Table of contents

Some issues in neuroscience can be addressed by building robot models of biological sensorimotor systems. What we can conclude from building models or simulations, however, is determined by a number of factors in addition to the central hypothesis we intend to test. These include the way in which the hypothesis is represented and implemented in simulation, how the simulation output is interpreted, how it is compared to the behaviour of the biological system, and the conditions under which it is tested. These issues will be illustrated by discussing a series of robot models of cricket phonotaxis behaviour.

This tutorial presents an architecture for autonomous robots to generate behavior in joint action tasks. To efficiently interact with another agent in solving a mutual task, a robot should be endowed with cognitive skills such as memory, decision making, action understanding and prediction. The proposed architecture is strongly inspired by our current understanding of the processing principles and the neuronal circuitry underlying these functionalities in the primate brain. As a mathematical framework, we use a coupled system of dynamic neural fields, each representing the basic functionality of neuronal populations in different brain areas. It implements goal-directed behavior in joint action as a continuous process that builds on the interpretation of observed movements in terms of the partner's action goal. We validate the architecture in two experimental paradigms: (1) a joint search task; (2) a reproduction of an observed or inferred end state of a grasping–placing sequence. We also review some of the mathematical results about dynamic neural fields that are important for the implementation work.

The magnitude of brain tissue micromotion relative to stationary brain implants and its impact on the viability and function of the surrounding brain tissue due to mechanical stresses is poorly understood. The central goal of this study is to characterize surface micromotion in the somatosensory cortex against stationary cylindrical implants. We used a differential variable reluctance transducer (DVRT) in adult rats (n = 6) to monitor micromotion normal to the somatosensory cortex surface. Experiments were performed both in the presence and in the absence of dura mater and displacement measurements were made at three different locations within craniotomies of two different sizes. In anesthetized rats, pulsatile surface micromotion was observed to be in the order of 10–30 µm due to pressure changes during respiration and 2–4 µm due to vascular pulsatility. Brain displacement values due to respiration were significantly lower in the presence of the dura compared to those without the dura. In addition, large inward displacements of brain tissue between 10–60 µm were observed in n = 3 animals immediately following the administration of anesthesia. Such significant micromotion can impact a wide variety of acute and chronic procedures involving any brain implants, precise neurosurgery or imaging and therefore has to be factored in the design of such procedures.

Long-term integration of neuroprosthetic devices is challenged by reactive responses that compromise the brain–device interface. The contribution of physical insertion parameters to immediate damage is not well described. We have developed an ex vivo preparation to capture real-time images of tissue deformation during device insertion using thick tissue slices from rat brains prepared with fluorescently labeled vasculature. Qualitative and quantitative assessments of damage were made for insertions using devices with different tip shapes inserted at different speeds. Direct damage to the vasculature included severing, rupturing and dragging, and was often observed several hundred micrometers from the insertion site. Slower insertions generally resulted in more vascular damage. Cortical surface features greatly affected insertion success; insertions attempted through pial blood vessels resulted in severe tissue compression. Automated image analysis techniques were developed to quantify tissue deformation and calculate mean effective strain. Quantitative measures demonstrated that, within the range of experimental conditions studied, faster insertion of sharp devices resulted in lower mean effective strain. Variability within each insertion condition indicates that multiple biological factors may influence insertion success. Multiple biological factors may contribute to tissue distortion, thus a wide variability was observed among insertions made under the same conditions.

This paper compares different ICA preprocessing algorithms on cross-validated training data as well as on unseen test data. The EEG data were recorded from 22 electrodes placed over the whole scalp during motor imagery tasks consisting of four different classes, namely the imagination of right hand, left hand, foot and tongue movements. Two sessions on different days were recorded for eight subjects. Three different independent components analysis (ICA) algorithms (Infomax, FastICA and SOBI) were studied and compared to common spatial patterns (CSP), Laplacian derivations and standard bipolar derivations, which are other well-known preprocessing methods. Among the ICA algorithms, the best performance was achieved by Infomax when using all 22 components as well as for the selected 6 components. However, the performance of Laplacian derivations was comparable with Infomax for both cross-validated and unseen data. The overall best four-class classification accuracies (between 33% and 84%) were obtained with CSP. For the cross-validated training data, CSP performed slightly better than Infomax, whereas for unseen test data, CSP yielded significantly better classification results than Infomax in one of the sessions.

The confluence of micropatterning, microfabricated multielectrode arrays, and low-density neuronal culture techniques make possible the growth of patterned neuronal circuits overlying multielectrode arrays. Previous studies have shown synaptic interaction within patterned cultures which was more active on average than random cultures. In our present study, we found patterned cultures to have up to five times more astrocytes and three times more neurons than random cultures. In addition, faster development of synapses is also seen in patterned cultures. Together, this yielded greater overall neuronal activity as evaluated by the number of active electrodes. Our finding of astrocytic proliferation within serum-free culture is also novel.

A novel measure of spectral distance is presented, which is inspired by the prediction residual parameter presented by Itakura in 1975, but derived from frequency domain data and extended to include autoregressive moving average (ARMA) models. This new algorithm is applied to electroencephalogram (EEG) data from newborn piglets exposed to hypoxia for the purpose of early detection of hypoxia. The performance is evaluated using parameters relevant for potential clinical use, and is found to outperform the Itakura distance, which has proved to be useful for this application. Additionally, we compare the performance with various algorithms previously used for the detection of hypoxia from EEG. Our results based on EEG from newborn piglets show that some detector statistics divert significantly from a reference period less than 2 min after the start of general hypoxia. Among these successful detectors, the proposed spectral distance is the only spectral-based parameter. It therefore appears that spectral changes due to hypoxia are best described by use of an ARMA- model-based spectral estimate, but the drawback of the presented method is high computational effort.

