LIST
OF ABSTRACTS
Synchronization and Multistability in Rings of Coupled Oscillators
Author: Sue Ann Campbell
Abstract:
Certain neural network
structures can be considered as rings of
coupled (near)-identical oscillators. Our recent work has shown that
synchronization is ubiquitous in such models, but can be
fragile
due to the presence of other stable asynchronous oscillations
(multistability). We will discuss how the structure of the network,
regardless of the form of the oscillator model, influences the presence
of synchronization and multistability.
Session I : Principles of spatiotemporal pattern formation, Friday 04-15-2005
Motifs in Brain Networks: Building Blocks for a Cognitive Architecture
Author: Olaf Sporns
Abstract: Large-scale networks of mammalian cortex balance dual requirements for functional segregation and functional integration, by maintaining high local clustering of connections combined with short average path lengths. An analysis of the spectrum of structural motifs of such brain networks reveals characteristic patterns that may point to the importance of some specific motifs in cortical processing. The contribution of a catalogue of motifs to information integration in extended networks will be discussed.
Session I : Principles of spatiotemporal pattern formation, Friday 04-15-2005
Critical Branching Captures Activity in Living Neural Networks and Places Constraints on Local Network Topology
Author: John Beggs
Abstract: To investigate dynamics in local cortical networks, we record from both acute slices and organotypic cultures of rat cortex using 60 channel microelectrode arrays. These networks are typically quiescent for several seconds, and then experience a spontaneous burst of local field potential activity that drives the voltage levels at some electrodes above a threshold. Within 20 ms, the activity at these electrodes may spread to other electrodes before the network returns to quiescence and starts another burst. The total number of electrodes activated gives the size of the burst. Interestingly, the distribution of burst sizes recorded over 10 hrs follows a power law, similar to the distribution of cluster sizes seen at the critical point in a continuous phase transition or the number of toppled sites in critical avalanche models. In addition, activity in these cortical networks does not propagate in a random manner, but in preferred paths that are repeated significantly more than chance over several hours. We use a parsimonious branching process to capture both the power law distribution of burst sizes and the significantly repeating activity patterns. Remarkably, simulations suggest that a critical branching process optimizes both information transmission and information storage. However, this branching process also places several constraints on functional network topology: It suggests that at the local network level “super nodes” should not exist, and that highly active nodes should not be connected to each other. Experiments now underway may allow us to test these hypotheses.
Session
I : Principles of spatiotemporal
pattern formation, Friday
04-15-2005
Symbiotic Relationship Between Neuronal Dynamics and Architectures
Author: Michael Breakspear
Abstract: This talk is concerned with the relationship between spatiotemporal patterns displayed by ensembles of neuronal oscillators and the architectures that support them. We begin with a visual (qualitative) and wavelet-based (quantitative) analysis of stationary and moving structures that spontaneously arise in two dimensional sheets of neural subsystems with simple coupling matrices. Following recent studies by Gong & Van Leeuwan (2003), we then examine what happens when the dynamics are permitted to influence the underlying architecture through a Hebbian learning rule. Specifically, we study the conditions under which the architecture evolves naturally from a random to a small-world network. At the same time, the dynamics also change from simple chaotic to complex itinerant and bursting behaviour. Such a "symbiotic" structure-function relationship can be understood through an analysis of the system's Lyapunov exponents and local neighbourhood motifs. Studying such phenomena in biologically plausible models permits us to make tentative inferences about the role of spatiotemporal dynamics in the brain.
Session I : Principles of spatiotemporal pattern formation, Friday 04-15-2005
Dynamic Causal models of ERPs
Author: Karl Friston
Abstract: We
present a radically new
approach to modelling event-related
responses measured with EEG or MEG. This approach uses a biologically
informed model to enable inferences about the underlying neuronal
networks generating the responses. The approach can be regarded as a
neurobiologically constrained source reconstruction scheme, in which
the parameters of the reconstruction have an explicit neuronal
interpretation. Specifically, these parameters encode, among other
things, the coupling among sources and how that coupling depends upon
stimulus attributes or experimental context.
Neuronally plausible,
generative or forward models are essential for
understanding how event-related fields (ERFs) and potentials (ERPs) are
generated. In previous work, we simulated ERPs using a hierarchical
neural mass model that embodied bottom-up, top-down and lateral
connections among remote regions. Here, we describe a Bayesian
procedure to estimate the parameters of this model using empirical
data. We demonstrate this procedure by characterising the role of
changes in cortico-cortical coupling, in the genesis of ERPs. In the
first experiment, ERPs recorded during the perception of faces and
houses were modelled as distinct cortical sources in the ventral visual
pathway. Category-selectivity, as indexed by the face-selective N170,
could be accounted for bycategory-specific differences in
forward
connections from sensory to higher areas in the ventral stream. We were
able to quantify and make inferences about these effects
using
conditional estimates of connectivity. This allowed us to identify
where, in the processing stream, category-selectivity emerged.
