Search

In complex work domains and organizations, understanding scheduleing dynamics can ensure objectives are reached and delays are mitigated. In the current paper, we examine the scheduling dynamics for NASA’s Exploration Flight Test 1 (EFT-1) activities. For this examination, we specifically modeled simultaneous change in percent complete and estimated duration for a given project as they were included in monthly reports over time. In short, we utilized latent change score mixture modeling to extract the attractor dynamics within the scheduling data. We found three primarily patterns: an attractor at low duration, low percent complete; a saddle that was attractive toward full completion and repelled duration away from five months, and an attractor at full completion and high duration. We replicated these three patterns using multilevel modeling. Then, we examined how task dependencies, in terms of the number of predecessors and successors, affected the probability of exhibiting a given pattern over time. Thus, we offer a flexible method for understanding the patterns that can characterize scheduling dynamics as well as other dynamical systems. Several recommendations for future directions are discussed.

In this article, for the purpose of improving neural network models applied in face recognition using single image per person, a bidirectional neural network inspired of neocortex functional model is presented. We have applied this novel adapting model to separate person and pose information. To increase the number of training samples in the classifier neural network, virtual views of frontal images in the test dataset are synthesized using estimated manifolds. Training classifier network via virtual images gives an accuracy rate of 85.45% which shows 14.55% improvement in accuracy of face recognition compared to training classifier with only frontal view images.

In this paper we show how non-linear attractor dynamics can be used as a framework to control teams of autonomous mobile robots that should navigate according to a predefined geometric formation. The environment does not need to be known a priori and may change over time. Implicit to the control architecture are some important features such as establishing and moving the formation, split and join of formations (when necessary to avoid obstacles). Formations are defined by a formation matrix. By manipulating this formation matrix it is also possible to switch formations at run time. Examples of simulation results and implementations with real robots (teams of Khepera robots and medium size mobile robots), demonstrate formation switch, static and dynamic obstacle avoidance and split and join formations without the need for any explicit coordination scheme. Robustness against environmental perturbations is intrinsically achieved because the behaviour of each robot is generated as a time series of asymptotically stable states, which contribute to the asymptotic stability of the overall control system., This work was supported through COOP-DYN (POSI/SRI/38051/2001), financed by the Portuguese Foundation for Science and Technology (FCT) and project fp6-IST2 EU-project JAST-Joint Action Science and Technology (project number 003747). We thank Miguel Vaz and Nzoji Hipolito for their help in the implementations with real robots. We would like also to thank the anonymous reviewers for their insightful comments which helped improving the paper.

There is broad consensus that the therapeutic alliance constitutes a core common factor for all modalities of psychotherapy. Meta-analyses corroborated that alliance, as it emerges from therapeutic process, is a significant predictor of therapy outcome. Psychotherapy process is traditionally described and explored using two categorically different approaches, the experiential (first-person) perspective and the behavioral (third-person) perspective. We propose to add to this duality a third, structural approach. Dynamical systems theory and synergetics on the one hand and enactivist theory on the other together can provide this structural approach, which contributes in specific ways to a clarification of the alliance factor. Systems theory offers concepts and tools for the modeling of the individual self and, building on this, of alliance processes. In the enactive perspective, the self is conceived as a socially enacted autonomous system that strives to maintain identity by observing a two-fold goal: to exist as an individual self in its own right (distinction) while also being open to others (participation). Using this conceptualization, we formalized the therapeutic alliance as a phase space whose potential minima (attractors) can be shifted by the therapist to approximate therapy goals. This mathematical formalization is derived from probability theory and synergetics. We draw the conclusion that structural theory provides powerful tools for the modeling of how therapeutic change is staged by the formation, utilization, and dissolution of the therapeutic alliance. In addition, we point out novel testable hypotheses and future applications.

