Fifth Annual Meeting of the BICA Society
November 7-9 (Friday-Sunday): Massachusetts Institute of Technology, Cambridge,MA 02139, USA (http://bicasociety.org/meetings/2014).
Sponsors: The BICA Society; The MIT; Elsevier B.V.; DOLLabs, Inc. Points of contact: Paul Robertson (firstname.lastname@example.org) and Alexei Samsonovich (email@example.com)
Selected Invited Speakers, Alphabetically
David W. Aha (keynote) (Adaptive Systems Section, NCARAI, Naval Research laboratory, USA)
Rajan Bhattacharyya (CNES, ISSL, HRL Laboratories, LLC, USA)
Michael Cox (Wright State Research Institute, USA)
Antonio Damasio (University of Southern California, USA)
Olivier Georgeon (Universite Claude Bernard Lyon 1, France)
Stephen Grossberg (Boston University, USA)
Eva Hudlicka (Psychometrix Associates & University of Massachusetts Amherst, USA)
Robert Laddaga (DARPA Information Innovation Office, USA)
Louis Lome (former MDA/BMDO Program Manager, USA)
Frank Ritter (Pennsylvania State University, USA)
Paul Robertson (Dynamic Object Language Laboratories, Inc., USA)
Sweitze Roffel (Elsevier B.V., The Netherlands)
Paul Rosenbloom (University of Southern California, USA)
Walter Schneider (University of Pittsburgh, USA)
Howard Schrobe (MIT Computer Science and Artificial Intelligence Laboratory, USA)
Terrence C. Stewart (University of Waterloo, Canada)
Troy D. Kelley (U.S. Army Research Laboratory, USA)
Junichi Takeno (Meiji University, Japan)
Giulio Tononi (University of Wisconsin Madison, USA)
Craig Vineyard (Cognitive Sciences, Sandia National Laboratories, USA)
Mary-Anne Williams (University of Technology, Sydney, Australia)
Patrick H. Winston (MIT Computer Science and Artificial Intelligence Laboratory, USA)
Truncated Abstracts of Accepted Submissions
Antonio Lieto. A Computational Framework for Concept Representation in Cognitive Systems and Architectures: Concepts as Heterogeneous Proxytypes
In this paper a general framework for the representation of concepts in cognitive artificial systems and cognitive architectures is provided. The proposed framework is inspired by the so called proxytype theory of concepts and combines it with the heterogeneity approach to concept representations in Cognitive Science, according to which concepts do not constitute a unitary phenomenon. The contribution of the paper is twofold: on one hand, it is aimed at providing a computational characterization of the problem of concept representation in cognitively inspired artificial systems and ...
Himanshu Joshi, Paul Rosenbloom and Volkan Ustun. Isolated Word Recognition in the Sigma Cognitive Architecture
Symbolic architectures are effective at complex cognitive reasoning, but typically are incapable of important forms of sub-cognitive processing – such as perception – without distinct modules connected to them via low-bandwidth interfaces. Neural architectures, in contrast, may be quite effective at the latter, but typically struggle with the former. Sigma has been designed to leverage the state-of-the-art hybrid (discrete + continuous) mixed (symbolic + probabilistic) capability of graphical models to provide in a uniform non-modular fashion effective forms of, and integration across, ...
Kazuteru Miyazaki and Junichi Takeno. A study on the necessity of a secondary system in the consciousness system
Our research purpose is to realize a consciousness system on computers. In this paper, we focus on the relationship between a primary system, that learns the input-output relation with an environment, and a secondary system, that is able to act against the primary system. We believe that consciousness is not constructed with only the primary system, and the presence of the secondary system is essential. The purpose of this paper is to clarify the importance of the secondary system. We show that the secondary system can follow more wider range of environmental changes than the primary system ...
Koki Kanazawa and Junichi Takeno. A proposal for a Pavlov robot
Ivan Petrovich Pavlov was awarded the 1904 Nobel Prize in Physiology and Medicine for his experiments. One of his famous experiments is related to dogs. When presented with their food, the dogs exhibited a physiological (mental) phenomenon in which their mouths began salivating. Also, a metronome would be rung at the same time the food was presented in the experiments, and gradually a phenomenon appeared in which the dogs would salivate only from the sound of the metronome. The authors believe that attempting to reproduce this phenomenon using a robot is crucial to understanding the ...
Paul Rosenbloom, Abram Demski and Volkan Ustun. Efficient Message Computation in Sigma’s Graphical Architecture
Human cognition runs at ~50 msec per cognitive cycle, implying that any biologically inspired cognitive architecture that strives for real-time performance needs to be able to run at this speed. Sigma is a cognitive architecture built upon graphical models – a broadly applicable state-of-the-art formalism for implementing cognitive capabilities – that are solved via message passing (with complex messages based on n-dimensional piecewise-linear functions). Earlier work explored optimizations to Sigma that reduced by an order of magnitude the number of messages sent per cycle. Here, ...
Wael Hafez. General Architecture for Perception-Action Information Flow Control
The perception-action (P-A) cycle is the process of receiving information about sensory events taking place in the environment through sensory systems, processing this information and triggering motor systems to initiate corresponding action events back into the environment. This information processing along the P-A pathway requires defining a certain level of dependency between the sensory events and their corresponding action events. This dependency is what determines the behavior of the system and its ability to implement its goals and to react to its environment. Information processing ...
Suhas E. Chelian, Ryan M. Uhlenbrock, Seth Herd and Rajan Bhattacharyya. Application of a neural network model of prefrontal cortex to emulate human probability matching behavior
Probability matching behavior occurs in a variety of decision-making domains that can be mapped to the n-arm bandit problem. Prefrontal cortex has been implicated in executive control over several tasks including the n-arm bandit problem. Previously the Prefrontal cortex Basal Ganglia Working Memory (PBWM) model has been used to replicate other decision-making functions of prefrontal cortex such as recognizing sequences of symbols or visual scenes. In this work, we emulate probability matching from human subjects using the PBWM model in n-arm bandit-like problems. Possible extensions to ...
