2015 3rd International Conference on Applied Computing and Information Technology/2nd International Conference

on Computational Science and Intelligence

Cognitive Computing:
A Brief Survey and Open Research Challenges
J. Octavio Gutierrez-Garcia Emmanuel López-Neri
Department of Computer Science Centro de Investigación, Innovación
Instituto Tecnológico Autónomo de México y Desarrollo Tecnológico
Mexico, DF 01080 Universidad del Valle de México
[email protected] Guadalajara, Jalisco 45601
[email protected]

Abstract—Cognitive computing is a multidisciplinary field of Szymanski and Hise [52] indicate that cognitive computing
research aiming at devising computational models and decision- is “an emerging field of inquiry that draws on principles from
making mechanisms based on the neurobiological processes of the behavioral, cognitive, computer, and related sciences”.
the brain, cognitive sciences, and psychology. The objective of
cognitive computational models is to endow computer systems Wang defines cognitive computing in terms of cognitive
with the faculties of knowing, thinking, and feeling. The major informatics, which is a multidisciplinary field that applies how
contributions of this survey include (i) giving insights into the brain processes information and copes with decision
cognitive computing by listing and describing its definitions, making to information sciences [63]. Then, Wang [61] defines
related fields, and terms; (ii) classifying current research on cognitive computing as “an emerging paradigm of intelligent
cognitive computing according to its objectives; (iii) presenting a computing methodologies and systems based on cognitive
concise review of cognitive computing approaches; and (iv) informatics that implements computational intelligence by
identifying the open research issues in the area of cognitive autonomous inferences and perceptions mimicking the
computing. mechanisms of the brain”.
Keywords—cognitive computing; computational intelligence; Modha et al. [38] describe cognitive computing by stating
artificial intelligence its purpose, which is “to develop a coherent, unified, universal
mechanism inspired by the mind’s capabilities”.
I. INTRODUCTION Nahamoo [41] defines cognitive computing as “a
Cognition is a process of mind in charge of creating fundamentally new computing paradigm for tackling real
knowledge [57] and understanding by obtaining an abstract world problems, exploiting enormous amounts of information
representation of a world [47]. Nevertheless, it should be noted using massively parallel machines that interact with humans
that there is no commonly accepted definition of cognition or and other cognitive systems.” In the same context, Evans et al.
of the mind itself [38] despite the extensive research into [23] state that the focus of cognitive computing is on
human cognition (see [3, 8, 51]). Hence, cognitive computing representing and processing information.
or (cognitive computation [55] as originally coined by Valiant Other researchers have opted to outline cognitive
in 1995 [68]) has been defined differently by several computing by stating its main principles, like Clark [9], who
researchers across time and contexts. states that in cognitive computing “there exist suitable ways to
Valiant [55] defines cognitive computing as “a discipline abstract detailed behavior, and to talk about goals, plans,
that links together neurobiology, cognitive psychology and constraints and methods at a high level”.
artificial intelligence”. The major contributions of this survey include (i) giving
Brasil et al. [4] state that cognitive computing is “a insights into cognitive computing by listing and describing its
collection of emerging technologies inspired by the biological definitions, related fields, and terms (Sections I-II); (ii)
processing of information in the nervous system, human classifying current research on cognitive computing according
reasoning, decision making, and natural selection”. However, to its objectives (Section III); (iii) presenting a concise review
it is noteworthy that natural selection, as defined by Hancock of cognitive computing approaches (Sections III.A-C); and (iv)
[29], is the process in which desirable inheritable traits of a identifying the open research issues in the area of cognitive
population predominate over undesirable traits across time, computing (Section IV). Finally, Section V includes some
which is very distant from the definition of cognition as a concluding remarks.
process of mind in charge of creating knowledge [57].
Preissl et al. [46] describe cognitive computing as “the II. COGNITIVE COMPUTING AND RELATED CONCEPTS
quest for approximating the mind-like function, low power, In order to gain a better understanding of cognitive
small volume, and real-time performance of the human brain”. computing, defining its related fields and terms is required.
