Biological cybernetics

This paper presents a control methodology for achieving orbital stabilization with simultaneous time synchronization of periodic trajectories in underactuated robotic systems. The proposed approach extends the classical transverse linearization framework to explicitly incorporate time-desynchronization dynamics. To stabilize the resulting extended transverse dynamics, we employ a combination of time-varying LQR and sliding-mode control. The theoretical results are validated experimentally through the implementation of both centralized and decentralized control strategies on a group of six Butterfly robots.

To uncover the underlying control structure of three-ball cascade juggling, we studied its spatiotemporal properties in detail. Juggling patterns, performed at fast and preferred speeds, were recorded in the frontal plane and subsequently analyzed using principal component analysis and serial correlation techniques. As was expected on theoretical grounds, the principal component analysis revealed that maximally four instead of the original six dimensions (3 balls  2 planar coordinates) are sucient for describing the juggling dynamics. Juggling speed was shown to aect the number of dimensions (four for the fast condition, two for the preferred condition) as well as the smoothness of the time evolution of the eigenvectors of the principal component analysis, particularly around the catches. Contrary to the throws and the zeniths, and regardless of juggling speed, consecutive catches of the same hand showed a markedly negative lag-one serial correlation, suggesting that the catches are timed so as to preserve the temporal integrity of the juggling act.

We examined the development of task-specific couplings among functional subsystems (i.e., ball circulation, respiration, and body sway) when learning to juggle a three-ball cascade, with a focus on learninginduced changes in the coupling between ball movements and respiration and the coupling between ball movements and body sway. Six novices practiced to juggle three balls in cascade fashion for one hour per day for twenty days. On specific days (7 in total), ball movements, center-of-pressure (CoP) trajectories and respiration traces were measured simultaneously. Discrete, time-continuous and spectral analyses revealed that the spatio-temporal variability of the juggling patterns decreased with practice and that the degree to which the task constraints were satisfied increased gradually. No conclusive evidence was found for ball movementrespiration coupling. In contrast, clear-cut evidence was found for the presence of 1:3 and 2:3 frequency locking between the vertical component of the ball trajectories and both the anterior-posterior and the medio-lateral components of the CoP. Incidence and expression of these mode locks varied across individuals and altered in the course of learning. Gradual changes in locking strength, appearances and disappearances of mode locks, as well as abrupt transitions between coupled states were observed. These results indicate that dissimilar learning dynamics may arise in the functional embedding of subsystems into a task-specific organization and that motor equivalence is an inherent property of such emerging task-specific organizations.

Ecol Psychol 7: 291±314] argued that the relative phase dynamics of rhythmic interlimb coordination may be attributed to the timing level in that the stability properties of the relative phase are largely independent of dynamical principles operating at the goal level, such as those related to the maintenance of a particular amplitude or target position. Yet, according to the coupling functions in the coupled oscillator model proposed by Biol Cybern 51: 347±356], the eect of frequency on the stability properties of relative phase is either wholly or partially mediated by frequencyinduced changes in amplitude, implying that the relative phase dynamics strongly depends on spatial factors. In order to distinguish between these contrasting interpretations of the organizational principles underwriting the phase dynamics of interlimb coordination, an experiment was conducted in which the eects of frequency and amplitude on the stability of relative phase were separated. Six subjects performed both in-phase and antiphase coordination patterns at seven dierent frequencies and three dierent amplitudes. Two measures of pattern stability were used, the standard deviation of relative phase and the exponent of the relaxation process following phasic perturbations of relative phase. According to both measures, pattern stability decreased with increasing frequency, whereas the amplitude manipulation only had a signi®cant eect on the standard deviation of relative phase. This result was interpreted to imply that the organizational principles at the (relative) timing level are aected only moderately by task constraints pertaining to the goal level, and that models of interlimb coordination in which amplitude coupling plays a partial or subordinate role should be preferred above models relying solely on amplitude coupling.

We focus on a generic multiterminal remote source coding scenario, appearing in a variety of real-world applications. Specifically, several noisy observations from a remote user/source signal should be quantized at some intermediate nodes prior to a forward transmission via multiple error-free and rate-limited links to a processing unit. To design the local quantizers, we follow the Information Bottleneck method, and devise a purely data-driven solution, which can be categorized as a Latent Variable Model in the context of generative AI. To that end, we derive a tractable variational lower-bound of the original objective functional, and present the pertinent learning architecture, over which, the design problem can be addressed by the joint training of the encoder DNNs and the decoder DNN, e.g., by some form of the Stochastic Gradient Descent. By several numerical investigations, we further show that this data-driven compression scheme performs (almost) on par with the SotA model-based approach, without requiring the prior knowledge of the (joint) statistics of input signals. This becomes quite important, especially, in those applications where the joint input statistics are either unavailable or hard to estimate.

The outer retina removes the first-order correlation, the background light level, and thus more efficiently transmits contrast. This removal is accomplished by negative feedback from horizontal cell to photoreceptors. However, the optimal feedback gain to maximize the contrast sensitivity and spatial resolution is not known. The objective of this study was to determine, from the known structure of the outer retina, the synaptic gains that optimize the response to spatial and temporal contrast within natural images. We modeled the outer retina as a continuous 2D extension of the discrete 1D model of Yagi et al. (Proc Int Joint Conf Neural Netw 1: 787-789, 1989). We determined the spatiotemporal impulse response of the model using small-signal analysis, assuming that the stimulus did not perturb the resting state of the feedback system. In order to maximize the efficiency of the feedback system, we derived the relationships between time constants, space constants, and synaptic gains that give the fastest temporal adaptation and the highest spatial resolution of the photoreceptor input to bipolar cells. We found that feedback which directly modulated photoreceptor calcium channel activation, as opposed to changing photoreceptor voltage, provides faster adaptation to light onset and higher spatial resolution. The optimal solution suggests that the feedback gain from horizontal cells to photoreceptors should be ~0.5. The model can be extended to retinas that have two or more horizontal cell networks with different space constants. The theoretical predictions closely match experimental observations of outer retinal function.