In the second experiment we used an auditory oddball paradigm to show the P300 can be explained by changes in connectivity. Specifically, using Bayesian model selection, we assessed changes in backward connections, above and beyond changes in forward connections. In accord with theoretical predictions, there was strong evidence for learning-related changes in both forward and backward coupling. These examples show that category- or context-specific coupling among cortical regions can be assessed explicitly, within a mechanistic inference framework.
Session II : Measuring connectivity and dynamics, Friday 04-15-2005
Functional Connectivity and The
Hemodynamic Response
Authors: Peter A. Bandettini, Rasmus M. Birn, Jason Diamond, and Anthony Boemio
Abstract: Functional MRI (fMRI) has emerged in the last decade as a
powerful technique for noninvasively mapping human brain function.
Because fMRI is based on the detection of hemodynamic changes, the
temporal resolution, spatial resolution, and interpretability of the
technique is limited by how well neuronal activity is represented by
changes in blood flow, volume, and oxygenation. The hemodynamic
response can be roughly characterized as a low pass temporal and
spatial filter to neuronal activity. This lecture will begin with an
assessment of the filter characteristics of the hemodynamic response
and efforts at calibrating the signal such that the resolution and
interpretability may be improved. Second, the use of fMRI towards
assessing functional connectivity during "rest" and activation will be
reviewed. Third, strategies for using fMRI in conjunction with other
techniques including Magnetoencephalography, electroencephalography,
and diffusion tensor imaging will be outlined. Lastly, other potential
sources of contrast in fMRI, perhaps more closely related to neuronal
activity, will be discussed.
Session II : Measuring connectivity and dynamics, Friday 04-15-2005
Independence and Coordination of Macroscopic Brain Dynamics
Author:
Scott Makeig
Appreciating human subjects as active, self-motivated, mnemonically
impressioned, and affectively sensitive participants in dynamic brain
imaging experiments (rather than modeling them as passively stimulated
'responders' to experimental events and instructions) motivates the
view that dynamics of 'top-down' as well as 'bottom-up' brain processes
are of essential interest to cognitive neuroscience. These include
shifts of attention and performance strategy, search for associations
or insight,and active recall, fantasy and imagination -- processes in
which past associations and affective responses play important roles.
At the same time, more and more evidence reveals that major portions of
EEG, MEG and BOLD signal dynamics reflect brain processes whose time
courses are not simply or linearly related to the timing of stimulus
appearance or to block changes in task demand, but to 'top-down' brain
response and organizational processes.
Independent component analysis (ICA), a signal processing approach developed during the 1990s, is now being applied in many application domains including analysis of EEG, MEG, fMRI, PET and SPECT data. Its power lies in its underlying objective to separate and characterize maximally independent sources of information within high-dimensional data -- sources that contribute maximally distinct information, in a particular sense. to the recorded data. By contrast, standard methods of statistics and signal processing attempt to characterize sources of summed signal variance, e.g. to separate predicted (e.g., mean) data ('signals') features from unpredicted features ('noise').
Applied to whole fMRI BOLD data, ICA isolates brain regions with coherent signal fluctuations and maximally distinct spatial distributions, while applied to whole EEG or MEG data, ICA isolates maximally distinct signals arising in separate cortical domains exhibiting locally (partially) synchronous local field activity. This has the wholly desirable consequence of minimizing the effects of volume conduction on measu s of coherence and cross-talk between processes. Applying ICA to brain imaging data appears to offer an important opportunity to learn more about the macroscopic activities of and interactions between brain regions supporting the so-called 'higher' brain processes fundamental to human cognition and awareness. I will illustrate with some examples and evolving perspective.
Session II : Measuring connectivity and dynamics, Friday 04-15-2005
Connectivity, Distance and Scaling in Whole Brain Functional Networks
Authors: Ed Bullmore, Sophie Achard, Alle-Meije Wink, Raymond Salvador
Abstract:We
have used
classical multivariate methods applied to human, resting-state fMRI
data to explore the neurophysiological architecture of whole brain
functional networks, including an examination of the effects of
anatomical distance between regions on their functional connectivity,
and a characterisation of the "small world" properties of network
topology [1]. Our more recent work has used Fourier and wavelet methods
to investigate the relationships between connectivity (partial
coherence or partial correlation), anatomical distance, and frequency
or scale of the time series components subtending connectivity. We will
show that most long-range functional connections are bilaterally
symmetric or left-sided intrahemispheric and demonstrate different
frequency dependence or scaling properties compared to local
connections. Methodological and biological implications will be
discussed.
