SOCIETY FOR AUTONOMOUS NEURODYNAMICS (SAND)
PRINCIPLES OF AUTONOMOUS NEURODYNAMICS 2004
Friday August 13, 2004
AN INTRODUCTION TO VOLUNTARY BEHAVIOR
Living molecules reverse entropy, in the sense that they maintain their structure in a universe where complex structures tend to break down. To maintain their structure, living molecules interact with their surrounding environment - the source of energy and structural materials. In multicellular animals, this interaction has evolved into behavior, which is usually defined as "the observable action of muscles and glands".
Generally speaking, three levels of behavior are recognized, the more complex having developed out of the simpler, and all three co-active in higher animals. The simplest behaviors are reflexes. These occur whether the animal is awake (conscious) or asleep, and involve no motivation. They are not based on learning. The next level of behavior is instinctive patterns. These occur only when the animal is awake (conscious) and they are motivated. They are often based on early learning. The most complex behaviors are purposive or goal-oriented behaviors. These occur only when the animal is awake (conscious) and they are motivated. They are based on a great deal of learning, and an internalized world model.
Instinctive patterns and purposive behaviors are often referred to as "voluntary behaviors". They represent the two ends of the spectrum of motivated behavior. In higher animals, activity in some of the circuits involved in motivated behaviors is referred to as "consciousness."
NEURONAL ACTIVITY AND THE CONTROL OF LIMB MOVEMENT
Body movement is not like that of an electromagnetic motor. Muscle unit activity varies notoriously among repeated trials of a movement, which is nonetheless behaviorally identical each time. The final resolution of forces occurs in the musculoskeletal 'plant', not within the nervous system. Accordingly, a rather diverse set of CNS outputs generates the same movement. A unique correspondence between a pattern of neuronal activity and a specific body movement cannot be expected. Since the variation in output sets is constrained by the necessity of 'tradeoff' cooperation among the set members, there is hope that some code can be deciphered to link neuronal activity and movements. Neural network processing of the discharge of multiple recorded neurons in sensorimotor areas of cerebral cortex to control robotic arms appears to demonstrate the existence of such a code. However, only for some cyclic movements such as locomotion, is there a reasonably established model of motor pattern generation, a model that relies on oscillatory activity sustained both by pacemaker membrane properties and sensory signals. Oscillatory activity-observed in local field potential or EEG recording-is also evident during primate/human limb movement. By analogy with locomotion, it may be presumed to play a dynamical role in shaping the motor pattern. Indeed, motor cortical beta rhythm (~20 Hz) correlates to motor unit firing at the same frequency, during sustained (isometric) muscle contractions. Theta, alpha and gamma rhythms have also been linked to hand movements. At the cortical level, oscillatory field potentials may not identify specific movements, but they may define movement phases. Multiple unit 'spike' activity is probably required to identify a specific movement.
FROM CELLULAR AUTOMATA TO AUTONOMOUS AGENTS: BREAKING AWAY FROM SYMMETRY, SEIZURES, AND CYBERNETICS
A fundamental problem with epilepsy is the potentially disastrous loss of freedom: (i) loss of freedom in the technical sense that there is a loss of "degrees of freedom" with respect to the computational network elements and (ii) in the broader sense of the term "freedom", in that the effects often extend far beyond the temporally restricted ictal period and can undermine the general autonomy of an individual. This loss of autonomy can manifest itself as a complete disconnection from the world or conversely an extreme coupling to an external stimulus. Epilepsy is thus an archetypal case of acute loss of autonomy within the world with chronic consequences.
Early cyberneticists tried to resolve such problems by focusing on control. I will argue that this perspective, though technically insightful, has too often failed us. Specifically, the control perspective is oriented in the wrong direction in terms of the alteration of dynamics - it points to dead objects rather than autonomous active agents. A variety of cellular automata and recurrent neural network modeling approaches will be presented. The data suggest that controlled cells, in the sense of quiescence, "stability", synchrony and even predictability are not only achievable but, ironically, may be the default case. Moreover, the dynamics and embodied behavior of such networks is highly suggestive of pathology. The challenge thus may not be one of control but rather to elucidate how the brain goes beyond fixed point and limit cycle activity to evoke more complex dynamics - the move from shallow to deep symmetry breaking. The approach presented here suggests that we may begin to address this move from synchronous activity to more complex dynamics by modeling, evolving and studying autonomous agents that can interact with the world.