We describe a new technique for the classification of motor imagery electroencephalogram (EEG) recordings in a brain computer interface (BCI) task. The technique is based on an adaptive time–frequency analysis of EEG signals computed using local discriminant bases (LDB) derived from local cosine packets (LCP). In an offline step, the EEG data obtained from the C₃/C₄ electrode locations of the standard 10/20 system is adaptively segmented in time, over a non-dyadic grid by maximizing the probabilistic distances between expansion coefficients corresponding to left and right hand movement imagery. This is followed by a frequency domain clustering procedure in each adapted time segment to maximize the discrimination power of the resulting time–frequency features. Then, the most discriminant features from the resulting arbitrarily segmented time-frequency plane are sorted. A principal component analysis (PCA) step is applied to reduce the dimensionality of the feature space. This reduced feature set is finally fed to a linear discriminant for classification. The online step simply computes the reduced dimensionality features determined by the offline step and feeds them to the linear discriminant. We provide experimental data to show that the method can adapt to physio-anatomical differences, subject-specific and hemisphere-specific motor imagery patterns. The algorithm was applied to all nine subjects of the BCI Competition 2002. The classification performance of the proposed algorithm varied between 70% and 92.6% across subjects using just two electrodes. The average classification accuracy was 80.6%. For comparison, we also implemented an adaptive autoregressive model based classification procedure that achieved an average error rate of 76.3% on the same subjects, and higher error rates than the proposed approach on each individual subject.

We introduce a simple model to produce EEG-like signals. The model is based on the assumption that the number of active nerve cells that generate an electric field, at a given time, is essentially chaotic. In accordance, we use the logistic equation together with a spike-like function to simulate the neuronal activity processes. With this model, we are able to generate EEG-like patterns, with quite a short time of calculation. Real pre-recorded neuronal and simulated signals, as well as their power spectra, are compared in terms of the main conventional EEG frequency peaks.

Neural engineering has grown substantially in the last few years and it is time to review the progress of the first journal in this field. Journal of Neural Engineering (JNE) is a quarterly publication that started in 2004. The journal is now in its third volume and eleven issues, consisting of 114 articles in total, have been published since its launch. The editorial processing times have been kept to a minimum, the receipt to first decision time is 41 days, on average, and the time from receipt to publication has been maintained below three months. It is also worth noting that it is free to publish in Journal of Neural Engineering—there are no author fees—and once published the articles are free online for the first month. The journal has been listed in Pubmed^® since 2005 and has been accepted by ISI^® in 2006.

Who is reading Journal of Neural Engineering? The number of readers of JNE has increased significantly from 8050 full-text downloads in 2004 to 14 900 in 2005 and the first seven months of 2006 have already seen 12 800 downloads. The top users in 2005 were the Microsoft Corporation, Stanford University and the University of Michigan. The list of top ten users also includes non-US institutions: University of Toronto, University of Tokyo, Hong Kong Polytechnic, National Library of China and University College London, reflecting the international flavor of the journal.

What are the hot topics in neural engineering? Based on the number of downloads and citations for 2004–2005, the top three topics are:

(1) Brain–computer interfaces (2) Visual prostheses (3) Neural modelling

Several other topics such as microelectrode arrays, neural signal processing, neural dynamics and neural circuit engineering are also in the top ten.

Where are Journal of Neural Engineering articles cited? JNE articles have reached a wide audience and have been cited in of some of the best journals in physiology and neuroscience such as Nature Neuroscience, Journal of Neuroscience, Trends in Neuroscience, Journal of Physiology, Proceedings of the National Academy of Science as well as in engineering and physics journals such as Annals of Biomedical Engineering, Physical Review Letters and IEEE Transactions on Biomedical Engineering. However, the number of citations in clinical journals is limited.

What is special about Journal of Neural Engineering? JNE has published two special issues: (1) The Eye and the Chip (visual prostheses) (vol. 2, (1), 2005) and (2) Sensory Integration: Role of Internal Models (vol. 2, (3), 2005). These special issues have attracted a lot of attention based on the number of article downloads. JNE also publishes tutorials intended to provide background information on specific topics such as classification, sensory substitution and cortical neural prosthetics. A series of tutorials from the 3rd Neuro-IT and Neuroengineering Summer School has been published with the first appearing in vol. 2 (4), 2005.

What is in the future for Journal of Neural Engineering? The goal of any journal should be to provide a particular field with the best venue for scientists and engineers to make their work available and noticeable to the rest of the community. In particular, attracting a strong readership base and high quality manuscripts should be the first priority. Providing accurate, reliable and speedy reviews should be the next. With an international board of experts in the field of neural engineering, a solid base of reviewers, readers and contributors, JNE is in a strong position to continue to serve the neural engineering community. However, this is still a small community and growth is essential for continued success in this area. There are two areas of expansion of great interest for the field of neural engineering currently poised between basic science on one hand and clinical implementation on the other: translational neuroscience and therapeutic neural engineering. We should strive to bridge the gap between basic neuroscience, clinical science and engineering by attracting contributions from neuroscientists and clinicians with an interest in neural engineering. I urge members of the neural engineering community to encourage their colleagues in these areas to consider JNE for publication of those manuscripts at the interface with neuroscience and engineering.

I would like to take this opportunity to acknowledge the work of the board members, the reviewers of the articles and the staff at the Institute of Physics Publishing for their contribution to the Journal of Neural Engineering.