The brain is able to interpret streams of high-dimensional ambiguous information and yield coherent percepts. The mechanisms governing sensory integration have been extensively characterized using time-varying visual stimuli (Britten et al. 1996; Roitman and Shadlen 2002), but some of the basic principles regarding the network dynamics underlying this process remain largely unknown. We captured the basic features of a neural integrator using three canonical one-dimensional models: (1) the Drift Diffusion Model (DDM), (2) the Perfect Integrator (PI) which is a particular case of the DDM where the bounds are set to infinity and (3) the double-well potential (DW) which captures the dynamics of the attractor networks (Wang 2002; Roxin and Ledberg 2008). Although these models has been widely studied (Bogacz et al. 2006; Roxin and Ledberg 2008; Gold and Shadlen 2002), it has been difficult to experimentally discriminate among them because most of the observables measured are only quantitatively different among these models (e.g. psychometric curves). Here we aim to find experimentally measurable quantities that can yield qualitatively different behaviors depending on the nature of the underlying network dynamics. We examined the categorization dynamics of these models in response to fluctuating stimuli of different duration (T). On each time step, stimuli are drawn from a Gaussian distribution N(μ, σ) and the two stimulus categories are defined by μ > 0 and μ < 0. Psychometric curves can therefore be obtained by quantifying the probability of the integrator to yield one category versus μ . We find however that varying σ can reveal more clearly the differences among the different integrators. In the small σ regime, both the DW and the DDM perform transient integration and exhibit a decaying stimulus reverse correlation kernel revealing a primacy effect (Nienborg and Cumming 2009; Wimmer et al. 2015) . In the large σ regime, the integration in the DDM becomes even more transient (i.e. the bound is reached earlier during the stimulus) but in the DW the rate of reversed classifications increases and the kernel changes from monotonically decreasing to monotonically increasing revealing a recency effect. In contrast the PI performs uniform integration and shows a constant kernel for all σ's. The discrimination sensitivity in the PI decreases monotonically as 1/σ just because the two stimulus categories become less separable in the stimulus space. In the DDM the decrease with σ is even more pronounced due to the reduction of the effective integration time (the average time to reach the bound). Unexpectedly however, the sensitivity in the DW shows a non-monotonic behavior with a local maximum for Ts beyond a critical value. This is because, provided there is enough time to reverse the initial classification, the increase in the rate of correcting reversals compensates for the decrease in stimulus separability. Finally, we also found that the three models behave differently with σ when the initial condition of the integrator is offset (e.g. representing that the prior probabilities of the two stimulus categories are uneven): increasing σ decreases the choice bias caused by the offset in the DW model, increases the bias for the DDM model and leaves it unchanged in the PI model. We propose specific ways to test these predictions in a random duration two alternative forced-choice (2AFC) task which could help us pinning down the basic principles of sensory integration dynamics.

In order to adapt to new situations, primates must be able to recognize these situations. How the cortex represents contingencies in its activity is the main subject of this thesis. First, complex new situations are often explained by the interaction between sensory, internal and motor information. Recent studies have shown that single-neuron activities referred to as mixed selectivity which are ubiquitous in the prefrontal cortex (PFC) are a possible mechanism to represent arbitrary interaction between information defining a contingency. In parallel, a recent area of reasearch referred to as Reservoir Computing has demonstrated that recurrent neural networks have the property of recombining present and past inputs into a higher dimensional space thereby providing a pre-coding of an essentially universal set of combinations which can then be selected and used arbitrarily for their relevance to the task at hand. Combining these two approaches we argue that the highly recurrent nature of local prefrontal connectivity is at the origin of dynamic form of mixed selectivity. Also, we attempt to demonstrate that a simple linear regression, implementable by a single neuron, can extract any information/ contingency encoded in these highly complex and dynamic combinations. In addition, previous inputs, whether sensory or motor, to these PFC networks must be maintained in order to influence current processing and behavioral demand. We argue that representations of contexts defined by these past inputs must be expressed explicitely and fed back to the local PFC networks in order to influence the current combinations at the origin of contingencies representation; Les primates doivent pouvoir reconnaître de nouvelles situations pour pouvoir s'y adapter. La représentation de ces situations dans l'activité du cortex est le sujet de cette thèse. Les situations complexes s'expliquent souvent par l'interaction entre des informations sensorielles, internes et motrices. Des activités unitaires dénommées sélectivité mixte, qui sont très présentes dans le cortex préfrontal (CPF), sont un mécanisme possible pour représenter n'importe quelle interaction entre des informations. En parallèle, le Reservoir Computing a démontré que des réseaux récurrents ont la propriété de recombiner des entrées actuelles et passées dans un espace de plus haute dimension, fournissant ainsi un pré-codage potentiellement universel de combinaisons pouvant être ensuite sélectionnées et utilisées en fonction de leur pertinence pour la tâche courante. En combinant ces deux approches, nous soutenons que la nature fortement récurrente de la connectivité locale du CPF est à l'origine d'une forme dynamique de sélectivité mixte. De plus, nous tentons de démontrer qu'une simple régression linéaire, implémentable par un neurone seul, peut extraire n'importe qu'elle information/contingence encodée dans ces combinaisons complexes et dynamiques. Finalement, les entrées précédentes, qu'elles soient sensorielles ou motrices, à ces réseaux du CPF doivent être maintenues pour pouvoir influencer les traitements courants. Nous soutenons que les représentations de ces contextes définis par ces entrées précédentes doivent être exprimées explicitement et retournées aux réseaux locaux du CPF pour influencer les combinaisons courantes à l'origine de la représentation des contingences

