Events

The anatomical and functional coupling between neurons and astrocytes a prominent type of glial cells of the cortex is an essential component of the brain physiology in health and disease. Astrocytes are fundamental in clearing glutamate from the extracellular space, preventing excitotoxicity by extracellular accumulation of this neurotransmitter. At the same time, astrocytes can also release glutamate into the extracellular space in an activity- dependent fashion, mediating neuromodulation. Although we may expect that both glutamatergic neuromodulation and glutamate uptake by astrocytes likely play a part in higher brain functions, the biophysical underpinnings accounting for this possibility remain elusive. Considering the case study of Alzheimer’s disease, I present experimental and modeling results pinpointing a bistable regulation of extracellular glutamate by astrocytes, in the early stages of the disease that prelude to cognitive impairment. Then, scaling up from the synaptic microenvironment to networks, I introduce theoretical arguments in support of the possibility that bistability could also ensue from astrocytic modulation of glutamatergic neurotransmission. In this context, a neuron-glial network model is developed to test the hypothesis that astrocytic neuromodulation could promote the emergence of persistent neural firing, with important implications for our current working memory framework. Finally, I conclude with an overview of some possible directions to complement and extend the above results, in the spirit of unifying computational approaches to harness the staggering complexity of the neuropil.

Discovering novel treatment strategies for complex chronic illnesses through traditional discovery pipelines is extremely expensive, carries a high probability of failure, and a lengthy cycle time. Furthermore, it is becoming clear that “one target, one treatment” solutions may not be capable of addressing difficult conditions. Repurposing Food and Drug Administration approved drugs offers a cost-effective solution with a significantly abbreviated timeline. Furthermore, combining multi-system modeling with these bioinformatics techniques can harness the regulatory dynamics of the human body to identify robust treatment courses that might produce lasting remission. Here it will be discussed how differentially expressed gene modules cross-referenced with drug atlas and pharmacogenomic databases can be used to identify targetable systems and agents. Based on these results it will be discussed how to construct a discrete ternary logic representation of signaling networks from physiological and biochemical literature to provide a qualitative description of multi-system behavior. By exploiting the regulatory dynamics of the resulting model through the application of a combinatorial optimization scheme and Monte Carlo simulation, it will be discussed how to predict treatment courses that might produce lasting disease remission. While the methods presented here can be applied generally, they will be discussed in the context of Myalgic Encephalomyelitis/Chronic Fatigue Syndrome and Gulf War Illness, debilitating chronic multi-symptom disorders for which there is no known treatment.

I will describe two recent studies on the use of strategies toward cognitively demanding tasks. One study used neural models to investigate the storage limitations of working memory. The other used graphical methods with functional magnetic resonance imaging to investigate sensorimotor adaptation. Both studies provide evidence that cognitive strategies are supported by the control of distributed neural dynamics, and that these dynamics explain group differences in task performance. I will discuss relationships between these and related tasks and methodologies, how they provide a foundation for investigating the neural bases of cognition, and their potential for clinical research on neurodegeneration and cognitive impairment.

Computational modeling of cognitive processes has largely focused on single cortical areas, but the engagement of multi-regional brain circuits in these processes is not completely understood. In the first part of my talk, I will present a circuit model to study the shared and unique roles of frontal and parietal cortices in a variety of cognitive tasks. We found that structural differences between these cortical areas map onto complementary dynamical regimes that subserve working memory and decision-making computations. Next, I consider the primate pulvinar, which is the largest part of the visual thalamus and is reciprocally connected to multiple visual and association cortical areas. I put forward a framework of pulvino-cortical interactions to clarify the pulvinar’s involvement in attention, confidence, and communication. I will also present a circuit model of thalamo-cortical interactions during motor planning, constrained by multisite recordings in the mouse. We propose that subcortical inputs to the thalamus selectively gate cortical ‘activity modes’ relevant for movement. Overall, the modeling results support the existence of computational principles for distributed and large-scale interactions in the brain that we are only beginning to uncover. At the end of the talk, I will suggest a roadmap towards ‘cognitive Deep Brain Stimulation’, which combines large-scale circuit modeling with a ‘virtual’ stimulation protocol to study the effects of exogenous stimulation on cognitive processes.

