Correlating Excitation and Inhibition in Visual Cortical Circuits: Functional Consequences and Biological Feasibility (PhD, 2006-01 / 2009-05)
Thèse de Doctorat de l’Université d’Aix-Marseille II Ecole Doctorale des Sciences de la Vie et de la Santé Marseille, France
en Cotutelle avec
Fakultät für Biologie Albert-Ludwigs-Universität Freiburg im Breisgau, Allemagne
- The goal of the FACETS (Fast Analog Computing with Emergent Transient States) project was to create a theoretical and experimental foundation for the realisation of novel computing paradigms which exploit the concepts experimentally observed in biological nervous systems. The continuous interaction and scientific exchange between biological experiments, computer modelling and hardware emulations within the project provides a unique research infrastructure that will in turn provide an improved insight into the computing principles of the brain. This insight may potentially contribute to an improved understanding of mental disorders in the human brain and help to develop remedies.
The primary visual cortex (V1) is one of the most studied cortical area in neuroscience. Together with the retina and the lateral geniculate nucleus (LGN), it forms the early visual system, which has become a common model for studying computational principles in the sensory systems. Simple artificial stimuli (such as drifting gratings (DG)) have given precious insights into the neural basis of visual processing. However, recently more researchers have used more complex natural images (NI) visual stimuli, arguing that the low dimensional artificial stimuli are not sufficient for a complete understanding of the visual system. For example, whereas the responses of V1 neurons to DG are dense but with variable spike timings, the neurons are activated with only few and precise spikes to NI. Furthermore, if linear receptive field models provide a good fit to responses during simple stimuli, they often fail during NI.
To investigate the mechanisms behind the stimulus dependent responses of cortical neurons we have built a biophysical, yet simple and comprehensible, model of the early visual system. We show how the spatial and temporal stimulus properties interact with the model architecture to give rise to differential response behaviour. Our results show in particular that during NI, the LGN afferents show epochs of correlated activity. These temporal correlations are necessary to induce transient excitatory synaptic inputs, and result in precise spike timings in V1. Furthermore, the sparseness of the responses to NI can be explained by a hardwired, correlated and lagging inhibitory conductance, or conductance temporal window, which is induced by the interactions of the thalamocortical circuit with the spatiotemporal correlations in the stimulus.
We continue by investigating the origin of nonlinear responses during NI in the temporal window, by comparing models of different complexity. Our results suggest first that adaptive processes shape the responses, depending on the temporal properties of the stimuli. The different spatial properties can result in nonlinear inputs through the recurrent cortical network. We then study the functional consequences of correlated excitatory and inhibitory condutances in more details in general models. These results show that: (1) spiking of individual neurons becomes sparse and precise, (2) the selectivity of signal propagation increases and the detailed delay allows to gate the propagation through feed-forward structures (3) and recurrent cortical networks are more stable and more likely to elicit in vivo type activity states. Lastly our work illustrates new advances in methods of constructing and exchanging models of neuronal systems by the means of a simulator independent description language (called PyNN). We use this new tool to investigate the feasibility of comparing software simulations with neuromorphic hardware emulations. The presented work give new perspectives on the way conductances can be used for computations and it opens the door for more elaborated models of visual system's mechanisms.
Le cortex visuel primaire (V1) est l'aire corticale la plus étudiée en neurosciences. En effet, ce système complété de la rétine et du corps genouillé latéral constitue le système visuel de bas niveau et constitue une référence pour l'étude de modèles de systèmes sensoriels. Des stimuli simples comme des réseaux sinusoïdaux en mouvement (DG) ont donné des informations fondamentales sur les bases neurales du traitement neural de l'information visuelle. Cependant, de nombreux chercheurs utilisent des signaux plus complexes basés sur des images naturelles (NI) car des signaux de faibles complexité ne sont pas pertinents pour une vision complète du système visuel. Par exemple, alors que les réponses des neurones de V1 sont denses et imprécises pour des réseaux (DG), elles sont parcimonieuses et de grande résolution temporelles pour des scènes naturelles (NI). De plus, le modèle d'un champ récepteur d'intégration linéaire décrit bien la réponse à ces premiers stimuli mais est en échec pour une réponses aux images naturelles.
Pour comprendre ces mécanismes corticaux dépendants du stimulus, nous avons construit un modèle biophysique simple et réaliste du système visuel de bas niveau. Nous montrons de cette façon comment les propriétés spatio-temporelles du stimulus interagissent au niveau de la structure du modèle afin de donner ces réponse différenciées. Nos résultats montrent en particulier que durant la stimulation NI, les afférents thalamiques montrent des phases d'activité corrélée. Ces corrélations temporelles sont nécessaires pour générer dans V1 une réponse synaptique excitatrice phasique qui cause une réponse temporelle précise. En particulier, la parcimonie de la réponse peut être expliquée par une phase inhibitrice corrélée et légèrement retardée, ou fenêtre temporelle de conductance, induite par un circuit thalamocortical spécialisé en interaction avec l'activité spatio-temporelle corrélée du stimulus entrant.