RECIPROCAL INTERACTIONS BETWEEN "FORMAT" AND "FLOW" -
Neurons process (filter) information by integrating spatially- and temporally-distributed inputs from cells synapsing upon them; as a result, altering the responsiveness of a neuron may change its filtering characteristics. Plasticity, broadly defined, refers to the modifiability of neuronal properties in response to activation, and is induced by particular combinations of stimulus properties (frequency, intensity, duration). Numerous mechanisms determine the nature of the resultant alterations in transmission efficacy, and how a neuron responds to a given stimulus will depend on which processes have acted recently and on how quickly they dissipate. As rapid-onset short-term effects may occur even within a spike train, the contributions of various types of plasticity to the response function of a neuron should change constantly during a series of spikes as different effects accumulate and decay. This will dynamically modify the computation (filtering) being performed by the neuron on the signal, which will in turn alter the induction of plasticity by spikes later in the train. Thus the transmission of information should alter both the signal and the substrate "on-line" and interactively. This implies that spike trains might "self-select" how they are processed based on their spatiotemporal parameters. I hypothesize that a spatiotemporally-distributed stimulus-evoked spike train interacts dynamically with the network through which it is transmitted to continuously modify the properties of both; the structure of an input thus determines how (temporally) and where (spatially) it will be processed.
RELIABLE TRANSMISSION OF SENSORY INPUTS INTO THE CORTEX
In vivo recordings from cat LGN and striate cortex neurons show reliable firing. This is surprising, given the presence of large amounts of noise from background synaptic inputs and stochastic synaptic release, and the fact that only about a tenth of a V1 layer 4 input cell's total synapses are directly from LGN. We investigated the ability of three spatial and temporal characteristics of a large population of synaptic inputs to influence spike timing reliability in a model V1 stellate neuron. These input population characteristics were - correlated firing, bursting, and dendritic spatial grouping of correlated synapses. We used a realistic multi-compartmental biophysical model of a reconstructed cat stellate cell with both exitatory and inhibitory synapses that exhibit short-term use-dependent dynamics of facilitation and depression. We found that reliable firing patterns could be obtained for a surprisingly low level of pre-synaptic temporal correlation, that specific pre-synaptic intra-burst frequencies and burst lengths were optimal for reliable post-synaptic firing, and that spatial grouping was in general detrimental to reliable firing. These results place morphological and physiological constraints on the LGN to V1 information flow and highlight interactions between the spatial distribution and timing of excitatory and inhibitory inputs.
HIPPOCAMPAL INVOLVEMENT IN A CHRONIC MODEL OF
Atypical absence seizures (AAS) are a pediatric malignant seizure disorder that involves the hippocampal and thalamocortical circuitry with slow spike-and wave discharges (SSWD) and associated cognitive dysfunction. Because the EEG, behavioural, pharmacological and developmental characteristics of clinical AAS are reproducible in the chronic AY-9944 (AY) model, we test the hypothesis that AY-induced SSWD in rats may produce or be associated with impaired hippocampal function, resulting in significant cognitive retardation. The hippocampal function of AY rats was examined with in vitro electrophysiology of synaptic plasticity and with a hippocampus-dependent radial arm maze (RAM) task to assess their spatial learning and memory. AY rats exhibited an impairment in long-term potentiation (LTP), paired-pulse facilitation (PPF) and synaptic depression (SD) in their hippocampal slices. AY rats also performed poorly with fewer perfect entries, higher number of errors and required longer time to learn the RAM task. Their perturbed spatial cognitive ability was not reversed by ethosuximide, which blocked their spontaneous seizures. Our data thus suggest that the deficit in learning and memory of AY rats with AAS appeared not to be seizure-induced
CELL DEATH AND NEUROGENESIS I: DO SEIZURES DAMAGE THE BRAIN?