How the brain combines information from different sensory modalities and of differing reliability is an important and still-unanswered question. Using the head direction (HD) system as a model, we explored the resolution of conflicts between landmarks and background cues. Sensory cue integration models predict averaging of the two cues, whereas attractor models predict capture of the signal by the dominant cue. We found that a visual landmark mostly captured the HD signal at low conflicts: however, there was an increasing propensity for the cells to integrate the cues thereafter. A large conflict presented to naive rats resulted in greater visual cue capture (less integration) than in experienced rats, revealing an effect of experience. We propose that weighted cue integration in HD cells arises from dynamic plasticity of the feed-forward inputs to the network, causing within-trial spatial redistribution of the visual inputs onto the ring. This suggests that an attractor network can implement decision processes about cue reliability using simple architecture and learning rules, thus providing a potential neural substrate for weighted cue integration.

During rest, the human brain performs essential functions such as memory maintenance, which are associated with resting-state brain networks (RSNs) including the default-mode network (DMN) and frontoparietal network (FPN). Previous studies based on spiking-neuron network models and their reduced models, as well as those based on imaging data, suggest that resting-state network activity can be captured as attractor dynamics, i.e., dynamics of the brain state toward an attractive state and transitions between different attractors. Here, we analyze the energy landscapes of the RSNs by applying the maximum entropy model, or equivalently the Ising spin model, to human RSN data. We use the previously estimated parameter values to define the energy landscape, and the disconnectivity graph method to estimate the number of local energy minima (equivalent to attractors in attractor dynamics), the basin size, and hierarchical relationships among the different local minima. In both of the DMN and FPN, low-energy local minima tended to have large basins. A majority of the network states belonged to a basin of one of a few local minima. Therefore, a small number of local minima constituted the backbone of each RSN. In the DMN, the energy landscape consisted of two groups of low-energy local minima that are separated by a relatively high energy barrier. Within each group, the activity patterns of the local minima were similar, and different minima were connected by relatively low energy barriers. In the FPN, all dominant local minima were separated by relatively low energy barriers such that they formed a single coarse-grained global minimum. Our results indicate that multistable attractor dynamics may underlie the DMN, but not the FPN, and assist memory maintenance with different memory states.

Different models have been able to account for different features of the data on grid cell firing properties, including the relationship of grid cells to cellular properties and network oscillations. This paper describes a model that combines elements of two major classes of models of grid cells: models using interference of oscillations and models using attractor dynamics. This model includes a population of units with oscillatory input representing input from the medial septum. These units are termed heading angle cells because their connectivity depends upon heading angle in the environment as well as the spatial phase coded by the cell. These cells project to a population of grid cells. The sum of the heading angle input results in standing waves of circularly symmetric input to the grid cell population. Feedback from the grid cell population increases the activity of subsets of the heading angle cells, resulting in the network settling into activity patterns that resemble the patterns of firing fields in a population of grid cells. The properties of heading angle cells firing as conjunctive grid-by-head-direction cells can shift the grid cell firing according to movement velocity. The pattern of interaction of oscillations requires use of separate populations that fire on alternate cycles of the net theta rhythmic input to grid cells, similar to recent neurophysiological data on theta cycle skipping in medial entorhinal cortex.