Stimulus representation is a functional interpretation of early sensory cortices. Early sensory cortices are subject to stimulus-induced modi®cations. Common models for stimulus-induced learning within topographic representations are based on the stimuli's spatial structure and probability distribution. Furthermore, we argue that average temporal stimulus distances re¯ect the stimuli's relatedness. As topographic representations re¯ect the stimuli's relatedness, the temporal structure of incoming stimuli is important for the learning in cortical maps. Motivated by recent neurobiological ®ndings, we present an approach of cortical self-organization that additionally takes temporal stimulus aspects into account. The proposed model transforms average interstimulus intervals into representational distances. Thereby, neural topography is related to stimulus dynamics. This oers a new time-based interpretation of cortical maps. Our approach is based on a wave-like spread of cortical activity. Interactions between dynamics and feedforward activations lead to shifts of neural activity. The psychophysical saltation phenomenon may represent an analogue to the shifts proposed here. With regard to cortical plasticity, we oer an explanation for neurobiological ®ndings that other models cannot explain. Moreover, we predict cortical reorganizations under new experimental, spatiotemporal conditions. With regard to psychophysics, we relate the saltation phenomenon to dynamics and interaction in early sensory cortices and predict further eects in the perception of spatiotemporal stimuli.

A neural network approach to stereovision is presented based on aliasing effects of simple disparity estimators and a fast coherencedetection scheme. Within a single network structure, a dense disparity map with an associated validation map and, additionally, the fused cyclopean view of the scene are available. The network operations are based on simple, biological plausible circuitry; the algorithm is fully parallel and non-iterative.

A neural network approach to stereovision is presented based on aliasing effects of simple disparity estimators and a fast coherencedetection scheme. Within a single network structure, a dense disparity map with an associated validation map and, additionally, the fused cyclopean view of the scene are available. The network operations are based on simple, biological plausible circuitry; the algorithm is fully parallel and non-iterative.

This paper concerns the processing of the outputs of the two opponent-color mechanisms in the human visual system. We present experimental evidence that opponent-color signals interact after joint modulation even though they are essentially independent under neutral steady adaptation and after exclusive modulation of each mechanism. In addition, prolonged modulation linearizes the response function of each mechanism. The changes in interaction serve to orthogonalize opponent signals with respect to the adapting modulation, and the changes in response functions serve to equalize the relative frequencies of different levels of response to the adapting modulation. Adaptive orthogonalization reduces sensitivity to the adapting color direction, improves sensitivity to the orthogonal direction, and predicts shifts in color appearance. Response equalization enhances effective contrast and explains the difference between the effects of adaptation to uniform versus temporally or spatially modulated stimuli.

We show that an assumption of rigidity or quasi-rigidity is not necessary, in principle, for the computation of three-dimensional structure and motion from changing retinal images. In particular, we show that the three-dimensional structure of certain nonrigid objects, namely objects whose texture elements rotate about a common axis but at varying angular velocities, can in principle be computed from three successive retinal images of four texture elements, or from four successive images of two texture elements. We then show that in both cases the computed structure matches the actual structure of the object with probability one.

The human brain is composed of two hemispheres. Even though most functions are represented in both, they differ in processing abilities, enabling the left hemisphere to speak and control learned motor sequences. One current hypothesis how the hemispheres differ is in their processing of relative frequencies of sensory stimuli (Ivry and Robertson, 1998; Flevaris et al., 2010). The Double-filtering-by- frequency (DFF) theory proposes that the left hemisphere has a pref- erence to process relative high frequencies and the right hemisphere relative low frequencies. The authors hypothesize that, hemispheric differences in sensory processing should transfer to the speech and motor domain. The goal of this thesis was to investigate frequency dependent hemispheric preferences for hand motor control. An fMRI and an MEG experiment were performed to answer the following questions: Is there a hemispheric preference for relative movement frequencies visible in behavioral measures? What are the co...

We designed four arborized neurons which are able to evaluate the exclusive-or (XOR) function from two inputs. The input neurons form exclusively excitatory synapses on a dendritic tree which is a patchwork of "passive" (ohmic) and "active" cable segments. The active segments are described by the Hodgkin-Huxley model. The dynamics of the neurons and their output are obtained by numerical integration of the cable equation. In neurons 1 and 2 the XOR function is based on the annihilation of colliding action potentials. In neuron No. 3 the design takes advantage of the refractory period of action potentials. In neuron No. 4 voltage inversion is used as it occurs for inactivated sodium conductance in the Hodgkin-Huxley model. In all cases the XOR function depends critically on an appropriate timing of the input signals and on delays of the voltage transients in different branches of the dendrite.

A multiple choice experiment with free flying bees trained to a color signal is described which allows for multidimensional scaling of color similarity. The choice proportions are analysed by metric non-metric (Kruskal 1964a, b) multidimensional scaling. The light reflected from the twelve color signals used differed in spectral composition, intensity, and the proportion of white light. Only two scales are necessary to reconstruct the experimental data. The interpretation of the scale values by Helmholtzcoordinates, derived from the chromaticity diagram for bees, shows that the main perceptual parameters are hue and saturation (or blue/greenness and UV/blue-greenness, respectively). Brightness is ignored by the bees in this choice situation. The total color difference is related to the differences on the two perceptual parameters by the city-block metric (Minkowski exponent p = 1).