Based on the different definitions of cognitive computing [4,

978-1-4673-9642-4/15 $31.00 © 2015 IEEE 328

DOI 10.1109/ACIT-CSI.2015.64
38, 41, 52, 55, 61, 63] (see Section I), related concepts to
cognitive computing are as follows: knowledge, neurobiology,
cognitive psychology, artificial intelligence, behavioral
sciences, cognitive sciences, computer sciences, biological
processing, nervous system, reasoning, decision making,
cognitive informatics, human brain and mind, and exploitation
of enormous amounts of information, i.e., big data [12].
According to Thornton [53], artificial intelligence and
cognitive psychology are members of the cognitive sciences.
Artificial intelligence is concerned with reasoning and its
underlying processes of thought in order to build intelligent
systems that (i) act and/or think like humans and (ii) think
and/or act rationally [49]. Reasoning is a series of steps that
allows obtaining conclusions (i.e., knowledge) from some
initial premises [58]. Knowledge can be defined as the
understanding of a subject obtained through study and/or
experience [6]. Big data focuses on gaining knowledge from
enormous volumes of data [48], which can be used to support
decision-making. Decision-making is defined as the process of
evaluating and selecting, based on a given criteria, the best
alternative from a set of two or more possible choices [44] in
order to reach a goal.
Cognitive psychology deals with how the brain perceives
and interprets external impressions [39]. Neurobiology studies
how connections among neurons influence the behavior of an
individual [33]. The nervous system, composed of neurons,
reacts to physical stimuli (e.g., external impressions) and
exchanges electrochemical signals among neurons to
coordinate body’s (re)actions. A biological process is a
chemical process in which a series of molecular-level events
[50] interact among each other to attain a biological goal (e.g., Fig. 1. A coarse taxonomy of cognitive computing research areas.
metabolism) [14].
According to [18, 27], the behavioral sciences include
cognitive computing models can be supported by big data; for
biology, sociology, anthropology, and psychology. Biology (as
instance, see the automatic knowledge extraction mechanism
the scientific study of body and cells) is highly and inherently
[24] of IBM Watson [25].
related to cognitive computing. In a similar manner, sociology,
anthropology, and psychology hold a high-level relationship
with cognitive computing due to their focus on the study of III. CURRENT RESEARCH ON COGNITIVE COMPUTING
human behavior either as an individual or as a society. In this survey, we categorize current research on cognitive
It is noteworthy that both brain and mind are commonly computing according to its objectives into those providing a
included in definitions of cognitive computing (see [61] and computer system with the faculties of (i) knowing, (ii) thinking
[38], respectively) and also commonly used interchangeably or (iii) feeling. See Fig. 1 for a coarse taxonomy of cognitive
(see [40, 42, 56]). However, mind and brain are different. Mind computing research areas, which are analyzed in Sections
refers to a set of activities that enable a human being to feel, III.A-C.
think, and know, whereas the brain is an organ that endows
human beings with the faculties of feeling, thinking, and A. Endowing Computer Systems with the Faculty of Knowing
knowing [31]. Hence, the focus of cognitive computing should Provided that the result of cognition is knowledge [57],
be on mimicking the mechanisms of the brain to endow several research efforts have been conducted to formally define
computer systems with the faculties of feeling, thinking and knowledge from a cognitive computing’s perspective.
knowing.
Tian et al. [54] formalize knowledge by making use of
In conclusion, cognitive computing is a multidisciplinary concept algebra [64] and real-time process algebra [67].
field of research aiming at devising computational models and Concept algebra is a mathematical framework supported by the
decision-making mechanisms based on the neurobiological object-attribute-relation theory for knowledge manipulation
processes of the brain, cognitive sciences, and psychology to [64]. Abstract concepts can be defined, related to each other,
endow computer systems with the faculties of knowing, compared and composed in order to create a network of
thinking, and feeling. In addition, it is noteworthy that concepts that formalizes knowledge. Real-time process algebra

329
is a mathematical notation system capable of describing system Computer machinery to enable thinking. Hardware has
architectures as well as the dynamic and static behaviors of been designed and implemented in order to attempt to recreate
their components [67]. Tian et al. [54] combine concept the human brain.
algebra with real-time process algebra to devise a knowledge
Imam et al. [32] and Arthur et al. [2] propose an event-
representation system capable of acquiring and manipulating
driven digital neurosynaptic core, which is composed of 256
knowledge.
