Published papers

Neurodevelopmental disorders, including autism spectrum disorder, have been intensively investigated at the neural, cognitive, and behavioral levels, but the accumulated knowledge remains fragmented. In particular, developmental learning aspects of symptoms and interactions with the physical environment remain largely unexplored in computational modeling studies, although a leading computational theory has posited associations between psychiatric symptoms and an unusual estimation of information uncertainty (precision), which is an essential aspect of the real world and is estimated through learning processes. Here, we propose a mechanistic explanation that unifies the disparate observations via a hierarchical predictive coding and developmental learning framework, which is demonstrated in experiments using a neural network-controlled robot. The results show that, through the developmental learning process, homogeneous intrinsic neuronal excitability at the neural level induced via self-organization changes at the information processing level, such as hyper sensory precision and overfitting to sensory noise. These changes led to multifaceted alterations at the behavioral level, such as inflexibility, reduced generalization, and motor clumsiness. In addition, these behavioral alterations were accompanied by fluctuating neural activity and excessive development of synaptic connections. These findings might bridge various levels of understandings in autism spectrum and other neurodevelopmental disorders and provide insights into the disease processes underlying observed behaviors and brain activities in individual patients. This study shows the potential of neurorobotics frameworks for modeling how psychiatric disorders arise from dynamic interactions among the brain, body, and uncertain environments.

Intelligent robots are expected to collaboratively work with humans in dynamically changing daily life environments. To realize successful human-robot collaboration, robots need to deal with latent spatiotemporal complexity in the workspace and the task. To overcome this crucial issue, three levels of adaptability-motion modification, action selection, and role switching-should be considered. This paper demonstrates that a single hierarchically organized neuro-dynamical system called a multiple timescale recurrent neural network can achieve these levels of adaptability by utilizing hierarchical and bidirectional information processing. The system is implemented in a humanoid robot and the robot is required to learn to perform collaborative tasks in which some parts must be performed by a human partner and others by the robot. Experimental results show that the robot can perform collaborative tasks under dynamically changing environments, including both learned and unlearned situations, thanks to different levels of adaptability acquired in the system.

Functional connectivity (FC) and neural excitability may interact to affect symptoms of autism spectrum disorder (ASD). We tested this hypothesis with neural network simulations, and applied it with functional magnetic resonance imaging (fMRI). A hierarchical recurrent neural network embodying predictive processing theory was subjected to a facial emotion recognition task. Neural network simulations examined the effects of FC and neural excitability on changes in neural representations by developmental learning, and eventually on ASD-like performance. Next, by mapping each neural network condition to subject subgroups on the basis of fMRI parameters, the association between ASD-like performance in the simulation and ASD diagnosis in the corresponding subject subgroup was examined. In the neural network simulation, the more homogeneous the neural excitability of the lower-level network, the more ASD-like the performance (reduced generalization and emotion recognition capability). In addition, in homogeneous networks, the higher the FC, the more ASD-like performance, while in heterogeneous networks, the higher the FC, the less ASD-like performance, demonstrating that FC and neural excitability interact. As an underlying mechanism, neural excitability determines the generalization capability of top-down prediction, and FC determines whether the model’s information processing will be top-down prediction-dependent or bottom-up sensory-input dependent. In fMRI datasets, ASD was actually more prevalent in subject subgroups corresponding to the network condition showing ASD-like performance. The current study suggests an interaction between FC and neural excitability, and presents a novel framework for computational modeling and biological application of a developmental learning process underlying cognitive alterations in ASD.

Creating autonomous robots that can actively explore the environment, acquire knowledge and learn skills continuously is the ultimate achievement envisioned in cognitive and developmental robotics. Importantly, if the aim is to create robots that can continuously develop through interactions with their environment, their learning processes should be based on interactions with their physical and social world in the manner of human learning and cognitive development. Based on this context, in this paper, we focus on the two concepts of world models and predictive coding. Recently, world models have attracted renewed attention as a topic of considerable interest in artificial intelligence. Cognitive systems learn world models to better predict future sensory observations and optimize their policies, i.e. controllers. Alternatively, in neuroscience, predictive coding proposes that the brain continuously predicts its inputs and adapts to model its own dynamics and control behavior in its environment. Both ideas may be considered as underpinning the cognitive development of robots and humans capable of continual or lifelong learning. Although many studies have been conducted on predictive coding in cognitive robotics and neurorobotics, the relationship between world model-based approaches in AI and predictive coding in robotics has rarely been discussed. Therefore, in this paper, we clarify the definitions, relationships, and status of current research on these topics, as well as missing pieces of world models and predictive coding in conjunction with crucially related concepts such as the free-energy principle and active inference in the context of cognitive and developmental robotics. Furthermore, we outline the frontiers and challenges involved in world models and predictive coding toward the further integration of AI and robotics, as well as the creation of robots with real cognitive and developmental capabilities in the future.