We describe two psychophysical experiments testing predictions of the square difference mechanism we have previously proposed for intensity-based stereo. Experiment 1 assesses the relative contributions of disparity and contrast to intensity-based stereo by measuring detection thresholds. The product of disparity and contrast at threshold is shown to be constant. In experiment 2, we measure quantitatively the global depth position perceived in stereograms of curved, smoothly shaded surfaces. The results show that disparity averaging over the surface involves a contrastdependent weighting function. The results from both experiments are consistent with predictions derived from the square difference mechanism. The relation of this mechanism to feature correspondence stereopsis and shape-from-shading is discussed and a general framework for assessing the modularity of stereopsis is presented.

Maintaining balance during quiet standing is a challenging task for the neural control mechanisms due to the inherent instabilities involved in the task. The feedback latencies and the lowpass characteristics of skeletal muscle add to the difficulty of regulating postural dynamics in real-time. Inverted-pendulum (IP) type robotic models have served as a popular paradigm to investigate control of postural balance. In this study, an in-depth neuromechanical postural control model is developed from physiological principles. The model comprises a single-segment IP robotic model, Hill-type muscle model, and proprioceptive feedback from the muscle spindle (MS) and golgi tendon organ (GTO). An optimal proportional-integral-derivative (PID) controller is proposed to realize effective postural control amid latencies in sensory feedback. The neural commands for postural stabilization are generated by a time-varying PID controller, tuned using linear quadratic regulator (LQR) principles. Computer simulations are used to assess the efficacy of the tuned PID-LQR controller. Sensitivity analysis of the controlled system shows a delay tolerance of 300ms. Preliminary empirical data in support of the mathematical model were obtained from perturbation experiments. The model response to perturbation torque, measured in terms of the center of mass (COM) excursion in the anterior-posterior (AP) direction, displays a high degree of correlation with the empirical data (ρ = 0.91).

What are the simplest search strategies that lead an animal to a particular target, what are their limitations, and what changes can be made to develop more effective strategies? To answer these questions a class of search strategies was examined that require an animal to have only a minimal capacity for spatial orientation; the effectiveness of such strategies in solving the following basic search problem was determined. The animal begins its search at a distance r 0 (starting distance) from a spatially fixed target. It detects the target when it has approached it to within a certain distance a (the detection radius). The analysed class of search strategies has the following characteristics: C 1. The animal uses the same search strategy in all regions it enters. Therefore it needs no information as to the actual location of the target. C2. Its search strategy is constant in time. The animal has only to detect whether it has reached the target or not. C3. Once the animal has chosen a direction, it continues in that direction for a certain distance. This is the only way in which the preceding parts of the search affect the animal's decision as to the direction in which it will search next. In the long term the animal's movement directions are independent, with no preference for any particular direction. Despite their extreme simplicity in application these "Brownian" search strategies are remarkably successful (Fig. ). Indeed, if the search is continued long enough the target is certain to be found. The success of the search depends in part on the search duration (Fig. ) or the search-path length S, the starting distance and the detection radius. On the other hand, an animal can have a decisive influence on its degree of success simply by adjusting the frequency with which it changes its walking direction to match its sensory abilities. That is, a not-too-short Brownian search (S>>ro) is most successful when the searching animal, between the points at which it changes direction, walks approximately straight for a distance equal to the detection radius (Figs. and). A further in-crease in search effectiveness is possible only by turning to another class of search strategies. These, however, demand that the animal either have more information about the position of its target at the beginning of the search or be able to organize its search behavior even over fairly long periods of time.

Nicotinic acetylcholine receptors are transmembrane oligomeric proteins that mediate interconversions between open and closed channel states under the control of neurotransmitters. Fast in vitro chemical kinetics and in vivo electrophysiological recordings are consistent with the following multi-step scheme. Upon binding of agonists, receptor molecules in the closed but activatable resting state (the Basal state, B) undergo rapid transitions to states of higher affinities with either open channels (the Active state, A) or closed channels (the initial Inactivatable and fully Desensitized states, I and D). In order to represent the functional properties of such receptors, we have developed a kinetic model that links conformational interconversion rates to agonist binding and extends the general principles of the Monod-Wyman-Changeux model of allosteric transitions. The crucial assumption is that the linkage is controlled by the position of the interconversion transition states on a hypothetical linear reaction coordinate. Application of the model to the peripheral nicotinic acetylcholine receptor (nAChR) accounts for the main properties of ligand-gating, including single-channel events, and several new relationships are predicted. Kinetic simulations reveal errors inherent in using the dose-response analysis, but justify its application under defined conditions. The model predicts that (in order to overcome the intrinsic stability of the B state and to produce the appropriate cooperativity) channel activation is driven by an A state with a K in the 50 nM range, hence some 140-fold stronger than the apparent affinity of the open state deduced previously. According to the model, recovery from the desensitized states may occur via rapid transit through the A state with minimal channel opening, thus without necessarily undergoing a distinct recovery pathway, as assumed in the standard 'cyclic' model. Tran-

A new neural network model with feedback based on the concept of information storage matrices is proposed. This model is similar to the Hopfield and spectral type neural networks but has a more general structure. The presentation gives a fully developed theory for first-order networks, including results on the formation of fixed points and their domains of attraction. These results are used to determine, in deterministic sense, the information storage capacity. The algorithm is applied to the DNA sequencing problem. It is demonstrated how a hidden genetic information in an arbitrary long DNA strand can be extracted. This paper presents a spectral type algorithm based on a global learning principle which determines a feedback dynamic system. The proposed paradigm is globally stable and exhibits a dissipative neurodynamics which can be applied in problems with either a heteroor auto-associative memory. The model presented focuses on a first-order discrete time, finite step, neural network architecture that operates on binary numbers. This approach is particularly appealing for silicon implementation. The continuous version is not treated in the paper. However, the presented results apply to the continuous case when difference equations are replaced by differential equations.