Clinically, the issue of whether seizures damage the brain is still being debated. Idiopathic epilepsies show no discernible lesion or structural abnormality - suggesting that in such cases, repeated seizures do not damage the brain. In symptomatic epilepsies, where there is evidence of malformation or damage of the brain, it is unclear whether the lesion is a cause or a consequence of the seizures.
Regarding this issue, experimental evidence from animal models is also inconclusive. It has been clearly shown that prolonged induced seizures (status epilepticus) can cause widespread brain damage, particularly seen in the hippocampus. It has also been reported, however, that in certain seizure models such as kindling or electroconvulsive shock (ECS), there is no hippocampal cell loss occurring. It is of interest that, in kindling and ECS, seizures are similar in duration to those sustained by most epilepsy patients - and therefore are more clinically relevant than status epilepticus.
The present review examines a number of studies that have investigated the relationship between seizures and cell loss, using animal models. Special emphasis will be paid to studies utilizing the kindling and ECS models. Our study, investigating whether amygdala kindled-seizures cause hippocampal cell loss, will also be discussed. We found that after 30 amygdala-kindled stage 5 seizures, no statistically significant hippocampal cell loss occurred, although interesting trends were observed.
Probing this question is of immense importance. Brain damage related to seizures could be the cause of a number of related co-morbidities of epilepsy, such as impairments in learning and memory, mood disorders and depression, and a general decline in cognitive functioning. If seizure-induced damage occurs, the development of novel neuro-protective epilepsy therapies is highly warranted. On the other hand, examining seizure-induced anatomical changes is of fundamental importance in the quest to gain a better understanding of the development of epilepsy, and the nature of seizures themselves. In this regard, the implications of seizure-induced cell loss in the computational modeling of epilepsy, in terms of dynamical neural systems, should be emphasized.
CELL DEATH AND NEUROGENESIS II:
Ongoing neuron production in the mammalian brain continues to occur in the adult dentate gyrus of the hippocampal formation throughout life. This neurogenesis is extremely sensitive to many environmental and physiological influences and is thought to be involved in some forms of hippocampal dependent learning and memory. Neurogenesis is upregulated in several pathological states including ischemia, brain trauma and seizures. We have been examining the effects of seizures on hippocampal neurogenesis using various seizure and epilepsy models in rats. Kindling, electroconvulsive shock and chemically-induced status epilepticus all profoundly increase neurogenesis, suggesting that this is a general response to seizures. It is currently unknown whether this enhancement of neurogenesis contributes in a significant way to the pathological changes known to occur in the hippocampus following seizures. We are ultimately interested in understanding the contribution that seizure-induced changes in hippocampal neurogenesis might make to cognitive deficits associated with epilepsy.
THE KETOGENIC DIET AND EPILEPSY: PROBING THE MECHANISMS OF ACTION
The ketogenic diet (KD) is a treatment used to control drug-resistant seizures in children. The diet appears to demonstrate broad-spectrum anticonvulsant actions and works when many anticonvulsant drugs have failed. Some reports suggest that the KD also has beneficial effects on behavior and cognition in children with intractable epilepsy.
The KD's mechanisms remain unclear, however, as they have proven difficult to unravel. There are several reasons for these difficulties. One is that the KD produces a relatively small elevation in seizure threshold which is difficult to measure and quantify. Another difficulty is that epileptic seizure models, both in vivo and in vitro, appear to be inadequate in reproducing the KD's effects. The major difficulty, however, may be that the KD induces extremely large number of changes in the brain and in the entire body. These include a switch from glucose to ketones as the main energy substrate, a shift in ion and water balance, a change in lipid levels, a change in cellular membrane fluidity, and many others. It is unclear whether just one, or multiple factors are required to produce the KD's therapeutic effects.
The authors will discuss a need for an array of different experimental approaches and different model systems to study the roles of single factors in isolation (reductionist approach) and the roles of multiple factors in combination (systemic approach). One of the major advantages of this project is that the KD is used clinically and some hypotheses can be ultimately probed in studies involving human patients.