Recently, applying computational models developed in cognitive science to psychiatric disorders has been recognized as an essential approach for understanding cognitive mechanisms underlying psychiatric symptoms. Autism spectrum disorder is a neurodevelopmental disorder that is hypothesized to affect information processes in the brain involving the estimation of sensory precision (uncertainty), but the mechanism by which observed symptoms are generated from such abnormalities has not been thoroughly investigated. Using a humanoid robot controlled by a neural network using a precision-weighted prediction error minimization mechanism, it is suggested that both increased and decreased sensory precision could induce the behavioral rigidity characterized by resistance to change that is characteristic of autistic behavior. Specifically, decreased sensory precision caused any error signals to be disregarded, leading to invariability of the robot's intention, while increased sensory precision caused an excessive response to error signals, leading to fluctuations and subsequent fixation of intention. The results may provide a system-level explanation of mechanisms underlying different types of behavioral rigidity in autism spectrum and other psychiatric disorders. In addition, our findings suggest that symptoms caused by decreased and increased sensory precision could be distinguishable by examining the internal experience of patients and neural activity coding prediction error signals in the biological brain.

We propose an imitative learning model that allows a robot to acquire positional relations between the demonstrator and the robot, and to transform observed actions into robotic actions. Providing robots with imitative capabilities allows us to teach novel actions to them without resorting to trial-and-error approaches. Existing methods for imitative robotic learning require mathematical formulations or conversion modules to translate positional relations between demonstrators and robots. The proposed model uses two neural networks, a convolutional autoencoder (CAE) and a multiple timescale recurrent neural network (MTRNN). The CAE is trained to extract visual features from raw images captured by a camera. The MTRNN is trained to integrate sensory-motor information and to predict next states. We implement this model on a robot and conducted sequence to sequence learning that allows the robot to transform demonstrator actions into robot actions. Through training of the proposedmodel, representations of actions, manipulated objects, and positional relations are formed in the hierarchical structure of the MTRNN. After training, we confirm capability for generating unlearned imitative patterns.

Neurodevelopmental disorders are characterized by heterogeneous and non-specific nature of their clinical symptoms. In particular, hyper- and hypo-reactivity to sensory stimuli are diagnostic features of autism spectrum disorder and are reported across many neurodevelopmental disorders. However, computational mechanisms underlying the unusual paradoxical behaviors remain unclear. In this study, using a robot controlled by a hierarchical recurrent neural network model with predictive processing and learning mechanism, we simulated how functional disconnection altered the learning process and subsequent behavioral reactivity to environmental change. The results show that, through the learning process, long-range functional disconnection between distinct network levels could simultaneously lower the precision of sensory information and higher-level prediction. The alteration caused a robot to exhibit sensory-dominated and sensory-ignoring behaviors ascribed to sensory hyper- and hypo-reactivity, respectively. As long-range functional disconnection became more severe, a frequency shift from hyporeactivity to hyperreactivity was observed, paralleling an early sign of autism spectrum disorder. Furthermore, local functional disconnection at the level of sensory processing similarly induced hyporeactivity due to low sensory precision. These findings suggest a computational explanation for paradoxical sensory behaviors in neurodevelopmental disorders, such as coexisting hyper- and hypo-reactivity to sensory stimulus. A neurorobotics approach may be useful for bridging various levels of understanding in neurodevelopmental disorders and providing insights into mechanisms underlying complex clinical symptoms.