A new neural network model with feedback based on the concept of information storage matrices is proposed. This model is similar to the Hopfield and spectral type neural networks but has a more general structure. The presentation gives a fully developed theory for first-order networks, including results on the formation of fixed points and their domains of attraction. These results are used to determine, in deterministic sense, the information storage capacity. The algorithm is applied to the DNA sequencing problem. It is demonstrated how a hidden genetic information in an arbitrary long DNA strand can be extracted. This paper presents a spectral type algorithm based on a global learning principle which determines a feedback dynamic system. The proposed paradigm is globally stable and exhibits a dissipative neurodynamics which can be applied in problems with either a heteroor auto-associative memory. The model presented focuses on a first-order discrete time, finite step, neural network architecture that operates on binary numbers. This approach is particularly appealing for silicon implementation. The continuous version is not treated in the paper. However, the presented results apply to the continuous case when difference equations are replaced by differential equations.

) have been proposed to work over different combinatorial problems. However, both MPPMX and ABC neither suffered from a very high computational time neither poor performance. Therefore, this work proposes a novel multiparent order crossover (MPOX) for solving several combinatorial optimization problems with reasonable amount of time. The MPOX extends the two-parent order crossover by modifying the recombination operator to recombine more than two parents and generates a new offspring. MPOX at first selects the crossover points and divides the parents into n substrings based on these points (where n is the number of parents). Then, MPOX copies a predefined number of elements from each parent into the offspring based on their order while checking the feasibility of the offspring. The performance of MPOX is tested on the traveling salesman problems and berth allocation problems, which are widely studied in the literature. Experimental results demonstrated that the MPOX significantly improves the OX in both problem domains and outperforms both ABC and MPPMX over the travelling salesman problem and the berth allocation problem with less computational time. These results indicate the effectiveness of MPOX over OX, ABC and MPPMX, and its capability for solving both problems.

The "Necker-Zeno model", a model for bistable perception inspired by the quantum Zeno effect, was previously used to relate three basic time scales of cognitive relevance to one another in a quantitative manner. In this paper, the model predictions are compared with experimental results obtained under discontinuous presentation of an ambiguous stimulus. In addition to earlier results for long inter-stimulus intervals, we show that the reversal dynamics according to the Necker-Zeno model is also in agreement with new results for short inter-stimulus intervals. Moreover, we refine the model in such a way that it accounts for the distribution of "dwell times" (inverse reversal rates). Finally, we indicate applications concerning the modification of cognitive time scales under conditions of psychopathological impairments and meditationinduced modes of awareness.

The origin of functional differences between motoneurons of varying size was investigated by employing a one-compartmental motoneuron model containing a slow K + conductance dependent on the intracellular calcium concentration. The size of the cell was included as an explicit parameter. Simulations show that motoneurons of varying size cannot be regarded as simple 'scaled' versions of each other. Rather, they are expected to have intrinsic properties that vary with cell size. These intrinsic properties refer to the membrane conductances per unit area and the dynamics of the intracellular calcium concentration.

Spike-timing-dependent plasticity (STDP) incurs both causal and acausal synaptic weight updates, for negative and positive time differences between pre-synaptic and postsynaptic spike events. For realizing such updates in neuromorphic hardware, current implementations either require forward and reverse lookup access to the synaptic connectivity table, or rely on memory-intensive architectures such as crossbar arrays. We present a novel method for realizing both causal and acausal weight updates using only forward lookup access of the synaptic connectivity table, permitting memory-efficient implementation. A simplified implementation in FPGA, using a single timer variable for each neuron, closely approximates exact STDP cumulative weight updates for neuron refractory periods greater than 10 ms, and reduces to exact STDP for refractory periods greater than the STDP time window. Compared to conventional crossbar implementation, the forward table-based implementation leads to substantial memory savings for sparsely connected networks supporting scalable neuromorphic systems with fully reconfigurable synaptic connectivity and plasticity.

Methods ot retrieval ot signiticant intormation in biomedical curves, based on the application ot the algorithms ot data compression , are described. These methods employ the a priori intormation about the morphology ot the type ot the curve under examination, emphasizing the standpoint ot the user. -The identitication ot the characteristic points and ot the intormative sections ot biomedical curves is paid closer attention. The procedure is illustrated by two actual algorithms, applied to the analysis ot ECG and EMG signals.

Recently various types of robots are being studied and developed, which can be classified into two groups: humanoid type and animal type. Since each group has its own merits and demerits, a new type of robot is expected to emerge with greater strengths and fewer weaknesses. In this paper we propose a new type of robot called the "Centaur Robot" by merging the concepts of these two types of robots. This robot has a human-like upper body and a four-legged animal-like lower body. Due to this basic architecture, the robot has several merits, including human-like behaviors and stable walking even on non-smooth ground. We describe its hardware and software architectures. Also we describe the experiments to evaluate its walking capability.

