This document provides an overview of cognitive science artificial intelligence (AI), which studies how to simulate human intelligence through AI to better understand the human mind and solve complex problems. Current research aims to develop human-level machine intelligence through projects that simulate human reasoning, decision-making, creativity, and goal management. Applications of this research could optimize design problems and advance cognitive science theories with implications for education and medicine. Future progress depends on improved understanding of human cognition and advanced computing technologies.
2. would not be distinguished from human intelligence, a from the intelligent agent’s environment in working memory
challenge known as the Turing test [3]. Because intelligent [6]. Because immediate sensory data is sometimes
agents often face situations with incomplete information, insufficient for decision-making in the real world, storing
encoding data for all possible situations is a limited approach previous situations is useful in differentiating between
to simulate human intelligence [4]. In other words, because situations that would otherwise appear identical to the
there are infinitely many situations that can arise in the real intelligent agent at a specific instance [7]. In short,
world, it is impossible to design an intelligent agent pre- maintaining memory of past events “makes it possible to not
programmed with solutions to all of the problems it may face. only make correct decisions but to learn the correct
Instead, an intelligent agent must be equipped with the decision” [7]. When the intelligent agent enters an impasse,
ability to make decisions based on the information it has and the agent can search its memory for a solution to the
re-evaluate its past solutions to improve future decisions. problem. If the problem is unique, the agent remembers its
Consequently, a more fundamental understanding about how actions in case the problem is encountered again [6]. In
the human mind learns and solves problems is necessary to summary, the SOAR cognitive architecture system relies on
design an intelligent agent with the same intelligence. maintaining information from decisions and outcomes in
Currently, numerous research projects are making progress past experiences to improve future decisions in simulating
in these goals of both simulating human intelligence to study human behavior. The SOAR system is a useful tool for using
the human mind as well as the simulation of human simulated human intelligence to solve complex problems.
intelligence to solve complex problems. C. Simulating Creativity
A. Simulating Theory of Mind In his papers on the simulation of human level
A central topic in cognitive science and psychology, intelligence in the decision process, Dr. Zadeh emphasizes
theory of mind refers to “one’s ability to infer and the importance of imitating creativity in intelligent agents.
understand the beliefs, desires, and intentions of others, Although knowledge of past experiences is a useful tool in
given the knowledge one has available” [5]. To investigate decision-making, Dr. Zadeh acknowledges that “creativity is
the various theories that explain how theory of mind takes a gifted ability of human beings in thinking, inference,
place on the cognitive level, Dr. Hiatt and Dr. Trafton use problem solving, and product development” [8]. In his
the ACT-R cognitive architecture to simulate how accurately formal definition of the unique ability, creativity is divided
children can predict the actions of others as they age, a prime into three categories: abstract, concrete, and artistic. More
example of using artificial intelligence to study the human relevant to engineering applications, concrete creativity
mind. ACT-R consists of modules associated with different involves generating new, innovative solutions in an
areas of the brain, buffers which each hold a symbolic item, environment limited by goals and available conditions [8].
and a pattern matcher that determines actions to be taken Aiming to equip intelligent agents with the creative ability of
based on the contents of the buffers. Furthermore, this core the human mind, Zadeh provides an outlined approach for
cognitive architecture has the ability to interface with the implementing the creative process in a computer program.
environment via visual, audio, motor, and aural modules and The ability of an intelligent agent to create new approaches
learn new facts and rules through reinforcement learning; to solving problems is vital for modeling human level
based on these capabilities, ACT-R is a suitable system for intelligence.
simulating the mind of a growing child [5]. Based on the D. Simulating Rationality
idea that children learn and mature as they grow, Dr. Hiatt
and Dr. Trafton include a maturation parameter associated The multi-agent recursive simulation technology
with the age of the simulated child. A higher level of for N-th order rational agents (MARS-NORA) is a procedure
maturity corresponds to a more advanced ability in the child developed by Dr. Mussavi Rizi and Dr. Latek to rationally
to select between their inferred beliefs about the beliefs and choose a course of action for multiple artificial intelligence
actions of others [5]. From simulating the theory of mind agents in a dynamic environment. Similar to how a human
development of numerous children, Dr. Hiatt and Dr. weighs the pros and cons of a decision, MARS-NORA
Trafton found evidence supporting the legitimacy of main requires agents to derive the probability distribution of
theories of how theory of mind is developed in existence utility gained for each possible course of action [9]. MARS-
today. NORA has two algorithms for determining the optimal
course of action once all possible algorithms are considered:
B. The SOAR Project myopic planning and non-myopic planning. In myopic
Dr. Laird, a professor in computer science at the planning, the zero-order agent chooses a random action.