What does a quantitative theory of cortex entail? What are the computational principles that underlie cortical dynamics? Despite the fast pace of discoveries and progress in disparate domains of neuroscience, the lack of unifying principles and fundamental theories of the cortex is vividly apparent. The key shortcoming is that the inherent nature of the brain as a complex adaptive system and multiscale aspects of information processing in neuronal networks are mostly ignored or sacrificed to fit the reductionist approach. To develop a theory of cortical computation, one must address collective information processing and understand ensemble pattern formation at multiple scales. In search of a global theory of cortex, I explored several aspects of neuro-signals at multiple scales and conditions. These included the variability of oscillatory patterns, oscillatory entrainment of ensemble spiking, wave propagation, ensemble excitation/ inhibition balance, and the emergence of network disorder (seizure). The insights gleaned from these collective computational dynamics provide the foundation for a multiscale cortical quantitative theory of cortex that will guide us in the design of the next generation of neuro-inspired computational algorithms and biomedical devices.

There are two major theories of how gamma oscillations arise in the brain: the stochastic population oscillator theory and coupled oscillator theory. In the former, the relevant oscillator is the local circuit comprised of a population of individual neurons in the fluctuation driven regime, whereas in the latter it is an oscillator in the mean driven regime. In the latter, the intrinsic oscillatory properties of individual neurons matter, in the former they do not. An important aspect of the intrinsic dynamics is whether the mathematical bifurcation giving rise to spiking is a subcritical Hopf leading to an abrupt onset of spiking at a nonzero frequency (type 1), or a saddle node on an invariant circle that allows arbitrarily slow firing (type 1). We and others have shown that PV+ fast spiking interneurons in layers 2 and 3 of the entorhinal cortex have Hodgkin’s type 2 excitability, whereas others have shown that those area CA1 have type 1 excitability. Here we explain using phase plane techniques that in some ways coupled type 1 inhibitory neurons synchronize better with shunting inhibition, whereas type 2 synchronize better with hyperpolarizing inhibition. We also analyzed the tendency of populations of coupled interneurons to break into clusters using phase response curve theory. Finally, we examine the evidence that interneurons are in the oscillatory rather than the subthreshold regime during gamma oscillations.

Lecturer: Frances Skinner (Krembil)

Lecturer: Milad Lankarany (Krembil)

Dr. Uludag and his postdoc, Dr. Soheila Nazari, will be presenting in the Valiante Lab Meeting.

Lecturer: Steve Prescott (Sickkids)

Lecturer: Taufik Valiante (Krembil)

Lecturer: Jeremie Lefebvre (Ottawa, Krembil)

Speaker: Luka Milosevic

Lecturer: John Griffths (CAMH)

The anatomical and functional coupling between neurons and astrocytes a prominent type of glial cells of the cortex is an essential component of the brain physiology in health and disease. Astrocytes are fundamental in clearing glutamate from the extracellular space, preventing excitotoxicity by extracellular accumulation of this neurotransmitter. At the same time, astrocytes can also release glutamate into the extracellular space in an activity- dependent fashion, mediating neuromodulation. Although we may expect that both glutamatergic neuromodulation and glutamate uptake by astrocytes likely play a part in higher brain functions, the biophysical underpinnings accounting for this possibility remain elusive. Considering the case study of Alzheimer’s disease, I present experimental and modeling results pinpointing a bistable regulation of extracellular glutamate by astrocytes, in the early stages of the disease that prelude to cognitive impairment. Then, scaling up from the synaptic microenvironment to networks, I introduce theoretical arguments in support of the possibility that bistability could also ensue from astrocytic modulation of glutamatergic neurotransmission. In this context, a neuron-glial network model is developed to test the hypothesis that astrocytic neuromodulation could promote the emergence of persistent neural firing, with important implications for our current working memory framework. Finally, I conclude with an overview of some possible directions to complement and extend the above results, in the spirit of unifying computational approaches to harness the staggering complexity of the neuropil.

Discovering novel treatment strategies for complex chronic illnesses through traditional discovery pipelines is extremely expensive, carries a high probability of failure, and a lengthy cycle time. Furthermore, it is becoming clear that “one target, one treatment” solutions may not be capable of addressing difficult conditions. Repurposing Food and Drug Administration approved drugs offers a cost-effective solution with a significantly abbreviated timeline. Furthermore, combining multi-system modeling with these bioinformatics techniques can harness the regulatory dynamics of the human body to identify robust treatment courses that might produce lasting remission. Here it will be discussed how differentially expressed gene modules cross-referenced with drug atlas and pharmacogenomic databases can be used to identify targetable systems and agents. Based on these results it will be discussed how to construct a discrete ternary logic representation of signaling networks from physiological and biochemical literature to provide a qualitative description of multi-system behavior. By exploiting the regulatory dynamics of the resulting model through the application of a combinatorial optimization scheme and Monte Carlo simulation, it will be discussed how to predict treatment courses that might produce lasting disease remission. While the methods presented here can be applied generally, they will be discussed in the context of Myalgic Encephalomyelitis/Chronic Fatigue Syndrome and Gulf War Illness, debilitating chronic multi-symptom disorders for which there is no known treatment.

