device. Cereb. Cortex 14:1185-­‐1199.


Single or multi-frequency generators in on-going MEG data

Joana Cabral

Department of Psychiatry, University of Oxford

Envelopes of band-limited ongoing MEG signals co-vary in consistent patterns across the whole brain and have been attracting a growing body of research. While the genesis of such envelopes remains under debate, we consider two distinct model scenarios: a classical one where each brain area can generate oscillations in one single frequency, and a novel one where each brain area generates oscillations in multiple frequency bands. The models share, as a common generator of damped oscillations, the normal form of a Hopf bifurcation operating at the critical border between the asynchronous state and the oscillatory regime. The envelopes of the simulated signals are compared with empirical MEG data using new methods to analise the envelope dynamics in terms of their phase coherence and stability across the spectrum of carrier frequencies.

Considering the whole-brain model with a single frequency generator in each brain area, we obtain the best fit with the empirical MEG data when the fundamental frequency is tuned at 12Hz. However, when multiple frequency generators are placed at each local brain area, we find the model largely outperforms the single one, leading to an improved fit of the spatio-temporal structure of on-going MEG data across all frequency bands.


An overview of numerical tools for simulating brain dynamics

Paula Sanz-Leon

Complex Systems Group, School of Physics, The University of Sydney

Mesoscopic-scale models such as neural masses and neural fields are widely used to study spatiotemporal dynamics of neural tissue. When analytic methods become intractable, the use of numerical simulations is essential to get a deeper understanding of brain activity. Currently, there are tools available to simulate from one single area (or tissue), to a few interconnected areas to the whole brain. In this talk I will give an overview of three different software tools (The Virtual Brain, Neural Field Simulator and Neurofield) followed by a brief discussion about their similarities, advantages and limitations.


A multi-scale spiking network model of macaque visual cortex

Maximilian Schmidt

Institute of Neuroscience and Medicine (INM-6) and Institute for Advanced Simulation (IAS-6), and JARA BRAIN Institute I, Jülich Research Centre, Jülich, Germany

I present a multi-scale spiking network model of all vision-related areas of macaque cortex that represents each area by a full-scale microcircuit with area-specific architecture based on a model of early sensory cortex [1]. The layer- and population-resolved network connectivity integrates axonal tracing data from the CoCoMac database with recent quantitative tracing data, and is systematically refined using dynamical constraints [2]. Gaps in the data are bridged by exploiting regularities of cortical structure such as the exponential decay of connection densities with inter-areal distance and a fit of laminar patterns versus logarithmized ratios of neuron densities. The resulting connectivity is refined within the error bounds of the experimental data using a mean-field approach that enables improving the attractors of the network while retaining their global stability.

Simulations reveal a stable asynchronous irregular ground state with heterogeneous activity across areas, layers, and populations. In the presence of large-scale interactions, the model reproduces longer intrinsic time scales in higher compared to early visual areas, similar to experimental findings [3]. Activity propagates preferentially in the feedback direction, mimicking experimental results associated with visual imagery [4]. Cortico-cortical interaction patterns agree well with fMRI resting-state functional connectivity [5]. The model bridges the gap between local and large-scale accounts of cortex, and clarifies how the detailed connectivity of cortex shapes its dynamics on multiple scales.

VSR computation time grant JINB33, Helmholtz Portfolio SMHB, EU Grant 269921 (BrainScaleS), EU Grant 604102 (Human Brain Project, HBP), SFB936 /A1, Z1 and TRR 169 /A2.

[1] Potjans TC, and Diesmann M: The cell-type specific cortical microcircuit: relating structure and activity in a full-scale spiking network model. Cereb Cortex 2014, 24:785-806.
[2] Schuecker J, Schmidt M, van Albada SJ, Diesmann M, Helias M: Fundamental activity constraints lead to specific interpretations of the connectome, arXiv preprint 2015, arXiv:1509.03162
[3] Murray JD, Bernacchia A, Freedman DJ, Romo R, Wallis JD, Cai X, Padoa-Schioppa C, Pasternak T, Seo H, Lee D, Wang X-J: A hierarchy of intrinsic timescales across primate cortex. Nat Neurosci 2014, 17:1661–3.
[4] Dentico D, Cheung BL, Chang J-Y, Guokas J, Boly M, Tononi G, Van Veen B: Reversal of cortical information flow during visual imagery as compared to visual perception. Neuroimage 2014, 100:237–243.
[5] Shen K, Bezgin G, Hutchison RM, Gati JS, Menon RS, Everling S, McIntosh AR: Information processing architecture of functionally defined clusters in the macaque cortex. J Neurosci 2012, 32:17465–76.