CONSCIOUSNESS AND ANTICONVULSANTS
Anticonvusant drugs are drugs that are used to suppress seizures. They are also called "antiepileptic drugs" or, more properly, "antiseizure drugs". The anticonvulsants have a long history, beginning with the bromides in the 19th Century and stretching up to the nine new drugs released in the past decade. The earliest drugs, the bromides and phenobarbital, had strong sedative side effects and were, in fact, used as daytime sedatives. They clearly impaired consciousness, and phenobarbital still does when used in adults. The discovery of phenytoin in 1938 was the result of an explicit search for an anticonvulsant with fewer sedative side effects than phenobarbital. The evolution of anticonvulsants since that time has generally been an evolution towards drugs that cause less sedation. It has been suggested, in fact, that the modern agents are no more effective than the old ones, but that they simply have fewer side effects. Continuing this tradition, the nine recently released drugs - with some exceptions - are believed to have a very good side effect profile. It should be noted, however, that this is based on clinical impression rather than double blind studies.
The question of sedation is important in itself, but has added significant in light of the fact that many epileptic children suffer from ADHD. Sedatives, of course, exacerbate ADHD, which may account for the "personality changes" seen in many children treated with anticonvulsants.
THE ENIGMA OF SEIZURE GENERATION AND SPREAD
Seizures represent entrained (synchronous) neuronal network activity with disruptive functional consequences. Epilepsy is the condition of unprovoked recurrent seizures. Ictal activity represents a seizure. Interictal activity is qualitatively different and brief. Important considerations:
Multiplicity of generators
Spread of seizure activity:
Neuronal clusters: dynamic, pathological
Receptivity of targets:
Transition to Seizure:
2. Discontinuous Process:
3. Or Both:
Duckrow and Spencer stated, "Epileptiform activity requires that large aggregates of neurons act synchronously... We conclude that the process of neuronal entrainment during seizure onset involves a transient interaction between brain regions, but the maintenance of this interaction is not required for sustained seizure activity. We hypothesize that seizures are a manifestation of coupled oscillators and that preventing the coupling will prevent seizures". From: Regional coherence and the transfer of ictal activity during seizure onset in the medial temporal lobe. EEG and Clinical Neurophysiology, 82: 415-422, 1992
Derchansky, Carlen, et al. hypothesize that seizures are a manifestation of coupled oscillators and that preventing the coupling will prevent seizures.
BIDIRECTIONAL SEIZURE ACTIVITY AS A MANIFESTATION OF COUPLED NETWORK OSCILLATORS
The concept of coupled neural network oscillators is central to our understanding of brain function. We illustrate, from two experimental models of epilepsy and from human epileptic intracranial EEG, that epileptic seizures in intact brain structures are a manifestation of coupled network oscillators. In the isolated intact mouse hippocampus (P8-13) exposed to low-Mg2+ artificial cerebral spinal fluid, recurrent, spontaneous seizures generated bidirectional and reverberatory epileptiform activity along the septotemporal axis, demonstrated by multiple electrode recordings and optical imaging. Uncoupling the system of network oscillators, by partial or complete cuts along the septotemporal hippocampal axis, resulted in independent bidirectional and reverberatory seizures in the separated tissue areas. These experiments revealed that epileptogenicity was more pronounced in the temporal as compared to the septal half. Within the same epileptic burst, low and high frequency components originated from different locations at different times, both in the intact and sectioned hippocampus. A second seizure model, in a more mature preparation (P15-27), using focal tetanic electrical stimulation, also exhibited sustained bidirectional discharges. Clinical data obtained from dual depth electrodes in the hippocampus of epileptic patients, similarly revealed bidirectional and reverberatory seizures. Conceptualizing epileptiform activity in a localized brain region as a manifestation of dynamically coupled network oscillators, as opposed to a single seizure generator with passive, cable-like propagation, provides a useful framework both for understanding and controlling epileptic seizures.
IS THE SUBICULUM A GENERATOR OF HIPPOCAMPAL RIPPLES?
