Search

The reported effectiveness of these home preparations, especially those with high cannabidiol (CBD) concentrations, has garnered the attention of the medical community. In particular cannabis sativa, known for its lack of a psychoactive effect and high CBD content, has become a target of medical research. The shift in public and political interest to medicinal applications of CBD demands renewed research into its efficacy. Pediatric populations in particular stand to benefit significantly from a better understanding of the safety and efficacy of this novel treatment. This review discusses the current state of CBD research and identifies areas that require further investigation as they pertain to pediatric epilepsy populations. It will especially cover those suffering from refractory epilepsies for which other methods of remediation have not sufficed.

Different pairs of adjacent sub-regions of rat hippocampal neurons were co-cultured within two-well devices on micro-electrode arrays (MEAs). Micro-tunnels allowed passage of axons, strong structural connectivity, and information transmission between wells. Paired-pulse stimuli were delivered at each electrode to increase the probability of the excitation without overstimulation. Because of the micro-channel construction, the electrodes inside these tunnels pick up more activity than those within each well. In particular, the strong axonal inputs from the sources of native feed-forward EC-DG and CA3-CA1 co-cultures evoked more target activity than the feed-back networks. In contrast, the evoked target activity in the native feedforward DG-CA3 networks was lower than the reverse flow of information.

In this work, a carbon nanofiber (CNF) based neural probe compliant with high magnetic field functional magnetic resonance imaging (fMRI) has been demonstrated. The CNF probe has high spatial resolution electrode patterns, which enable extracellular deep brain stimulation (DBS) and detection. As the probe is radio frequency transparent, in-situ DBS and brain imaging can be performed simultaneously. CNF shows 5 times higher neural density in-vitro. 4.7T fMRI has been performed on the CNF neural probes implanted in a rat's brain ex-vivo and no image distortion has been observed under spin or gradient echo modes.

To better understand encoding and decoding of stimulus information in two specific hippocampal sub-regions, we isolated and co-cultured rat primary dentate gyrus (DG) and CA3 neurons within a two-chamber device with axonal connectivity via micro-tunnels. We tested the hypothesis that, in these engineered networks, decoding performance of stimulus site information would be more accurate when stimuli and information flow occur in anatomically correct feed-forward DG to CA3 vs. CA3 back to DG. In particular, we characterized the neural code of these sub-regions by measuring sparseness and uniqueness of the responses evoked by specific paired-pulse stimuli. We used the evoked responses in CA3 to decode the stimulation sites in DG (and vice-versa) by means of learning algorithms for classification (support vector machine, SVM). The device was placed over an 8 × 8 grid of extracellular electrodes (micro-electrode array, MEA) in order to provide a platform for monitoring development, self-organization, and improved access to stimulation and recording at multiple sites. The micro-tunnels were designed with dimensions 3 × 10 × 400 μm allowing axonal growth but not migration of cell bodies and long enough to exclude traversal by dendrites. Paired-pulse stimulation (inter-pulse interval 50 ms) was applied at 22 different sites and repeated 25 times in each chamber for each sub-region to evoke time-locked activity. DG-DG and CA3-CA3 networks were used as controls. Stimulation in DG drove signals through the axons in the tunnels to activate a relatively small set of specific electrodes in CA3 (sparse code). CA3-CA3 and DG-DG controls were less sparse in coding than CA3 in DG-CA3 networks. Using all target electrodes with the three highest spike rates (14%), the evoked responses in CA3 specified each stimulation site in DG with optimum uniqueness of 64%. Finally, by SVM learning, these evoked responses in CA3 correctly decoded the stimulation sites in DG for 43% of the trials, significantly higher than the reverse, i.e., how well-recording in DG could predict the stimulation site in CA3. In conclusion, our co-cultured model for the in vivo DG-CA3 hippocampal network showed sparse and specific responses in CA3, selectively evoked by each stimulation site in DG.

