Tibetan pramāṇa theory rests first on the conviction that there are reliable ways of knowing, and it classifies these into two fundamental types of valid cognition. Direct perception is described as a fresh, non-conceptual awareness that is not mistaken with respect to its object; it arises immediately through the senses, the mental faculty, self-awareness, or yogic realization. Inference, by contrast, is conceptual and discursive: it knows what is not directly evident by relying on a valid sign or reason that is connected to the object to be established. Both forms of cognition are considered valid only when they are non-deceptive, genuinely new with respect to the knower, and endowed with certainty rather than doubt. In this way, epistemology is not an abstract exercise but a disciplined training in distinguishing clear knowing from confusion.

Within this framework, Tibetan logic gives a very precise account of how reasoning must be structured if it is to yield trustworthy knowledge. A standard syllogism has three parts: a thesis to be proven, a reason that supports it, and an example that illustrates the connection between reason and thesis. For the reason to function as a valid sign, it must satisfy three stringent conditions: it must be present in the subject under discussion, present in similar cases where the predicate is known to hold, and absent in dissimilar cases where the predicate is known not to hold. These three modes ensure both forward entailment and counter-entailment, so that the sign is inseparably linked with what it is supposed to prove. Debate training revolves around testing and refining these connections, exposing any lapse where a sign fails one of these conditions.

Debate itself becomes a living laboratory for pramāṇa. Monastics adopt the roles of challenger and defender, articulating theses, reasons, and examples, and then probing them through consequences that reveal contradictions or hidden implications. The method of consequence (prasaṅga) is particularly valued for showing that an opponent’s position leads to untenable results, thereby clearing the ground for more accurate views. Throughout this process, the classification of objects of knowledge plays a guiding role: some phenomena are manifest and can be directly perceived, others are hidden and must be known by inference, and still others are extremely hidden and can only be approached through reliable testimony. By weaving together valid perception, rigorous inference, and disciplined debate, Tibetan logic serves as a path for transforming ordinary, uncertain cognition into a more stable and liberating understanding.