neurons, 1,024 axons and 262,144 synapses (i.e., the junction
Similar to Tian et al. [54], ElBedwehy et al. [21] also make between two given neurons). There are three types of synapses:
use of real-time process algebra [67] to design a cognitive strong excitatory, weak excitatory and inhibitory. Imam et al.
semantic model of knowledge consisting of four stages. In the argue that the neuromorphic chip (i) is relatively compact
first stage, a prototype of knowledge is created. The second occupying only 4.2mm2 of silicon and (ii) requires low energy
stage consists of discriminating between relevant and irrelevant consumption, namely, 45 pJ per spike [37]. Arthur et al.
knowledge. In the third stage, knowledge is generalized, and empirically demonstrated that the neurosynaptic core is highly
finally, the fourth stage consists of developing an algorithm configurable (due to its large number of potential connections
composed of a set of symbolic rules derived from the previous among neurons) by using four test cases: robot navigation,
stages. As in [21], Xiao and Lan [69] also emphasize the need implementation of a virtual player of pong, digit recognition,
for knowledge reduction, i.e., obtaining a sufficient number of and an associative memory. In addition, Esser et al. [22]
attributes to describe an object or concept in order to simplify designed and implemented many other algorithms that take full
the cognition process. advantage of networks of neurosynaptic cores. Among these
new algorithms [22] are music recognition and eye detection.
Peña-Ayala and Mizoguchi [45] model knowledge by using
causal reasoning [30] and fuzzy logic [71]. Concepts and their In the same vein as [2, 32, 37], Preissl et al. [46] propose
relationships are formalized by means of ontologies. Then, Compass, a multi-threaded engine that simulates TrueNorth
causal relationships (expressed as fuzzy rules) are defined for [36], a brain-inspired digital chip with 4,096 neurosynaptic
the concepts, which are integrated into a cognitive map, i.e., a cores [2, 32, 37]. Compass is a massively scalable simulator
model of cognition [19]. The cognitive map represents the final capable of attaining near-optimal scaling performance. The
model of knowledge. cognitive architecture of Compass is composed of a number of
neurons comparable to that of the human brain. In addition,
Wang [60, 62] proposes inference algebra, a denotational Compass is provided with novel PGAS [70] communication
mathematics to formally define causation and different types of primitives that allow for efficient communication between
inferences, which are as follows: conditional, causal, neurons.
numerical, event-driven, time-driven, logical, and fuzzy
inferences. In addition, Wang also proposes a set of operators Software frameworks to enable thinking. Advances in
to manipulate inferences, which are used to represent cognitive software to enable thinking have focused on either:
knowledge.
1) Taking advantage of brain-like computer machinery, or
IBM Watson [25] makes use of semantic technologies,
which led IBM Watson to outperform top-ranked human 2) Determining causal relationships among concepts of a
players at Jeopardy, a question and answer TV show [28]. The given domain knowledge base.
semantic technologies of IBM Watson consists of more than To take advantage of the brain-like computer machinery
100 natural language processing techniques [26, 28], which are proposed in [2, 32, 37], Amir et al. [1] designed a
mostly probabilistic reasoning approaches [26]. To represent programming language for TrueNorth [36]. Amir et al. argue
knowledge, IBM Watson uses both semi-structured (e.g., that hardware architectures mimicking the human brain
taxonomies) and structured approaches (e.g., ontologies). composed of a set of interconnected neurosynaptic cores
However, in order to answer open-domain questions, IBM require a different programming approach to efficiently
Watson is endowed with an automatic knowledge extraction execute tasks. Hence, Amir et al. devised an object-oriented
mechanism [24]. This mechanism is capable of conducting a programming language called Corelet. A program written in
lexical analysis on a large collection of web documents. The Corelet requires an abstract representation of the network of
analysis results in a set of objects (i.e., concepts) and the neurosynaptic cores, along with well-defined interfaces
relationships among them. consisting of inputs and outputs (of the network), which
receive and send spikes, respectively. Corelet enables
B. Endowing Computer Systems with the Faculty of Thinking programmers to construct and compose a high-level network of
Once knowledge has been formally modeled, cognitive networks of neurosynaptic cores. Amir et al. also contribute a
computing systems are able to solve problems autonomously repository of more than 100 Corelet algorithms, which can be
by interpreting the causal relationships and concepts of a given used to create new programs.