We study an unusual but robust phenomenon that appears in an example system of four coupled phase oscillators. The coupling is preserved under only one symmetry, but there are a number of invariant subspaces and degenerate bifurcations forced by the coupling structure, and we investigate these. We show that the system can have a robust attractor that responds to a specific detuning ∆ between a certain pair of the oscillators by a breaking of phase locking for arbitrary ∆ > 0 but not for ∆ ≤ 0. As the dynamical mechanism behind this is a particular type of heteroclinic cycle, we call this a 'heteroclinic ratchet' because of its dynamical resemblance to a mechanical ratchet.

1) Five possibilities of defining a coefficient of facilitation and inhibition are described. 2) It is shown that the application of these definitions to the same spike train activity eventually leads to considerably different results, e.g., a response which is inhibitory according to one definition sometimes is facilitatory according to another definition. 3) To find the most reliable definition the theoretical differences between the five alternatives are examined, whereby a coefficient is considered reliable if it is reproducible and independent of external experimental parameters, such as the record length. 4) As an experimental example spike trains were recorded from mitral cells in the olfactory bulb of the goldfish. We divide every response into two sections: an initial reaction and a steady state reaction. In this way each response can be uniquely classified. 5) The most reliable definition of a coefficient of facilitation turns out to be based on the steady state levels of an averaged peristimulus time histogramme. Under certain conditions this corresponds to considering the means of the sample mean rates of a stochastic point process before and after the stimulus application, whereby the initial reactions are neglected. These should be classified by other methods and notions.

The yaw movements of both distal eyestalks of the shore crab Carcinus maenas in response to a sinusoidally oscillated striped pattern were recorded simultaneously. The control of eye movements under various experimental conditions is interpreted on the basis of a linear model of the optokinetic system (Fig. ). The dynamics of the open-loop response, combined with the results of other authors, lead to a description of the motion-detecting mechanism of the crab (Fig. ): The crab processes movement in three velocity tuned channels in parallel. Each channel can be described by a correlation-model with adequate time constants. The channel tuned to rapid movements habituates very fast. The long time constants of the channel tuned to very slow movements constitute the "optokinetic memory". The dynamical properties of eye coupling can be described by a parallel shunted lowpass-filter of the order 0.5 (Fig. , Table ). With decreasing illumination the motion-detecting mechanism remains unchanged. The overall gain, however, decreases and the delays increase (Table ). Nonlinear properties of the system become apparent when it is stimulated with large amplitudes. The dynamics and the design of the optokinetic system of the crab are discussed as a compromise between the needs of optimal information exploitation and prevention of instabilities. The results are compared with known electrophysiological data.

Neuronal dynamics refers to the temporal behaviors that nervous systems exhibit. By coding we ask what and how does the neuronal activity represent our sensory experience, emotions, mental decisions and actions. A measure of activity or information that ignores temporal structure, say the average over time, gives a zeroth order indication of our world or our interpretation/representation of it. It falls upon on us to reach deeper, to consider the potential for encoding that is offered by the dynamical features of activity. What resources from the brain's wetware enable coding in a dynamical fashion? What are the basic components, the neurons, synapses, circuits dynamically capable of, how do they respond to dynamic stimuli? How does the brain exploit these resources to extract dynamical features and to encode information in spike trains, in multi-unit spatio-temporal activity patterns and across brain areas and via transformations from the periphery to higher associative areas? A dynamical understanding of brain function builds on knowledge of how systems respond to inputs, in particular, biologically significant input. In the early 1960's, the stimulation paradigm of the day was heavily influenced by the emerging field of cybernetics with its use of linear systems theory. The relevance of this knowledge to true brain function always depends on how well one can relate it to the in vivo situation. From this point of view, sensory systems, along with motor systems, have always presented a distinct advantage because one has a relatively clear idea of what constitutes a relevant input, and what the output is used for-in contrast e.g. to cells buried deep into associative cortices.

We present a biologically plausible model of processing intrinsic to the basal ganglia based on the computational premise that action selection is a primary role of these central brain structures. By encoding the propensity for selecting a given action in a scalar value (the salience), it is shown that action selection may be recast in terms of signal selection. The generic properties of signal selection are de®ned and neural networks for this type of computation examined. A comparison between these networks and basal ganglia anatomy leads to a novel functional decomposition of the basal ganglia architecture into `selection' and `control' pathways. The former pathway performs the selection per se via a feedforward o-centre on-surround network. The control pathway regulates the action of the selection pathway to ensure its eective operation, and synergistically complements its dopaminergic modulation. The model contrasts with the prevailing functional segregation of basal ganglia into `direct' and `indirect' pathways.

Swimming in vertebrates such as eel and lamprey involves the coordination of alternating left and right activity in each segment. Forward swimming is achieved by a lag between the onset of activity in consecutive segments rostrocaudally along the spinal cord. The intersegmental phase lag is approximately 1% of the cycle duration per segment and is independent of the swimming frequency. Since the lamprey has approximately 100 spinal segments, at any given time one wave of activity is propagated along the body. Most previous simulations of intersegmental coordination in the lamprey have treated the cord as a chain of coupled oscillators or well-defined segments. Here a network model without segmental boundaries is described which can produce coordinated activity with a phase lag. This 'continuous' pattern-generating network is composed of a column of 420 excitatory interneurons (E1 to E420) and 300 inhibitory interneurons (C1 to C300) on each half of the simulated spinal cord. The interneurons are distributed evenly along the simulated spinal cord, and their connectivity is chosen to reflect the behavior of the intact animal and what is known about the length and strength of the synaptic connections. For example, E100 connects to all interneurons between E51 and E149, but at varying synaptic strengths, while E101 connects to all interneurons between E52 and E150. This unsegmented E-C network generates a motor pattern that is sampled by output elements similar to motoneurons (M cells), which are arranged along the cell column so that they receive input from seven E and five C interneurons. The M cells thus represent the summed excitatory and inhibitory input at different points along the simulated spinal cord and can be regarded as representing the ventral root output to the myotomes along the spinal cord. E and C interneurons have five simulated compartments and Hodgkin-Huxley based dynamics. The simulated network produces rhythmic output over a wide range of frequencies (1-11 Hz) with a phase lag constant over most of the length, with the exception of the 'cut' ends due to reduced synaptic input. As the inhibitory C interneurons in the simulation have more extensive caudal than rostral projections, the output of the simulation has positive phase lags, as occurs in forward swimming. However, unlike