University of Michigan, developed the SOAR system, a Each proceeding agent chooses its optimal course of action
cognitive architecture programming structure with the goal based on the actions of agents of lower order, overall
of simulating a human brain, as a unique, alternative resulting in the on-average optimal action of the multi-agent
approach to the traditional and restricted hard-coding data [9]. Because the actions of previous agents limit the actions
approach. The SOAR system stores information retrieved of agents of higher order, myopic planning is not suitable for
3. situations in which the multi-agent acts asynchronously with intelligence agent’s behavior to achieve top-level goals in a
other multi-agents. Myopic planning also fails if the multi- dynamic environment.
agent wishes to derive multiple optimal courses of action or As seen in these current research projects, cognitive
takes inconsistent amounts of time to complete each action; science artificial intelligence can be used to supplement
instead, non-myopic planning can be used [9]. Because research in cognitive science and vice versa. Furthermore,
asynchronous multi-agents’ actions influence each other, the these works contribute to achieving improved human-level
non-myopic planning algorithm considers three situations. intelligence simulations in the cognitive science artificial
First, in the event that a multi-agent has both a higher order intelligence field. Although no artificial intelligence has
of rationality and a longer planning horizon than the other come close to achieving the goal of human-level intelligence,
multi-agent, the stronger agent selects its optimal course of intelligent agents are consistently being re-evaluated and
action while the latter agent accepts a short term loss and improved.
returns to a synchronous state with the stronger agent [9].
The second situation involves a multi-agent that has a higher III. APPLICATIONS AND THE IMPORTANCE OF
order of rationality than its opposing multi-agent but a short COGNITIVE SCIENCE ARTIFICIAL INTELLIGENCE
planning horizon. In this situation, the multi-agent with the As seen in the goals of the previously mentioned
shorter planning time is “locked” into their path of actions researchers in the field, there are numerous, important, real
and will not make optimal decisions. The third situation world applications of cognitive science artificial intelligence
involves multi-agents with relatively equal orders of research. In our society, engineers and architects constantly
rationality and planning horizon length. In this case, the face tasks, such as constructing a highway or designing a
agents have similar cognitive abilities and can cooperate to traffic light, that require optimizing a design despite physical
optimize their actions [9]. With two algorithms for deriving and financial limitations. For instance, in the traffic light
an optimal choice of action for multi-agents, MARS-NORA example, an engineer must consider the tradeoff economics
allows agents to behave rationally by following the decision between using stronger materials and the price of these
process of humans during the action selection process. materials or calculate statistics on the large amounts of
E. Achieving Top-down Goals traffic data available for the intersection to determine light
timing. With the ability to consider large amounts of
The ICARUS Architecture is a cognitive
information and design considerations in a short period of
architecture comparable to SOAR. The architecture supports
time, advanced intelligence can be developed to solve these
top-level goals by guiding the agent’s behavior to
types of complex logic problems [2]. In this way, the use of
accomplish its tasks while maintaining reactivity. However, artificial intelligence as a tool for engineers could make the
because ICARUS does not support adding, deleting, or
design process faster, more efficient, and more accurate.
reordering top-level goals, the ability to manage multiple
top-level goals in this cognitive architecture is somewhat In addition, the creation of a human level intelligent
limited, especially since the goals of a human are often agent provides a “better mirror” of the human mind that is
changed and prioritized [10]. Dr. Choi at Stanford University easier to study than the human brain for cognitive science
addresses this limitation in his extension of the ICARUS researchers. By studying realistic simulations of human
architecture. In his revision of the goal management system, cognition, theories can be drawn about humans’ nature and
each general goal now includes a goal description and cognitive limitations. Furthermore, researchers can achieve
relevance conditions, used to prioritize goals based on the specific cognitive science goals, such as understanding how
current state of the agent [10]. The new system receives intelligence develops in the brain or how damage to different
information about its surroundings during each “cycle.” parts of the brain affect cognition [2]. Progress in these areas
Once information about the environment is retrieved, the can powerfully impact how the human mind is understood,
goal management system can add, remove, or re-prioritize with the potential of leading to improvements in present
goals based on the agent’s “belief state” through a goal society. For instance, a better understanding of how the
nomination process. Top-level goals are prioritized based on human mind learns and retains information can lead to
initial priority and relevancy to the current state of the improved learning methods implemented in schools to
environment [10]. Furthermore, when selecting actions to accelerate human progress. In the same way, improved
achieve goals, Dr. Choi’s extension retrieves the agent’s theories on how different areas of the brain affect behavior
skills relevant to the current goal and generates a plan to can help develop medical solutions for victims of brain
accomplish the goal, utilizing the non-primitive skills first trauma [3]. The useful applications of pursuing research in
[10]. This goal management system design is more realistic cognitive science artificial intelligence continue to grow as
to a human’s behavior as goals change as the surrounding research in the field continues.
