|This article does not cite any references or sources. (January 2007)|
The Pramāṇa-samuccaya ("A Compendium of Validities") is a work by Dignāga, the early medieval Indian Buddhist logician and epistemologist, which may be seen as the definitive statement of his epistemological work. The work comprises an outline in the highly elliptical verse format typical of early Indian philosophical texts and an explanatory auto-commentary.
Structure of the work 
Chapter 1 opens with the statement that there are only two means of knowledge: direct perception and inference. Corresponding to these we have two objects: particulars and universals.
Direct perception is knowledge which excludes conceptual thought (kalpanā). This only reveals the bare features of an object via the senses. This knowledge is inexpressible in words, relating to real objects and ultimate reality. Errors of perception arise through misinterpretations by conceptual thought. Each item of sense perception is unique. Dignāga does not specify what the nature of the object of perception is, but implies that although it is not atomic or otherwise, it is existent. It is real because it is causally efficient (artha-kriyā).
Chapter 2 deals with "inference for oneself" (sva-artha-anumāṇa). This is knowledge of what can be inferred through a middle term (liṅga), which has the three characteristics for a valid middle term, namely, that it is concomitantly present in the thesis, present in a similar example and absent from a dissimilar example. According to Dignāga, inference only deals with universals and is always dependent upon the subject/object relation.
Chapter 3 deals with "inference for other" (para-artha-anumāṇa), the process by which one makes public what one knows, by formal means, using a syllogistic means of argument. This typically takes the following form:
- Thesis: Sound is impermanent
- Reason: Because it is created
- Exemplification: Whatever is created is known to be impermanent
- Similar example: As in the case of a pot
- Dissimilar example: As not in the case of space
Chapter 4 deals with examples and how they are to be used and how to select relevant examples. In Dignāga's method of syllogistic logic, agreeing and different examples are needed to establish concomitance of the middle term.
Chapter 5 deals with the "exclusion of other" (anya-apoha). Here, Dignāga first eliminates authority as a separate valid means of knowledge, stating that it is a kind of inference. Authority is acceptable only if it does not contradict one's own perception and inference. Dignāga states that conceptual knowledge derives from words, but asks what do words mean. He proposes that words express their meaning by exclusion of opposite meanings. Words do not denote real universals, as there is no necessary connection between words and universals. Instead, words only express imaginary concepts and vice versa. In Dignāga's view, words do not produce knowledge by referring to particular objects, but only demark X from non-X. That is, the word "white" does not bring about knowledge of all white objects, but only demarks white from non-white. In this way, some form of classification is possible in the mind through this process of distinctions. This is done by the internal application of agreement and difference, so he maintains that speech derives from inference. However, he adds the proviso that this process is aided by direct perception which helps avoid fallacies. Thus, the imagined world of universals can be made to fit the real world by correcting contradictions in the light of direct perception.
After Dignāga's lifetime, Dharmakīrti compiled his Pramāṇa-vārttika, which may be seen as an extended commentary on Dignāga's Pramāṇa-samuccaya, with the purpose of correcting philosophical difficulties that had subsequently become apparent.
Anyapoha : The Context of the Discussion of the Doctrine (Apohavada) in Pramanasamuccaya : At the outset in the Pramanasamuccaya Dingnaga has stated that the main idea behind writing the work is to remove misapprehensions about the acquisition of knowledge. These misapprehensions are that there are more than two means of acquiring valid knowledge. According to him the objects of knowledge (prameya) are divided into two : the unique self-characteristic (sva-lakaksana ) and the generic class-characteristics (samanya- laksana ), and with reference to these two kinds of prameyas, two means of valid knowledge, perception (pratyaksa) and inference (anumana) are requisitioned, by the former we apprehend the svalka–a‡a aspect of things which alone is perceptible (pratyaksa) and by the latter the samanya laksana which is imperceptible (paroksa). As there can not be more than two kinds of prameyas, there can not be more than two pramanas : pratyaksa and anumana. Any belief that is not founded upon these two must be discarded as an unwarranted belief.7 The discussion on means, if any, of acquiring knowledge outside sensation and well founded reasoning is the main theme of the fifth chapter viz; apoha of Pramanasamuccaya''. The main object behind writing this chapter is to enforce upon the idea that apart from sensation and reasoning all other means of acquiring knowledge named by others, such as through 'words' can be reduced to forms of reasoning. The reduction of the validity of 'words to the validity of reasoning made this philosophy controversial. Dingnaga emphasised that a 'word' had no validity without reason. He says that "the statements of credible persons are inference insofar as they have the common character of not being false. Present in the object of inference and in what is similar to it, and absent in their absence. Di×nÈga, perhaps, was the first Buddhist philosopher to point out that this philosophy was contradictory to the traditional BrÈhma‡ical philosophers.