Swimming in vertebrates such as eel and lamprey involves the coordination of alternating left and right activity in each segment. Forward swimming is achieved by a lag between the onset of activity in consecutive segments rostrocaudally along the spinal cord. The intersegmental phase lag is approximately 1% of the cycle duration per segment and is independent of the swimming frequency. Since the lamprey has approximately 100 spinal segments, at any given time one wave of activity is propagated along the body. Most previous simulations of intersegmental coordination in the lamprey have treated the cord as a chain of coupled oscillators or well-defined segments. Here a network model without segmental boundaries is described which can produce coordinated activity with a phase lag. This 'continuous' pattern-generating network is composed of a column of 420 excitatory interneurons (E1 to E420) and 300 inhibitory interneurons (C1 to C300) on each half of the simulated spinal cord. The interneurons are distributed evenly along the simulated spinal cord, and their connectivity is chosen to reflect the behavior of the intact animal and what is known about the length and strength of the synaptic connections. For example, E100 connects to all interneurons between E51 and E149, but at varying synaptic strengths, while E101 connects to all interneurons between E52 and E150. This unsegmented E-C network generates a motor pattern that is sampled by output elements similar to motoneurons (M cells), which are arranged along the cell column so that they receive input from seven E and five C interneurons. The M cells thus represent the summed excitatory and inhibitory input at different points along the simulated spinal cord and can be regarded as representing the ventral root output to the myotomes along the spinal cord. E and C interneurons have five simulated compartments and Hodgkin-Huxley based dynamics. The simulated network produces rhythmic output over a wide range of frequencies (1-11 Hz) with a phase lag constant over most of the length, with the exception of the 'cut' ends due to reduced synaptic input. As the inhibitory C interneurons in the simulation have more extensive caudal than rostral projections, the output of the simulation has positive phase lags, as occurs in forward swimming. However, unlike

This study addresses mechanisms for the generation and selection of visual behaviors in anamniotes. To demonstrate the function of these mechanisms, we have constructed an experimental platform where a simulated animal swims around in a virtual environment containing visually detectable objects. The simulated animal moves as a result of simulated mechanical forces between the water and its body. The undulations of the body are generated by contraction of simulated muscles attached to realistic body components. Muscles are driven by simulated motoneurons within networks of central pattern generators. Reticulospinal neurons, which drive the spinal pattern generators, are in turn driven directly and indirectly by visuomotor centers in the brainstem. The neural networks representing visuomotor centers receive sensory input from a simplified retina. The model also The online version of this article (doi:10.1007/s00422-012-0524-4) contains supplementary material, which is available to authorized users. This article forms part of a special issue of Biological Cybernetics entitled "Lamprey, Salamander Robots and Central Nervous System".

Current computational neuroscience models often struggle to account for the continuous, value-oriented human drive that is not satisfied by achieving discrete goals. This paper introduces a new conceptual framework, the Unified Dynamic Model of the Mind (UDMM), which posits the mind as an inferential system proactively shaping its environment. Central to UDMM is the concept of a "saturated world," reconceptualized here not as an achievable goal-state but as a virtual attractor-a hypothetical probabilistic distribution that cannot be fully attained, yet serves as a consistent compass guiding system behavior. We propose an initial mathematical formulation grounded in the Free Energy Principle and Active Inference (Friston, 2010), where the agent does not simply minimize prediction error to reach a stable state. Instead, it continuously minimizes the Kullback-Leibler (KL) Divergence between its expected future state distribution and the ideal distribution representing the saturated world. This approach produces behavior resembling a stable orbit around higher values. The paper outlines a roadmap for developing computational models based on this principle and opens new avenues for modeling meaning, purpose, and mental disorders as dysfunctions in the dynamic relationship with this virtual attractor.

This study introduces an innovative, AI-driven nanobot engineered for systemic navigation within the human bloodstream, designed to achieve simultaneous diagnostic and therapeutic functions against cancerous cells and viral infections. This immune-evasive device incorporates personalized histological data to differentiate pathological tissues from healthy ones with high precision. Powered autonomously by intracellular ATP and hydrogen micro-fuel cells, the nanobot performs targeted interventions while promoting tissue regeneration inspired by biological models. This concept opens a new frontier in personalized, regenerative, and minimally invasive medicine.

A model for monocular line perception by human Ss is based on three basic assumptions: (a) the line's inclination is coded by the maximally excited orientation detector's number; (b) the inclination of the perceived line is equivocally determined by the excitation vector in the subjective space; (c) the analyzer has a maximum differential sensitivity over the whole range of the line inclinations. This simple model for the line inclination analyzer, taking into account the optimization of its sensitivity, provides incomplex explanations for a wide range of psychophysical and neurophysiological data obtained from human and animal experiments.

