What is Cognitive Science?

한국어 버전:

This provides an overview of cognitive science. Many parts of this post refer to the book <Cognitive Science: Interdisciplinary Convergence and Applications>¹.

What is cognitive science?

Cognitive science originates from questions about the self, exploring how individual minds operate. All questions that can be asked about the mind and intelligence (of animals, humans, or machines) are the domain of cognitive science, highlighting the complexity and interdisciplinary nature of understanding cognition. It emphasizes that no single discipline can fully answer these profound questions. But, we can learn about the human mind one by one through an interdisciplinary approach that crosses the natural sciences, humanities, and social sciences.

Cognitive science explores the relationships between (1) the brain, (2) the mind, (3) technological innovations such as computers, and (4) other artifacts, including both tangible hardware like robots and smartphones, and intangible systems such as language, economy, and administrative structures. It is defined as a multidisciplinary and interdisciplinary field that examines the informational and cognitive aspects of knowledge formation and utilization.

Information processing paradigm

From the start, cognitive science has employed the information processing paradigm as a foundational scientific framework to examine the human mind, its physical manifestation in the brain, and the computational analogs found in computers. This paradigm assumes that the mind functions are an information processing system, engaging in a series of processes such as sensation, perception, learning, memory, language, thinking, and emotion. It inquires into the specific types of information processing that occur within these processes, how each process interacts, and aims to describe the underlying informational (knowledge) structures—or representational structures—that facilitate each process.

Two important concepts in the information processing paradigm are computationalism and representationalism.

• Computationalism: The process of processing, utilizing, and converting information is called computation. The information processing paradigm is the activity of uncovering the computational processes of the mind and computer.

What is information? The definitions of ‘information’ in information theory and ‘information’ in the information processing approach are different. Information theory deals with the amount of information, not the meaning of information. On the other hand, when expressing cognitive phenomena, the concept of ‘information’ is generally used in relation to the concept of message meaning (the basic unit of knowledge).

• Representationalism: It is believed that various data are stored in the form of representations in the mind or computer.

What is representation? Rather than treating the object itself as is (presentation), the mind re-expresses (re-) it in some symbol or form, that is, abstracts it and stores and utilizes it. This abstracted content about an object is called a representation.

cognitive processes

Cognitive processes covered by cognitive science are various. The figure above shows topics and trends in cognitive science.²

The representative cognitive processes I think are as follows.

• Sensation
• Perception
• Cognitive Control
• Attention
• Reward / Punishment
• Value
• Learning
• Homeostasis
• Memory
• Language
• Decision Making
• Knowledge
• Thinking
• Feeling / Emotion / Affect
• Social Cognition
• etc…

Characteristics of Cognitive Science

1. Cognitive science views humans and the mind as an information processing system.
2. Cognitive science’s position is that the basic function of an information processing system is the conversion of information.
3. Representationalism, which states that information processing systems deal with representations, is another basic position in cognitive science.
4. The foundation of information processing systems emphasizes the neuroscience foundation of the nervous system.
5. Cognitive scientific information processing systems emphasize the need for a multidisciplinary approach.

Levels of approach in cognitive science

The level of approach taken when explaining complex human mental phenomena may vary from researcher to researcher. The theoretical framework for access (explanation) levels is based on the ideas of David Marr (Marr’s Three Levels), who proposed a theory of computational vision. ³

level	detail	related fields
Computational theory	What is the goal of the computation, why is it appropriate, and what is the logic of the strategy by which it can be carried out?	Psychology, linguistics, anthropology, etc.
Representation and algorithm	How can this computational theory be implemented? In particular, what is the representation for the input and output, and what is the algorithm for the transformation?	Computational neurology, computational linguistics, computational modeling
Hardware implementation	How can the representation and algorithm be realized physically?	Biology, neuroscience, electrical and electronic engineering

These three levels are independent of each other. This does not mean that one level of explanation is necessarily reducible to another level, but rather that all three levels of explanation may be necessary and play complementary roles.

History of Cognitive Science

Cognitive science has continued to change since its inception in the 1950s.⁴

figure source: Anna Riedl website

Early Years 1: Late 1950s - Early 1970s

Cognitive science, which has become a basic science in the West since the 1950s, was a scientific revolution. With the emergence of computationalism centered on Turing machines, the information processing paradigm began to gain attention. The core of the starting point of cognitive science is that “the computational process of a computer and the operating process of the human mind can be considered as a group under the common concept of information processing.” In other words, humans and computers have different hardware, but they are considered the same type of system.