[1] Salvador R, Suckling J,
Coleman MR, Pickard JD, Menon DK,
Bullmore ET (2005) Neurophysiological architecture of functional
magnetic resonance images of human brainb. Cereb Cortex 5 Jan, [Epub
ahead of print]
Session II : Measuring connectivity and dynamics, Friday 04-15-2005
Assessing Directions of Neural Interactions with Granger Causality Spectra
Authors: Mingzhou Ding
Abstract: Commonly used interdependency measures such as cross correlation and spectral coherence do not yield directional information. Phase spectra may be used for that purpose only under very ideal conditions. Recent work has begun to explore the use of causal measures to further dissect the interaction patterns among neural signals. In this talk I will describe the concept of Granger Causality and introduce Geweke's causality spectra. The technique will then be applied to the analysis of multichannel local field potentials recorded from behaving monkeys performing sensorimotor and selective attention tasks.
Session III : Measuring connectivity and dynamics, Saturday 04-16-2005
Using Large-Scale Neural Models to Help Determine the Neural Substrates of Functional and Effective Connectivity
Author: Barry Horwitz
Abstract: The view that cognitive functions are mediated by networks of interacting brain regions now plays a central role in interpreting neuroscientific data, particularly that acquired using functional brain imaging. Numerous neuroimaging studies assess these interregional interactions using techniques that evaluate interregional functional and effective connectivity. The neurobiological substrates of functional and effective connectivity are, however, uncertain (Horwitz, NeuroImage, 2003).
We have constructed neurobiologically realistic models for visual and auditory object processing with multiple interconnected brain regions that perform delayed match-to-sample (DMS) tasks (Tagamets & Horwitz, Cerebral Cortex, 1998; Husain et al., NeuroImage, 2004). Here, we used these models to investigate the relations between neural activity and several measures of functional and effective connectivity evaluated between fMRI timeseries. We found that timeseries correlations between integrated synaptic activities between anterior temporal and prefrontal cortex were larger during the DMS task than during a control task, as one would have expected based on the underlying neural dynamics of the models. These results were less clear when the integrated synaptic activity was hemodynamically convolved to generate simulated fMRI activity.
We (Lee et al., submitted) also explored the validity of inferences made using Dynamic Causal Modeling (DCM) (Friston et al., NeuroImage, 2003) about the intrinsic connectivity structure and where bilinear modulatory effects act. A simple three-area model with linear hierarchical connectivity and a complex five-area model with feedback loops were examined using DCM and Bayesian model selection (Penny et al, NeuroImage, 2004). This approach revealed strong evidence for those models with correctly specified intrinsic connectivity. Furthermore, Bayesian model comparison favored those models where the specification of bilinear modulatory effects corresponded to their implementation in the large-scale neural model. Taken together, these results illustrate the usefulness of using biologically realistic neural models to evaluate fMRI-based network analysis methods.
Session III : Measuring connectivity and dynamics, Saturday 04-16-2005
Inter-hemispheric connectivity: what is the functional role of the corpus callosum?
Authors: Klaas Enno Stephan
Abstract: Understanding neural systems requires mathematical descriptions of their dynamics, and these must take into account the structural connectivity of the system. Knowing the structural connectivity alone, however, is not sufficient to predict functional principles. A good example of this is the corpus callosum: even though its connectivity is understood fairly well anatomically, we lack a precise and universally accepted theory of its functional role. Based on behavioural and neurophysiological studies, two candidate functions have been proposed, i.e. transfer of information and setting the cognitive processing mode, but decisive evidence is lacking.
In this talk, I will suggest how this issue can be investigated systematically, using a combination of fMRI and modelling techniques. Specifically, I will present dynamic causal models that describe inter-hemispheric effective connectivity in the visual system as a function of task demands and/or visual field. A range of competing models, derived from systematic combinations of experimental factors, embody different hypotheses about the functional role of callosal connections. The optimal model, and thus the most likely hypothesis, is determined by a Bayesian model selection technique. Our results show that the functional role of the corpus callosum differs between visual sub-systems: in the ventral stream, there is clear evidence that the corpus callosum subserves context-dependent information transfer whereas in the dorsal stream it appears to be more concerned with controlling the cognitive processing mode. Beyond the specific example presented, I will discuss to what extent the kind of model presented enables inferences about mechanistic processes at the neural level. I will briefly touch on some current nonlinear extensions of these models and discuss their potential clinical applications.