Suhas E. Chelian, Matthias D. Ziegler, Peter Pirolli and Rajan Bhattacharyya. Learning to prognostically forage in a neural network model of the interactions between neuromodulators and prefrontal cortex
Neuromodulatory systems and prefrontal cortex are involved in a number of decision-making contexts. In this work, we adapt a recent neural network model that simulates interactions between neuromodulatory and prefrontal areas to the problem of prognostic foraging--that is choosing information to update or form a hypothesis. In the context of an adversarial game, the model assesses a number of decision variables and strategies to choose actions that maximize information utility akin to information foraging theory. The model is also capable of modeling biases in decision making such as ...
Matthew Phillips, Suhas E. Chelian, Peter Pirolli and Rajan Bhattacharya. Forensic foraging of change detection in opponent strategies with a neural model of the interactions between temporal and prefrontal cortex
Detecting change, and the choice of which information to attend to, are key research problems relevant to understanding adaptive behavior. Rational analyses of change detection have been developed in optimal foraging theory (McNamara & Houston, 1987; Stephens, 1987) and psychology (Gallistel et al., 2014). Information foraging models (Pirolli, 2007) have been developed to predict the optimal choice of information, and when to give up collecting further information. Many real-world tasks depend crucially on changing behavior when the world changes. For example, intelligence analysts track ...
Muneo Kitajima and Makoto Toyota. Hierarchical structure of human action selection - An update of Newell's time scale of human action
What we observe as each individual's physical behavior is the results of a multi- ple processing with a PDP system, not with a single unied system. This PDP system is organized evolutionally, and realized as a neural network system, in- cluding the brain, the spinal nerves, and the peripheral nerves. This paper illustrates a matrix representation of the relationships between the hierarchical structure of cognition under Two Minds and the hierarchical structure of the neural network system under PDP.
Muneo Kitajima and Makoto Toyota. Topological Considerations of Memory Structure
The human memory system is an integration of three distributed memory sys- tems associated with respective autonomous organic systems; the perceptual system that takes care of sensory input from the environment, the conscious system that performs deliberate decision making, and the unconscious system that carries out action selections in the environment. The memory system works as a memory component in the comprehensive brain model, MHP/RT , which is capable of simulating human daily behavior considering the real time constraints that should dene strong mutual dependencies among ...
Keitarou Yoshida and Junichi Takeno. An attempt to build a computer model of mental trauma using consciousness modules
A discussion must be begun without delay about problems related to the “mentality of robots.” When attempting to be able to realize in a robot such human-like attributes as advanced cognitive functions, evolutional learning, thought, sense, will, emotion and feelings, and use of experience, then research on brain disease in humans should be considered a very important key area of focus for the healthy development of both humans and robots. The authors of this paper first discuss the well-known mental trauma illnesses of post-traumatic stress disorder (PTSD) and dissociation. While ...
Muneo Kitajima and Makoto Toyota. The role of consciousness in memorization : Asymmetric functioning of consciousness in memory encoding and decoding
One can see each individual’s daily life as a sequence of events, each of which should be associated with his/her conscious and deliberate activities of decision- making, and unconscious and automatic activities of action selection. In contin- uation of the discussion we provided in  concerning event memory creation and utilization on the basis of the architecture model the authors have developed for simulating human beings’ in situ action selection, Model Human Processor with Realtime Constraints (MHP/RT), this paper provides a deeper understanding of the role of consciousness in ...
Harmen De Weerd, Rineke Verbrugge and Bart Verheij. Higher-order theory of mind in Tacit Communication Game
To understand and predict the behaviour of others, people regularly reason about their beliefs, goals, and intentions. People can even use this theory of mind recursively, and form beliefs about the way others reason about the beliefs, goals, and intentions of others. Although the evolutionary origins of this cognitively demanding ability are unknown, the Vygotskian intelligence hypothesis suggests that higher-order theory of mind allows individuals to cooperate more effectively. In this paper, we investigate this hypothesis through the Tacit Communication Game. In this game, two agents ...
Paul Kogut, Jonathan Darvill, David Rosenbluth and David Morgenthaler. Top Down Bottom Up Brain Models
A recent paper by Eliasmith and Trujillo includes an interesting discussion about the role of and approaches for building large-scale brain models. The authors emphasize connecting the model to behavior and compare top down and bottom up approaches to modeling. In this paper we describe our experience developing the FRAMES model with a mixed top down and bottom up brain modeling approach. The FRAMES model focuses on general purpose high level cognitive behavior. The specific goal of the FRAMES model was to study the mechanisms of biases in sensemaking. The FRAMES model was grounded in the ...
Christian Huyck, Giuseppe Primiero and Franco Raimondi. Programming the MIRTO Robot with Neurons
MIRTO is a new, inexpensive, open-source robot. The specification, the necessary libraries, and sample code are freely available. It has been used to teach undergraduate students programming and as an extensible base platform for students engineering robots. While it is typically programmed in traditional programming languages, it has also been driven by simulated neurons; point neurons have been used to follow a line. Since neurons are the basis of animal cognition, using them as the basis of a cognitive architecture is a promising idea. The neural MIRTO line following system can ...
Mukta Gahlawat, Amita Malik and Poonam Bansal. Natural Sounding Speech Synthesizer Using Hybrid Approach for Blinds
When we talk about synthesized speech, the two challenges are faced by the researchers. The first is intelligible and second is naturalness. By intelligible, we mean easily understandable and naturalness means the quality of speech is very near to human speech. Human speech is very dynamic in nature as it consists of lots of variation because of emotional state of mind and not bound to one place while speaking. Same content of speech in different emotional situations are having different prosodic parameters. Even our facial expressions also changes if we speak in different mood like happy, ...