A simple neural network model is proposed for kindling -the phenomenon of generating epilepsy by means of repeated electrical stimulation. The model satisfies Dale's hypothesis, incorporates a Hebb-like learning rule and has low periodic activity in absence of shocks. Many of the experimental observations are reproduced and some new experiments are suggested. It is proposed that the main reason for kindling is the formation of a large number of excitatory synaptic connections due to learning.

In spite of the fact that the participation of well defined ionic particles in generating convulsive unit discharges is established, there is a gap between the data on ionic movements and on first-order statistics of firing patterns. Our aim was to tight this gap by studying the effectiveness of functionally separated electrical conductances of membrane during the generation of consecutive interspike interval histograms (IIHs) of unitary discharges. On account of the nonstationarity of the process curve fitting analysis which based on the simple modifications of the integrate-andfire model has been implemented in the sequential interspike interval histogram procedure (SIIH). The experimental data were recorded from cat cortex treated with 3-Aminopyridine (3-Ap) by glass microelectrodes during nembutal anesthezia. Assuming the normal distribution of input parameters it is concluded, that the efficiency of the fluctuations of the active spike-generating conductance go and the passive diffusional conductance g~ may increase during the generation of the unimodal IIHs and the first mode of the bimodal IIHs. The simple conductance coupling gz=a.go+b may participate in go activation, moreover, the reciprocally coupled mechanism gg=c/g~ may be driven by gz activation (a, b, c are the coupling constants). A temporal separation of processes governed by go or gl respectively was observed. The timeindependent occurrences of the reciprocally coupled conductance processes may be involved in the unit activities represented by the prolonged IIHs and second modes of the bimodal IIHs.

Many songbirds develop remarkably large vocal repertoires, and this has prompted questions about how birds are able to successfully learn and use the often enormous amounts of information encoded in their various signal patterns. We have studied these questions in nightingales (Luscinia megarhynchos), a species that performs more than 200 dierent types of songs (strophen), or more than 1000 phonetically dierent elements composing the songs. In particular, we investigated whether and how both song repertoires and song performance rules of nightingales were coded by auditory stimuli presented in serial learning experiments. Evaluation of singing episodes produced by our trained birds revealed that nightingales cope well with an exposure to even long strings of master song-types. They can readily acquire information encoded within and between the dierent master songs, and they memorize, for example, which master song-types they have experienced in the same learning context. Imitations of such song-types form distinct sequential associations that are termed ``context groups''. Additionally, nightingales develop other song-type associations that are smaller in size and termed ``package groups''. Package formation results from constraints of the acquisition mechanisms which obviously lead to a segmentation of auditorily perceived master song sequences. Further experimentation validated that the song memory of nightingales is organized in a hierarchical manner and holding information about ``context groups'' composed of packages, ``package groups'' composed of songs, and songs composed of song elements. The evidence suggests that implementation of such a hierarchical organization facilitates a quick retrieval of particular songs, and thereby provides an essential prerequisite for a functionally appropriate use of large vocal repertoire is in songbirds.

The properties of multi-peaked "fitness landscapes" have attracted attention in a wide variety of fields, including evolutionary biology. However, relatively little attention has been paid to the properties of the landscapes themselves. Herein, we suggest a framework for the mathematical treatment of such landscapes, including an explicit mathematical model. A central role in this discussion is played by the autocorrelation of fitnesses obtained from a random walk on the landscape. Our ideas about average autocorrelations allow us to formulate a condition (satisfied by a wide class of landscapes we call AR(1) landscapes) under which the average autocorrelation approximates a decaying exponential. We then show how our mathematical model can be used to estimate both the globally optimal fitnesses of AR(1) landscapes and their local structure. We illustrate some aspects of our method with computer experiments based on a single family of landscapes (Kauffman's "N-k model"), that is shown to be a generic AR(I) landscape. We close by discussing how these ideas might be useful in the "tuning" of combinatorial optimization algorithms, and in modelling in the experimental sciences.

Holland's "hyperplane transform" of a "fitness landscape", a random, real valued function of the verticies of a regular finite graph, is shown to be a special case of the Fourier transform of a function of a finite group. It follows that essentially all of the powerful Fourier theory, which assumes a simple form for commutative groups, can be used to characterize such landscapes. In particular, an analogue of the Karhunen-Lorve expansion can be used to prove that the Fourier coefficients of landscapes on commutative groups are uncorrelated and to infer their variance from the autocorrelation function of a random walk on the landscape. There is also a close relationship between the Fourier coefficients and Taylor coefficients, which provide information about the landscape's local properties. Special attention is paid to a particularly simple, but ubiquitous class of landscapes, so-called "AR(1) landscapes".

In natural visual scenes, there is considerable rapidly changing information. The ability to detect these changes in the environment is crucial. Previous studies suggested that the neural representation of change detection is based on the fast transient responses of neurons. In visual area MT, transient responses are demonstrated to be highly informative about motion direction and stimulus change. Moreover, the transient responses are shown to be correlated with behaviour performance of change detection tasks. In the current study, we proposed a model with a simple circuit to investigate the underlying computational mechanisms of transient responses. We further applied model and mathematical analysis to physiological data recorded in area MT. After fitting the model to experimental data, the results showed that the model is capable of reproducing the time courses of transient responses under different conditions of speed change. To investigate the attentional modulation on transient responses, visual attention is implemented as a multiplicative modulation on the input to the model. By mathematical analyses of the model, we predicted that the slopes of transient responses would be consistently modulated by attention, leading to faster transient responses to stimulus change when attention is on the stimulus. The prediction is supported by the united results of analysing the data recorded in two different physiological studies. Finally, by comparing stochastic realization of a simulated transient response trace, we reproduced the correlation of shorter latencies of transient responses to shorter RTs of change detection tasks. Thus, through modelling, analyses of the model and comparing it to physiological data, we provide a potential framework explaining response dynamics of MT neurons. Moreover, we provided strong evidences showing that the slopes of transient responses are highly correlated with attentional modulation on transient responses and with the attention-induced better behaviour performance.