Anatomical and behavioral studies, and in vivo and slice electrophysiology of the hippocampus suggest specific functions of the dentate gyrus (DG) and the CA3 subregions, but the underlying activity dynamics and repeatability of information processing remains poorly understood. To approach this problem, we engineered separate living networks of the DG and CA3 neurons that develop connections through 51 tunnels for axonal communication. Growing these networks on top of an electrode array enabled us to determine whether the subregion dynamics were separable and repeatable. We found spontaneous development of polarized propagation of 80% of the activity in the native direction from DG to CA3 and different spike and burst dynamics for these subregions. Spatial-temporal differences emerged when the relationships of target CA3 activity were categorized with to the number and timing of inputs from the apposing network. Compared to times of CA3 activity when there was no recorded tunnel input, DG input led to CA3 activity bursts that were 7× more frequent, increased in amplitude and extended in temporal envelope. Logistic regression indicated that a high number of tunnel inputs predict CA3 activity with 90% sensitivity and 70% specificity. Compared to no tunnel input, patterns of >80% tunnel inputs from DG specified different patterns of first-to-fire neurons in the CA3 target well. Clustering dendrograms revealed repeating motifs of three or more patterns at up to 17 sites in CA3 that were importantly associated with specific spatial-temporal patterns of tunnel activity. The number of these motifs recorded in 3 min was significantly higher than shuffled spike activity and not seen above chance in control networks in which CA3 was apposed to CA3 or DG to DG. Together, these results demonstrate spontaneous input-dependent repeatable coding of distributed activity in CA3 networks driven by engineered inputs from DG networks. These functional configurations at measured times of activation (motifs) emerge from anatomically accurate feed-forward connections from DG through tunnels to CA3.

In this study we created four network topologies composed of living cortical neurons and compared resultant structural-functional dynamics including the nature and quality of information transmission. Each living network was composed of living cortical neurons that were created using microstamping of adhesion promoting molecules and each was “designed” with different levels of convergence embedded within each structure. Networks were cultured over a grid of electrodes that permitted detailed measurements of neural activity at each node in the network. Of the topologies we tested, the “Random” networks in which neurons connect based on their own intrinsic properties transmitted information embedded within their spike trains with higher fidelity relative to any other topology we tested. Within our patterned topologies in which we explicitly manipulated structure, the effect of convergence on fidelity was dependent on both topology and time-scale (rate versus temporal coding). At long time scales associated with rate based coding the simple linear 2D or “line” networks produced the highest fidelity during transmission among neurons (nodes) followed by 4D (4 connections per node) and 8D (8 connections per node) grid networks. By contrast, at more precise temporal scales it was the increased convergence within the 8D topology that now resulted in the highest fidelity followed by 4D and 2D networks. A more detailed examination using tools from network analysis revealed that these changes in fidelity were also associated with a number of other structural properties including a node’s degree, degree-degree correlations, path length, and clustering coefficients. Whereas information transmission was apparent among nodes with few connections, the greatest transmission fidelity was achieved among the few nodes possessing the highest number of connections (high degree nodes or putative hubs). These results provide a unique view into the relationship between structure and its affect on transmission fidelity, at least within these small neural populations with defined network topology. They also highlight the potential role of tools such as microstamp printing and microelectrode array recordings to construct and record from arbitrary network topologies to provide a new direction in which to advance the study of structure-function relationships.

Objective Functions ascribed to the hippocampal sub-regions for encoding episodic memories include the separation of activity patterns propagated from the entorhinal cortex (EC) into the dentate gyrus (DG) and pattern completion in CA3 region. Since a direct assessment of these functions is lacking at the level of specific axonal inputs, our goal is to directly measure the separation and completion of distinct axonal inputs in engineered pairs of hippocampal sub-regional circuits. Approach We co-cultured EC-DG, DG-CA3, CA3-CA1 or CA1-EC neurons in a two-chamber PDMS device over a micro-electrode array (MEA60), inter-connected via distinct axons that grow through the micro-tunnels between the compartments. Taking advantage of the axonal accessibility, we quantified pattern separation and completion of the evoked activity transmitted through the tunnels from source into target well. Since pattern separation can be inferred when inputs are more correlated than outputs, we first compared the correlations among axonal inputs with those of target somata outputs. We then compared, in an analog approach, the distributions of correlation distances between rate patterns of the axonal inputs inside the tunnels with those of the somata outputs evoked in the target well. Finally, in a digital approach, we measured the spatial population distances between binary patterns of the same axonal inputs and somata outputs. Main results We found the strongest separation of the propagated axonal inputs when EC was axonally connected to DG, with a decline in separation to CA3 and to CA1 for both rate and digital approaches. Furthermore, the digital approach showed stronger pattern completion in CA3, then CA1 and EC. Significance To the best of our knowledge, these are the first direct measures of pattern separation and completion for axonal transmission to the somata target outputs at the rate and digital population levels in each of four stages of the EC-DG-CA3-CA1 circuit.

