Pragmatism is presented as anti-representationalism, a critique of the possibility of a discipline which could provide foundation for knowledge and truth. The pragmatist position is based on a functional interpretation of knowledge, according to which our assertions (scientific or not) about the world are rules for interaction with phenomena (and thus do not represent a perfect or ultimate reality). In the context of pragmatist thought, then, the problem of validation of knowledge does not belong to the field of the rules established by an epistemological discipline; it is simply a matter of conversation. The pragmatist argument is related to a functional approach to language which results in considering languagem as a form of action rather than as a system for representation of reality. Such a view of language emphasizes the arbitrary and conventional basis of every language use and points to the impossibility of a private language; it also suggests that a functional use of language cannot be taken as descriptive of a thing to which a single individual has access. Skinner's radical behaviorism is discussed in the light of the pragmatism-representationalism opposition. Skinner's conception of knowledge is considered in three different instances: knowledge in general, scientific knowledge, and knowledge of oneself. It is argued that Skinner develops a pragmatist conception of knowledge, related to a functional approach to language, when dealing with knowledge in general and scientific knowledge. In these cases knowledge is taken to be action or rules for action, and is judged according to the interaction it provides with a given enviromental set. When treating the issue of knowing oneself, however, it is suggested that Skinner adopts a different position; language is considered in terms of refference to events accessible to single individuals, and knowledge is discussed in terms of (in )accuracy of self descriptions. What makes this interpretation possible is Skinner's argument that "private reports" are not reliable, and cannot enter scientific study of human behavior, because they do not correspond accurately to what happens inside the individual's body; they are not under stimulus control of those events. Correspondence, and not functionality of those reports, is the criterion for assigning knowledge to them; a criterion that cannot be reached in terms of the empirical evidence required by a science of behavior. Skinner is then closer to representationalism than to pragmatism. Such a change is considered a result of the dualism that stands in Skinner's distinction between public and private events and his belief that there is a "private in itself", independent of language use, and which language should accurately describe in order to be reliable.O pragmatismo é apresentado como uma atitude anti-representacionista, isto é, como uma crítica à idéia de fundamentação do conhecimento ou da verdade. Esta crítica torna-se possível a partir de uma interpretação funcional para o problema do conhecimento, segundo a qual as asserções (científicas ou no) constituem-se de recursos para ínteração com os fenômenos (elas não representam, portanto, uma realidade última ou perfeita. A tese pragmatista está articulada com uma concepção igualmente funcional de linguagem, sendo esta então interpretada enquanto forma de ação, e não enquanto sistema de representação do mundo. Tal concepção de linguagem implica, ainda, a impossibilidade de uma linguagem privada. O behaviorismo radical skinneriano é discutido à luz da oposição pragmatismo-representacionismo, examinando-se suas concepções de conhecimento em geral, de conhecimento científico e de conhecimento de si mesmo. Argumenta-se que Skinner desenvolve uma concepção pragmatista de conhecimento em geral e de conhecimento científico. No que diz respeito à noção de conhecimento de si mesmo, sustenta-se que Skinner distancia-se de uma postura pragmatista e de uma concepção funcional de linguagem. Sugere-se que tal distanciamento está articulado com o dualismo que persiste em sua distinção público-privado e com a suposição de que os "relatos sobre si mesmo" são imprecisos e carecem de valor científico.