Liang Zhang, Hung Ling Huang, Marjan Nassiri Asl and Chiping Wu
During slow wave sleep and consummatory behaviors, the rodent hippocampus exhibits intermittent sharp waves (SPW) that are superimposed with ~200 Hz oscillations called ripples. The SPW-ripples are thought to be important for memory consolidation processes and the subiculum plays a key role in transferring these signals to neocortical areas. While previous in vivo studies have revealed subiculum ripples, it is unclear how these oscillatory signals are transferred from the CA1 to the subiculum. Moreover, CA1 pyramidal neurons do not fire in high frequency during slow wave sleep, raising the question that SPW-associated discharges of CA1 pyramidal neurons may not be fully responsible the ripples seen in the subiculum.
Because subiculum pyramidal neurons have recurrent excitatory connections, we hypothesize that the subiculum may generate its own ripples. To test this we used thick (1 mm) mouse hippocampal slices that contained CA1-subiculum areas only. We found: 1) spontaneous rhythmic field potentials (SRFPs) could occur coherently in the CA1 and subiculum areas. The subiculum SRFP led CA1 SRFPs by several ms and the former was often associated with ripples. 2) SRFPs could persist without CA1 rhythm. 3) CNQX, bicuculline or a high Mg perfusate blocked these SRFPs, suggesting that their generation requires network interactions of GABA-A and AMPA glutamate receptors. 4) Coherent CA1 and subiculum SRFPs became independent after a surgical cut that separated these two areas.
We propose that the subiculum acts not only as a relay but also an amplifier in transferring hippocampal SPW-ripples to the neocortical areas.
ELECTRICAL OSCILLATIONS IN HIPPOCAMPAL NETWORKS:
We propose a hybrid neural network where artificial neural networks are used, in an in-silico epilepsy model of biological neural networks, a) to predict the onset of state transitions from higher complexities, possibly chaotic, to lower complexity possibly rhythmic activities, b) to predict the onset of state transitions from higher complexity to lower complexity activities, and c) to restore the original higher complexity activity. A coupled nonlinear oscillators model is used to represent the spontaneous epileptiform oscillations of CA3 hippocampal neurons to illustrate the prediction and control schemes of these state transition onsets. Our prediction scheme consists of a recurrent neural network having radial-basis-function nonlinearities. When the onset of lower complexity activity is predicted in the in-silico model, then our control scheme consists of applying a small perturbation to a system variable (i.e., the amplitude signal) when it is sufficiently close to the unstable higher complexity manifold. The system state can be restored back to its previous mode utilizing the forces of the system's vector field. We have also investigated the applicability of our prediction scheme for the onsets of state transitions into ictal-like activities using a low magnesium in vitro epilepsy model of field recordings from the rat hippocampus.
ACTIVE DENDRITES AND SPIKE PROPAGATION IN
It is well known that interneurons are heterogeneous in their morphologies, biophysical properties, pharmacological sensitivities and electrophysiological responses, but it is unknown how best to understand this diversity. Given their critical roles in shaping brain output, it is important to try to understand the functionality of their computational characteristics. It has been shown that long-term potentiation is induced specifically on oriens-lacunosum/moleculare (O-LM) interneurons in hippocampus CA1 and that these same cells contain the highest density of dendritic sodium and potassium conductances measured to date. We speculate that the highly active dendrites of these interneurons endow them with a specialized function within the hippocampal circuitry by allowing them to regulate direct and indirect signally pathways within the hippocampus. In this way the O-LM interneurons act as 'gate-keepers', allowing certain signaling pathways to be dominant during different physiologically relevant population rhythms. I will discuss several models of O-LM interneurons, with focus on the types and distributions of ion channels along the somato-dendritic tree, spike initiation and propagation, frequency preferences and the unique role of the cell in the hippocampal network.
EPILEPSY AND AUTONOMY: A DEVELOPMENTAL PERSPECTIVE
The child and family growing up with epilepsy often experience repeated episodes of loss of autonomy and control. Depending on the developmental age of the child, this can present as discontinuities in the sense of being a person, and of being in control of one's body and one's mind. Disruptions in learning and social relatedness can result. In addition, there can be shifts from more mature to more developmentally early modes of thought and action, as well as difficulties in storage and retrieval of memory. Early developmental-psychoanalytic neurology-oriented physicians, including Freud, Sperling, and Schilder, proposed prototypical neural networks to account for some of these phenomena. Freud's 1891 and 1895 neural network models continue to serve as a framework for thinking about ways that the child and family can promote healthy mental functioning, while reducing susceptibility to regression and disorganization. Case material will be discussed to illustrate the various points.