Lee Scheffler. NeurOS(tm) and NeuroBlocks(tm): A Neural/Cognitive Operating System and Building Blocks
NeurOS is an open platform for accelerating research, development and hosting execution of intelligent applications. A NeurOS application is a directed "neural graph" of components connected by signal paths, similar to biological brain connectivity and functional block diagrams of neural pathways. Built-in reusable modules (NeuroBlocks) provide a wide range of general- and special-purpose capabilities: inputs/senses, outputs/effectors, processing, memory, pattern learning and recognition, visualization/instrumentation, custom module development, interfacing to and integrating external ...
Christian Huyck and Ian Mitchell. Building Neuromorphic Embodied Cell Assembly Agents that Learn
Neuromorphic embodied cell assembly agents that learn are one application being developed for the Human Brain Project (HBP). The HBP will build tools, available for all researchers, for building brain simulations. Existing simulated neural agents will be translated to the platforms provided by the HBP; these agents will then run on neuromorphic chips instead of a von Neumann based computer. The initial translation is relatively straight forward but non-trivial software engineering, as any computational system can be programmed from simulated neurons by setting the neural and ...
Catherine Schuman, J. Douglas Birdwell and Mark Dean. Spatiotemporal Classification Using Neuroscience-Inspired Dynamic Architectures
We discuss a neuroscience-inspired dynamic architecture (NIDA) and associated design method based on evolutionary optimization. NIDA networks designed to perform anomaly detection tasks and control tasks have been shown to be successful in previous work. In particular, NIDA networks perform well on tasks that have a temporal component. We present methods for using NIDA networks on classification tasks in which there is no temporal component, in particular, the handwritten digit classification task. The approach we use for both methods produces useful subnetworks that can be combined to ...
Jaehyon Paik, Yunfeng Zhang and Peter Pirolli. Counterfactual Reasoning as a Key for Explaining Adaptive Behavior in a Changing Environment
It is crucial for animals to detect changes in their surrounding environment, and reinforcement learning is one of the well-known processes to explain the change detection behavior. However, reinforcement learning itself cannot fully explain rapid, relatively immediate changes in strategy in response to abrupt environment changes. A previous model employed reinforcement learning and counterfactual reasoning to explain adaptive behavior observed in a changing market simulation environment. In this paper, we used the same model mechanisms to simulate data from two additional tasks that require ...
Eric Bigelow, Daniel Scarafoni, Lenhart Schubert and Alex Wilson. On the Need for Imagistic Modeling in Story Understanding
There is ample evidence that human understanding of ordinary language relies in part on a rich capacity for imagistic mental modeling. We argue that genuine language understanding in machines will similarly require an imagistic modeling capacity enabling fast construction of instances of prototypical physical situations and events, whose participants are drawn from a wide variety of entity types, including animate agents. By allowing fast evaluation of predicates such as `can-see', `under', and `inside', these model instances support coherent text interpretation. Imagistic modeling is thus a ...
Akshay Maan, Dinesh Kumar and Alex James. Memristive Threshold Logic Face Recognition
A new hardware based implementation of face recognition application in real time has been proposed that is inspired by the cortical neuron firing in human brain. This paper presents a face recognition method implemented using reconfigurable network of memristive threshold logic cells that can be practically realised in a secondary plane to the pixel arrays. Among the most distinguishing features of the presented system are a) an early detection and storage of only the relevant information directly from the sensors, b) a parallel, scalable information storage and detection architecture ...
Jonathan Vitale, Mary-Anne Williams and Giuseppe Boccignone. Affective Facial Expression Processing via Simulation: A Probabilistic Model
Understanding the mental state of other people is an important skill for intelligent agents and robots to operate within social environments. However, the mental processes involved in 'mind-reading' are complex. One explanation of such processes is Simulation Theory - it is supported by a large body of neuropsychological research. Yet, determining the best computational model or theory to use in simulation-style emotion detection, is far from being understood. In this work, we use Simulation Theory and neuroscience findings on Mirror-Neuron Systems as the basis for a novel computational ...
Denis Kleyko and Evgeny Osipov. On bidirectional transitions between localist and distributed representations: The case of common substrings search using Vector Symbolic Architecture
The contribution of this article is twofold. First, it presents an encoding approach for seamless bidirectional transitions between localist and distributed representation domains. Second, the approach is demonstrated on the example of using Vector Symbolic Architecture for solving a problem of finding common substrings. The proposed algorithm uses elementary operations on long binary vectors. For the case of two patterns with respective lengths L1 and L2 it requires O(L1 + L2 - 1) operations on binary vectors, which is equal to the suffix trees approach – the fastest algorithm for this ...
Agnese Augello, Ignazio Infantino, Giovanni Pilato, Riccardo Rizzo and Filippo Vella. Creativity evaluation in a cognitive architecture
Evaluation is a key factor of creativity: for this reason it should be integrated into a cognitive architecture of a creative artificial agent. The approach illustrated in this paper uses the Psi model, and describes the framework for introducing internal and external evaluations, and how them influence demands and motivation of the artificial agent. Internal evaluation mechanisms drive the creative process, and influence competence of the creative agent. External evaluation acts through certainty, and requires interaction with human users that express both opinions and some subjective ...
Jalal Karam. Global Threshold and Level Dependent Threshold For Speech Compression
The effect of different compression constraints and schemes presented in a new and flexible paradigm to achieve high compression ratios and acceptable signal to noise ratios of speech signals. Compression parameters are computed for variable frame sizes of different levels Discrete Wavelet Transform (DWT) representation of the signals for different analyzing mother wavelet functions. Results are obtained and compared for Global threshold and level dependent threshold techniques. Comparisons are drawn with four different parameters Signal to Noise Ratios, Peak Signal to Noise Ratios ...
