The whole idea behind cognitive neuroscience is the once radical notion that the mysteries of human thought can be explained in much the same way as everything else in science -- by reducing a complex phenomenon (cognition) into simpler components (the underlying biological mechanisms of the brain). This process is just reductionism, which has been and continues to be the standard method of scientific advancement across most fields. For example, all matter can be reduced to its atomic components, which helps to explain the various properties of different kinds of matter, and the ways in which they interact. Similarly, many biological phenomena can be explained in terms of the actions of underlying DNA and proteins.
Although it is natural to think of reductionism in terms of physical systems (e.g., explaining cognition in terms of the physical brain), it is also possible to achieve a form of reductionism in terms of more abstract components of a system. Indeed, one could argue that all forms of explanation entail a form of reductionism, in that they explain a previously inexplicable thing in terms of other, more familiar constructs, just as one can understand the definition of an unfamiliar word in the dictionary in terms of more familiar words.
There have been many attempts over the years to explain human cognition using various different languages and metaphors. For example, can cognition be explained by assuming it is based on simple logical operations? By assuming it works just like a standard serial computer? Although these approaches have borne some fruit, the idea that one should look to the brain itself for the language and principles upon which to explain human cognition seems more likely to succeed, given that the brain is ultimately responsible for it all. Thus, it is not just reductionism that defines the essence of cognitive neuroscience -- it is also the stipulation that the components be based on the physical substrate of human cognition, the brain. This is physical reductionism.
As a domain of scientific inquiry matures, there is a tendency for constructs that play a role in that domain to become physically grounded. For example, in the biological sciences before the advent of modern molecular biology, ephemeral, vitalistic theories were common, where the components were posited based on a theory, not on any physical evidence for them. As the molecular basis of life was understood, it became possible to develop theories of biological function in terms of real underlying components (proteins, nucleic acids, etc.) that can be measured and localized. Some prephysical theoretical constructs accurately anticipated their physically grounded counterparts; for example, Mendel's theory of genetics anticipated many important functional aspects of DNA replication, while others did not fare so well.
Similarly, many previous and current theories of human cognition are based on constructs such as ``attention'' and ``working memory buffers'' that are based on an analysis of behaviors or thoughts, and not on physical entities that can be independently measured. Cognitive neuroscience differs from other forms of cognitive theorizing in that it seeks to explain cognitive phenomena in terms of underlying neurobiological components, which can in principle be independently measured and localized. Just as in biology and other fields, some of the nonphysical constructs of cognition will probably fit well with the underlying biological mechanisms, and others may not [ChurchlandChurchland1986Churchland86, e.g.,,]. Even in those that fit well, understanding their biological basis will probably lead to a more refined and sophisticated understanding (e.g., as knowing the biological structure of DNA has for understanding genetics).