domain knowledge base. In this regard, scientific research
Djurfeldt [15, 16] proposes a formal language called
efforts aiming at endowing computer systems with the faculty
connection-set algebra for structural connections of parallel
of thinking can be categorized into (i) low level computer
neural networks similar to Corelet [1]. Connection-set algebra
machinery (see for instance [2, 32, 37]) and (ii) high level
consists of a group of connection operators based on set
software frameworks (see for instance [1, 34, 59]).
algebra and matrix algebra, which create a graph representation
of the neural network, e.g., an adjacency matrix. However,

330
connection-set algebra is a high-level formalism focused on component). Finally, Orozco et al. designed a set of matrix
defining declarative connection patterns among neural operations that evolve (over time) self-regulation of avatars by
networks, and hence it does not take into account brain-like taking into account the vector of personality traits, the vector of
hardware architecture. universal emotions, and the vector of basic moods. The
resultant emotional intelligence model allows virtual avatars to
In order to determine causal relationships among concepts show empathy for humans’ emotional situations.
of a given domain knowledge base, Wang [59] proposes a
model of human attention (as a cognitive process) formalized In order for computers to feel, they need to perceive its
by means of denotational mathematics [60, 62]. According to environment [65]. Wang [65] proposes an automatic perception
Wang, attention is both a conscious and a subconscious engine, along with a formal inference engine, which allow
cognitive process that allows perceiving external events and computers to perceive through their behaviors and experiences.
reacting accordingly. Attention supports sensory processes, Both the perception engine and the inference engine are
memory processes, perception processes and action formalized by using descriptive mathematics [64, 66, 67].
intelligence, which are among the functions of the cerebellum
Provided that emotions (i.e., strong feelings) are a central
and some cerebral cortexes. In addition, Wang [59] states that
element for decision-making [5], Cambria et al. [5] combine an
the brain can be seen as a real-time, parallel system with
opinion mining engine and a facial expression classifier in
multiple threads of thought, which are interrupted and switched
order to determine the emotions of humans. By using a
based on an attention engine. Thus, attention allows cognitive
conversational avatar capable of perceiving and fusing
computational systems to select and focus on a set of parallel
information from user-generated written texts, voice, and
cognitive processes according to external stimuli.
video, the affective state of a human (interacting with the
Lawniczak and Di Stefano [34] propose a hierarchical 5- avatar) is extracted. Then, the avatar makes use of a facial
layer architecture for cognitive agents composed of a expression generator, a speech generator and a body gestures
perceptual layer, a reasoning layer, a judging layer, a response generator in order to create an empathic response.
layer, and a learning layer. The perceptual layer is in charge of
modeling sensors and creating an abstract representation of the IV. RESEARCH AGENDA
external world. The reasoning layer receives as input (i) a
model of the world and (ii) sensors’ outputs. Then, by using Based on the overview and analysis of current research on
fuzzy inference mechanisms, cognitive agents can obtain the cognitive computing (see Section III), we have identified three
logical rules of the world. The judgment layer uses the fuzzy main research opportunities in the area. The first opportunity is
rules provided by the reasoning layer in order to (i) extract to bridge the gap among cognitive approaches designed to
features from available data, (ii) create a rule-based model for endow computer systems with the faculties of knowing,
decision-making, and (iii) obtain estimations in the presence of thinking and feeling (Section IV.A). The second opportunity
uncertainty. The response layer consists of a collection of consists of integrating software-based cognitive approaches
automata, which define agents’ actions over the environment. and brain-like computer machinery (Section IV.B). The third
Finally, the learning layer implements a feedback loop to opportunity is to devise computational models for social
generate new knowledge from the interaction of agents with cognitive agents (see Section IV.C for details).
their environment.
A. Bridging the Gap among Knowing, Thinking and Feeling
C. Endowing Computer Systems with the Faculty of Feeling A myriad of cognitive approaches have been proposed to
As stated by Hoffman [31], the brain endows human beings capture and model knowledge. Both computer machinery and
with the faculty of feeling. In this regard, within the cognitive software frameworks aiming at endowing computer systems
computing community, there are several research efforts with the faculty of thinking have been designed. Mathematical,
oriented toward devising computational models of feelings (or psychological and data-oriented models of feelings have been
emotions), which are as follows. devised. However, it is noteworthy that even though the
faculties of knowing, thinking and feeling are interconnected
Orozco et al. [43] propose an emotional intelligence model [35], there is little work on integral cognitive architectures that
supported by the emotional competence framework, which is take all of them into account. The relationship between
composed of three components: self-consciousness, social knowledge and thinking has been somewhat studied.