Microelectrode arrays (MEAs) are commonly utilized for stimulating and recording extracellular electrical signals including both local field and action potentials in both in-vitro and in-vivo neural studies. This work demonstrates the feasibility of 3D microelectrode arrays using a novel carbon nanomaterial, electrospun carbon nanofiber (CNF). CNF MEAs impedance is characterized and compared to that of carbon nanotube MEAs and commercial TiN MEAs. An in-vitro culture of CNF electrodes are performed using E18 cortical neurons and analyzed for cell interaction. With these electrodes we are able to detect extracellular neural signals including action potentials from single neurons from an array of CNF electrodes embedded within the substrate of an MEA.

Seven serologically related, but antigenically different, botulinum toxins (BoNTs) have been identified including types A, B, C, D, E, F, and G. The bacterium Clostridium botulinum along with some strains of Clostridium baratti and Clostridium butyricum are known to produce botulinum toxins responsible for 4 forms of botulism poisoning including food-borne botulism, inhalation botulism, wound botulism, and infant botulism. Botulism toxins consist of a heavy chain (100 kDa), responsible for binding to target cells, and a light chain (50 kDa) responsible for catalytic protein cleaving activity. Light chain has been identified as a zinc endopeptidase that cleaves proteins forming the synaptic vesicle docking and fusion complex (Simpson 1996; Lacy and Stevens 1997). The standard for detection of BoNT toxins is the mouse bioassay, which is able to detect as little as 0.02 ng of toxin. Strengths of the mouse bioassay include conceptual simplicity and sensitivity. While the non-selectivity of the mouse bioassay enables it to detect any BoNT serotype, additional neutralization assays are necessary to determine serotype. Other limitations of the mouse bioassay include expense, expertise related to maintaining mouse-rearing facilities, and time, because as much as 4 d may be required to obtain results (Hallis and others 1996; Witcome and others 1999). Several attempts to replace the mouse bioassay have been made. Methods that have been developed and hold promise for future replacement of the mouse bioassay include mass spectroscopy, immunoassays, polymerase chain reaction (PCR) assays, and assays based upon protease activities of BoNTs. Currently, no single assay appears to be capable of replacing the broadly applicable mouse bioassay.

We report the design and application of a Micro Electro Mechanical Systems (MEMs) device that permits investigators to create arbitrary network topologies. With this device investigators can manipulate the degree of functional connectivity among distinct neural populations by systematically altering their geometric connectivity in vitro. Each polydimethylsilxane (PDMS) device was cast from molds and consisted of two wells each containing a small neural population of dissociated rat cortical neurons. Wells were separated by a series of parallel micrometer scale tunnels that permitted passage of axonal processes but not somata; with the device placed over an 8 × 8 microelectrode array, action potentials from somata in wells and axons in microtunnels can be recorded and stimulated. In our earlier report we showed that a one week delay in plating of neurons from one well to the other led to a filling and blocking of the microtunnels by axons from the older well resulting in strong directionality (older to younger) of both axon action potentials in tunnels and longer duration and more slowly propagating bursts of action potentials between wells. Here we show that changing the number of tunnels, and hence the number of axons, connecting the two wells leads to changes in connectivity and propagation of bursting activity. More specifically, the greater the number of tunnels the stronger the connectivity, the greater the probability of bursting propagating between wells, and shorter peak-to-peak delays between bursts and time to first spike measured in the opposing well. We estimate that a minimum of 100 axons are needed to reliably initiate a burst in the opposing well. This device provides a tool for researchers interested in understanding network dynamics who will profit from having the ability to design both the degree and directionality connectivity among multiple small neural populations.