Nous poursuivons en étudiant l'origine des réponses non-linéaires observées pour les images naturelles en comparant des modèles de complexités croissantes. Nos résultats suggèrent premièrement que des processus adaptatifs modèlent le stimulus en fonction des propriétés temporelles du stimulus. Le propriétés spatiales peuvent ainsi générer des effets non-linéaires amplifiés par l'intermédiaire du réseau cortical récurrent que nous modélisons. Nous étudions alors les conséquences fonctionnelles de la phase corrélée des conductances excitatrices et inhibitrices dans des modèles génériques. Nous montrons que: (1) des neurones individuels deviennent plus parcimonieux et précis, (2) la sélectivité de la propagation de l'information dans une structure de type "en-avant" peut être contrôlée finement grâce au délai dans la fenêtre temporelle. (3) La réponse d'un modèle de réseau cortical récurrent est plus robuste et est compatible avec les états corticaux observés in vivo.
En complément, ce travail illustre des avancées méthodologiques pour construire et échanger des modèles neuraux grâce à un langage de description indépendant de l'architecture appelé PyNN. Nous utilisons cet outil pour développer ces modèles sur différentes solutions logicielles mais aussi sur des circuits intégrés neuromorphiques. En conclusion, ce travail ouvre des perspectives sur le rôle computationnel générique des conductances neurales et en particulier pour la mise en place de modèle plus élaborés pour comprendre les mécanismes de la vision.
List of publications as a first author
- Jens Kremkow, Laurent U Perrinet, Cyril Monier, Jose-Manuel Alonso, Ad Aertsen, Yves Fregnac, Guillaume S Masson. Push-pull receptive field organization and synaptic depression: Mechanisms for reliably encoding naturalistic stimuli in V1, URL URL2 URL3 . Frontiers in Neural Circuits, 2016 abstractNeurons in the primary visual cortex are known for responding vigorously but with high variability to classical stimuli such asdrifting bars or gratings. By contrast, natural scenes are encoded more efficiently by sparse and temporal precise spiking responses. We used a conductance-based model of the visual system in higher mammals to investigate how two specific features of the thalamo-cortical pathway, namely push-pull receptive field organization and synaptic depression, can contribute to this contextual reshaping of V1 responses. By comparing cortical dynamics evoked respectively by natural vs. artificial stimuli in a comprehensive parametric space analysis, we demonstrate that the reliability and sparseness of the spiking responses during natural vision is not a mere consequence of the increased bandwidth in the sensory input spectrum. Rather, it results from the combined impacts of synaptic depression and push-pull inhibition, the later acting for natural scenes as a form of “effective” feed-forward inhibition as demonstrated in other sensory systems. Thus, the combination of feedforward-like inhibition with fast thalamo-cortical synaptic depression by simple cells receiving a direct structured input from thalamus composes a generic computational mechanism for generating a sparse and reliable encoding of natural sensory events.
- Jens Kremkow, Laurent U. Perrinet, Guillaume S. Masson, Ad Aertsen. Functional consequences of correlated excitatory and inhibitory conductances in cortical networks, URL . Journal of Computational Neuroscience, 28(3):579-94, 2010 abstractNeurons in the neocortex receive a large number of excitatory and inhibitory synaptic inputs. Excitation and inhibition dynamically balance each other, with inhibition lagging excitation by only few milliseconds. To characterize the functional consequences of such correlated excitation and inhibition, we studied models in which this correlation structure is induced by feedforward inhibition (FFI). Simple circuits show that an effective FFI changes the integrative behavior of neurons such that only synchronous inputs can elicit spikes, causing the responses to be sparse and precise. Further, effective FFI increases the selectivity for propagation of synchrony through a feedforward network, thereby increasing the stability to background activity. Last, we show that recurrent random networks with effective inhibition are more likely to exhibit dynamical network activity states as have been observed in vivo. Thus, when a feedforward signal path is embedded in such recurrent network, the stabilizing effect of effective inhibition creates an suitable substrate for signal propagation. In conclusion, correlated excitation and inhibition support the notion that synchronous spiking may be important for cortical processing..