environment changes. Improved by modifying the IV. THE FUTURE OF COGNITIVE SCIENCE
architecture to better resemble a human’s goal management ARTIFICIAL INTELLIGENCE
process, Dr. Choi’s extension of the ICARUS cognitive
architecture is an effective system for guiding an artificial
4. Although there have been many breakthroughs in addresses the goals of both cognitive science and artificial
the cognitive science artificial intelligence field, researchers intelligence. As research in the field continues, improved
are continually working to improve intelligent agents. The intelligent agents will be developed with the ability to
human mind has the impressive capability of preforming simulate human-level intelligence in the final cognitive
numerous mental and physical tasks with little mental strain science research goal of fully understanding the human mind
[2]. On the other hand, computer simulated intelligence is or to address important, complex problems of mankind
limited by the speed and capacity of hardware for through artificial intelligence.
performing computations. The development of advanced
nanotechnology to increase hardware speed and memory
will reduce this restraint on simulating human level REFERENCES
intelligence [11]. Furthermore, while theories of cognitive [1] Thagard, Paul. (2009). Cognitive Science. The Stanford Encyclopedia
science artificial intelligence have fostered improved of Philosophy (Fall 2008 Edition), Edward N. Zalta (ed.).
understanding of the human mind, advancements in the [2] Simon, H. (2010). Cognitive Science: Relationship of AI to Psychology
psychology and cognitive science fields, that help better and Neuroscience. AAAI.
understand human behavior, can be used to further improve [3] Wang, Y. (2008). Proceedings of the Seventh IEEE International
intelligent agents [3]. Finally, an issue more acknowledged Conference on Cognitive Informatics: ICCI 2008: August 14-16, 2008,
Stanford University, California, USA. [Piscataway, N.J.]: IEEE
by the public than researchers in the field, include ethical Xplore.
and organizational concerns with the coevolution of humans [4] Bickhard, M., and Terveen, L. (1995). Foundational Issues in Artificial
and intelligent systems; these issues may one day have to be Intelligence and Cognitive Science. Elsevier Science Publishers.
addressed [11]. For instance, society would have to address [5] Hiatt, L. M., and Trafton, J. G. (2010). A Cognitive Model of Theory of
restrictions on how a human-simulating robot can behave. Mind. 10th International Conference on Cognitive Modeling: ICCM
2010: August 5-8, 2010, Philadelphia, PA, USA.
Because intelligent agents are still far from [6] Lehman, J.F., Laird, J., and Rosenbloom, P. (2006). A Gentle
achieving artificial intelligence goals, such as passing the Introduction to SOAR, an Architecture for Human Cognition: 2006
Turing test, or cognitive science goals, such as achieving Update.
human level intelligence or improving the present [7] Laird, J.E., and Wang, Y. (2007). The Importance of Action History in
Decision Making and Reinforcement Learning. Proceedings of the
understanding of the human mind, there are still many Eighth International Conference on Cognitive Modeling. Ann Arbor,
opportunities for research achievements in the cognitive MI.
science artificial intelligence field. This room for growth [8] Zadeh, L. (2008). On Cognitive Foundations of Creativity and the
shows great potential for developing technology to increase Cognitive Process of Creation. Proceedings of the Seventh IEEE
International Conference on Cognitive Informatics: ICCI 2008 :
the progress of mankind. August 14-16, 2008, Stanford University, California,
V. CONCLUSIONS USA. [Piscataway, N.J.]: IEEE Xplore.
[9] Latek, M., and Mussavi Rizi, S.M. (2010). Plan, replan and plan to
Artificial intelligence is an extremely useful tool for replan Algorithms for robust courses of action under strategic
cognitive science research of both fundamental and high uncertainty. BRIMS 2010: March 21-24, 2010, Charleston, SC, USA.
level understanding of the human mind by simulating the [10] Choi, D. (2010). Nomination and Prioritization of Goals in a Cognitive
human mind. In the same way, cognitive science theories Architecture. 10th International Conferenceon Cognitive Modeling:
ICCM 2010: August 5-8, 2010, Philadelphia, PA, USA.
provide useful insight on human cognition that can be
[11] Jacobstein, N. (2005). The Prospects for AI. IT Conversations.
encoded into artificial intelligence. Cognitive science <http://itc.conversationsnetwork.org/shows/detail713.h tml>.
artificial intelligence is a powerful research area that