A linear canal-otolith interaction model to describe the human vestibulo-ocular reflex

A control systems model of the vestibulo-ocular reflex (VOR) originally derived for yaw rotation about an eccentric axis (Crane et al. 1997) was applied to data collected during ambulation and dynamic posturography. The model incorporates a linear summation of an otolith response due to head translation scaled by target distance, adding to a semi-circular canal response that depends only on angular head rotation. The results of the model were compared with human experimental data by supplying head angular velocity as determined by magnetic search coil recording as the input for the canal branch of the model and supplying linear acceleration as determined by flux gate magnetometer measurements of otolith position. The model was fit to data by determining otolith weighting that enabled the model to best fit the data. We fit to the model experimental data from normal subjects who were: standing quietly, walking, running, or making active sinusoidal head movements. We also fit data obtained during dynamic posturography tasks of: standing on a platform sliding in a horizontal plane at 0.2 Hz, standing directly on a platform tilting at 0.1 Hz, and standing on the tilting platform buffered by a 5-cm thick foam rubber cushion. Each task was done with the subject attending a target approximately 500, 100, or 50 cm distant, both in light and darkness. The model accurately predicted the observed VOR response during each test. Greater otolith weighting was required for near targets for nearly all activities, consistent with weights for the otolith component found in previous studies employing imposed rotations. The only exceptions were for vertical axis motion during standing, sliding, and tilting when the platform was buffered with foam rubber. In the horizontal axis, the model always fit near target data better with a higher otolith component. Otolith weights were similar with the target visible and in darkness. The model predicts eye movement during both passive whole-body rotation and free head movement in space implying that the VOR is controlled by a similar mechanism during both situations. Factors such as vision, proprioception, and efference copy that are available during head free motion but not during whole-body rotation are probably not important to gaze stabilization during ambulation and postural stabilizing movement. The linearity of the canal-otolith interaction was tested by re-analysis of the whole body rotation data on which the model is based (Crane et al. 1997). Normalized otolith-mediated gain enhancement was determined for each axis of rotation. This analysis uncovered minor non-linearities in the canal-otolith interaction at frequencies above 1.6 Hz and when the axis of rotation was posterior to the head. Received: 11 March 1998 / Received in revised form: 1 March 1999     

Is the adaptation model a valid description of the vestibulo-ocular reflex?

The pendulum model of the vestibulo-ocular reflex, including the effects of adaptation, has been evaluated using the responses of 36 normal subjects to impulsive stimuli of 128 and 256°/s. Estimates of the model parameters such as the time constants, the slow velocity threshold, and the minimum stimulus required to produce an after-nystagmus have been obtained using a new analytical technique. Although some of the data support the validity of the adaptation model, evidence is presented to demonstrate that the overall applicability of the model is limited.     

A matrix analysis for a conjugate vestibulo-ocular reflex

The technique of matrix analysis is used to compare the connectivity between vestibular neurons and oculomotor neurons of the two eyes that would generate a conjugate vestibulo-ocular reflex (VOR). The technique shows that the connectivity is normally anatomically symmetric. The technique is also used to determine the types and loci of adaptation within the VOR that will maintain conjugacy. Adaptation is divided into1) that evoked by changes in visual feedback, which requires VOR or system-specific changes and2) that produced by changes in the canals or muscles, which requires deficit-specific adaptation. In the former case, the adaptation could best be achieved by an additive alteration of the vestibularmotoneuron projections. In the latter case, the appropriate adaptations would be serial, multiplicative changes, applied at the level of the vestibular neurons when the canals are at fault or at the level of the motoneurons of the eye whose muscles are impaired. The analysis thus suggests multiple loci of plasticity within the VOR, specialized for adapting to different deficits.     

A learning network model of the neural integrator of the oculomotor system

Certain premotor neurons of the oculomotor system fire at a rate proportional to desired eye velocity. Their output is integrated by a network of neurons to supply an eye positon command to the motoneurons of the extraocular muscles. This network, known as the neural integrator, is calibrated during infancy and then maintained through development and trauma with remarkable precision. We have modeled this system with a self-organizing neural network that learns to integrate vestibular velocity commands to generate appropriate eye movements. It learns by using current eye movement on any given trial to calculate the amount of retinal image slip and this is used as the error signal. The synaptic weights are then changed using a straightforward algorithm that is independent of the network configuration and does not necessitate backwards propagation of information. Minimization of the error in this fashion causes the network to develop multiple positive feedback loops that enable it to integrate a push-pull signal without integrating the background rate on which it rides. The network is also capable of recovering from various lesions and of generating more complicated signals to simulate induced postsaccadic drift and compensation for eye muscle mechanics.     