We have developed a new method for culturing cells that maintains their health and sterility for many months. Using conventional techniques, primary neuron cultures seldom survive more than 2 months. Increases in the osmotic strength of media due to evaporation are a large and underappreciated contributor to the gradual decline in the health of these cultures. Because of this and the ever-present likelihood of contamination by airborne pathogens, repeated or extended experiments on any given culture have until now been difficult, if not impossible. We surmounted survival problems by using culture dish lids that form a gas-tight seal, and incorporate a transparent hydrophobic membrane (fluorinated ethylene–propylene) that is selectively permeable to oxygen (O2) and carbon dioxide (CO2), and relatively impermeable to water vapor. This prevents contamination and greatly reduces evaporation, allowing the use of a non-humidified incubator. We have employed this technique to grow dissociated cortical cultures from rat embryos on multi-electrode arrays. After more than a year in culture, the neurons still exhibit robust spontaneous electrical activity. The combination of sealed culture dishes with extracellular multi-electrode recording and stimulation enables study of development, adaptation, and very long-term plasticity, across months, in cultured neuronal networks. Membrane-sealed dishes will also be useful for the culture of many other cell types susceptible to evaporation and contamination. © 2001 Elsevier Science B.V. All rights reserved.

The brain is perhaps the most advanced and robust computation system known. We are creating a method to study how information is processed and encoded in living cultured neuronal networks by interfacing them to a computer-generated animal, the Neurally-Controlled Animat, within a virtual world. Cortical neurons from rats are dissociated and cultured on a surface containing a grid of electrodes (multi-electrode arrays, or MEAs) capable of both recording and stimulating neural activity. Distributed patterns of neural activity are used to control the behavior of the Animat in a simulated environment. The computer acts as its sensory system providing electrical feedback to the network about the Animat's movement within its environment. Changes in the Animat's behavior due to interaction with its surroundings are studied in concert with the biological processes (e.g., neural plasticity) that produced those changes, to understand how information is processed and encoded within a living neural network. Thus, we have created a hybrid real-time processing engine and control system that consists of living, electronic, and simulated components. Eventually this approach may be applied to controlling robotic devices, or lead to better real-time silicon-based information processing and control algorithms that are fault tolerant and can repair themselves.

Four experiments examined transfer of differential outcome performances to new choice responses in pigeons. Experiments 1A and 1B showed that new responses trained off a matching-to-sample baseline readily substituted for the choice alternatives in differential outcome matching, provided that they shared the same outcome associations as the alternatives they replaced. Experiment 2 showed that comparison responses trained on baseline, but in a task in which their different outcomes occurred equally often following each sample (viz., one-to-many matching), substituted for the choices in a standard, differential outcome task. Experiment 3 showed, somewhat surprisingly, that the choices in the latter task were likewise effective substitutes in one-to-many matching. These results pose separate challenges for standard two-process theory and for the bidirectional account of differential outcome performance, and they suggest other cues that pigeons may use to predict outcomes.

Pigeons were trained on a pair-comparison task in which left versus right choices were reinforced following different sequences of two center-key stimuli. Choice accuracy was higher when retention intervals occurred after the entire sequence than when they separated the two stimuli comprising it, and this effect occurred independently of whether the initial and terminal stimuli came from the same or different dimensions. The initial stimulus from the prior trial was a source of proactive interference only in groups for which the retention interval separated the two sequence stimuli. By contrast, differential delay-interval behavior was observed only in groups for which the retention interval followed presentation of the entire sequence. These results indicate that coding processes in delayed discriminations are influenced by the location of the retention interval, and that response mediation affects retention performances if the reinforced choice can be determined prior to the interval.

The coding processes occurring in many-to-one matching-to-sample with hue, response, and food samples were assessed in retention tests following initial acquisition and in mediated transfer tests given after pigeons had learned to match one of the original sample sets to new comparison stimuli. In retention, working memory performances were more accurate when different samples associated with the same correct comparison were presented prior to the retention interval than when the same sample was presented repeatedly. In addition, asymmetrical gradients were observed for each sample dimension: accuracy over delays fell much less rapidly with samples of no food, 1 response, and green than with samples of food, 20 responses, and red. In transfer, matching performances generalized to novel sample-comparison relations after interim training in which the hue samples were associated with new comparisons. Transfer was not apparent, however, when interim training involved response or food samples. These data provide additional evidence for common coding in many-to-one matching and also suggest that retention is an unreliable index of coding processes.