Yasuo Kinouchi. A model of consciousness and attention aimed at speedy autonomous adaptation
An advanced model is proposed that can explain the function in short term adaptation. Associative temporal memory and the function of top-down attention are adopted in the model, in which consciousness and attention are defined as clearly different functions; consciousness is mainly defined as a process of learning as a whole system, and attention is defined as functions that quickly select resources with priority based on the information for learning. These functions work by complementing each other via associative temporal memory. As a result, consciousness and attention are explained as ...
Zihan Xu, Abinash Mohanty, Pai-Yu Chen, Binbin Lin, Deepak Kadetotad, Jieping Ye, Sarma Vrudhula, Shimeng Yu, Jae-Sun Seo and Yu Cao. Parallel programming of resistive cross-point array for synaptic plasticity
This paper proposes a parallel programming scheme for the cross-point array with resistive random access memory (RRAM). Synaptic plasticity in unsupervised learning is realized by tuning the conductance of each RRAM cell. Inspired by the spike-timing-dependent-plasticity (STDP), the programming strength is encoded into the spike firing rate (i.e., pulse frequency) and the overlap time (i.e., duty cycle) of the pre-synaptic node and post-synaptic node, and simultaneously applied to all RRAM cells in the cross-point array. Such an approach achieves parallel programming of the entire RRAM ...
Salah Al-Majeed, Lela Mirtskhulava, Gillian Pearce, Mohamed Al-Mulla Al-Mulla and Julian Wong. Blood Clotting Analysis Based Neural Networks Modeling and Sensors Measurement
A blood clot is a mass of thickened blood. Clotting is a mechanism used by the body to stop bleeding. A blood clot becomes harmful when it blocks an artery or vein and stops blood flow. The World Health Organization reports that 15 million people worldwide suffer stroke, and of these, 5 million die and a further 5 million are left permanently disabled, many severely impaired. Consequently stroke is a major cause of mortality world-wide. Most strokes are caused by a blood clot that occludes an artery in the cerebral circulation. Prediction of patients’ physiological health status is ...
Mark Waser. Bootstrapping a Structured Self-Improving & Safe Autopoietic Self
After nearly sixty years of failing to program artificial intelligence (AI), it is now time to grow it using an enactive approach instead. Critically, however, we need to ensure that it matures with a “moral sense” that will ensure the safety and well-being of the human race. Consciousness and conscience can lead the way towards creating safe and cooperative machine entities.
Waqas Mughal, Bhaskar Choubey and Luiz Carlos Paiva Gouveia. On Threshold Comparing Biomorphic Image Sensors
CMOS image sensors have become the principal image sensors for the vast majority of digital cameras currently in market. The market popular sensor is a typical linear sensor which can capture 3-4 decades of illumination intensity, compared to 6-7 decades captured by the human eye. This has inspired research into biomorphic image sensors for over two decades by various group leading to a number of adaptive pixels, threshold comparing pixel with neurons like firing mechanism as well as logarithmic pixels utilising sub-threshold transistors. However, these have met little commercial success, ...
Michael Miller, Sheldon Linker and Ghassan Azar. The Equilibration of Neural Propositions
Jean Piaget explained equilibration as the process by which individuals create cognitive structures to compensate for disturbances encountered during their interactions with the world. When an individual has a failed prediction, an unsatisfied urge, or a failed plan, equilibrative processes compensate for these disturbances by building new structures that circumvent, incorporate, or transform the disturbance. Using Neural Propositions as its basic unit of representation, the Piagetian Autonomous Modeler (PAM.P2) is a cognitive system that employs equilibration to refine its world model as it ...
Romain Cazé. Threshold gates should include one, two, or three dendritic subunits.
Threshold gates, which integrate inputs linearly, are still the golden standard to model neurons in artificial neural networks. But neurons do not behave that way; rather, they possess dendrites that allow them to non-linearly integrate their inputs within dendritic subunits. Non-linear dendrites increase the computational capacity of a single neuron. They potentially enable a neuron to compute all possible classifications if it possesses a sufficient number of dendritic subunits. In practise, this observation is insufficient because a neuron has a finite number of subunits; it will ...
Balaji Balasubramaniam. Investigating Cognitive intelligence in Text processing
Human understanding is a fascinating factor and its investigation in natural language communications is being discussed here. Having enormous knowledge shared in the form of text data nowadays, the importance of textual processing has never been so demanding. Especially, in social and professional environments such as social networking like twitter or Facebook, academia while publishing research contributions or proposals, in industry while documentation or framing policies, etc. The paradigm is overwhelming and the necessity for inferring these data automatically is one of the challenging ...
Ben Goertzel. Characterizing Human-Like Consciousness: An Integrative Approach
Synthesizing concepts and findings from a number of recent models of human consciousness, a unified model of the key properties characterizing human consciousness is outlined. Six key properties are emphasized: Dynamical representation of the focus of consciousness, Focusing of energetic resources and focusing of informational resources on a subset of system knowledge, Global Workspace dynamics as outlined by Bernard Baars in his cognitive theory of consciousness, Integrated Information as emphasized by Tononi, and correlation of attentional focus with self-modeling. It is proposed that the ...
Ben Goertzel, David Hanson and Gino Yu. A Software Architecture for Generally Intelligent Humanoid Robotics
This paper summarizes the authors' thinking regarding the design of a software framework for interfacing between general-intelligence-oriented software systems and complex mobile robots, including humanoid robots. The framework describes incorporates perception synthesis, action orchestration, and high level control, and is designed to effectively leverage existing relevant software frameworks such as ROS and Blender. An initial case study motivating this work is the use of the OpenCog AGI (Artificial General Intelligence) software framework to help control humanoid robots created by ...
Ben Goertzel, Cosmo Harrigan, Matthew Ikle and Gino Yu. Guiding Probabilistic Logical Inference with Nonlinear Dynamical Attention Allocation
A series of experiments combining probabilistic logical inference with an artificial economics based attention allocation mechanism is demonstrated , as an exemplification of an integrative AGI design (OpenCog/CogPrime) utilizing ``cognitive synergy'' among various cognitive algorithms as a strategy for overcoming the challenge of combinatorial explosion. In the system under study, uncertain knowledge is represented in structured form as a weighted labelled hypergraph capable of representing arbitrarily complex expressions, and is manipulated via the PLN probabilistic logic engine. An ...