The conceptual (theoretical) frameworks and empirical studies that emerged around this time and directly or indirectly influenced cognitive science are as follows.

• 1936, Alan Turing’s Automata
• 1943, Warren McCulloch and Walter Pitts’ neural network computational model
• 1945, Von Neumann architecture
• 1948, Claude Shannon’s Information Theory
• 1948, Norbert Wiener’s Cybernetics
• 1957, Herbert Simon and Alan Newell’s general purpose symbol manipulation system theory
• 1957, Noam Chomsky’s Syntactic Structures
• Experimental results from cognitive psychological research
• Neurological studies and single cell activity recording studies in brain-injured people
• Emergence of new philosophy of science and philosophy of mind

These conceptual frameworks and empirical research results emerged differently and were not integrated, but became a new scientific paradigm starting with the Symposium on Information Theory held at MIT in 1956. At the Dartmouth Workshop held at Dartmouth College in the summer of the same year, the name Artificial Intelligence (AI) was first officially used by John McCarthy, marking the start of the field of artificial intelligence.

Early Years 2: Late 1970s - Early 1980s

Within the framework of cognitive science, multidisciplinary exchange of ideas continued centered around information processing machines called computers. Also, in 1967, Ulrik Neisser established a new discipline called ‘cognitive psychology’. In addition, various theories and frameworks are proposed in various disciplines such as linguistics, artificial intelligence, and philosophy. However, during this period, very small-scale research centered on molecular biology emerged in neuroscience and distanced itself from cognitive science.

Adolescence 1: Mid-1980s

From the 1980s, the mind’s operating process, which was once viewed as a computer’s calculation process, began to be viewed as the ‘brain’ operating process. Neural network theory based on the brain and nervous system (the hardware of living organisms) emerged and a connection with neuroscience emerged. It may not have been a coincidence that a landmark paper in the field of artificial neural networks was published in 1986.⁵ At this time, cognitive science based on connectionism emerged as a new movement.

Adolescence 2: Late 1980s - Early 1990s

Cognitive science further developed with a focus on brain research, and a field called Cognitive Neuroscience was officially launched. The development of functional magnetic resonance imaging (fMRI), which is still widely used today as a tool to measure functional activity of the brain, also began around this time.⁶ Advances in cognitive neuroscience began to accelerate with the development of tools to measure which parts of the brain are activated when living people perform specific tasks.

It was also a time when evolutionary approaches were attempted along with neuroscience. Based on Darwin’s theory of natural selection, cross-species research in behavioral biology, brain modeling and evolution research, creation of genetic algorithms, and computer simulation research of the evolutionary process appear.

Adolescence 3: Mid-1990s

Research on ‘consciousness’ and ‘emotion’, which had been rejected in the early years of cognitive science, is also being revived during this period. Additionally, the concept arises that the human mind changes due to cultural, social and environmental factors. This began with the goal of breaking away from the limitations of approaches that explained the mind only through existing cognitive neuroscience or proposition-centered logical systems. By emphasizing the interaction between mind and environment, we begin to reveal that human cognitive structure is inseparable from the environment.

Adolescence 4: After 2000

With the introduction of the dynamical approach in physics, the Dynamical Systems approach, which emphasizes the dynamic changes in information processing characteristics and representations on the flowing time axis, is attracting attention, contrary to the traditional cognitivism of the past that emphasized information processing in static time. In addition, cognitive neuroscientific and evolutionary approaches, as well as embodied approaches that emphasize the mind based on the body, are emerging as new paradigms in cognitive science.

Major subfields of cognitive science

• Philosophy: epistemology, mental theory, philosophy of science, logic
• Psychology: Research topics, experimental methods, mind, cognition, cognitive engineering
• Neuroscience: Study of the brain (nervous system), brain/abnormal injury phenomena, neural structures and processes
• Artificial intelligence: digital computers, computer analogies, stored programming concepts, programming
• Linguistics: formal theory, internal rule computational linguistics, mental abilities
• Anthropology: Species/Cultural (Social) Determinants, Evolutionary Theory

In addition, mathematics, physics, communication studies, and sociology can also be considered subfields of cognitive science. More broadly, it can be expanded to include electrical engineering, robotics, information science, medicine, biology, architecture, music, literature, art, aesthetics, political science, economics, public administration, education, and computer engineering.

