Goldman's early book, A Theory of Human Action (a revised version of his Ph.D. thesis), presents a systematic way of classifying and relating the many actions we perform at any time. Its influence was broad and can be found in, among other writings, John Rawls' book A Theory of Justice. Goldman's early work in action theory soon gave way to work in other branches of philosophy, most influentially epistemology.
Goldman's accounts of knowledge and justified belief, using notions like causation and reliability instead of normative concepts like permissibility and obligation, contributed to a philosophical approach that came to be known in the 1970s as naturalized epistemology. The first of these emerged as part of the efforts in the 1960s to find a "fourth" condition in response to the Gettier challenge to the account of knowledge as "justified true belief." In his 1967 paper, "A Causal Theory of Knowing", Goldman proposed that knowledge amounts to true belief appropriately caused by the fact that makes it true. Later, he claimed knowledge amounts to true belief that is produced by a reliable process.
More recently, Goldman has focused his epistemological efforts to questions of social epistemology and has applied his approach to epistemology to such issues as the law (especially evidence), voting and media. He attempts to provide (in his words) a less radical view of social epistemology than those suggested by cultural theorists and postmodernists under that name. His approach uses tools of analytic philosophy especially formal epistemology to analyze problems in social knowledge. Some of this work is summarized in his book Knowledge in a Social World.
Goldman has devoted significant time to showing how research in cognitive science is relevant to a variety of branches of philosophy including epistemology. Much of this work appears in his books Epistemology and Cognition, Philosophical Applications of Cognitive Science, and Simulating Minds.