David Marr was a British neuroscientist whose early work was a theory of the cerebellar cortex and later an explanation of visual information processing from the retina to the visual cortex. He died of leukemia in 1980.

His book Vision: A computational investigation into the human representation and processing of visual information, was published posthumously in 1982 by Marr's colleagues at MIT.

The introduction to Vision says "vision is an information processing task,"

"that our brains must somehow be capable of representing this information," and that the work is "an inquiry into the nature of the internal representations by which we capture this information and thus make it available as a basis for decisions about our thoughts and actions."

Vision, p.3

It continues...

"the fundamental point is that in order to understand a device that performs an information processing task, one needs many different kinds of explanations," that "each problem has to be addressed from several points of view — as a problem in representing information, as a computation capable of deriving the representation, and as a problem in the architecture of a computer capable of carrying out both things quickly and reliably."

Vision, pp.4-5

Marr's "computational" vision of "representations" doing "information processing" was based on the earlier work of many MIT scientists who saw the brain as a logical machine capable of doing computations, starting with Warren McCulloch and Walter Pitts's 1943 paper "A Logical Calculus of the Ideas Immanent in Nervous Activity."

A decade later there was the MIT Symposium on Information Theory on September 11, 1956. It marked the end of the behaviorism of John D. Watson and B.F.Skinner and the beginning of "cognitive science."

George Miller presented his seminal paper "The Magical Number Seven, Plus or Minus Two," outlining the limits of human short-term memory capacity.

Herbert Simon and Allen Newell presented the Logic Theorist and General Problem Solver, early computer programs capable of proving mathematical theorems. These became the model for future artificial intelligence.

Noam Chomsky described his transformational generative grammar concepts that shifted linguistics toward cognitive brain "deep structures." Chomsky characterized behaviorism and cognitivism as empiricist and rationalist, arguing that something beyond pure observation is needed to explain the mind.

Researchers began viewing the mind as an information-processing system—treating the brain as hardware and mental processes as software.

George Miller years later called this the birth of cognitive science...

Cognitive science is a child of the 1950s, the product of a time when psychology, anthropology and linguistics were redefining themselves and computer science and neuroscience as disciplines were coming into existence. Psychology could not participate in the cognitive revolution until it had freed itself from behaviorism, thus restoring cognition to scientific respectability. By then, it was becoming clear in several disciplines that the solution to some of their problems depended crucially on solving problems traditionally allocated to other disciplines. Collaboration was called for: this is a personal account of how it came about.

TRENDS in Cognitive Sciences Vol.7 No.3 March 2003

The Psychology Section of the MIT Department of Economics and Social Science became the Department of Brain and Cognitive Sciences (BCS)

In a sense, of course, the brain is a computer, but to say this without qualification is misleading, because the essence of the brain is not simply that it is a computer, but that it is a computer which is in the habit of performing some rather peculiar computations. The term computer usually refers to a machine with a rather standard type of instruction set that usually runs serially but nowadays sometimes in parallel, under the control of programs that have been stored in memory...

[A] critical point is that understanding computers is different from understanding computations. To understand a computer, one has to study that computer. To understand an information-processing task, one has to study that information-processing task. To understand fully a particular machine carrying out a particular information-processing task, one has to do both things. Neither alone will suffice.

Vision, p.5

From a philosophical point of view, the approach that I describe is an extension of what have been called representational theories of mind [cf., Jerry Fodor...]

Modern representational theories conceive the mind as having access to a system of internal representations; mental states are characterized by asserting what the internal representations currently specify, and mental processes by how such internal representations are obtained and how they interact.

Vision, p.6

Marr's "Epilogue" to the Vision book contains an imagined conversation between him, his colleague Tomaso Poggio, and Francis Crick. It began with the work of Horace Barlow.

Can we begin with the levels-of-explanation idea, since you attribute so much importance to it? How is it related to ideas about feature detectors and in particular to Horace Barlow’s first dogma (1972, p. 380), which states, “A description of the activity of a single nerve cell which is transmitted to and influences other nerve cells, and of a nerve cell’s response to such influences from other cells, is a complete enough description for functional understanding of the nervous system?"

Here, of course, I must disagree with Barlow’s formulation, although I do agree with one of the thoughts behind this dogma, namely, that there is nothing else looking at what the cells are doing — they are the ultimate correlates of perception. However, the dogma fails to take level one analysis — the level of the computational theory — into account.

Vision, p.336