awareness, and self-regulation. Orozco et al. devised (i) a Nevertheless, to the best of our knowledge, the relationship
knowledge base ontology for self-consciousness, (ii) a facial between knowledge and feelings has not been established.
expression classifier for social awareness, and (iii) a Cognitive computational models taking into account this
computational model that combines personality, emotions and relationship may provide insights into how feelings should
moods for self-regulation. Self-regulation is supported by (i) influence decision-making or vice versa.
the OCEAN model of personality [10] consisting of five
dimensions: openness, conscientiousness, extroversion, B. Integration of Software-based Cognitive Approaches and
agreeableness, and neuroticism; (ii) the universal emotions Brain-like Computer Machinery
[20]: anger, disgust, fear, happiness, sadness, and surprise;
and (iii) three basic moods: good, neutral, and bad. The values Software-based cognitive approaches focused either on
for each personality dimension, emotion, or mood, range representing knowledge, designing thinking or feeling
between 1 and -1 (representing the intensity level of a given capabilities are detached from brain-like computer machinery.
Networks of neurosynaptic cores are designed to efficiently

331
compute a large number of machine instructions. However, the [3] J. D. Bransford, “Human cognition: Learning, understanding and
network of neurosynaptic cores does not explain (i) how remembering,” Belmont, CA: Wadsworth, 1979.
domain-specific knowledge is represented in the brain [7] or [4] L. M. Brasil et al., “Hybrid expert system for decision supporting in the
medical area: complexity and cognitive computing,” International
(ii) the neurological relationship between reasoning and Journal of Medical Informatics, vol. 63, no. 1, pp. 19-30, 2001.
feelings [13]. On the one hand, there is a need for specialized [5] E. Cambria et al., “Sentic blending: Scalable multimodal fusion for the
neurosynaptic cores not only focused on efficient computations continuous interpretation of semantics and sentics,” in Proc. of the IEEE
but also focused on capturing knowledge and feelings. On the Symp. on Computational Intelligence for Human-like Intelligence, 2013,
other hand, there is a need for integration of cognitive software pp. 108-117.
focused on representing knowledge and modeling feelings into [6] Cambridge Dictionaries Online, “Knowledge definition, meaning – what
brain-like computer machinery. is knowledge in the British English Dictionary and Thesaurus,”
available at
http://dictionary.cambridge.org/dictionary/british/knowledge, 2015.
C. Social Cognitive Computing [7] A. Caramazza and J. Shelton, “Domain-specific knowledge systems in
There is scientific evidence in support of the development the brain: The animate-inanimate distinction,” Journal of Cognitive
of the brain due to the influence of complex social activities, Neuroscience, vol. 10, no. 1, pp. 1-34, 1998.
not only in human societies [11] but also in primate societies [8] K. Christoff and J. D. Gabrieli, “The frontopolar cortex and human
cognition: Evidence for a rostrocaudal hierarchical organization within
[17]. Hence, there is a need for social computational models the human prefrontal cortex,” Psychobiology, vol. 28, no. 2, pp. 168-
capable of evolving cognitive capabilities of agents. Cognitive 186, 2000.
agents interacting with each other not only should be capable [9] D. Clark, “A new vision for network architecture. private
of evolving their feelings and knowledge but also should be communication,” private communication, September 2002.
capable of improving their thinking skills. [10] P. T. Costa and R. R. McCrae, “Normal personality assessment in
clinical practice: The NEO Personality Inventory,” Psychological
Recalling that Brasil et al. [4] indicated that cognitive Assessment, vol. 4, no. 1, pp. 5-13, 1992.
computing is inspired by natural selection, improving thinking [11] L. Cozolino, “The Neuroscience of Human Relationships: Attachment
skills of a society of cognitive computing systems may be and the Developing Social Brain,” New York, NY: W. W. Norton &
achieved by using natural selection mechanisms in which Company, 2014.
desirable inheritable traits of a population predominate over [12] A. Cuzzocrea et al., “Analytics over large-scale multidimensional data:
undesirable traits across time. the big data revolution!,” in Proc. of the ACM 14th international
workshop on Data Warehousing and OLAP, 2011, pp. 101-104.