Neural network models of velocity storage in the horizontal vestibulo-ocular reflex

The vestibulo-ocular reflex (VOR) produces compensatory eye movements by utilizing head rotational velocity signals from the semicircular canals to control contractions of the extraocular muscles. In mammals, the time course of horizontal VOR is longer than that of the canal signals driving it, revealing the presence of a central integrator known as velocity storage. Although the neurons mediating VOR have been described neurophysiologically, their properties, and the mechanism of velocity storage itself, remain unexplained. Recent models of integration in VOR are based on systems of linear elements, interconnected in arbitrary ways. The present study extends this work by modeling horizontal VOR as a learning network composed of nonlinear model neurons. Network architectures are based on the VOR arc (canal afferents, vestibular nucleus (VN) neurons and extraocular motoneurons) and have both forward and lateral connections. The networks learn to produce velocity storage integration by forming lateral (commissural) inhibitory feedback loops between VN neurons. These loops overlap and interact in a complex way, forming both fast and slow VN pathways. The networks exhibit some of the nonlinear properties of the actual VOR, such as dependency of decay rate and phase lag upon input magnitude, and skewing of the response to higher magnitude sinusoidal inputs. Model VN neurons resemble their real counterparts. Both have increased time constant and gain, and decreased spontaneous rate as compared to canal afferents. Also, both model and real VN neurons exhibit rectification and skew. The results suggest that lateral inhibitory interactions produce velocity storage and also determine the properties of neurons mediating VOR. The neural network models demonstrate how commissural inhibition may be organized along the VOR pathway.     

Relating reflex gain modulation in posture control to underlying neural network properties using a neuromusculoskeletal model

During posture control, reflexive feedback allows humans to efficiently compensate for unpredictable mechanical disturbances. Although reflexes are involuntary, humans can adapt their reflexive settings to the characteristics of the disturbances. Reflex modulation is commonly studied by determining reflex gains: a set of parameters that quantify the contributions of Ia, Ib and II afferents to mechanical joint behavior. Many mechanisms, like presynaptic inhibition and fusimotor drive, can account for reflex gain modulations. The goal of this study was to investigate the effects of underlying neural and sensory mechanisms on mechanical joint behavior. A neuromusculoskeletal model was built, in which a pair of muscles actuated a limb, while being controlled by a model of 2,298 spiking neurons in six pairs of spinal populations. Identical to experiments, the endpoint of the limb was disturbed with force perturbations. System identification was used to quantify the control behavior with reflex gains. A sensitivity analysis was then performed on the neuromusculoskeletal model, determining the influence of the neural, sensory and synaptic parameters on the joint dynamics. The results showed that the lumped reflex gains positively correlate to their most direct neural substrates: the velocity gain with Ia afferent velocity feedback, the positional gain with muscle stretch over II afferents and the force feedback gain with Ib afferent feedback. However, position feedback and force feedback gains show strong interactions with other neural and sensory properties. These results give important insights in the effects of neural properties on joint dynamics and in the identifiability of reflex gains in experiments.     

A synaptic learning rule for exploiting nonlinear dendritic computation

1. Download : Download high-res image (231KB)
2. Download : Download full-size image

     

An unsupervised neural network model for the development of reflex co-ordination

In this paper, we present a model for the development of connections between muscle afferents and motoneurones in the human spinal cord. The model consists of a limb with six muscles, one motoneurone pool, one pooled (Ia-like) afferent for each muscle and a central programme generator. The weights of the connections between the afferents and the motoneurone pools are adapted during centrally induced movements of the limb. The connections between the afferents and the motoneurone pools adapt in a hebbian way, using only local information present at the synapses. This neural network is tested in two examples of a limb with two degrees of freedom and six muscles. Despite the simplifications, the model predicts the pattern of autogenic and heterogenic monosynaptic reflexes quite realistically.     

Control of bioreactors using a neural network model

Control of a continuous bioreactor based on a artificial neural network (ANN) model is carried out theoretically. The ANN model is identified, from input-output data of a bioreactor, using a three-layer feedforward network trained by a back propagation algorithm. The performance of the controller designed on the ANN model is compared with that of a conventional PI controller.     