L. Andrew Coward. Brain computational primitives
The brain uses computational primitives that are analogous with but qualitatively different from the computational primitives used in electronic computer systems. The primary computational primitives of the brain are described, and their implementation in anatomy and physiology discussed. Combinations and sequences of these primitives implement cognitive tasks. Many of the primitives have also been implemented electronically. The brain is a very effective general learning system, and although an artificial general intelligence system will be required to learn a different range of behaviours ...
Lee Scheffler. NeurOS and NeuroBlocks: Rapid cognitive software development
Here is a short description suitable for inclusion in conference program materials: "Demonstration will show interactive building and testing of cognitive applications using NeurOS and NeuroBlocks, and demonstrations of some already-built applications. This is a companion to the conference presentation on NeurOS and NeuroBlocks." --------------------------------------- Here is the original technology demonstration proposal: Depending on available facilities (table(s), chair(s}, projector/monitor, screen/wall, ambient noise, lighting, space, etc.) I may demonstrate items including the ...
Rony Novianto and Mary-Anne Williams. Operant Conditioning in ASMO Cognitive Architecture
An environment provides feedback for agents to correct their behaviours through operant conditioning learning. Models of operant conditioning mainly involve learning over states that can be difficult to determine. In this paper, we describe a novel operant conditioning mechanism that is not based on states, implemented in ASMO cognitive architecture, implemented in a physical agent and that models the influences to attention. The capability of this mechanism is demonstrated in the `Smokey robot companion' experiment. Results show that Smokey can learn its behaviours based on the feedback ...
Alireza Goudarzi and Darko Stefanovic. Towards a Calculus of Echo State Networks
Reservoir computing is a recent trend in neural networks which uses the dynamical perturbations on the phase space of a system to compute a desired target function. We present how one can formulate an expectation of system performance in a simple class of reservoir computing called echo state networks. In contrast with previous theoretical frameworks, which only reveal an upper bound on the total memory in the system, we analytically calculate the entire memory curve as a function of the structure of the system and the properties of the input and the target function. We demonstrate the ...
Christopher Rozell, Mengchen Zhu, Adam Charles, Han Lun Yap and Marissa Norko. The role of sparsity in visual perception
A critical component of cognitive architectures is the processing of sensory data to form percepts about the environment. Despite decades of research in biological vision, it is still unclear what fundamental information representation is used by these systems. Efficient coding models have have long proposed that optimal representations should reduce the redundancy in sensory signals to represent the essential information content. One concrete example of this is sparse coding, where signals are encoded using as few elements from a dictionary as possible. More generally, the signal ...
Olga D. Chernavskaya. An architecture of the cognitive system: The role of emotions and the sense of humor
In the present work, it is shown that the human emotions, being a subjective estimate of the current/future problems, and, simultaneously, the product of the neural intermediators, are inherently embedded into our architecture. Our concept is that the emotions could be simulated as the dynamical variations of the noise amplitude, and these variations control the activity of two subsystems. Accounting for the neural physiology, this model requires including an additional variable which corresponds to the level of global intermediators. The system of coupled equations for the noise amplitude and intermediator variable is proposed, which provides in particular the description of the stress effect
Piotr Boltuc. Advanced Self-Conscious Robots
Turing was right that soon there would be no functionalities that robots would be unable to perform. The Turing Test was met for closed domains and recently for a teenager; advanced programs, such as LIDA are just the beginning of advanced robot-to-human interphases. We even have discovery machines and new cognitive architectures abound. This is all fascinating but is there a limit? Philosophers take first-person awareness more and more seriously (Chalmers, Block, Nagel), so do some psychologists. Can a robot have non-reductive first person awareness that is not just functional. Can we ...
Vishwanathan Mohan, Ajaz Ahmad Bhat, Giulio Sandini and Pietro Morasso. From Object-Action to Property-Action: Learning causally dominant properties through cumulative explorative interactions
Emerging studies from neuroscience in relation to organization of sematic memory in the brain provide converging evidence suggesting that conceptual knowledge is organized in a distributed fashion in “property specific” cortical networks that directly support perception and action (and were active during learning). Though learning ‘object-action’ affordances and using such knowledge for prediction and planning is an active topic in cognitive robotics, this article urges to go beyond and look at “property-action” networks instead. To this effect, a brain guided framework for ...
Peter Ford Dominey. Reservior Computing in Higher Cognitive Functions
One of the origins of reservoir computing can be found in the primate fronto-striatal system. Local recurrent connections dominate cortico-cortical connections and thus implement the reservoir. The projection from cortex to the striatum is modifiable under the control of reward-related dopamine, thus forming the basis for the readout. We first specified this architecture in 1995 (J. Cog. Neuroscience) in the context of attempting to interpret prefrontal cortical neurons recorded in the behaving primate during a sensorimotor sequencing task. For the first time, a reservoir network ...
Usef Faghihi. A Cognitive Model Fleshes Out Kahneman’s Fast and Slow Systems
Daniel Kahneman  posits two main processes that characterize thinking: “System 1” is a fast decision making system responsible for intuitive decision making based on emotions, vivid imagery, and associative memory. “System 2” is a slow system that observes System 1’s outputs, and intervenes when “intuition” is insufficient. Such an intervention occurs “when an event is detected that violates the model of the world that System 1 maintains” . Here, we propose specific underlying mechanisms for Kahneman’s Systems 1 and 2, in terms of the LIDA model, a broad, ...
Usef Faghihi. How Gamification Techniques Applies for educational Purpose specially with College Algebra
Gamification is the “use of game design elements in non-game contexts" (Deterding et al, 2011, p.1). Gaming environments have been used to teach mathematics topics such as addition and division in a fun manner . However, when it comes to college level mathematical concepts such as the use of the quadratic formula, there are very few software that explain these concepts in a fun way. In this paper, we present a first step towards using video game ele-ments and Artificial Intelligence Tutoring system techniques to teach mathe-matical concepts such as factoring and the quadratic formula. ...