In this paper we address the problem of humanrobot joint transportation of large payloads. The human brings to the task knowledge on the goal destination and global path planning. The robot has no prior knowledge of the environment and must autonomously help the human, while simultaneously avoiding static and/or dynamic obstacles that it encounters. For this purpose a dynamic control architecture, formalized as a coupled system of non-linear differential equations, is designed to control the behavior of the mobile manipulator in close loop with the acquired sensorial information. Verbal communication is integrated that allows the robot to communicate its limitations. Results show the robot’s ability to generate stable, smooth and robust behavior in unstructured and dynamic environments. Furthermore, the robot is able to explain the difficulties it encounters and thus contribute to success of the task and to enhance the human-robot physical interaction., FP6-IST2-EU-project JAST (project no 003747), Portuguese Science and Technology Foundation (FCT) and FEDER project COOPDYN (POSI/SRI/38081/2001).

The speed-accuracy trade-off (SAT) is ubiquitous in decision tasks. While the neural mechanisms underlying decisions are generally well characterized, the application of decision-theoretic methods to the SAT has been difficult to reconcile with experimental data suggesting that decision thresholds are inflexible. Using a network model of a cortical decision circuit, we demonstrate the SAT in a manner consistent with neural and behavioural data and with mathematical models that optimize speed and accuracy with respect to one another. In simulations of a reaction time task, we modulate the gain of the network with a signal encoding the urgency to respond. As the urgency signal builds up, the network progresses through a series of processing stages supporting noise filtering, integration of evidence, amplification of integrated evidence, and choice selection. Analysis of the network's dynamics formally characterizes this progression. Slower buildup of urgency increases accuracy by slowing down the progression. Faster buildup has the opposite effect. Because the network always progresses through the same stages, decision-selective firing rates are stereotyped at decision time.

Este trabalho tem como objectivo a aplicação da abordagem baseada em atractores de sistemas dinâmicos, usada para o controlo do movimento de um veículo aéreo aerostático semi-autónomo. Foi projectada uma arquitectura de controlo formalizada como um sistema dinâmico não linear. Esta foi inicialmente validada através de simulações computacionais. Numa primeira fase usámos um simulador construído no programa MATLAB e a ferramenta SIMULINK desse programa. Numa fase posterior a simulação foi feita em linguagem C++ e a ferramenta gráfica OpenGL. Desta forma o cálculo do modelo do dirigível passou a ser realizado com recurso ao algoritmo de resolução de equações diferenciais por métodos numéricos, sendo usado o método de quarta ordem de Runge-Kuta. Finalmente fizemos a implementação física da arquitectura de controlo no dirigível adquirido, tendo para isso sido construído um interface entre o computador e o comando dos motores. A informação sensorial provém de uma câmara de vídeo CMOS instalada na gôndola do dirigível. O processamento de imagem dá-nos a direcção, a distância e a diferença de altitude à qual os obstáculos e os alvos estão localizados em relação à direcção frontal do dirigível. Esta informação é usada para gerar a dinâmica comportamental que será a responsável pela actuação dos motores., The goal of this work is the application of the approach based on attractors dynamics to the motion control of a semi-autonomous aerostatic air vehicle. The designed control architecture is fully formalized as a non-linear dynamical system. This was first validated through computational simulations. In order to do so, a simulator built with Matlab Simulink was used. In a latter stage a new simulator was built using C + + and OpenGL. In the second simulator, the Simulink space state model of the airship was replaced by differential equations solved with a fourth order Runge-Kuta method. The last step was the physical implementation of the control architecture on a real airship. To made this possible, an interface was designed and built so we could communicate with the airship thrusters using the computer. Sensory information comes from a CMOS camera in the airship gondola. Image processing gives us direction, distance and the altitude diference at which targets and obstacles lie relatively to the airship’s frontal direction. This information is then used to generate the behavioral dynamics which will be responsible for the actuation of the thrusters., Fundação para a Ciência e a Tecnologia (FCT) - projecto POSI/SRI/38051/2001