Plasticity in the adult vestibulo-ocular reflex arc.

Human subjects with maintained reversal of their horizontal field of vision exhibit very substantial adaptive changes in their 'horizontal' vestibulo-ocular reflex (v.o.r.). Short durations (8 min) of vision reversal during natural head movement led to 20% v.o.r. attenuation while long periods (4 weeks) eventually led to approximate reversal of the reflex. The reversed condition is approached by a complex, but highly systematic, series of changes in gain and phase of the reflex response relative to normal. Recovery after return to normal vision exhibits a similar duration, but different pattern, to that of the original adaptation. A chronic cat preparation with long-term optical reversal of vision has now been developed and shows similar adaptive and recovery changes at low test stimulus amplitudes, but different patterns of adaptive response at high amplitudes. An adaptive neural model employing known vestibulo-ocular pathways is proposed to account for these experimentally observed plastic changes. The model is used to predict the adapted response to patterns of stimulation extending beyond the range of experimental investigation.     

A hybrid neural learning algorithm using evolutionary learning and derivative free local search method

In this paper we investigate a hybrid model based on the Discrete Gradient method and an evolutionary strategy for determining the weights in a feed forward artificial neural network. Also we discuss different variants for hybrid models using the Discrete Gradient method and an evolutionary strategy for determining the weights in a feed forward artificial neural network. The Discrete Gradient method has the advantage of being able to jump over many local minima and find very deep local minima. However, earlier research has shown that a good starting point for the discrete gradient method can improve the quality of the solution point. Evolutionary algorithms are best suited for global optimisation problems. Nevertheless they are cursed with longer training times and often unsuitable for real world application. For optimisation problems such as weight optimisation for ANNs in real world applications the dimensions are large and time complexity is critical. Hence the idea of a hybrid model can be a suitable option. In this paper we propose different fusion strategies for hybrid models combining the evolutionary strategy with the discrete gradient method to obtain an optimal solution much quicker. Three different fusion strategies are discussed: a linear hybrid model, an iterative hybrid model and a restricted local search hybrid model. Comparative results on a range of standard datasets are provided for different fusion hybrid models.     

Calculating the target exposure index using a deep convolutional neural network and a rule base

PurposeThe objective of this study is to determine the quality of chest X-ray images using a deep convolutional neural network (DCNN) and a rule base without performing any visual assessment. A method is proposed for determining the minimum diagnosable exposure index (EI) and the target exposure index (EIt).MethodsThe proposed method involves transfer learning to assess the lung fields, mediastinum, and spine using GoogLeNet, which is a type of DCNN that has been trained using conventional images. Three detectors were created, and the image quality of local regions was rated. Subsequently, the results were used to determine the overall quality of chest X-ray images using a rule-based technique that was in turn based on expert assessment. The minimum EI required for diagnosis was calculated based on the distribution of the EI values, which were classified as either suitable or non-suitable and then used to ascertain the EIt.ResultsThe accuracy rate using the DCNN and the rule base was 81%. The minimum EI required for diagnosis was 230, and the EIt was 288.ConclusionThe results indicated that the proposed method using the DCNN and the rule base could discriminate different image qualities without any visual assessment; moreover, it could determine both the minimum EI required for diagnosis and the EIt.     

Vestibuloocular reflex arc analysis using an experimentally constrained neural network

The primary function of the vestibuloocular reflex (VOR) is to maintain the stability of retinal images during head movements. This function is expressed through a complex array of dynamic and adaptive characteristics whose essential physiological basis is a disynaptic arc. We present a model of normal VOR function using a simple neural network architecture constrained by the physiological and anatomical characteristics of this disynaptic reflex arc. When tuned using a method of global optimization, this network is capable of exhibiting the broadband response characteristics observed in behavioral tests of VOR function. Examination of the internal units in the network show that this performance is achieved by rediscovering the solution to VOR processing first proposed by Skavenski and Robinson (1973). Type I units at the intermediate level of the network possess activation characteristics associated with either pure position or pure velocity. When the network is made more complex either through adding more pairs of internal units or an additional level of units, the characteristic division of unit activation properties into position and velocity types remains unchanged. Although simple in nature, the results of our simulations reinforce the validity of bottom-up approaches to modeling of neutral function. In addition, the architecture of the network is consistent with current ideas on the characteristics and site of adaptation of the reflex and should be compatible with current theories regarding learning rules for synaptic modification during VOR adaptation.     

