Cognitive Science
Cognitive science is the interdisciplinary study of the mind and intelligence, combining psychology, linguistics, computer science, neuroscience, philosophy and anthropology around the idea that thinking can be understood as the processing of information and representations. Its distinctive method is to build explicit models, often as computer programs, and to test them against human behavior and brain evidence. It took shape between the late 1940s and the late 1970s, and it is the field where artificial intelligence and the science of the human mind most directly meet.
The usual birthdate is 11 September 1956, the second day of a symposium on information theory at MIT. The papers presented there included Allen Newell and Herbert Simon's report on their Logic Theory Machine, a program that proved theorems from Whitehead and Russell's Principia Mathematica; Noam Chomsky's "Three Models for the Description of Language", arguing that finite-state models cannot capture natural language syntax; and an early version of George Miller's paper on the limits of short-term memory. Miller later recalled the symposium as the moment when he became convinced that experimental psychology, theoretical linguistics and computer simulation were parts of a single larger enterprise. The date is a convenient marker rather than a founding act, since the Dartmouth workshop on AI the same summer and the 1948 Hixon Symposium on cerebral mechanisms in behavior, with Karl Lashley's critique of behaviorism, belong to the same current.
The revolt against behaviorism had specific, citable turning points. Chomsky's 1959 review of B. F. Skinner's Verbal Behavior argued that stimulus and response notions were too vague to explain the creativity and structure of language, and that children acquire grammar in ways no reinforcement account could cover. In 1960 George Miller and Jerome Bruner founded the Harvard Center for Cognitive Studies, and Miller, Eugene Galanter and Karl Pribram published Plans and the Structure of Behavior, which analyzed behavior in terms of hierarchically organized plans modeled on computer programs. Ulric Neisser's 1967 book Cognitive Psychology gave the field a name and a textbook, covering perception and memory as information processing.
Institutionally the field consolidated in the late 1970s. The journal Cognitive Science started in 1977, an influential 1978 report prepared for the Alfred P. Sloan Foundation pictured the field as a set of linked disciplines, and the Cognitive Science Society held its first conference at La Jolla in 1979. Programs and departments followed. The six-way map is often drawn as a hexagon, with the lines between disciplines labeled by the hybrid subfields that arose, such as psycholinguistics and neurolinguistics; the figure persists even though the exact membership has shifted.
Each parent discipline contributed a distinct tool. Psychology supplied experiments and reaction-time methods; linguistics, formal grammars and the competence-performance distinction; computer science, programs as precise models and the vocabulary of algorithms and data structures; neuroscience, constraints from physiology; philosophy, questions about representation, meaning and consciousness; and anthropology, attention to culture. David Marr's three levels of analysis, set out in his 1982 book Vision, became a common organizing frame: what problem a system solves (the computational level), how it does so (algorithm and representation), and how that is physically realized. Marr's separation helps explain why a model can be right about one level and silent about another.
The field has repeatedly changed its own assumptions. In 1986 David Rumelhart, James McClelland and the PDP Research Group published Parallel Distributed Processing, defending connectionist networks as models of cognition against the symbolic view that dominated the first decades. Later movements stressed embodied and situated cognition, Bayesian models of inference, and cognitive neuroscience with brain imaging. The result is a plural field rather than a single theory, and its debates, such as symbols versus networks or innate structure versus learning, reappear unchanged in discussions of modern language models.
Common misconceptions concern what the field is. Cognitive science is not the same as AI: AI builds capable systems, while cognitive science wants models that explain human performance, including errors, and a system that beats people may be useless as a theory. It is not simply cognitive psychology with computers either, since linguistics and philosophy shaped its questions. And it did not make behaviorism obsolete overnight; behaviorist methods of controlled experiments and learning theory survived inside the new field, and reinforcement learning in AI inherits directly from them.
Read cognitive science as the bargain struck around 1956 and formalized by 1979: explain the mind by building and testing precise information-processing models, with every discipline bringing the tools it has.