LIVING WITH EPILEPSY: APRIL'S LIFE SENTENCE
My name is Claudia Megna and my sister "April" (alias) has epilepsy. I have never studied epilepsy nor do I know all the scientific terms. However, what I do know from a personal account is that epilepsy has robbed my sister of her autonomy. Moreover, epilepsy is like a domino effect, it not only robs the patient's autonomy but affects the autonomy of the family members as well. My parents love my sister and would do anything for her, but it has come at a cost in the sense that they have been unable to fully enjoy raising three girls in a "typical" childhood. Each passing year brought a new seizure, a new trigger, a new personality, and we were left to wonder many times if this behaviour was my sister's personality or the seizure itself. My sister Sofia and I have always felt shame and guilt because "April" was a prisoner in her own body. We were able to enjoy the highlights of life - friends, going out, playing. However, many times "April's" behaviour was misunderstood because various actions were bizarre and appeared to be attention seeking.
DEVELOPING AN ANIMAL MODEL FOR THE REPRODUCTIVE DYSFUNCTIONS OF EPILEPSY
There is a greater proportion of reproductive dysfunction in the epileptic population than in the general population. Women with epilepsy, especially women with temporal lobe epilepsy, are more likely to suffer from imbalances in reproductive hormone levels, disruptions in menstrual cyclicity, and disruptions in ovarian physiology. All of these negative changes lead to a significant reduction in fertility in temporal-lobe epileptic women. Results from past experiments in our lab have shown that the amygdala-kindling model in the female rat is a promising potential model for the reproductive dysfunctions accompanying complex partial seizures of temporal lobe origin in women. Many of the reproductive abnormalities seen in women with temporal lobe epilepsy are similar to or are also seen in the amygdala-kindled female rat. Our present goal is to better understand this interictal effect of kindling through several planned parametric kindling studies, which will be discussed.
EPILEPSY: A DISORDER OF LOSS OF CONTROL - A CLINICIAN'S PERSPECTIVE
Epilepsy is characterized by recurrent spontaneous paroxysmal discharges of neuronal aggregates. Since Hippocrates, the genetic basis of its hereditary nature has been postulated but only lately have epilepsy genes linked to epilepsy syndromes been identified. Often the loss of control in complex partial seizures was referred to as "the dual consciousness" of psychomotor epilepsy (Hughlings Jackson). Recent studies of Gloor et al suggest loss of consciousness occurs when the ictal discharge spreads from limbic structures to the neocortex, resulting in loss of control and autonomy.
Much of the disability of epilepsy is attributed to comorbidity and its impact on the quality of life: sudden onset, often without warning; unpredictability of onset (with the occasional exception of reflex epilepsy), loss of contact with the environment and responsiveness to others and a perceived loss of autonomy (driving. Operation of powered tools, holding down critical jobs) and unemployment. Some of the disability of epilepsy may be ameliorated by increasing the degree of perceived control: lessening the probability of seizures (with medications or vitamins), biofeedback and employment of neuroaugmentation methods that increase autonomous actions (e.g. magnetic stimulation of implanted deep brain or vagus nerve stimulators).
The future promises new avenues of episodic self-control as better understanding of the basic mechanisms underlying the epilepsies and their consequences is revealed. The application of these basic principles to epilepsy and its control is discussed.