Template learning of cellular neural network using genetic programming

A new learning algorithm for space invariant Uncoupled Cellular Neural Network is introduced. Learning is formulated as an optimization problem. Genetic Programming has been selected for creating new knowledge because they allow the system to find new rules both near to good ones and far from them, looking for unknown good control actions. According to the lattice Cellular Neural Network architecture, Genetic Programming will be used in deriving the Cloning Template. Exploration of any stable domain is possible by the current approach. Details of the algorithm are discussed and several application results are shown.     

Analysis of reflex modulation with a biologically realistic neural network

In this study, a neuromusculoskeletal model was built to give insight into the mechanisms behind the modulation of reflexive feedback strength as experimentally identified in the human shoulder joint. The model is an integration of a biologically realistic neural network consisting of motoneurons and interneurons, modeling 12 populations of spinal neurons, and a one degree-of-freedom musculoskeletal model, including proprioceptors. The model could mimic the findings of human postural experiments, using presynaptic inhibition of the Ia afferents to modulate the feedback gains. In a pathological case, disabling one specific neural connection between the inhibitory interneurons and the motoneurons could mimic the experimental findings in complex regional pain syndrome patients. It is concluded that the model is a valuable tool to gain insight into the spinal contributions to human motor control. Applications lay in the fields of human motor control and neurological disorders, where hypotheses on motor dysfunction can be tested, like spasticity, clonus, and tremor. Action Editor: Karen Sigvardt     

Anti-Hebbian learning in a non-linear neural network

The Hebbian rule (Hebb 1949), coupled with an appropriate mechanism to limit the growth of synaptic weights, allows a neuron to learn to respond to the first principal component of the distribution of its input signals (Oja 1982). Rubner and Schulten (1990) have recently suggested the use of an anti-Hebbian rule in a network with hierarchical lateral connections. When applied to neurons with linear response functions, this model allows additional neurons to learn to respond to additional principal components (Rubner and Tavan 1989). Here we apply the model to neurons with non-linear response functions characterized by a threshold and a transition width. We propose local, unsupervised learning rules for the threshold and the transition width, and illustrate the operation of these rules with some simple examples. A network using these rules sorts the input patterns into classes, which it identifies by a binary code, with the coarser structure coded by the earlier neurons in the hierarchy.     

Control and modulation of canal driven vestibulo-ocular reflex.

It has been well known that the canal driven vestibulo-ocular reflex (VOR) is controlled and modulated through the central nervous system by external sensory information (e.g. visual, otolithic and somatosensory inputs) and by mental conditions. Because the origin of retinal image motion exists both in the subjects (eye, head and body motions) and in the external world (object motion), the head motion should be canceled and/or the object should be followed by smooth eye movements. Human has developed a lot of central nervous mechanisms for smooth eye movements (e.g. VOR, optokinetic reflex and smooth pursuit eye movements). These mechanisms are thought to work for the purpose of better seeing. Distinct mechanism will work in appropriate self motion and/or object motion. As the results, whole mechanisms are controlled in a purpose-directed manner. This can be achieved by a self-organizing holistic system. Holistic system is very useful for understanding human oculomotor behavior.     

An avoidance learning situation. A neural net model

A simple avoidance situation is considered in terms of a neural net learning model. Data for the control situation can be represented by an expression having three parameters which determine the initial and the steady state activities together with the transient aspects. The introduction of a learning parameter then allows one to calculate satisfactorily the results obtained in the experimental situation in which shock is applied. This research was supported in part by the United States Air Force through the Air Force Office of Scientific Research of the Air Research Development Command under Grant No. AF AFOSR 370-63 and in part by the United States Public Health Service Grant RCA GM K6 18,420.     

Optimization of fed-batch bioreactor using neural network model

An algorithm using feedforward neural network model for determining optimal substrate feeding policies for fed-batch fermentation process is presented in this work. The algorithm involves developing the neural network model of the process using the sampled data. The trained neural network model in turn is used for optimization purposes. The advantages of this technique is that optimization can be achieved without detailed kinetic model of the process and the computation of gradient of objective function with respect to control variables is straightforward. The application of the technique is demonstrated with two examples, namely, production of secreted protein and invertase. The simulation results show that the discrete-time dynamics of fed-batch bioreactor can be satisfactorily approximated using a feedforward sigmoidal neural network. The optimal policies obtained with the neural network model agree reasonably well with the previously reported results.