Behavioral recovery in animal models of human CNS syndromes suggests that transplanted stem cell derivatives can augment damaged neural networks but the mechanisms behind potentiated recovery remain elusive. Here we use microelectrode array (MEA) technology to document neural activity and network integration as rat primary neurons and rat hippocampal neural progenitor cells (NPCs) differentiate and mature. The natural transition from neuroblast to functional excitatory neuron consists of intermediate phases of differentiation characterized by coupled activity. High-frequency network-wide bursting or "superbursting" is a hallmark of early plasticity that is ultimately refined into mature stable neural network activity. Microelectrode array (MEA)-plated neurons transition through this stage of coupled superbursting before establishing mature neuronal phenotypes in vitro. When plated alone, adult rat hippocampal NPC-derived neurons fail to establish the synchronized bursting activity that neurons in primary and embryonic stem cell-derived cultures readily form. However, adult rat hippocampal NPCs evoke re-emergent superbursting in electrophysiologically mature rat primary neural cultures. Developmental superbursting is thought to accompany transient states of heightened plasticity both in culture preparations and across brain regions. Future work exploring whether NPCs can re-stimulate developmental states in injury models would be an interesting test of their regenerative potential.

Two experiments investigated whether differential outcomes in matching-to-sample (MTS) would enhance acquisition even in cases in which the outcomes could not be anticipated at the beginning of a trial. In Experiment 1, food and no-food outcomes were differential with respect to both hue and line comparisons in one-to-many matching-to-sample (MTS) but were nondifferential with respect to the samples. Overall acquisition and acquisition with each comparison set were faster in relation to controls that received each outcome equally often on all trials. In Experiment 2, hue and line comparisons were associated with either different probabilities of food (p = 1.0 vs. 0.2) or with the same probability (p =.6). Again, matching acquisition was more rapid in the differential group

An understanding of the in vivo spatial emergence of abnormal brain activity during spontaneous seizure onset is critical to future early seizure detection and closed-loop seizure prevention therapies. In this study, we use Granger causality (GC) to determine the strength and direction of relationships between local field potentials (LFPs) recorded from bilateral microelectrode arrays in an intermittent spontaneous seizure model of chronic temporal lobe epilepsy before, during, and after Racine grade partial onset generalized seizures. Our results indicate distinct patterns of directional GC relationships within the hippocampus, specifically from the CA1 subfield to the dentate gyrus, prior to and during seizure onset. Our results suggest sequential and hierarchical temporal relationships between the CA1 and dentate gyrus within and across hippocampal hemispheres during seizure. Additionally, our analysis suggests a reversal in the direction of GC relationships during seizure, from an abnormal pattern to more anatomically expected pattern. This reversal correlates well with the observed behavioral transition from tonic to clonic seizure in time-locked video. These findings highlight the utility of GC to reveal dynamic directional temporal relationships between multichannel LFP recordings from multiple brain regions during unprovoked spontaneous seizures.

While temporal lobe epilepsy (TLE) has been treatable with anti-seizure medications over the past century, there still remain a large percentage of patients whose seizures remain untreatable pharmacologically. To better understand and treat TLE, our laboratory uses several in vivo analytical techniques to estimate connectivity in epilepsy. This paper reviews two different connectivity-based approaches with an emphasis on application to the study of epilepsy. First, we present effective connectivity techniques, such as Granger causality, that has been used to assess the dynamic directional relationships among brain regions. These measures are used to better understand how seizure activity initiates, propagates, and terminates. Second, structural techniques, such as magnetic resonance imaging, can be used to assess changes in the underlying neural structures that result in seizure. This paper also includes in vivo epilepsy-centered examples of both effective and anatomical connectivity analysis. These analyses are performed on data collected in vivo from a spontaneously seizing animal model of TLE. Future work in vivo on epilepsy will no doubt benefit from a fusion of these different techniques. We conclude by discussing the interesting possibilities, implications, and challenges that a unified analysis would present.