Matthieu Gilson

Center for Brain Cognition, Dept of Technology and Information, Universitat Pompeu Fabra, Barcelona

Human behavior entails a flexible and context-sensitive interplay between brain areas to integrate information according to goal-directed requirements. Nevertheless, the neural principles governing the integration and entrainment of functionally specialized brain areas remains poorly understood. Here, we introduce a theoretical framework that uncovers the interactions between cortical regions to process information when performing a task. We use a dynamic model of the whole cortex to reproduce functional connectivity (FC). We find that (co)variances of the BOLD signals convey information that allows for the discrimination between two tasks: 1) watching a black screen in silence and 2) watching a movie while listening to the soundtrack. More precisely, the model estimates both the inputs to the network and its connectivity by extracting spatio-temporal information from the FC. This allows for the analysis of information flow between cortical regions and the underlying mechanistic changes between the two conditions. Our results suggest that changes in the FC variances are mainly due to the inputs rather than the cortico-cortical connectivity. Nevertheless, estimated changes in the connectivity appear essential to counterbalance the increase of activity due to inputs in the movie condition, as compared to the black-screen condition. Our results suggest a dynamic reconfiguration of the cortical interactions to select pathways for information processing through a specific hierarchy of cortical regions.


The relation between anatomical connection strength and inter-areal functional connectivity through rhythmic synchronization

Andre Bastos

Bastos A.M.1,2, Vezoli J.1, Lewis C.1, Bosman C.A. 3,4, Kennedy H. 5, Fries P. 1,3
1 Ernst Strüngmann Institute (ESI) for Neuroscience in Cooperation with Max Planck Society, Deutschordenstraße. 46, 60528 Frankfurt, Germany;

2 Picower Institute for Learning and Memory, MIT, Cambridge, MA, 02139, USA;

3 Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Kapittelweg 29, 6535 EN Nijmegen, Netherlands;

4 Swammerdam Institute for Life Sciences, Center for Neuroscience, Faculty of Science, University of Amsterdam, Sciencepark 904, 1098 XH Amsterdam, Netherlands;

5 Inserm U846, Stem Cell and Brain Research Institute, 18 avenue du Doyen Lépine,69500 Bron, France; Université de Lyon, Lyon1, UMR-S 846, 69003 Lyon, France

The relationship between the strength of rhythmic neuronal synchronization and sensory stimulation, attention, or other cognitive factors is well established. Yet, the degree to which neuronal synchronization between brain areas depends on the strength of their anatomical connectivity has not been studied systematically. We obtained quantitative measures of anatomical connection strength for many pairs of cortical areas in the macaque. In two macaque monkeys, we recorded with 252 channel electrocorticographic grids from the left hemisphere, while animals performed a visual attention task. We determined frequency-resolved coherence, Granger-causality and power-power correlations for 78 area pairs. This shows that across pairs of cortical areas, inter-areal functional connectivity is correlated with inter-areal anatomical connection strength. This correlation holds when it is partialized for distance, either physical distance, distance along the dural surface (relevant for potential volume conduction) or distance through the white matter. The functional-anatomical correlation was highly significant for theta, beta and gamma bands, which were prominent in this dataset. When the beta and gamma networks were investigated separately, this revealed distinct topographical differences. Consistent with the recent insight that gamma (beta) predominates in feedforward (feedback) signaling, we found the gamma network to be strongest among early visual areas, and the beta network among fronto-parietal areas. This is likely related to laminar differences in the origin of the respective anatomical projections. In summary, while inter-areal rhythmic neuronal synchronization can be strongly modulated by stimuli and task parameters, it is also structured by the backbone of cortico-cortical connections.