[13] A. Damasio, “Descartes’ error: Emotion, reason and the human brain,”
V. CONCLUSION New York, NY: Penguin, 1994.
The intent of this survey is to promote further research on [14] G. Dennis Jr. et al., “DAVID: database for annotation, visualization, and
cognitive computing by presenting a concise review of integrated discovery,” Genome biol, vol. 4, no. 5, P3, 2003.
literature on cognitive computing and identifying the open [15] M. Djurfeldt, “The connection-set algebra—a novel formalism for the
representation of connectivity structure in neuronal network models,”
research issues in the area. In addition, we also contribute a Neuroinformatics, vol. 10, no. 3, pp. 287-304, 2012.
taxonomy that categorizes cognitive computing research
[16] M. Djurfeldt, “The Connection-set Algebra: a formalism for the
according to its objectives into those providing a computer representation of connectivity structure in neuronal network models,
system with the faculties of knowing, thinking or feeling. implementations in Python and C++, and their use in simulators,” BMC
Moreover, we provide a general characterization of cognitive Neuroscience, vol. 12, no. 1, P80, 2011.
computing by discussing its multiple definitions and describing [17] R. I. Dunbar and S. Shultz, “Evolution in the social brain,” Science, vol.
their related fields and terms. 317, no. 5843, pp. 1344-1347, 2007.
[18] R. B. Dwore and J. Matarazzo, “The Behavioral Sciences and Health
Research on cognitive computing has had outstanding Education: Disciplines with a Compatible Interest?,” Health education,
advances in the last decade either on endowing computer vol. 12, no. 3, pp. 4-7, 1981.
systems with the faculties of knowing, thinking or feeling. [19] C. Eden, “On the nature of cognitive maps,” Journal of management
However, there is still a long journey ahead before fully studies, vol. 29, no. 3, pp. 261-265, 1992.
mimicking how brain processes information, generates [20] P. Ekman et al., “Emotion in the human face: Guidelines for research
knowledge, makes decisions and creates feelings. and an integration of findings,” New York, NY: Pergamon, 1972.
[21] M. N. ElBedwehy et al., “A computational knowledge representation
model for cognitive computers,” Neural Computing and Applications,
ACKNOWLEDGMENTS vol. 25, no. 7-8, pp. 1517-1534, 2014.
The first author acknowledges the support provided by [22] S. K. Esser et al., “Cognitive computing systems: Algorithms and
applications for networks of neurosynaptic cores,” in Proc. of the Int.
Asociación Mexicana de Cultura, A.C. and CONACYT under Joint Conf. on Neural Networks, 2013, pp. 1-10.
grant no. 216101. The second author wishes to thank UVM
[23] C. Evans et al., “Utilizing artificial neural networks and genetic
Laureate International Universities for their support. algorithms to build an algo-trading model for intra-day foreign exchange
speculation,” Mathematical and Computer Modelling, vol. 58, no. 5, pp.
1249-1266, 2013.
REFERENCES
[24] J. Fan et al., “Automatic knowledge extraction from documents,” IBM
[1] A. Amir et al., “Cognitive computing programming paradigm: a corelet Journal of Research and Development, vol. 56, no. 3.4, pp. 5-1, 2012.
language for composing networks of neurosynaptic cores,” in Proc. of
the Int. Joint Conf. on Neural Networks, 2013, pp. 1-10. [25] D. A. Ferrucci, “Introduction to ‘this is watson’,” IBM Journal of
Research and Development, vol. 56, no. 3.4, pp. 1-1, 2012.
[2] J. V. Arthur et al., “Building block of a programmable neuromorphic
substrate: A digital neurosynaptic core,” in Proc. of the Int. Joint Conf. [26] D. A. Ferrucci et al., “Building Watson: An overview of the DeepQA
on Neural Networks, 2012, pp. 1-8. project,” AI magazine, vol. 31, no. 3, pp. 59-79, 2010.

332
[27] H. Gintis, “Towards the unity of the human behavioral sciences,” [51] K. Shaw-Williams, “The social trackways theory of the evolution of
Politics, Philosophy & Economics, vol. 3, no. 1, pp. 37-57, 2004. human cognition,” Biological Theory, vol. 9, no. 1, pp. 16-26, 2014.