To enable floating robots to autonomously reach for a target position while avoiding obstacles we have generalized the attractor dynamics approach established for wheeled mobile robots to motion generation in blimps or lighter-than-air vehicles. In this approach the level of modelling is at the level of behaviours. A “dynamics'' of behaviour is defined over a state space of behavioural variables (heading direction, forward velocity and altitude). The environment is also modelled in these terms by representing task constraints as attractors (i.e. asymptotically stable states) or reppelers (i.e. unstable states) of behavioural dynamics. Attractors and repellers are combined into a vector field that governs the blimp’s behaviour. The resulting dynamical systems that generate the flying behaviour is non-linear and presents several attractors and reppelers (typically few) . By design the dynamic systems are tuned so that the behavioural variables are always very close to one attractor. Thus the motion of the airship is controlled by a time series of asymptotically stable states. Computer simulations that integrate the dynamic control architecture and the blimp’s physical model indicate that if parameter values are chosen within reasonable ranges, then the over all system works quite well even in cluttered environments. The stability properties of the dynamic control architecture enable the floating robot to remain robust against perturbations.

We describe a model of invariant visual object recognition in the brain that incorporates feedback biasing effects of top–down attentional mechanisms on a hierarchically organized set of visual cortical areas with convergent forward connectivity, reciprocal feedback connections, and local intra-area competition. The model displays space-based and object-based covert visual search by using attentional top–down feedback from either the posterior parietal or the inferior temporal cortex (IT) modules, and interactions between the two processing streams occurring in V1 and V2. The model explains the gradually increasing magnitude of the attentional modulation that is found in fMRI experiments from earlier visual areas (V1, V2) to higher ventral stream visual areas (V4, IT); how the effective size of the receptive fields of IT neurons becomes smaller in natural cluttered scenes; and makes predictions about interactions between stimuli in their receptive fields.

Head direction (HD) cell responses are thought to be derived from a combination of internal (or idiothetic) and external (or allothetic) sources of information. Recent work from the Jeffery laboratory shows that the relative influence of visual versus vestibular inputs upon the HD cell response depends on the disparity between these sources. In this paper, we present simulation results from a model designed to explain these observations. The model accurately replicates the Knight et al. data. We suggest that cue conflict resolution is critically dependent on plastic remapping of visual information onto the HD cell layer. This remap results in a shift in preferred directions of a subset of HD cells, which is then inherited by the rest of the cells during path integration. Thus, we demonstrate how, over a period of several minutes, a visual landmark may gain cue control. Furthermore, simulation results show that weaker visual landmarks fail to gain cue control as readily. We therefore suggest a second longer term plasticity in visual projections onto HD cell areas, through which landmarks with an inconsistent relationship to idiothetic information are made less salient, significantly hindering their ability to gain cue control. Our results provide a mechanism for reliability-weighted cue averaging that may pertain to other neural systems in addition to the HD system.