In order to study the motor unit action potential a computer simulation model was developed. It is based on the superposition of single muscle fibre potentials of the fibres belonging to the motor unit. The parameters which characterize each fibre (spatial position, diameter, and a dispersion of arrival time of the potential at the electrode) are chosen from statistical distributions which can be derived from anatomical and physiological data. The electrode type, position and dimensions can be specified. Simulated motor unit action potentials are presented in the time and frequency domain. The simulation results refer to (1) the influence of the electrode position and dimensions with respect to the motor unit territory, (2) the meaning of this model for the study of pathological phenomena, (3) the variability of some parameters characterizing the motor unit, (4) the selectivity of uni-and bipolar electrodes and finally (5) the influence of the geometrical situation of the motor end-plates within the muscle, on the shape of motor unit action potentials.

The paper explores the challenging task of performing a non-prehensile manipulation of several balls synchronously rolling on the curved hands of Butterfly robots. Each Butterfly robot represents a standard benchmark hardware setup, comprising a DC motor rotating a butterfly-shaped frame in a vertical plane, with a ball moving freely upon it, equipped with integrated computer vision, communication, programmable control, and computation interfaces. The combined dynamics of the considered system, consisting of N ≥ 2 such robots, is inherently underactuated, characterized by N active and N passive degrees of freedom, as well as N independent unilateral constraints that model the interactions between the frames and the balls, assuming no slipping. We focus on designing a model-based centralized feedback controller to achieve synchronized rotations of the balls. We assume the accuracy of our mathematical model and the feasibility of implementing a discretized version of the proposed continuoustime controller with a sufficiently small sampling time, that, in particular, is necessary for numerical differentiation. Relying on orbital stability of nominal periodic solution of the closedloop system, we will experimentally check robustness to various inevitable challenges such as noises, disturbances, uncertainties, and communication delays. Hence, our concentration lies in designing an orbitally stabilizing controller for the underactuated models. The primary contribution is proposing one set of transverse coordinates, enabling transverse-linearizationbased controller design, accompanied by pertinent closed-loop system analysis tools, thereby enhancing the efficacy of solving the manipulation task. Analytical and model-based arguments are validated through successful simulations and experiments conducted on two Butterfly robots, thereby emphasizing the validity and practicality of the proposed approach.

Reaction time (RT) and error rate that depend on stimulus duration were measured in a luminance-discrimiaation reactioa time task̲ Two patches of light with different luminance were presented to participants for ' short'(150 ms)or ' long'(I s)period on each trial.Vyhen the stimulus duration was ' short' ,the participants responded more ra, pidIy with poorer discrimination performance than they did in the longer duration̲The results suggested that different sensory responses in the visual cortices were responsible for the dependence of response speed and a, ccuracy on the stimulus duration during the luminance-discrimination reaction time task.It was shown that the simplewinnertake-a1 1-type neural network model receiving transient and sustained stimulus information from the primary visual cortex (V1)successfu通 y reproduced RT distributions for correct responses and error rates.Moreover,temporal spike sequences obtained from the model network closely resembled to the neura」 activity in the monkey prefrontal or parietal area during other visua」 decision tasks such as motion disa imination and oddball detection tasks̲

The yaw movements of both distal eyestalks of the shore crab Carcinus maenas in response to a sinusoidally oscillated striped pattern were recorded simultaneously. The control of eye movements under various experimental conditions is interpreted on the basis of a linear model of the optokinetic system (Fig. ). The dynamics of the open-loop response, combined with the results of other authors, lead to a description of the motion-detecting mechanism of the crab (Fig. ): The crab processes movement in three velocity tuned channels in parallel. Each channel can be described by a correlation-model with adequate time constants. The channel tuned to rapid movements habituates very fast. The long time constants of the channel tuned to very slow movements constitute the "optokinetic memory". The dynamical properties of eye coupling can be described by a parallel shunted lowpass-filter of the order 0.5 (Fig. , Table ). With decreasing illumination the motion-detecting mechanism remains unchanged. The overall gain, however, decreases and the delays increase (Table ). Nonlinear properties of the system become apparent when it is stimulated with large amplitudes. The dynamics and the design of the optokinetic system of the crab are discussed as a compromise between the needs of optimal information exploitation and prevention of instabilities. The results are compared with known electrophysiological data.

The tangential neurons in the lobula plate region of the flies are known to respond to visual motion across broad receptive fields in visual space. When intracellular recordings are made from t angential neurons while the intact animal is stimulated visually with moving natural imagery, we find that neural response depends upon speed of motion but is nearly invariant with respect to variations in natural scenery. We refer to this invariance as velocity constancy. It is remarkable because natural scenes, in spite of similarities in spatial structure, vary considerably in contrast, and contrast dependence is a feature of neurons in the early visual pathway as well as of most models for the elementary operations of visual motion detection. Thus, we expect that operations must be present in the processing pathway that reduce contrast dependence in order to approximate velocity constancy. We consider models for such operations, including spatial filtering, motion adaptation, saturating nonlinearities, and nonlinear spatial integration by the tangential neurons themselves, and evaluate their effects in simulations of a tangential neuron and precursor processing in response to animated natural imagery. We conclude that all such features reduce interscene variance in response, but that the model system does not approach velocity constancy as closely as the biological tangential cell.