- Jens Kremkow. Correlating Excitation and Inhibition in Visual Cortical Circuits: Functional Consequences and Biological Feasibility, URL . 2009 abstractThe primary visual cortex (V1) is one of the most studied cortical area in the brain. Together with the retina and the lateral geniculate nucleus (LGN) it forms the early visual system. Artificial stimuli (i.e. drifting gratings (DG)) have given insights into the neural basis of visual processing. However, recently researchers have started to use more complex natural visual stimuli (NI), arguing that the low dimensional artificial stimuli are not sufficient for a complete understanding of the visual system.For example, whereas the responses of V1 neurons to DG are dense but with variable spike timings, the neurons respond with only few but precise spikes to NI. Furthermore, linear receptive field models provide a good fit to responses during simple stimuli, however, they often fail during NI. To investigate the mechanisms behind the stimulus dependent responses of cortical neurons we have built a biophysical model of the early visual system.Our results show that during NI the LGN afferents show epochs of correlated activity, resulting in precise spike timings in V1. The sparseness of the responses to NI can be explained by correlated inhibitory conductance. We continue by investigating the origin of stimulus dependent nonlinear responses, by comparing models of different complexity. Our results suggest that adaptive processes shape the responses, depending on the temporal properties of the stimuli. Lastly we study the functional consequences of correlated excitatory and inhibitory condutances in more details in generic models.The presented work gives new perspectives on the processing of the early visual system, in particular on the importance of correlated conductances..
- Jens Kremkow, Laurent Perrinet, Guillaume S. Masson, Ad Aertsen. Functional consequences of correlated excitation and inhibition on single neuron integration and signal propagation through synfire chains, URL . In Eighth Göttingen Meeting of the German Neuroscience Society, pages T26-6B. 2009 abstractNeurons receive a large number of excitatory and inhibitory synaptic inputs whose temporal interplay determines their spiking behavior. On average, excitation (Gexc) and inhibition (Ginh) balance each other, such that spikes are elicited by fluctuations . In addition, it has been shown in vivo that Gexc and Ginh are correlated, with Ginh lagging Gexc only by few milliseconds (6ms), creating a small temporal integration window [2,3]. This correlation structure could be induced by feed-forward inhibition (FFI), which has been shown to be present at many sites in the central nervous system.To characterize the functional consequences of the FFI, we first modeled a simple circuit using spiking neurons with conductance based synapses and studied the effect on the single neuron integration. We then coupled many of such circuits to construct a feed-forward network (synfire chain [4,5]) and investigated the effect of FFI on signal propagation along such feed-forward network.We found that the small temporal integration window, induced by the FFI, changes the integrative properties of the neuron. Only transient stimuli could produce a response when the FFI was active whereas without FFI the neuron responded to both steady and transient stimuli. Due to the increase in selectivity to transient inputs, the conditions of signal propagation through the feed-forward network changed as well. Whereas synchronous inputs could reliable propagate, high asynchronous input rates, which are known to induce synfire activity , failed to do so. In summary, the FFI increased the stability of the synfire chain.Supported by DFG SFB 780, EU-15879-FACETS, BMBF 01GQ0420 to BCCN Freiburg Kumar A., Schrader S., Aertsen A. and Rotter S. (2008). The high-conductance state of cortical networks. Neural Computation, 20(1):1--43.  Okun M. and Lampl I. (2008). Instantaneous correlation of excitation and inhibition during ongoing and sensory-evoked activities. Nat Neurosci, 11(5):535--7. Baudot P., Levy M., Marre O., Monier C. and Fr\'egnac (2008). submitted.  Abeles M. (1991). Corticonics: Neural circuits of the cerebral cortex. Cambridge, UK  Diesmann M., Gewaltig M-O and Aertsen A. (1999). Stable propagation of synchronous spiking in cortical neural networks. Nature, 402(6761):529--33.  Kumar A., Rotter S. and Aertsen A. (2008), Conditions for propagating synchronous spiking and asynchronous firing rates in a cortical network model. J Neurosci 28 (20), 5268--80.,.