The brain is perhaps the most advanced and robust computation system known. We are creating a method to study how information is processed and encoded in living cultured neuronal networks by interfacing them to a computer-generated animal, the Neurally-Controlled Animat, within a virtual world. Cortical neurons from rats are dissociated and cultured on a surface containing a grid of electrodes (multi-electrode arrays, or MEAs) capable of both recording and stimulating neural activity. Distributed patterns of neural activity are used to control the behavior of the Animat in a simulated environment. The computer acts as its sensory system providing electrical feedback to the network about the Animat’s movement within its environment. Changes in the Animat’s behavior due to interaction with its surroundings are studied in concert with the biological processes (e.g., neural plasticity) that produced those changes, to understand how information is processed and encoded within a living neural network. Thus, we have created a hybrid real-time processing engine and control system that consists of living, electronic, and simulated components. Eventually this approach may be applied to controlling robotic devices, or lead to better real-time silicon-based information processing and control algorithms that are fault tolerant and can repair themselves.

In vitro neural networks of cortical neurons interfaced to a computer via multichannel microelectrode arrays (MEA) provide a unique paradigm to create a hybrid neural computer. Unfortunately, only rudimentary information about these in vitro network's computational properties or the extent of their abilities are known. To study those properties, a liquid state machine (LSM) approach was employed in which the liquid (typically an artificial neural network) was replaced with a living cortical network and the input and readout functions were replaced by the MEA-computer interface. A key requirement of the LSM architecture is that inputs into the liquid state must result in separable outputs based on the liquid's response (separation property). In this paper, high and low frequency multi-site stimulation patterns were applied to the living cortical networks. Two template-based classifiers, one based on Euclidean distance and a second based on a cross-correlation were then applied to measure the separation of the input-output relationship. The result was over a 95% (99.8% when nonstationarity is compensated) input reconstruction accuracy for the high and low frequency patterns, confirming the existence of the separation property in these biological networks.

Botulinum toxin (BoNT) is a potent neurotoxin produced by toxigenic strains of Clostridium botulinum. Botulinum toxin poses a major threat since it could be employed in a deliberate attack on the U.S. food supply. Furthermore, BoNT may be liberated in any insufficiently processed food containing a reduced oxygen atmosphere. Hence, rapid and reliable detection of BoNT in foods is necessary to reduce risks posed through food contamination. We present a BoNT biosensor employing living neural cultures grown in vitro on microelectrode arrays (MEAs). An MEA is a culture dish with a grid of electrodes embedded in its surface, enabling extracellular recording of action potentials of neural cultures grown over the array. Pharmaceutical grade BoNT A was applied to the media bath of mature cortical networks cultured on MEAs. Both spontaneous and evoked activities were monitored over 1 wk to quantify changes in the neural population produced by BoNT A. Introduction of BoNT A resulted in an increased duration and number of spikes in spontaneous and evoked bursts relative to control cultures. Increases were significant within 48 h of BoNT A dosage (P < 0.05). Application of BoNT A also induced unique oscillatory behavior within each burst that is reminiscent of early developmental activity patterns rather than the mature cultures used here. Three or more activity peaks were observed in 50% of the BoNT dosed cultures. Control cultures exhibited only a single activity peak. Thus activity of these cortical networks measured with MEAs could provide a valuable substrate for BoNT detection.

A total of 32 microwire electrodes were implanted bilaterally into the hippocampus of Sprague-Dawley rats, which were then stimulated in the manner prescribed for the chronic limbic epilepsy model. After the initial seizure brought on by the stimulation, the animals were recorded at a high sampling rate (approximately 12 kHz) for the entire duration of the latent period. Coherence was calculated across channels in both stimulated (and later seizing) animals and non-stimulated (and thus non-seizing control) animals. Average coherence over time was greatest in intrahemispherical electrode pairs in both stimulated and non-stimulated animals. However, the 200-800 Hz band displays increased coherence interhemispherically and up to 200 Hz band displays decreased coherence interhemispherically: this occurs only in stimulated animals.

The advanced and robust computational power of the brain is shown by the complex behaviors it produces. By embodying living cultured neuronal networks with a robotic or simulated animal (animat) and situating them within an environment, we study how the basic principles of neuronal network communication can culminate into adaptive goal-directed behavior. We engineered a closed-loop biological-robotic drawing machine and explored sensory-motor mappings and training. Preliminary results suggest that real-time performance-based feedback allowed an animat to draw in desired directions. This approach may help instruct the future design of artificial neural systems and of the algorithms to interface sensory and motor prostheses with the brain.

