In the mathematically-predictive hard sciences, citation is usually viewed as a necessary evil. Developing arguments 'from fundamentals' is more desirable but often impossible as the long chains of logic are harder to follow and remember. Accordingly, some reliance on authoritative prior scientific consensus is the norm, either with citation or not, e.g. a paper citing "F=MA" does not in general include a formal citation to Isaac Newton, although that's implied. It is more recent or controversial work that will in general require citations, and thus reliance on a very few such works is advised by most scientists, to avoid building on a still-shifting foundation.
In the more model-driven 'soft' or 'human' sciences, where prediction and experiment and controls are less common, citation is viewed somewhat differently. Terminolog] rather than logic is the key to an effective peer review, and so citation establishes the glossary and the definitions which the reviewers should keep in mind while reading. The number of citations should still be few, as there is risk of some 'name space clash', resulting in confusion or inexact application of abstractions to concretes. This constraint tends to make papers in the soft sciences more prone to falling into a 'school of thought' and less able to stand on their own without some body of prior knowledge.
Modern scientists are sometimes judged by the number of times their work is cited by others - this is actually a key indicator of the relative importance of a work in science. Accordingly, while the scientist is motivated to have his work cited early and often and as widely as possible, all other scientists are motivated to have total numbers of citations kept low in order to retain the integrity of this means of judgment. A formal citation index tracks which referred and reviewed papers have referred which other such papers.
Disciplined citation of prior works in mathematics and science is known at least as far back as Euclid. Late in the first millennium, Islamic scholars developed their practice of isnah, or "backing", which established the validity of sayings of Muhammad in the hadith. The Asharite school of early Muslim philosophy extended this into fiqh or jurisprudence, while the Mutazilite school used the traditional methods and applied them to science. Roman Catholic practice included constant quotation of Saints or Apostles and citing incidents of their lives as moral examples.
In some form, then, achieving authority by constant citation is thus a near-universal idea among the peoples of the Mediterranean, whose educated people were exposed to one or other of these practices well before the European Renaissance and the emergence of scientific method.
In patent law the citation of previous works, or prior art, helps establish the uniqueness of the invention being described. However, the focus in this practice is to claim originality for commercial purposes, and so the author is strongly motivated to avoid citing works that cast doubt on its uniqueness. This is thus not "scientific" citation, although Baruch Lev and other advocates of accounting reform consider the number of times a patent is cited to be a significant metric of patent quality and thus of innovation.
See also: citation, citation index, peer review, prior art, scientific method, philosophy of science