AbstractThe mechanism underlying the emergence of emotional categories from visual facial expression information during the developmental process is largely unknown. Therefore, this study proposes a system-level explanation for understanding the facial emotion recognition process and its alteration in autism spectrum disorder (ASD) from the perspective of predictive processing theory. Predictive processing for facial emotion recognition was implemented as a hierarchical recurrent neural network (RNN). The RNNs were trained to predict the dynamic changes of facial expression movies for six basic emotions without explicit emotion labels as a developmental learning process, and were evaluated by the performance of recognizing unseen facial expressions for the test phase. In addition, the causal relationship between the network characteristics assumed in ASD and ASD-like cognition was investigated. After the developmental learning process, emotional clusters emerged in the natural course of self-organization in higher-level neurons, even though emotional labels were not explicitly instructed. In addition, the network successfully recognized unseen test facial sequences by adjusting higher-level activity through the process of minimizing precision-weighted prediction error. In contrast, the network simulating altered intrinsic neural excitability demonstrated reduced generalization capability and impaired emotional clustering in higher-level neurons. Consistent with previous findings from human behavioral studies, an excessive precision estimation of noisy details underlies this ASD-like cognition. These results support the idea that impaired facial emotion recognition in ASD can be explained by altered predictive processing, and provide possible insight for investigating the neurophysiological basis of affective contact.

We propose a tool-use model that enables a robot to act toward a provided goal. It is important to consider features of the four factors; tools, objects actions, and effects at the same time because they are related to each other and one factor can influence the others. The tool-use model is constructed with deep neural networks (DNNs) using multimodal sensorimotor data; image, force, and joint angle information. To allow the robot to learn tool-use, we collect training data by controlling the robot to perform various object operations using several tools with multiple actions that leads different effects. Then the tool-use model is thereby trained and learns sensorimotor coordination and acquires relationships among tools, objects, actions and effects in its latent space. We can give the robot a task goal by providing an image showing the target placement and orientation of the object. Using the goal image with the tool-use model, the robot detects the features of tools and objects, and determines how to act to reproduce the target effects automatically. Then the robot generates actions adjusting to the real time situations even though the tools and objects are unknown and more complicated than trained ones.

Collaboration with a human partner is a challenging task expected of intelligent robots. To realize this, robots need the ability to share a particular goal with a human and dynamically infer whether the goal state is changed by the human. In this paper, we propose a neural network-based computational framework with a gradient-based optimization of the goal state that enables robots to achieve this ability. The proposed framework consists of convolutional variational autoencoders (ConvVAEs) and a recurrent neural network (RNN) with a long short-term memory (LSTM) architecture that learns to map a given goal image for collaboration to visuomotor predictions. More specifically, visual and goal feature states are first extracted by the encoder of the respective ConvVAEs. Visual feature and motor predictions are then generated by the LSTM based on their current state and are conditioned according to the extracted goal feature state. During collaboration after the learning process, the goal feature state is optimized by gradient descent to minimize errors between the predicted and actual visual feature states. This enables the robot to dynamically infer situational (goal) changes of the human partner from visual observations alone. The proposed framework is evaluated by conducting experiments on a human–robot collaboration task involving object assembly. Experimental results demonstrate that a robot equipped with the proposed framework can collaborate with a human partner through dynamic goal inference even when the situation is ambiguous.

We propose a tool-use model that can detect the features of tools, target objects, and actions from the provided effects of object manipulation. We construct a model that enables robots to manipulate objects with tools, using infant learning as a concept. To realize this, we train sensory-motor data recorded during a tool-use task performed by a robot with deep learning. Experiments include four factors: (1) tools, (2) objects, (3) actions, and (4) effects, which the model considers simultaneously. For evaluation, the robot generates predicted images and motions given information of the effects of using unknown tools and objects. We confirm that the robot is capable of detecting features of tools, objects, and actions by learning the effects and executing the task.

Robots with flexible joints have recently been attracting attention from researchers because such robots can passively adapt to environmental changes and realize dynamic motion that uses inertia. In previous research, body-model acquisition using deep learning was proposed and dynamic motion learning was achieved. However, using the end-effector position as a visual feedback signal to train a robot limits what the robot can know to only the relation between the task and itself, instead of the relation between the environment and itself. In this research, we propose to use images as a feedback signal so that the robot can have a sense of the overall situation within the task environment. This motion learning is performed via deep learning using raw image data. In an experiment, we let a robot perform task motions once to acquire motor and image data. Then, we used a convolutional auto-encoder to extract image features from raw image data. The extracted image features were used in combination with motor data to train a recurrent neural network. As a result, motion learning through deep learning from image data allowed the robot to acquire environmental information and conduct tasks that require consideration of environmental changes, making use of its advantage of passive adaptation.