Cognitive Science Methodology

Cognitive science research methods use all existing research methods from various disciplines.

• Philosophy: logical and formal analysis
• Psychology: Experimental methods, simulations, natural observation methods
• Neuroscience: neuroanatomy, brain injury, brain imaging
• Artificial Intelligence: Simulation, language reporting analysis methods
• Linguistics: formal analysis, natural observation, experimental methods
• Anthropology: people’s livelihood methods, natural observation methods

(1) Experimental method

• Reaction time methods: As a central method in cognitive psychology, it is believed that information processing time (speed) changes as the cognitive task changes qualitatively or quantitatively, and it is determined from the difference in reaction time that occurs as the problem situation slightly changes. It is a method of exploring whether information processing processes/structures exist, are involved, and progress, and how their characteristics differ from other processes or structures.
• Psychophysics
  • Classic psychophysical method: This is a method of exploring how the sensation and perceived degree of a physical stimulus, that is, the degree of psychological change, changes as the size or intensity of the physical stimulus changes. German psychologist Fechner created the psychophysical method, which primarily measures thresholds.
  • Signal detection theory: The basic position of signal detection theory is that the sensory response to a human stimulus is not simply determined as a function of the size of the stimulus, but rather involves the human subjective decision process. We look at it in two stages: (1) collecting stimulus information and (2) evaluating it and deciding on a response.

(2) Computer simulation

In computer simulations, the detailed information processing operations of certain psychological processes are logically and rationally inferred, specified, and analyzed based on the individual’s external reactions and introspection reports. When this is converted into a program and inserted into a computer, does the computer behave the same as a human? By observing (whether output is produced), it is verified whether the program, that is, the information processing operations assumed by the researcher, are theoretically sound.

(3) Protocol analysis

When scientific research on the mind began in the 19th century, not only external observations but also data in which subjects reported their subjective inner experiences were used for analysis. However, this introspection method was rejected by behaviorist psychologists on the grounds that it was unscientific, but was revised and revived after the information processing paradigm of the 20th century. Verbal report analysis, a type of introspection method, is a method of analyzing introspective verbal reports about the process of performing a task by a subject performing a certain task. This method first presents a stimulus and then verbally reports the information processing that occurs in the mind.

(4) Cognitive neuroscience methods

This is an illustration comparing spatial resolution and temporal resolution at a glance for a measurement tool used in neuroscience.⁷

(a) Neuroimaging

Imaging of electrical signals

name	detail
Single cell recording	This method provides information about the activity of a neuron by injecting an electrode directly near a single neuron in the animal’s brain and recording the electrical activity of this neuron on a computer.
Electroencephalogram (EEG)	A method of recording the electrical activity of the brain by attaching multiple electrodes to the outside of the scalp. Brain waves are drawn by recording potential differences between various points in the brain.
Event-related potentials (ERP)	This is a method of repeatedly presenting the same stimulus event and recording the measurement by obtaining the average potential of these events.
Magnetoencephalography (MEG)	A picture of magnetic signals generated in the brain is called a magnetoencephalogram. It is a method of measuring the process of generating electrical signals when the brain is active and then creating magnetic signals again using a superconducting sensor. This method overcomes the limitations of temporal resolution and spatial resolution to some extent, but has limitations in accurately finding the area performing complex cognitive tasks, so it is usually used in conjunction with other methods such as fMRI.