Ulysses Bernardet and Steve Dipaola. Affective response patterns as indicators of personality in virtual characters
Inter-individual differences in cognition, emotion, and behaviour are pervasive mediators of social interaction. Higgins and Scholer (2008) suggest that personality is revealed through motivated preferences and biases in the way people interact with their environment. We investigate personality attribution using a paradigm where a human participant is tasked with assessing the personality of an autonomous virtual characters (AVC) responding to affective stimuli. We evoke different impressions of personality by varying the characteristics of the mapping between affective quality of the ...
Vladislav Veksler, Kevin Gluck, Christopher Myers, Jack Harris and Thomas Mielke. Alleviating the curse of dimensionality -- A psychologically-inspired approach.
Various combinations of perceptual features are relevant for learning and action-selection. However, the storage of all possible feature combinations presents computationally impractical, and psychologically implausible, memory requirements in non-trivial environments due to a state-space explosion. Some psychological models suggest that feature combinations, or chunks, should be generated at a conservative rate. Other models suggest that chunk retrieval is based on statistical regularities in the environment, i.e. recency and frequency. We present a computational model for chunk learning ...
Matthias Ziegler, Suhas Chelian, James Benvenuto, Jeffrey Krichmar, Randall O'Reilly and Rajan Bhattacharyya. A model of proactive and reactive cognitive control with anterior cingulate cortex and neuromodulatory systems
Proactive and reactive cognitive control is often associated with anterior cingulate cortex (ACC). How ACC affects processing in other brain areas, however, is often not explicitly delineated. In this work, we describe a model of how ACC computes measures of conflict and surprise that are in turn relayed to the basal forebrain (BF) and locus coeruleus (LC) in that order. BF and LC signals then respectively sharpen posterior cortical processing and trigger the reframing of prefrontal cortical decision-making frames. We implemented this theory in a large-scale neurocognitive model that ...
Javier Snaider and Stan Franklin. Vector LIDA
The representation paradigm used by a cognitive architecture helps to determine the kind of processes that it can perform more efficiently. Vector LIDA is a variation of the LIDA cognitive architecture that employs high-dimensional Modular Composite Representation (MCR) vectors as its main representation model and Integer Sparse Distributed Memory as its main memory implementation technology. The advantages of this new model include a more realistic and biologically plausible model, better integration with its episodic memory, better integration with other low level perceptual processing ...
Steve Dipaola. Computer Modelling Fine Art Painting using a Cognitive Correlative Heuristics Approach
We begin to model cognitive and perceptual mechanisms or ‘cognitive correlates’ which correspond and relate to artists’ techniques and conceptions regarding fine art painting in general and portraiture in particular. As our starting point, we analyze an extensive corpus of art-theory literature to identify broadly accepted understandings and techniques, which might be relevant to human perception and cognition. We further condense this passed down artistic knowledge into a concise set of heuristics, which are suitable for parameterization and algorithmic implementation, and then ...
Maryam Saberi, Ulysses Bernardet and Steve Dipaola. An Architecture for Personality-Based, Nonverbal Behavior in Affective Virtual Humanoid Character
As humans we perceive other humans as individually different based – amongst other things – on a consistent pattern of affect, cognition, and behavior. Here we propose a biologically and psychologically grounded cognitive architecture for the control of nonverbal behavior of a virtual humanoid character during dynamic interactions with human users. Key aspects of the internal states and overt behavior of the virtual character are modulated by high-level personality parameters derived from the scientific literature. The virtual character should behave naturally and consistently while ...
Steve Dipaola. Using a Contextual Focus Model for an Automatic Creativity Algorithm to Generate Art Work
We sought to determine whether incorporating cognitive based contextual focus into a genetic programming fitness function would play a crucial role in enabling the computer system to generate art that humans find "creative" (i.e. possessing qualities of novelty and aesthetic value typically ascribed to the output of a creative artistic process).We implemented contextual focus in the evolutionary art algorithm by giving the program the capacity to vary its level of fluidity and control over different phases of the creative process in response to the output it generated creative domain of ...
Tarek Richard Besold. Formal Foundations for Cognitive AI Systems and the Need for Structure-Based Approximation
The recognition that human minds/brains are finite systems with limited resources for computation has led researchers in cognitive science to propose that human cognitive capacities are constrained by computational tractability. Transferring and adapting this thesis to a human-level AI context may give rise to insights that can help in progressing towards meeting the goals of (re)creating intelligence and capacities inspired by the human mind. In this talk I report on work towards the development of a framework for the application of formal methods of analysis to cognitive systems and models ...
Blerim Emruli, Fredrik Sandin and Jerker Delsing. Towards a vector space architecture for emergent interoperability of ubiquitous systems
Internet-connected embedded systems are expected to vastly outnumber people on the planet in the near future, leading to new challenges in software engineering and automation in application domains involving complex and evolving systems. Several decades of artificial intelligence research suggests that present approaches to making such systems automatically interoperable using handcrafted "semantic" descriptions of services and information are difficult to apply. We outline a bio-inspired learning approach to creating interoperable systems, which does not require handcrafted semantic ...
Lola Cañamero. Cognitive Architectures to Bridge Interdisciplinary Gaps in Emotion Research
Emotions are a fundamental aspect of cognition and interaction, and their importance has been broadly acknowledged by both the ``sciences of the natural" (e.g., neuroscience, psychology, biology) and those of ``the artificial" (e.g., artificial intelligence, cognitive science / robotics, artificial life). Emotions provide an ideal framework for inter- and cross-disciplinary research since, due to their complex multi-faceted nature, they cannot be properly understood from the perspective of a single discipline. In this abstract/presentation, I would argue that the use of robots as both ...