Drawing is a representative human cognitive ability and may mirror cognitive characteristics including those associated with psychiatric symptoms. Therefore, analysis of drawing data collected from various populations such as healthy people and psychiatric patients may be beneficial for better understanding human cognition. However, collecting such large-scale data about the relationship between drawing and cognitive/personality traits offline-in a laboratory-is a difficult issue. To overcome this issue, we devised a novel experimental paradigm involving a goal-directed drawing task conducted online-on the eb-with participants recruited via a crowdsourcing platform. With the assistance of 1155 participants with differing levels of psychiatric symptoms, we collected a total of 194, 040 trajectory data and answers to seven different self-report psychiatric symptom questionnaires comprising 181 items. We visualized the collected trajectory data and performed an exploratory factor analysis on the correlation matrix of the psychiatric symptom questionnaire items. Our results suggest that there were associations between psychiatric symptoms represented by specific psychiatric factors and atypical behavior observed while performing the goal-directed drawing task. This indicates the efficacy of a dimensional approach to large-scale online experiments with respect to clinical psychiatry.

Adaptation and planning are crucial for both biological and artificial agents. In this study, we treat these as an inference problem that we solve using a gradient-based optimization approach. We propose adaptation and planning by gradient descent (APGraDe), a gradient-based computational framework with a hierarchical recurrent neural network (RNN) for adaptation and planning. This framework computes (counterfactual) prediction errors by looking back on past situations based on actual observations and by looking ahead to future situations based on preferred observations (or goal). The internal state of the higher level of the RNN is optimized in the direction of minimizing these errors. The errors for the past contribute to the adaptation while errors for the future contribute to the planning. The proposed APGraDe framework is implemented in a humanoid robot and the robot performs a ball manipulation task with a human experimenter. Experimental results show that given a particular preference, the robot can adapt to unexpected situations while pursuing its own preference through the planning of future actions.

Human sensing enables intelligent vehicles to provide driver-adaptive support by classifying perceived workload into multiple levels. Objective of this study is to classify driver workload associated with traffic complexity into five levels. We conducted driving experiments in systematically varied traffic complexity levels in a simulator. We recorded driver physiological signals including electrocardiography, electrodermal activity, and electroencephalography. In addition, we integrated driver performance and subjective workload measures. Deep learning based models outperform statistical machine learning methods when dealing with dynamic time-series data with variable sequence lengths. We show that our long short-term memory based recurrent neural network model can classify driver perceived-workload into five classes with an accuracy of 74.5%. Since perceived workload differ between individual drivers for the same traffic situation, our results further highlight the significance of including driver characteristics such as driving style and workload sensitivity to achieve higher classification accuracy.

Recently, the importance of the application of computational models utilized in cognitive neuroscience to psychiatric disorders has been recognized. This study utilizes a recurrent neural network model to test aberrant sensory precision, a normative theory of autism spectrum disorder. We particularly focus on the effects of increased and decreased sensory precision on adaptive behavior based on a prediction error minimization mechanism. To distinguish dysfunction at the behavioral and network levels, we employ a humanoid robot driven by a neural network and observe ball-playing interactions with a human experimenter. Experimental results show that behavioral rigidity characteristic of autism spectrum disorder - including stopping movement and repetitive behavior - was generated from both increased and decreased sensory precision, but through different processes at the network level. These results may provide a system-level explanation of different types of behavioral rigidity in psychiatric diseases such as compulsions and stereotypies. The results also support a system-level model for autism spectrum disorder that suggests core deficits in estimating the uncertainty of sensory evidence.

We propose a tool-use model that can select tools that require neither labeling nor modeling of the environment and actions. With this model, a robot can choose a tool by itself and perform the operation that matches a human command and the environmental situation. To realize this, we use deep learning to train sensory motor data recorded during tool selection and tool use as experienced by a robot. The experience includes two types of selection, namely according to function and according to size, thereby allowing the robot to handle both situations. For evaluation, the robot is required to generate motion either in an untrained situation or using an untrained tool. We confirm that the robot can choose and use a tool that is suitable for achieving the target task.