THE GENETIC EPILEPSIES AND COMMON COMORBIDITIES
Miles D Thompson1, Maire E Percy3, David EC Cole 1and Paul A Hwang 3
Developmental and psychiatric disorders are increasingly being recognized as sequelae of epilepsy. As with other epilepsies, 20% of the genetic familial idiopathic epilepsies are refractory to treatment and a third of these lead to psychiatric and/or developmental comorbidities. Depression, anxiety, attention-deficit/hyperactivity disorder (ADHD), and psychotic disorders are all common comorbidities of epilepsy. Currently, the comorbid developmental and psychiatric symptoms are often treated separately from the seizures. The first line of treatment involves the use of antiepilepsic drugs (AEDs) in order to attempt to control the EEG abnormalities that are associated with seizure phenotypes, but in many cases AED treatment may not be fully successful. Many patients, however, continue to suffer comorbidities whether or not their seizures are controlled. Here, we highlight the ways in which novel therapeutic interventions for epilepsy and comorbid developmental/psychiatric disorders may become available once the genes that are disrupted genetic epilepsies become known. Novel AEDs may selectively target pathways, or specific proteins, that are disrupted in a genetic disease. An understanding of the molecular genetics of the epilepsies may benefit the field of epilepsy pharmacogenomics - a research field that focuses on the development of drug therapies that target specific proteins based on a knowledge of the structure and function of the encoding gene. This strategy may result in better treatment of comorbidities common in both genetic and non-genetic forms of epilepsy.
WOMEN AND EPILEPSY: AN INTEGRATIVE APPROACH
Seven years ago, Epilepsy Ontario hosted the "Women and Epilepsy" conference here at the University of Toronto. The issues discussed pertained to the effect of seizure disorders on women's lives, both as caregivers to children with epilepsy and as individuals with epilepsy. Research in the field has mainly been divided into psychological, physiological and sociological issues. In truth, these issues interact and are interdependent. The literature has expanded in recent years to cover the issues of endocrine disorders, affective disorders, menopause and pregnancy. However, many topics remain insufficiently studied. Many women with epilepsy remain frustrated with the lack of or inaccessibility of information on the advances in this field. As researchers and health care professionals, it is a fine balance between understanding the science of disorders and understanding of their impact on life. To integrate our knowledge and improve our research and care, we must develop the dialogue between the bench and the bedside. Our objective is to hold a second "Women and Epilepsy" conference to revive discussion, evaluate our progress, and consider our next steps in research, patient care and advocacy.
BISTABLE DYNAMICS OF EPILEPTIC PHENOMENA
Piotr Suffczynski1,2, Stiliyan Kalitzin1, Fernando Lopes da Silva1,3
Seizures are accompanied by EEG paroxysms that start and end abruptly. The mechanisms of the spontaneous transitions between "normal" EEG activity and paroxysmal discharges are not well understood. Based on a computational model of the thalamocortical network we hypothesized that some types of epileptic transitions represent bifurcations occurring in a bistable system. In such system normal and epileptic states exist simultaneously for the same set of system's parameters. Transitions between the two states may occur due to random inputs and/or parameter fluctuations or can be assigned to gradual change of one or more system parameters. Information about the dynamical processes that are responsible for the transitions can be extracted from the duration distributions of normal and paroxysmal epochs. We analyzed experimental time sequences from various animal models of absence epilepsy, human epileptic subjects and in vitro data. In a number of cases the null hypothesis of a bistable system with purely random transitions corresponding to the Poisson process, could not be rejected. In remaining cases deterministic time-dependent mechanisms were involved in seizure termination while seizure initiation is associated with a random walk process. The conclusions are as follows:
AUTONOMOUS DYNAMIC MODELS OF CLINICAL AND EXPERIMENTAL EPILEPSY
Neural network models attempting to copy real biological systems sooner or later meet an enormous complexity barrier. We cannot incorporate in a model all the features of the neurons and their networks, neither can we measure all the properties of the biological systems in order to fully validate the models. Our alternative approach is based on studying some general classes of dynamic behavior, such as parameter random walk, bistability and intermittence. Analytic techniques may allow us, in some cases, to recognize these classes in data measured from the real world. In epilepsy research that could mean understanding the generic forms of transitions between interictal ("normal") and ictal states. An essential ingredient in our models is that ictal transitions are not controlled by an external parameter tuning but emerge as an autonomous feature in the system. Depending on the dynamic patterns of these transitions, optimal strategies for "forecasting" seizures and perhaps even dynamic control of the latter might be feasible. In this talk we consider some basic dynamic scenarios that can be modeled and compared with available experiments. Empiric results gathered from clinical research, still awaiting their "dynamic match" are also discussed.
Please send comments to: firstname.lastname@example.org