Matthew Lewis and Lola Cañamero. Modulating Perception with Pleasure for Action Selection
Persistence and opportunism are two key features of cognitive action selection architectures. For an autonomous robot that has to satisfy multiple conflicting survival-related needs, it is crucial to persist in the execution of behaviors for long enough to get sufficient benefit. Opportunism concerns the initiation of actions, and occurs when an agent chooses to consume a resource that might not satisfy its most pressing need, but which is available now and might not be available later. The degree to which a robot should show persistence and opportunism depends on multiple factors; generally ...
Tarek Richard Besold. What neural-symbolic networks can(not) do in 2014
In this introductory talk, a short review of the history and an assessment of the status quo of the development of neural-symbolic networks (with an emphasis on their applications and relations to the modeling of cognitive capacities) is given, before the focus is shifted towards identifying some important but (as of 2014) unsolved challenges as possible basis for future work.
Joscha Bach. Computationalism and Cognitive Science
Cognitive architectures rest on the notion of treating the mind as a machine, or more precisely: a computational system. The implications of this fact are rarely reflected upon in psychology, and give rise to much controversy in within philosophy of mind. Here, I will defend the computationalist agenda, and discuss different degrees of computationalist commitments: ontological computationalism, epistemological computationalism, computationalism of mental events and weak computationalism, as well as possibly necessary relaxations of current notions of computation in the context of cognitive ...
Manuel Caro, Darsana Josyula, Michael Cox and Jovani Alberto Jiménez Builes. MISM: a metamodel of computational metacognition
Computational metacognition is a technical area of artificial intelligence whose aim is to increase the degree of autonomy and awareness an intelligent system has about its own reasoning and learning. In the literature, different models of metacognition are applied to artificial intelligent systems. However many of the-se models have a narrow focus, because they do not address comprehensively the elements of metacognition. This paper presents an analysis of metacognitive models discussed in the literature in order to discover the common (invariants) and varying (variants) elements. The main ...
Michael Brady. A Neural Field Model for Babble-Feedback Learning in Speech
A theory of motor-feedback learning based on a bidirectional-edge graph is summarized. Each vertex of the graph corresponds to a dynamic neural field and each edge of the graph maps to a pair of reciprocal adaptive filters between neural fields. Output from the network drives an articulatory speech synthesizer. The network is first exposed to an ongoing stream of speech sounds where neural fields tune themselves to respond to repetitive patterns in the input (phonetic features). The network is then made to actuate motors with babble feedback learning in an effort to train the synthesizer to ...
Frank Ritter and Christopher Dancy. An update on tying cognitive architectures to physiological architectures
In this report we will give an update on tying the ACT-R cognitive architecture to the HumMod physiological architecture. We will describe the combined architecture and a model of how an audio interruption leads to a startle response while performing a serial (repeated) subtraction task. The talk will describe the physiology effects, the physiology-cognition pathways, and the cognitive effects of the startle and repetitions of it on the physiology and the behavior. If time allows, we will discuss other ways of realizing these effects using other architectures and approaches, and how we ...
Dustin Dannnenhauer, Michael Cox, Shubham Gupta, Matthew Paisner and Don Perlis. Toward Meta-Level Control of Autonomous Agents
Metareasoning is an important capability for autonomous systems, particularly for those being deployed on long duration missions. An agent with increased self-observation and the ability to control itself in response to changing environments will be more capable in achieving its goals. This is essential for long-duration missions where system designers will not be able to, theoretically or practically, predict all possible problems that the agent may encounter. In this paper we describe preliminary work that integrates the metacognitive architecture MIDCA with an autonomous TREX agent, ...
Norifumi Watanabe, Fumihiko Mori and Takashi Omori. Walk Assistance Interface by Sensory Superposition Fusion of Vison and Somatosensor
We have mobile adaptive information terminal to navigate like to present information during walking. And it is important to behavior support by by measuring human behavior in real time and feedback information intuitive. Human walking is affected by vision, vestibular, somatic and other various sensations that come through the sensory-motor loop. But detail of the sensory-motor loop is not clear. In this research, we examined a possible affect of self motion sensation by an optic flow stimulus in peripheral vision with a decayed somato-sensory feeling by a vibration stimulus on leg and foot ...
James Eilbert. The Vertebrate Strategy for Brain Evolution
In the study of Biologically Inspired Cognitive Architectures (BICA), relatively little attention has been paid to the evolutionary development of cognitive capabilities. There are two successful evolutionary strategies, i.e. resilient and cheap to reproduce individuals vs. more capable individuals that are necessarily expensive to reproduce. The vertebrate family is the prime example of the “evolve more capable individual” strategy. A sequence of increasingly complex and cognitively capable individuals has resulted from the vertebrate strategy. At the heart of the vertebrate strategy ...
Olivier Georgeon and Amélie Cordier. Inverting the Interaction Cycle to Model Embodied Agents
The interaction between an agent and an environment is traditionally represented as a cycle in which the agent alternatively receives input data from the environment and sends output data to the environment. In most models, this cycle begins with the agent receiving input data representing the environment, and ends with the agent sending output data to the environment. This paper suggests doing the opposite: the agent initiates the cycle by sending output data that specify an experiment to perform in the environment, and, in return, receives input data that represents the result of this ...
Dhireesha Kudithipudi and Cory Merkel. On the Role of Cortico-thalamic Feedback in Neuromemristive Vision Systems
he human thalamus is often regarded as the gateway to the mind. Uniquely positioned between the cerebrum and brain stem, the thalamus forwards afferent signals from sensory systems to the brain's cortical areas. However, recent studies suggest that the thalamus also plays an active role in processing sensory data, with compelling evidence given by the large degree of cortico-thalamic feedback projections. The present study proposes a neuromemristive system (NMS) for vision-related applications which leverages the computational principles of the thalamo-cortico-thalamic topology. Inspired ...