- Jens Kremkow, Laurent Perrinet, Alexandre Reynaud, Ad Aertsen, Guillaume S. Masson, Frédéric Chavane. Dynamics of non-linear cortico-cortical interactions during motion integration in early visual cortex: A spiking neuron model of an optical imaging study in the awake monkey, URL URL2 . In Eighteenth Annual Computational Neuroscience Meeting: CNS*2009 Berlin, Germany. 18–23 July 2009, pages 10(Suppl 1):P176. 2009 abstract
- Jens Kremkow, Laurent Perrinet, Pierre Baudot, Manu Levy, Olivier Marre, Cyril Monier, Yves Fregnac, Guillaume Masson, Ad Aertsen. Control of the temporal interplay between excitation and inhibition by the statistics of visual input: a V1 network modelling study, URL . In Proceedings of the Society for Neuroscience conference, 2008 abstractIn the primary visual cortex (V1), single cell responses to simple visual stimuli (gratings) are usually dense but with a high trial-by-trial variability. In contrast, when exposed to full field natural scenes, the firing patterns of these neurons are sparse but highly reproducible over trials (Marre et al., 2005; Fr\'egnac et al., 2006). It is still not understood how these two classes of stimuli can elicit these two distinct firing behaviours. A common model for simple-cell computation in layer 4 is the ``push-pull'' circuitry (Troyer et al. 1998). It accounts for the observed anti-phase behaviour between excitatory and inhibitory conductances in response to a drifting grating (Anderson et al., 2000; Monier et al., 2008), creating a wide temporal integration window during which excitation is integrated without the shunting or opponent effect of inhibition and allowed to elicit multiple spikes. This is in contrast to recent results from intracellular recordings in vivo during presentation of natural scenes (Baudot et al., submitted). Here the excitatory and inhibitory conductances were highly correlated, with inhibition lagging excitation only by few milliseconds (~6 ms). This small lag creates a narrow temporal integration window such that only synchronized excitatory inputs can elicit a spike, similar to parallel observations in other cortical sensory areas (Wehr and Zador, 2003; Okun and Lampl, 2008). To investigate the cellular and network mechanisms underlying these two different correlation structures, we constructed a realistic model of the V1 network using spiking neurons with conductance based synapses. We calibrated our model to fit the irregular ongoing activity pattern as well as in vivo conductance measurements during drifting grating stimulation and then extracted predicted responses to natural scenes seen through eye-movements. Our simulations reproduced the above described experimental observation, together with anti-phase behaviour between excitation and inhibition during gratings and phase lagged activation during natural scenes. In conclusion, the same cortical network that shows dense and variable responses to gratings exhibits sparse and precise spiking to natural scenes. Work is under way to show to which extent this feature is specific for the feedforward vs recurrent nature of the modelled circuit. ,.
- Jens Kremkow, Laurent U. Perrinet, Ad Aertsen, Guillaume S. Masson. Functional properties of feed-forward inhibition, URL . In Proceedings of the second french conference on Computational Neuroscience, Marseille, 2008 abstract
- Jens Kremkow, Laurent Perrinet, Arvind Kumar, Ad Aertsen, Guillaume Masson. Synchrony in thalamic inputs enhances propagation of activity through cortical layers, URL URL2 . In Sixteenth Annual Computational Neuroscience Meeting: CNS*2007, Toronto, Canada. 7--12 July 2007, 2007 abstractSensory input enters the cortex via the thalamocortical (TC) projection, where it elicits large postsynaptic potentials in layer 4 neurons . Interestingly, the TC connections account for only 15% of synapses onto these neurons. It has been therefore controversially discussed how thalamic input can drive the cortex. Strong TC synapses have been one suggestion to ensure the strength of the TC projection ("strong-synapse model"). Another possibility is that the excitation from single thalamic fibers are weak but get amplified by recurrent excitatory feedback in layer 4 ("amplifier model"). Bruno and Sakmann  recently provided new evidence that individual TC synapses in vivo are weak and only produce small excitatory postsynaptic potentials. However, they suggested that thalamic input can activate the cortex due to the synchronous firing and that cortical amplification is not required. This would support the "synchrony model" proposed by correlation analysis .Here, we studied the effect of correlation in the TC input, with weak synapses, to the responses of a layered cortical network model. The connectivity of the layered network was taken from Binzegger et al. 2004 . The network was simulated using NEST  with the Python interface PyNN  to enable interoperability with different simulators. The sensory input to layer 4 was modelled by a simple retino-geniculate model of the transformation of light into spike trains , which was implemented by leaky integrate-and-fire model neurons.We found that introducing correlation into TC inputs enhanced the likelihood to produce responses in layer 4 and improved the activity propagation across layers. In addition, we compared the response of the cortical network to different noise conditions and obtained contrast response functions which were in accordance with neurophysiological observations. This Work is supported by the 6th RFP of the EU (grant no. 15879-FACETS) and by the BMBF grant 01GQ0420 to the BCCN Freiburg.1. Chung S, Ferster D: Strength and orientation tuning of the thalamic input to simple cells revealed by electrically evoked cortical suppression. Neuron 1998, 20:1177-1189. 2. Bruno M, Sakmann B: Cortex is driven by weak but synchronously active thalamocortical synpases. Science 2006, 312:1622-1627. 3. Alonso JM, Usrey WM, Reid RC: Precisely correlated firing in cells of the lateral geniculate nucleus. Nature 1996, 383:815-819. 4. Binzegger T, Douglas RJ, Martin KAC: A quantitative map of the circuit of the cat primary visual cortex. J Neurosci 2004, 24:8441-8453. 5. NEST http://www.nest-initiative.org6. PyNN http://neuralensemble.org/PyNN7. Gazeres N, Borg-Graham LJ, Fr\'egnac Y: A phenomenological model of visually evoked spike trains in cat geniculate nonlagged X-cells.Vis Neurosci 1998, 15:1157-1174..