We propose an efficient algorithm to compute the smoothed correlogram for the detection of temporal relationship between two spike trains. Unlike the conventional histogram-based correlogram estimations, the proposed algorithm operates on continuous time and does not bin either the spike train nor the correlogram. Hence it can be more precise in detecting the effective delay between two recording sites. Moreover, it can take advantage of the higher temporal resolution of the spike times provided by the current recording methods. The Laplacian kernel for smoothing enables efficient computation of the algorithm. We also provide the basic statistics of the estimator and a guideline for choosing the kernel size. This new technique is demonstrated by estimating the effective delays in a neuronal network from synthetic data and recordings of dissociated cortical tissue.

In this paper we present three related areas of research we are pursuing to study neural computation in vitro. Rat cortical neurons cultured on 60 channel multielectrode array (MEA) allow the researcher to measure from and stimulate sixty different sites across a small population of neurons grown in vitro. Using this system we can send stimulation patterns into the network and study how these living neural networks compute by measuring its outputs. Our first series of studies uses chaotic control techniques to study the dynamics and potentially control the behavior of cortical network. At the same time, we are beginning to apply a model of computation called the liquid state machine or LSM model developed by Wolfgang Maass to provide a firm mathematical framework from which to proceed with our investigations. Each of these components is integrated into a third area investigating the role of computation and feedback using a real-time sensory-motor feedback robotic flight system.

By combining MEA electrophysiology with long-term time-lapse imaging, it is possible to make correlations between changes in network function and changes in neuronal morphology. By re-embodying dissociated cultured networks, network function can be mapped onto behavior, and in vitro research can now make use of a new kind of behavioral studies that include detailed (submicron) imaging not possible in vivo. By closing the sensory-motor loop around MEA cultures, they are more likely to shed light on the mechanisms of learning, memory, and information processing in animals.

High Frequency Oscillations (HFO) in limbic epilepsy represent a marked difference between abnormal and normal brain activity. Faced with the difficult of visually detecting HFOs in large amounts of intracranial EEG data, it is necessary to develop an automated process. This paper presents Teager Energy as a method of finding HFOs. Teager energy is an ideal measure because unlike conventional energy it takes into account the frequency component of the signal as well as signal amplitude. This greatly aids in the dissection of HFOs out of the noise and other signals contained in the EEG. Therein, Teager energy analysis is able to detect high- frequency, low-amplitude components that conventional energy measurements would miss.

Epilepsy is a common neurological disorder that can have damaging effects in the brain including over 50% loss of neuronal activity in the hippocampal regions of the CA1 and CA3. The pre-ictal period was studied in an animal model of limbic epilepsy using Shannon entropy and correlation analysis. The primary aim was to uncover underlying relative changes in signals between the Dentate Gyrus and CA1 areas of the bilateral hippocampus. Preliminary entropy analysis results included dynamical changes between channels in the Dentate Gyrus and channels in the CA1 region at and around the time of the seizure. Epilepsy affects 3-5% of the population worldwide. Epilepsy is a neurological disorder characterized by recurrent and unprovoked seizures. An individual loses awareness when experiencing a complex partial seizure due to the spread of the seizure through both temporal lobes and subsequently impairing memory. (1) Of all cases, approximately 60% respond favorably to anti-epileptic drugs. (2) Regardless of age, sex or race, the harmful effects of limbic Epilepsy can be caused by past infections, vascular malformations, hamartomas and gliomas. Head trauma in the form of hemorrhaging or contusion in the brain often leads to the development of limbic Epilepsy after a number of months to years. This span of time is known as a latent period when cellular and network changes are thought to occur precipitating the onset of seizures. In epileptogenesis over 50% of the neurons in the hippocampal regions of the CA1 and CA3 are lost. Neuronal loss also occurs with the granule cells in the Dentate Gyrus; accompanying these changes is a loss of inhibitory neurons, excitatory neurons and excitatory axonal sprouting. (1) The Chronic Limbic Epilepsy rat model imitates human limbic epilepsy with the initial insult to the brain quiescent period and resultant seizures later in life. The manner in which these seizures develop is thought to be a result of structural changes in the brain such as the strengthening of excitatory networks, loss of inhibitory neurons or suppression of GABA receptors. (3) Since little is known about the time period over which the changes occur, it is proposed that detectable changes occur gradually within the brain over the latent period eventually causing the later hypersynchronous seizure activity. The link between the nature of the pre-ictal period and the electrical changes manifested in the brain are not well characterized. The abnormal mode of communication, a characteristic of seizures, is demonstrated by large- amplitude wave discharges occurring over a large hemisphere of the brain. The preictal period in an animal model of limbic epilepsy is studied using Shannon entropy measurements. The goal of this research is to characterize underlying changes in signals between the Dentate Gyrus and CA1 areas of the bilateral hippocampus.