I will describe two recent studies on the use of strategies toward cognitively demanding tasks. One study used neural models to investigate the storage limitations of working memory. The other used graphical methods with functional magnetic resonance imaging to investigate sensorimotor adaptation. Both studies provide evidence that cognitive strategies are supported by the control of distributed neural dynamics, and that these dynamics explain group differences in task performance. I will discuss relationships between these and related tasks and methodologies, how they provide a foundation for investigating the neural bases of cognition, and their potential for clinical research on neurodegeneration and cognitive impairment.

Computational modeling of cognitive processes has largely focused on single cortical areas, but the engagement of multi-regional brain circuits in these processes is not completely understood. In the first part of my talk, I will present a circuit model to study the shared and unique roles of frontal and parietal cortices in a variety of cognitive tasks. We found that structural differences between these cortical areas map onto complementary dynamical regimes that subserve working memory and decision-making computations. Next, I consider the primate pulvinar, which is the largest part of the visual thalamus and is reciprocally connected to multiple visual and association cortical areas. I put forward a framework of pulvino-cortical interactions to clarify the pulvinar’s involvement in attention, confidence, and communication. I will also present a circuit model of thalamo-cortical interactions during motor planning, constrained by multisite recordings in the mouse. We propose that subcortical inputs to the thalamus selectively gate cortical ‘activity modes’ relevant for movement. Overall, the modeling results support the existence of computational principles for distributed and large-scale interactions in the brain that we are only beginning to uncover. At the end of the talk, I will suggest a roadmap towards ‘cognitive Deep Brain Stimulation’, which combines large-scale circuit modeling with a ‘virtual’ stimulation protocol to study the effects of exogenous stimulation on cognitive processes.

What does a quantitative theory of cortex entail? What are the computational principles that underlie cortical dynamics? Despite the fast pace of discoveries and progress in disparate domains of neuroscience, the lack of unifying principles and fundamental theories of the cortex is vividly apparent. The key shortcoming is that the inherent nature of the brain as a complex adaptive system and multiscale aspects of information processing in neuronal networks are mostly ignored or sacrificed to fit the reductionist approach. To develop a theory of cortical computation, one must address collective information processing and understand ensemble pattern formation at multiple scales. In search of a global theory of cortex, I explored several aspects of neuro-signals at multiple scales and conditions. These included the variability of oscillatory patterns, oscillatory entrainment of ensemble spiking, wave propagation, ensemble excitation/ inhibition balance, and the emergence of network disorder (seizure). The insights gleaned from these collective computational dynamics provide the foundation for a multiscale cortical quantitative theory of cortex that will guide us in the design of the next generation of neuro-inspired computational algorithms and biomedical devices.

There are two major theories of how gamma oscillations arise in the brain: the stochastic population oscillator theory and coupled oscillator theory. In the former, the relevant oscillator is the local circuit comprised of a population of individual neurons in the fluctuation driven regime, whereas in the latter it is an oscillator in the mean driven regime. In the latter, the intrinsic oscillatory properties of individual neurons matter, in the former they do not. An important aspect of the intrinsic dynamics is whether the mathematical bifurcation giving rise to spiking is a subcritical Hopf leading to an abrupt onset of spiking at a nonzero frequency (type 1), or a saddle node on an invariant circle that allows arbitrarily slow firing (type 1). We and others have shown that PV+ fast spiking interneurons in layers 2 and 3 of the entorhinal cortex have Hodgkin’s type 2 excitability, whereas others have shown that those area CA1 have type 1 excitability. Here we explain using phase plane techniques that in some ways coupled type 1 inhibitory neurons synchronize better with shunting inhibition, whereas type 2 synchronize better with hyperpolarizing inhibition. We also analyzed the tendency of populations of coupled interneurons to break into clusters using phase response curve theory. Finally, we examine the evidence that interneurons are in the oscillatory rather than the subthreshold regime during gamma oscillations.

Lecturer: Frances Skinner (Krembil)

Lecturer: Milad Lankarany (Krembil)

Dr. Uludag and his postdoc, Dr. Soheila Nazari, will be presenting in the Valiante Lab Meeting.

Lecturer: Steve Prescott (Sickkids)

Lecturer: Taufik Valiante (Krembil)

Lecturer: Jeremie Lefebvre (Ottawa, Krembil)

Speaker: Luka Milosevic

Lecturer: John Griffths (CAMH)