[28] A. Gliozzo et al., “Semantic technologies in IBM Watson,” available at [52] D. M. Szymanski and R. T. Hise, “E-satisfaction: an initial
http://www.aclweb.org/anthology/W13-3413, 2013. examination,” Journal of Retailing, vol. 76, no. 3, pp. 309-322, 2000.
[29] A. M. Hancock and A. Di Rienzo, “Detecting the genetic signature of [53] J. E. Thornton, “Life-span learning: A developmental perspective,” The
natural selection in human populations: models, methods, and data,” International Journal of Aging and Human Development, vol. 57, no. 1,
Annual review of anthropology, vol. 37, pp. 197-217, 2008. pp. 55-76, 2003.
[30] D. Heckerman and R. Shachter, “Decision-theoretic foundations for [54] Y. Tian et al., “A formal knowledge representation system (FKRS) for
causal reasoning,” Journal of Artificial Intelligence Research, vol. 3, pp. the intelligent knowledge base of a cognitive learning engine,”
405-430, 1995. International Journal of Software Science and Computational
[31] U. J. Hoffman, “Brain and Mind,” in The Science of the Mind Applied to Intelligence, vol. 3, no. 4, pp. 1-17, 2011.
Teaching: Including the Human Temperaments and Their Influence [55] L. G. Valiant, “Cognitive computation,” in Proc. of the IEEE 54th Annu.
Upon the Mind; the Analysis of the Metnal Faculties, and how to Symp. on Foundations of Computer Science, 1995, pp. 2-2.
Develop and Train Them; the Theory of Education and the School; and [56] L. B. Vaughn, “Feeling at Home: Law, Cognitive Science, and
Methods of Instruction and School Management, 1st ed., New York, NY, Narrative,” McGeorge L. Rev., vol. 43, pp. 999, 2012.
US: Fowler & Wells Company, 1885, pp. 379.
[57] E. Von Glasersfeld, “Cognition, construction of knowledge, and
[32] N. Imam et al., “A digital neurosynaptic core using event-driven qdi teaching,” Synthese, vol. 80, no. 1, pp. 121-140, 1989.
circuits,” in Proc. of the 18th IEEE Int. Symp. on Asynchronous Circuits
and Systems, 2012, pp. 25-32. [58] D. N. Walton, “What is reasoning? What is an argument?,” The Journal
of Philosophy, vol. 87, no. 8, pp. 399-419, 1990.
[33] W. B. Kristan, “The neurobiology of swimming in the leech,” Trends in
Neurosciences, vol. 6, pp. 84-88, 1983. [59] Y. Wang et al., “The cognitive process and formal models of human
attentions,” International Journal of Software Science and
[34] A. T. Lawniczak and B. N. Di Stefano, “Computational intelligence Computational Intelligence, vol. 5, no. 1, pp. 32-50, 2013.
based architecture for cognitive agents,” Procedia Computer Science,
vol. 1, no. 1, pp. 2227-2235, 2010. [60] Y. Wang, “On denotational mathematics foundations for the next
generation of computers: Cognitive computers for knowledge
[35] J. LeDoux, “The emotional brain: The mysterious underpinnings of processing,” Journal of Advanced Mathematics and Applications, vol. 1,
emotional life,” New York, NY: Simon and Schuster, 1998. no. 1, pp. 121-133, 2012.
[36] P. A. Merolla et al., “A million spiking-neuron integrated circuit with a [61] Y. Wang, “Towards the synergy of cognitive informatics, neural
scalable communication network and interface,” Science, vol. 345, no. informatics, brain informatics, and cognitive computing,” in Cognitive
6197, pp. 668-673, 2014. Informatics for Revealing Human Cognition: Knowledge Manipulations
[37] P. A. Merolla et al., “A digital neurosynaptic core using embedded in Natural Intelligence, 1st ed., Hershe PA, USA, IGI Global, 2012, ch.
crossbar memory with 45pJ per spike in 45nm,” in Proc. of the IEEE 1, pp. 1-19.