Understanding biological information processing systems is key in overcoming the limitations of traditional engineered systems. The advent of the liquid state machine (LSM) as a computational model for cortical processing, as well as the ability to experiment with dissociated cortical tissue (DCT) cultures in a controlled environment provide new opportunities in advancing our understanding of such systems. In this paper we examine the possibilities for modeling the behavior of the DCT cultures in the LSM framework. We show that the LSM framework has the capability to model the spontaneous activity and burstiness of the DCT cultures. Finally, multifractal measures are used to characterize the long range dependencies of the recorded and simulated data. Though the detrended fluctuation analysis (DFA) used to estimate the long range dependencies in the data does not show wholly consistent similarities, the endogenously active neurons that drive the biological and model networks are found to have similar fractal structure.

The brain is perhaps one of the most robust and fault tolerant computational devices in existence and yet little is known about its mechanisms. Microelectrode arrays have recently been developed in which the computational properties of networks of living neurons can be studied in detail. In this paper we report work investigating the ability of living neurons to act as a set of neuronal weights which were used to control the flight of a simulated aircraft. These weights were manipulated via high frequency stimulation inputs to produce a system in which a living neuronal network would "learn" to control an aircraft for straight and level flight.

Delivery of pharmacological agents in vitro can often be a difficult, time consuming and costly process. In this paper, we describe an economical method for in vitro delivery using a hydrogel of poly hydroxyethyl methacrylate (PHEMA) that can absorb up to 50% of its weight of any water-solubilized pharmacological agent. This agent will then passively diffuse into surrounding media upon application in vitro. An in vitro test of PHEMA as a drug delivery device was conducted using dissociated rat-cortical neurons cultured on micro-electrode arrays. These micro-electrode arrays permit the real-time measurement of neural activity at 60 different sites across a network of neurons. Neural activity was compared during the application of PHEMA saturated with cell culture media and PHEMA saturated with bicuculline, a widely used pharmacological agent with stereotypical effects on neural activity patterns. Application of PHEMA saturated with bicuculline produced a gradual increase in concentration in vitro. When the minimum effective concentration of bicuculline was reached, which was found to be 0.59 microM using the diffusion properties of PHEMA, it produced the rapid almost periodic synchronized bursting characteristically associated with this agent. In contrast, the application of PHEMA saturated in culture media alone had no effect on neural activity reinforcing its inherent inert properties. Since PHEMA is nontoxic, can be molded into a variety of shapes, quickly manufactured in any laboratory and is inexpensive to produce, the material represents a promising alternative to drug delivery systems on the market today.

In this study, pigeons learned 2 separate one-to-many conditional discriminations in which they matched form samples to line and hue comparisons. Correct choices within each comparison dimension yielded differential (food vs. no-food) outcomes that were not predictable from the samples alone. At asymptote, latency to make a correct choice was shorter when food was the contingent outcome than when no food was the outcome. More important, when the samples from each task were subsequently exchanged, comparison choice varied systematically as a function of the sample and the set of new comparison alternatives that followed them. Together, these results indicate that choices were cued by differential outcome expectancies arising from serial compounds consisting of each sample and the dimensional characteristics of the comparisons.

We present a software suite, MeaBench, for data acquisition and online analysis of multi-electrode recordings, especially from micro-electrode arrays. Besides controlling data acquisition hardware, MeaBench includes algorithms for real-time stimulation artifact suppression and spike detection, as well as programs for online display of voltage traces from 60 electrodes and continuously updated spike raster plots. MeaBench features real-time output streaming, allowing easy integration with stimulator systems. We have been able to generate stimulation sequences in response to live neuronal activity with less than 20 ms lag time. MeaBench is open-source software, and is available for free public download at http://www.its.caltech.edu/~pinelab/wagenaar/meabench.html.