Joel Parthemore. Navigating between concepts and consciousness
Computer models of mind are one way of exploring and coming to a better understanding of human consciousness by creating an external “picture” of what we take our consciousness to be. Because human beings are _embodied_ in a complex physical form and _embedded_ in an environment with which they richly interact – if not, as the enactivists would have it, ultimately continuous with – it marks a natural step to extend that project to robotics. Although the goal should be clearly focused on modelling and not creation of consciousness, still there is no reason, in principle, why ...
David Aha. Goal Reasoning: Progress and Needs
Goal reasoning (GR) is the capability of an agent to autonomously and dynamically assume the responsibility of identifying and pursuing its goals/objectives. It is a required competency of fully autonomous intelligent agents. GR is not a new topic; it has been addressed by several researchers, under a variety of names, in the context of cognitive architectures, robotics, planning, and game AI, for example. However, it has usually been studied in the context of other commitments (e.g., to a specific cognitive architecture's models and other reasoning methods). More recently, a few groups, ...
Dhireesha Kudithipudi, Cory Merkel, Qutaiba Saleh and Colin Donahue. Memristive Reservoir Computing Architecture for Epileptic Seizure Detection
Echo state networks (ESN) or reservoirs, are random, recurrent neural network topologies that integrate temporal data over short time windows by operating on the edge of chaos. Recently, there is a significant effort on the mathematical modeling and software topologies of the reservoirs. However, hardware reservoir fabrics are essential to deploy in energy constrained environments. In this paper, we investigate a hardware reservoir with bi-stable memristive synapses. In particular, we demonstrate a scalable hardware model for detecting real-time epileptic seizures in human models . The ...
Craig Vineyard, Stephen Verzi and James Aimone. Quantication of Neural Computation
Neural computing has many innate advantages over conventional computing such as parallelism, distributed representations, extensive connectivity, dynamic functionality, and low power consumption. Effectively, neural inspired algorithms and architectures serve as exciting approaches to problems such as pattern recognition, speech processing, and computer vision. However, as an alternative computing paradigm it does not make sense to evaluate neural computing algorithms and architectures using the same benchmarks as standard computers. Neurons do not perform floating point operations, nor do ...
Stevo Bozinovski. Modeling Mechanisms of Cognition-Emotion Interaction in Artificial Neural Networks, since 1981
The paper describes modeling of cognition-emotion interaction implemented in a neural network named Crossbar Adaptive Array in 1981. The architecture was proposed to meet two challenges: solving the delayed reinforcement learning problem for neural networks, and building a self-learning system (no advice and no reinforcement from environment) based on a neural network. The architecture introduced computation of feelings, and their interaction with learning and decision making mechanisms. It also introduced genetic environment as provider of initial emotions to the neural network. ...
Robert Laddaga. Active Perception
The talk presents a neuroscience inspired theory about perception that highlights multiple uses of feedback, including feedback from results of action to perceptual interpretation, and feedback from interpretation to low level control of sensors and signal processing. It further discusses the application of this theory to cyber defense, and Intrusion Detection and Response specifically. The talk also discusses Active Perception as a foundation theory for a cognitive architecture.
Daniel Hammerstrom. The Future of Embedded Computing: A DARPA Perspective
Today’s Defense missions rely on massive amounts of sensor data collected by intelligence, surveillance and reconnaissance (ISR) platforms. Not only has the volume of sensor data increased exponentially, there has also been a dramatic increase in the complexity of analysis required for applications such as target identification and tracking. The digital processors used for ISR data analysis are limited by power requirements, potentially limiting the speed and type of data analysis that can be done. Furthermore, as Moore’s Law slows down, power scaling has more or less stopped. The ...
Kamilya Smagulova and Aigerim Sametkhanova. Threshold Logic Object Recognizer
Threshold logic is an alternative approach to designing binary systems which are capable of implementing complex pattern matching problems. Scalability of threshold logic to main stream computing is a challenging problem. In this research, we present a pattern matching network using binary threshold logic networks consisting of CMOS devices and memories standard technologies to implement a word matching processor. The matching stage of any pattern matcher includes a similarity measure that has the capability to compare features and reduce the feature level score to a global similarity ...
Klaus Raizer and Ricardo Gudwin. A Neuroscience Inspired Gated Learning Action Selection Mechanism
This paper presents an algorithm for action selection, in the context of intelligent agents, capable of learning from rewards which are sparse in time. Inspiration for the proposed algorithm was drawn from computational neuroscience models of how the human prefrontal cortex (PFC) works. We have observed that this abstraction provides some advantages, such as the representation of solutions as trees, making it human-readable, and turning the learning process into a combinatorial optimization problem. Results for it solving the 1-2-AX working memory task are presented and discussed. We also ...
Allen King. The HaveNWant Schemata -- Nodes and Links for Representing Experiences Reactively
This Schemata generates a reactive semantic network with incremental learning abilities. It captures coincidences experienced by adding new nodes that then affect future reactions. Built into each node are algorithms which collectively allow the network to detect repeated coincidences, recognize equivalent states to form a compact state space, and then record state transitions to form a predictive model. The cognitive functions of several networks with dozens or hundreds of elements will be demonstrated. A short introductory video is at https://www.youtube.com/watch?v=0anyyX0SrTA.
Alexei V. Samsonovich and Paul Robertson. Preface: A Forum at the Dawn of the Era of Biologically Inspired Intelligent Machines
The emergence of biologically inspired cognitive architectures (BICA) challenges researchers across many disciplines with a new frontier: computational replication of the human mind, taken in all its essential aspects, as a functional unit in a team or a society. The mission of the international conference series on BICA is to facilitate interaction and collaboration among researchers who devoted themselves to solution of this BICA Challenge, by bridging cross-disciplinary and cross-cultural barriers. BICA annual conference series are now seven years old. Initially organized under the auspices of AAAI Fall Symposia, the conference grew up into a world-wide forum coordinated and organized by the BICA Society ...