Session III : Measuring connectivity and dynamics, Saturday 04-16-2005
Imaging
Oscillatory Activity with MEG
Author: Richard M. Leahy
Abstract: We use magnetoencephalography to investigate event-related changes in oscillatory activity in cortical networks. Linear minimum norm inverse methods are combined with wavelet based time-frequency decomposition to map magnetic fields onto cerebral cortex. Since oscillations are typically not phase-locked to the stimulus, we compute average signal power over multiple epochs in time-frequency regions of interest. To detect significant changes in power in these time-frequency regions we compute a test statistic from matched pre- and post-stimulus regions [1]. We then use a permutation test to establish a threshold on this statistic that controls the family wise error rate over the entire cortical surface [2]. The resulting maps reveal regions of cerebral cortex in which there are significant changes in oscillatory activity as a function of both frequency and time. These regions can then be used for subsequent multivariate analysis to detect large scale cortical interactions.
[1] D. Pantazis, C. Dale, D. Weber, TE Nichols, GV Simpson, RM Leahy, Imaging of oscillatory behavior in event-related MEG studies, IS&T/SPIE's 17th Annual Symposium: Electronic Imaging 2005
[2] D. Pantazis, T. E Nichols, S. Baillet, RM Leahy, A Comparison of Random Field Theory and Permutation Methods for the Statistical Analysis of MEG data, Neuroimage, in press
Session III : Measuring
connectivity and dynamics,
Saturday
04-16-2005
Anatomical and Electrophysiological Insights into Seizure Propagation in the Human Brain
Authors: J.G. Milton, J.D. Hunter, S. A. Chkhenkeli and V. L. Towle
Abstract: Experimental and computational investigations demonstrate that self-maintained seizure propagation in populations of inter-connected excitatory neurons takes the form of traveling spiral waves. However, measurements of EEG coherence between pairs of subdural electrodes suggest that the propagation of seizures in intact brain has little semblance with wave propagation in excitable media. This observation is not unexpected. First, in brain there are a number of different pathways available for the spread of neural activity with differing conduction velocities. For example, anatomical considerations suggest that the fastest route for seizure generalization is not necessarily the most direct route: a fast route for seizure generalization involves spread from cortex to sub-cortical structures, such as the thalamus, then back to cortex via thalamocortical axons. In this case, rapid seizure generalization is the result of the combination of fast conduction along large diameter, myelinated axons that interconnect thalamus and cortex, and the diffuseness of thalamocortical projections. Second, the co-existence of inhibitory and excitatory mechanisms makes it possible for multiple distinct regions of incoherent seizure activity to exist simultaneously. Finally, clinical observations concerning the evolution of patients with medically intractable temporal lobe epilepsy suggest that this disease is dynamic entity which involves the formation of an epileptic system, i.e. a spatially distributed system of groups of neurons located in the cortical and subcortical regions of brain that cooperate to control the onset, propagation, and arrest of epileptic seizures. Determining the nature of propagation within these systems represents a major challenge for computational neuroscientists.
Session IV: Clinical
aspects of
anatomical and functional
connectivity,
Saturday
04-16-2005
Neural
Signatures of Behavioral Dynamics
Revealed Using Functional Magnetic Resonance Imaging
Author: Kelly Jantzen
Abstract:
A
central aim in behavioral
neuroscience is to understand the brain mechanisms of dynamic
coordinated actions fundamental to the formation of complex human
cognition and behavior. Coordination dynamics provides a
conceptual and theoretical framework to model, explore and understand
the dynamic creation and dissolution of stable coordinative states
expressed at the level of both brain and behavior. Within
this
framework, sensorimotor coordination provides an eloquent entry-point
for characterizing and modeling the dynamics of simple coordinated
action and for connecting dynamics of coordination at the behavioral
level with the evolution of neural activity at the level of the
cortex.
Relative phase has been
identified as a key order parameter, capable
of reducing the degrees of freedom at the macroscopic behavioral level
and providing a low dimensional description of complex coordinated
action. Stability, defined as variability in relative phase,
has
provided a critical metric for characterizing the ability of the
nervous system to maintain a complex behavioral pattern and in
predicting spontaneous switches between coordination
patterns. A
remaining critical question when investigating the neural basis of
coordination is whether the quantities of relative phase and
coordinative stability, derived using a non-linear dynamic systems
approach, are of importance to the nervous system in the formation of
coordinated movement patterns. In a broader context this begs
the
question, what information is critical to the brain for the control of
coordinated behaviors? How do brain networks interact to
generate
stable behavioral patterns and how are these patters subsequently
dissolved to allow for switching to new patterns?