Custom Integrated Circuits Conference, 2011, pp. 1-4. [62] Y. Wang, “Inference Algebra (IA): A Denotational Mathematics for
[38] D. S. Modha et al., “Cognitive computing,” Communications of the Cognitive,” in Cognitive Informatics for Revealing Human Cognition:
ACM, vol. 54, no. 8, pp. 62-71, 2011. Knowledge Manipulations in Natural Intelligence, 1st ed., Hershe PA,
[39] T. V. Moore, “Cognitive psychology,” Philadelphia, Lippincott, 1939. USA, IGI Global, 2012, ch. 1, pp. 159-177.
[40] M. C. Nagel, “Nurturing a Healthy Mind: Doing what Matters Most for [63] Y. Wang and V. Chiew, “On the cognitive process of human problem
Your Child's Developing Brain,” Wollombi: Existle Publishing Pty Ltd, solving,” Cognitive Systems Research, vol. 11, no. 1, pp. 81-92, 2010.
2012. [64] Y. Wang, “On concept algebra: A denotational mathematical structure
[41] D. Nahamoo, “Cognitive computing journey,” in Proc. of the 1st for knowledge and software modeling,” International Journal of
workshop on parallel programming for analytics applications, 2014, pp. Cognitive Informatics and Natural Intelligence, vol. 2, no. 2, pp. 1-19,
63-64. 2008.
[42] V. Nazemoff, “Your Brain, Mind, and Business Transformation,” In The [65] Y. Wang, “Cognitive informatics: towards future generation computers
Four Intelligences of the Business Mind, Apress, 2014, pp. 1-16. that think and feel,” in Proc. of the 5th IEEE Int. Conf. on Cognitive
Informatics, 2006, pp. 3-7.
[43] H. Orozco et al., “Making empathetic virtual humans in human–
computer interaction scenarios,” in Proc. of the 11th Computer Graphics [66] Y. Wang, “On the informatics laws and deductive semantics of
International, 2010, pp. 1-4. software,” IEEE Transactions on Systems, Man, and Cybernetics, Part
C: Applications and Reviews, vol. 36, no. 2, pp. 161-171, 2006.
[44] T. Özcan et al., “Comparative analysis of multi-criteria decision making
methodologies and implementation of a warehouse location selection [67] Y. Wang, “The real-time process algebra (RTPA),” Annals of Software
problem,” Expert Systems with Applications, vol. 38, no. 8, pp. 9773- Engineering, vol. 14, no. 1-4, pp. 235-274, 2002.
9779, 2011. [68] J. Wiedermann, “The computational limits to the cognitive power of the
[45] A. Peña-Ayala and R. Mizoguchi, “Intelligent decision-making neuroidal tabula rasa,” in Algorithmic Learning Theory, 1st ed., Berlin,
approach based on fuzzy-causal knowledge and reasoning,” in Advanced Germany, Springer Berlin-Heidelberg, LNCS, vol. 1720, 1999, pp. 63-
Research in Applied Artificial Intelligence, 1st ed., Berlin, Germany, 76.
Springer Berlin Heidelberg, LNCS, vol. 7345, 2012, pp. 534-543. [69] T. Xiao and S. Lan, “Knowledge Reduction based on Cognitive Model
[46] R. Preissl et al., “Compass: A scalable simulator for an architecture for of Granular Computing,” Advances in Information Sciences & Service
cognitive computing,” in Proc. of the Int. Conf.on High Performance Sciences, vol. 5, no. 6, pp. 733-740, 2013.
Computing, Networking, Storage and Analysis, 2012, pp. 1-11. [70] K. Yelick et al., “Productivity and performance using partitioned global
[47] E. Rosch et al., “The embodied mind: Cognitive science and human address space languages,” in Proc. of the Int. Workshop on Parallel
experience,” Cambridge, MA: MIT press, 1992. Symbolic Computation, 2007, pp. 24-32.
[48] I. Rubinstein, “Big data: the end of privacy or a new beginning?,” [71] L. A. Zadeh, “Fuzzy logic = computing with words,” IEEE Transactions
International Data Privacy Law, vol. 3, no. 2, pp. 1-14, 2013. on Fuzzy Systems, vol. 4, no. 2, pp. 103-111, 1996.
[49] S. Russell and P. Norvig, “Artificial intelligence: A modern approach,”
Egnlewood Cliffs, Prentice-Hall, 1995.
[50] V. Saxena et al., “Absolute enrichment: gene set enrichment analysis for
homeostatic systems,” Nucleic acids research, vol. 34, no. 22, pp. e151-
e151, 2006.

333