Analogy (_upamana_) is accepted by Mima@msa in a sense which is different from that in which Nyaya took it. The man who has seen a cow (_go_) goes to the forest and sees a wild ox (_gavaya_), and apprehends the similarity of the gavaya with the _go,_ and then cognizes the similarity of the _go_ (which is not within the limits of his perception then) with the _gavaya._ The cognition of this similarity of the _gavaya_ in the _go,_ as it follows directly from the perception of the similarity of the _go_ in the _gavaya,_ is called upamana (analogy). It is regarded as a separate prama@na, because by it we can apprehend the similarity existing in a thing which is not perceived at the moment. It is not mere remembrance, for at the time the _go_ was seen the _gavaya_ was not seen, and hence the similarity also was not seen, and what was not seen could not be remembered. The difference of Prabhakara and Kumarila on this point is that while the latter regards similarity as only a quality consisting in the fact of more than one object having the same set of qualities, the former regards it as a distinct category.
_Arthapatti_ (implication) is a new prama@na which is admitted by the Mima@msa. Thus when we know that a person Devadatta is alive and perceive that he is not in the house, we cannot reconcile these two facts, viz. his remaining alive and his not being in the house without presuming his existence somewhere outside the house, and this method of cognizing the existence of Devadatta outside the house is called _arthapatti_ (presumption or implication).
The exact psychological analysis of the mind in this arthapatti cognition is a matter on which Prabhakara and Kumarila disagree. Prabhakara holds that when a man knows that Devadatta habitually resides in his house but yet does not find him there, his knowledge that Devadatta is living (though acquired previously by some other means of proof) is made doubtful, and the cause of this doubt is that he does not find Devadatta at his house. The absence of Devadatta from the house is not the cause of implication, but it throws into doubt the very existence of Devadatta, and thus forces us to imagine that Devadatta must remain somewhere outside. That can only be found by implication, without the hypothesis of which the doubt cannot be removed. The mere absence of Devadatta from the house is not enough for making the presumption that he is outside the house, for he might also be dead. But I know that Devadatta was living and also that he was not at home; this perception of his absence from home creates a doubt as regards my first knowledge that he is living, and it is for the removal of this doubt that there creeps in the presumption that he must be living somewhere else. The perception of the absence of Devadatta through the intermediate link of a doubt passes into the notion of a presumption that he must then remain somewhere else. In inference there is no element of doubt, for it is only when the smoke is perceived to exist beyond the least element of doubt that the inference of the fire is possible, but in presumption the perceived non-existence in the house leads to the presumption of an external existence only when it has thrown the fact of the man's being alive into doubt and uncertainty [Footnote ref 548].
Kumarila however objects to this explanation of Prabhakara, and says that if the fact that Devadatta is living is made doubtful by the absence of Devadatta at his house, then the doubt may as well be removed by the supposition that Devadatta is dead, for it does not follow that the doubt with regard to the life of Devadatta should necessarily be resolved by the supposition of his being outside the house. Doubt can only be removed when the cause or the root of doubt is removed, and it does not follow that because Devadatta is not in the house therefore he is living. If it was already known that Devadatta was living and his absence from the house creates the doubt, how then can the very fact which created the doubt remove the doubt? The cause of doubt cannot be the cause of its removal too. The real procedure of the presumption is quite the other way. The doubt about the life of Devadatta being removed by previous knowledge or by some other means, we may presume that he must be outside the house when he is found absent from the house. So there cannot be any doubt about the life of Devadatta. It is the certainty of his life associated with the perception of his absence from the house that leads us to the presumption of his external existence. There is an opposition between the life of Devadatta and his absence from the house, and the mind cannot come to rest without the presumption of his external existence. The mind oscillates between two contradictory poles both of which it accepts butcannot reconcile, and as a result of that finds an outlet and a reconciliation in the presumption that the existence of Devadatta must be found outside the house.
Well then, if that be so, inference may as well be interpreted as presumption. For if we say that we know that wherever there is smoke there is fire, and then perceive that there is smoke in the hill, but no fire, then the existence of the smoke becomes irreconcilable, or the universal proposition of the concomitance of smoke with fire becomes false, and hence the presumption that there is fire in the hill. This would have been all right if the universal concomitance of smoke with fire could be known otherwise than by inference. But this is not so, for the concomitance was seen only in individual cases, and from that came the inference that wherever there is smoke there is fire. It cannot be said that the concomitance perceived in individual cases suffered any contradiction without the presumption of the universal proposition (wherever there is smoke there is fire); thus arthapatti is of no avail here and inference has to be accepted. Now when it is proved that there are cases where the purpose of inference cannot be served by arthapatti, the validity of inference as a means of proof becomes established. That being done we admit that the knowledge of the fire in the hill may come to us either by inference or by arthapatti.
So inference also cannot serve the purpose of arthapatti, for in inference also it is the hetu (reason) which is known first, and later on from that the sadhya (what is to be proved); both of them however cannot be apprehended at the same moment, and it is exactly this that distinguishes arthapatti from anumana. For arthapatti takes place where, without the presumption of Devadatta's external existence, the absence from the house of Devadatta who is living cannot be comprehended. If Devadatta is living he must exist inside or outside the house. The mind cannot swallow a contradiction, and hence without presuming the external existence of Devadatta even the perceived non-existence cannot be comprehended. It is thus that the contradiction is resolved by presuming his existence outside the house. Arthapatti is thus the result of arthanupapatti or the contradiction of the present perception with a previously acquired certain knowledge.
It is by this arthapattiprama@na that we have to admit that there is a special potency in seeds by which they produce the shoots, and that a special potency is believed to exist in sacrifices by which these can lead the sacrificer to Heaven or some such beneficent state of existence.