We report on recent functional imaging data that supports the viewpoint
that dynamic features of coordinated action are represented within
broad cortical subcortical motor networks. We use a
parametric
approach to dissociate between neural networks that show global changes
in BOLD intensity with changes in (1) movement rate and (2) the
stability of the coordination pattern produced. .
Such
results suggest that a broadly distributed coupled network may mediate
intrinsic differences in coordinative stability expressed
behaviorally. Moreover, the results indicate that specific
brain
networks may be recruited in a functionally specific manner in response
the task demands defined by the relevant order parameter (relative
phase) and changes in the variability of this quantity
(stability).
Session IV: Clinical
aspects of
anatomical and functional
connectivity,
Saturday
04-16-2005
Neural Connectivity in the Lesioned Brain
Author:
Tomáš
Paus
Abstract: Our studies of neural connectivity in the human brain employ a combination of transcranial magnetic stimulation (TMS) with positron emission tomography (PET) or with electroencephalography (EEG); TMS is used to perturb neural activity in a region of interest while PET or EEG are used to measure region-specific consequences of such a perturbation (Paus 2002). In this talk, I will report results of our recent TMS/EEG and TMS/PET work carried out, respectively, in patients with Parkinson’s disease (PD) and patients with stroke. In the first study, we pursued our initial observations of TMS-elicited oscillations in the primary motor cortex (Paus et al. 2001) and asked whether a lesion of the thalamus, created surgically to treat tremor in PD patients, disrupts such oscillations in the ipsilateral hemisphere. In the second study, we used TMS/PET to map connectivity of the primary motor cortex ipsi- and contra-lateral to a subcortical stroke that occurred at least 12 months previously. We repeated the same TMS/PET protocol before and after two-weeks of constrain-induced therapy (Taub et al. 1993) in order to reveal possible neural mechanisms mediating therapy-induced improvements in motor functions.
Paus, T. Combination of
Transcranial Magnetic Stimulation with Brain
Imaging. In: J. Mazziotta, A. Toga (Eds). Brain Mapping: The Methods.
Second Edition Academic Press, pp. 691-705, 2002.
Paus T., Sipila P.K., Strafella A.P. Synchronization of neuronal
activity in the human sensori-motor cortex by transcranial magnetic
stimulation: a combined TMS/EEG study. Journal of Neurophysiology
86:1983-1990, 2001.
Taub E, Miller NE, Novack TA, Cook EW 3rd, Fleming WC, Nepomuceno CS,
Connell JS, Crago JE. Technique to improve chronic motor deficit after
stroke.
Arch Phys Med Rehabil. 74:347-54, 1993.
Session
IV: Clinical
aspects of
anatomical and functional
connectivity, Saturday
04-16-2005
Wave
Formation and Control in Brain
Author: Steven Schiff
Abstract: Plane waves in 1D and Spiral waves in 2D are a basic feature of excitable systems. We show in a 1D model and experiments in coronal cortical brain slices how electrical fields can modulate wave propagation. Using tangential slices of the middle layers of mammalian cortex, we demonstrate using voltage sensitive dye in this nearly isotropic 2D system a variety of wave types: ring, plane, spiral, and chaotic. Spiral waves occurred spontaneously and alternated with plane, ring, and irregular waves. A small ( 128 um) phase singularity occurred at the center of the spirals, about which were observed oscillations of widely distributed phases. The phase singularity drifted slowly across the tissue ( 1 mm/10 turns). We introduced a computational model of a cortical layer that predicted and replicated many of the features of these spirals. We speculate that rotating spiral waves may provide a spatial framework to organize cortical oscillations.
Huang, X, Troy, WC, Yang Q, Ma H, Laing, CR, Schiff, SJ, Wu J-Y, Spiral Waves in Disinhibited Mammalian Neocortex, Journal of Neuroscience, 24: 9897-9902, 2004.
Richardson, KA, Schiff, SJ, and Gluckman, BJ, Control of Traveling Waves in Mammalian Cortex, Physical Review Letters, 94, 028103, 2005
Schiff, S.J., Sauer, T., Kumar, R. Weinstein, S.L., Neuronal Spatiotemporal Pattern Discrimination: The Dynamical Evolution of Seizures. NeuroImage, in press 2005.
Session IV: Clinical aspects of anatomical and functional connectivity, Saturday 04-16-2005