Imaging of brain structure and function

name	detail
Computed tomography (CT: computerized tomography)	A method of imaging the brain structure in three dimensions using X-rays.
Single photon emission computed tomography (SPECT)	Functional brain imaging that observes changes in distribution of regional cerebral blood flow.
Positron emission tomography (PET)	Radioactive isotopes are injected into blood vessels and images are created with a camera that detects the radioactivity emitted by this substance. Before fMRI, it was widely used in brain and cognitive function research, but recently it is widely used to image substances at the molecular level in the brain.
Magnetic resonance imaging (MRI)	A method of imaging the anatomical structure of the brain using magnetic properties.
Functional magnetic resonance imaging (fMRI)	fMRI uses MRI equipment and repeatedly measures blood oxygen level dependent (BOLD) signals to image the degree of brain activation. Although brain function is examined indirectly through the change in blood flow resulting from this rather than the direct activity of nerve cells, it provides the highest quality images in terms of spatial resolution than any other imaging technology and is non-invasive, so it can be used to examine the relationship between the brain and cognitive function. This is the most used method when researching.
MRS (magnetic resonance spectroscopy)	A method of imaging not the water molecules that make up 80% of brain matter, but other substances (micromolecules, cell membranes, glial cells, etc.) that make up the remaining 20%.
Functional near infrared spectroscopy (fNIS)	A method of measuring blood flow using near infrared rays. It measures oxygen consumption and increase in blood flow by measuring the concentration of hemoglobin bound to oxygen and hemoglobin not bound to oxygen.
DTI (diffusion tensor imaging)	A method of visualizing the connectivity of nerve fibers, it detects the directional movement of water molecules and images the path of the nerve fibers. We look at the connectivity of white matter, not gray matter.

(b) Brain stimulation method

name	detail
Transcranial magnetic stimulation (TMS)	By intentionally stimulating a specific area of ​​the brain, information processing ongoing in that area can be temporarily interrupted or activated. When an electric current is passed through a series of magnetic field pulses through a magnetic coil to a specific area outside the skull, it stimulates or inhibits neural activity in the underlying brain area.
Intracranial brain stimulation	An invasive method of inserting an electric current directly into the brain. This method intentionally generates action potentials in neurons by inserting electrodes.

(c) Neuropsychological tests

A type of psychological test that evaluates whether and to what extent cognitive and behavioral problems have occurred due to brain damage. It is mainly used in clinical settings.

(d) Computational modeling of neural processes

This approach holds that the brain is essentially made up of many simple elements called neurons, which work in parallel and systematically connected. Biologically and functionally, the physiological mechanisms and changes of nerve cells or nervous systems are described through formulas and focus on mathematical modeling. Rather than higher-level cognitive processes, we are mainly interested in lower-level cognitive processes such as attention, sensation, perception, and basic memory processes in humans and animals, and use simulation methods to analyze and model how these cognitive processes occur.

(5) Non-experimental methods

• Case study method: A longitudinal study of cases that demonstrate extreme cognitive phenomena.
• Natural observation method: This is a method of observing and analyzing things as they are in natural situations, not in a laboratory, without intervening or controlling the situation.
• Discourse analysis: This is a method of analyzing the characteristics revealed in sentences spoken during conversation.

The future of cognitive science

To what extent will cognitive science be able to answer questions about ‘me’ in the future? In which direction is the recent rapid development of artificial intelligence heading and what impact will it have on us in the future?

• UX/UI (User eXperience / User Interface): Integrating human cognitive, emotional, and motivational characteristics and reflecting them in software and hardware development
• Robotics: Developing a concrete, embodied intelligence system rather than an abstract artificial intelligence system
• BCI (Brain-Computer Interface): Development of an interface that complements and strengthens brain functions by connecting the computer to the brain

---

1. Lee Jeong-mo. (2009). Cognitive science: Interdisciplinary convergence and applications. Sungkyunkwan University Press. ↩

2. Rothe A., Rich A. S., & Li Z. (2018). Topics and Trends in Cognitive Science. In T. T. Rogers, M. Rau, X. Zhu, & C. W. Kalish (Eds.), Proceedings of the 40th Annual Conference of the Cognitive Science Society (pp. 979-984). Austin, TX: Cognitive Science Society. (https://anselmrothe.github.io/publication/2018cogsci_dtm/) ↩

3. Marr, D. (1982). Vision: A Computational Approach. Freeman & Co. ↩

4. Boden, M. A. (2008). Mind as machine: A history of cognitive science. Oxford University Press. ↩

5. Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986). Learning representations by back-propagating errors. nature, 323(6088), 533-536. ↩

6. Ogawa, S., Lee, T. M., Nayak, A. S., & Glynn, P. (1990). Oxygenation‐sensitive contrast in magnetic resonance image of rodent brain at high magnetic fields. Magnetic resonance in medicine, 14(1), 68-78. ↩

7. Sejnowski, T. J., Churchland, P. S., & Movshon, J. A. (2014). Putting big data to good use in neuroscience. Nature neuroscience, 17(11), 1440-1441. ↩