What we see, how we see it and what emotions may arise from stimuli has long been studied by philosophers, psychologists, medical doctors and neuroscientists. This thesis work investigates a particular view on the possible dynamics, utilizing computational models of spiking neural attractor networks. From neurological studies on humans and other primates, we know visual perception and recognition of objects occur partly along the visual ventral stream, from V1 to V2, V4, IT and downstream to other areas. This visual awareness can be both conscious and unconscious and may also trigger an emotional response. As seen from many psychophysical experiments in backward masking (BM) and attentional blink (AB), some spatial and temporal dynamics can determine what becomes visually conscious and what does not. To explore this computationally, biophysical models of BM and AB were implemented and simulated to mimic human experiments, with the assumption that neural assemblies as attractor networks activate and propagate along the ventral stream and beyond. It was observed that attractor interference between percepts in sensory and associative cortex can occur during this activity. During typical human AB experimental trials in which two expected target symbols amongst distractors are presented less than 500 ms apart, the second target is often not reported as seen. When simulating this paradigm as two expected target neural attractors amongst distractors, it was observed in the present work that an initial attractor in associative cortex can impede the activation and propagation of a following attractor, which mimics missing conscious perception of the second target. It was also observed that simulating the presence of benzodiazepines (GABA agonists) will slow cortical dynamics and increase the AB, as previously shown in human experiments. During typical human BM experimental trials in which a brief target stimulus is followed by a masking stimulus after a short interval of less than 100 ms, recognition of the target can be impaired when in close spatial proximity. When simulating this paradigm using a biophysical model of V1 and V2 with feedforward and feedback connections, attractor targets were activated in V1 before imposition of a proximal metacontrast mask. If an activating target attractor in V1 is quiesced enough with lateral inhibition from a mask, or not reinforced by recurrent feedback from feedforward activation in V2, it is more likely to burn out before becoming fully active and progressing through V2 and beyond. BM was also simulated with an increasing stimulus interval and with the presence and absence of feedback activity. This showed that recurrent feedback diminishes BM effects and can make conscious perception more likely. To better understand possible emotional components of visual perception and early regulation, visual signaling pathways to the amygdala were investigated and proposed for emotional salience and the possible onset of fear. While one subcortical and likely unconscious pathway (before amydala efferent signaling) was affirmed via the superior colliculus and pulvinar, four others traversed through the ventral stream. One traversed though IT on recognition, another via the OFC on conditioning, and two other possibly conscious pathways traversed though the parietal and then prefrontal cortex, one excitatory pathway via the ventral-medial area and one regulatory pathway via the ventral-lateral area. Predicted latencies were determined for these signaling pathways, which can be experimentally testable. The conscious feeling of fear itself may not occur until after interoceptive inspection. A pathology of attractor dynamics was also investigated, which can occur from the presence of a brain tumor in white matter. Due to degradation from tumor invasion of white matter projections between two simulated neocortical patches, information transfer between separate neural attractors degraded, leading first to recall errors and later to epileptic-like activity. Neural plasticity could partially compensate up to a point, before transmission failure. This suggests that once epileptic seizures start in glioma patients, compensatory plasticity may already be exhausted. Interestingly, the presence of additional noise could also partially compensate for white matter loss. QC 20180305

We demonstrate bistable attractor dynamics in a spiking neural network implemented with neuromorphic VLSI hardware. The on-chip network consists of three interacting populations (two excitatory, one inhibitory) of leaky integrate-and-fire (LIF) neurons. One excitatory population is distinguished by strong synaptic self-excitation, which sustains meta-stable states of “high” and “low”-firing activity. Depending on the overall excitability, transitions to the “high” state may be evoked by external stimulation, or may occur spontaneously due to random activity fluctuations. In the former case, the “high” state retains a “working memory” of a stimulus until well after its release. In the latter case, “high” states remain stable for seconds, three orders of magnitude longer than the largest time-scale implemented in the circuitry. Evoked and spontaneous transitions form a continuum and may exhibit a wide range of latencies, depending on the strength of external stimulation and of recurrent synaptic excitation. In addition, we investigated “corrupted” “high” states comprising neurons of both excitatory populations. Within a “basin of attraction,” the network dynamics “corrects” such states and re-establishes the prototypical “high” state. We conclude that, with effective theoretical guidance, full-fledged attractor dynamics can be realized with comparatively small populations of neuromorphic hardware neurons.

Preprint, We propose and demonstrate how attractor dynamics can be used to design and implement a distributed dynamic control architecture that enables a team of two robots, without force/torque sensors and equipped solely with low-level sensors, to carry a long object and simultaneously avoid obstacles. The explicit required communication between robots is minimal. The robots have no prior knowledge of their environment. Experimental results in indoor environments show that if parameter values are chosen within reasonable ranges then the overall system works quite well even in cluttered environments. The robots’ behavior is stable and the generated trajectories are smooth.

This paper addresses the problem underlying the control and coordination of multiple autonomous airships that must travel maintaining a desired geometric formation and simultaneously avoid collisions with moving or stationary obstacles. The control architecture is based on the attractor dynamics approach to behaviour generation. The airship physical model is presented and the mathematical background for the control architecture is explained. Simulations (with perturbations) with formations of two and three autonomous airships are presented in order to validate the architecture., (undefined)