paricCheda 1 - pratyakshasya anumAna bAdhyatvam

In the previous chapters, the opponent's arguments supporting the supremacy of perception relative to other means of knowledge were refuted. In this chapter, the siddhikAra establishes that perception is capable of being sublated by inference.
अथ प्रत्यक्षस्यानुमानबाध्यत्वम् | Now, the sublatability of perception by inferential cognition is taken up.

किञ्च विपक्षबाधकसचिवमनुमानमपि प्रत्यक्षबाधकम् | An inferential cognition that is supported by a logical argument that disproves its contradiction, is capable of overruling perception.

The opponent objects to this by arguing that inference is inherently incapable of overruling perception, and if anyone insisted that it was possible, it would contradict the rules of scriptural interpretation laid down in mImAmsa.
ननु - एवमपि 'औदुम्बरीं स्पृष्ट्वा उद्गायेत्', 'ऐन्द्र्या गार्हपत्यमुपतिष्ठते', 'शरमयं बर्हिर्भवतीति' श्रुतित्रयग्राहि प्रत्यक्षं यथाक्रमं 'औदुम्बरी सर्वा वेष्टयितव्ये' ति स्मृतिरूपेण सर्ववेष्टनश्रुत्यनुमानेन, 'कदाचन स्तरीरसि नेन्द्र सश्चसि दाशुष' इति मन्त्रसामर्थ्यलक्षणेनेन्द्रशेषत्वश्रुत्यनुमानेन, चोदनालिङ्गरूपेण कुशश्रुत्यनुमानेन च बाध्येत इति सर्वमीमांसोन्मृदिता स्यादिति

He says:
ननु - एवमपि...इति सर्वमीमांसोन्मृदिता स्यात् (Holding that inference can overrule perception) would contradict the rules of exegesis established by mImAmsa. Why?

'औदुम्बरीं स्पृष्ट्वा उद्गायेत्','ऐन्द्र्या गार्हपत्यमुपतिष्ठते', 'शरमयं बर्हिर्भवतीति' श्रुतित्रयग्राहि प्रत्यक्षं
In the following three instances, the directly perceived shruti would end up being overruled by inference:
1) "(the sAma veda priest) should recite while touching the post made from a fig tree",
2) "the gArhapatya fire is to be worshipped with the aindri rik",
3) "spread the shara-grass (reed) as the ceremonial seat, barhis".

यथाक्रमं 'औदुम्बरी सर्वा वेष्टयितव्ये' ति स्मृतिरूपेण सर्ववेष्टनश्रुत्यनुमानेन, 'कदाचन स्तरीरसि नेन्द्र सश्चसि दाशुष' इति मन्त्रसामर्थ्यलक्षणेनेन्द्रशेषत्वश्रुत्यनुमानेन, चोदनालिङ्गरूपेण कुशश्रुत्यनुमानेन च बाध्येत
1) The first shruti would be overruled by the smRti which says "the fig post is to be fully covered with cloth". This smRti leads one to conclude that the fig post is to be fully covered up. If the smRti overruled the shruti, the fig post would need to be covered, and the priest will not be able to directly touch the post while reciting the sAma. However, according to siddhAnta, the shruti prevails.

2) The second shruti would be overruled by the inference that since the name of the rig mantra is aindri, and the word Indra occurs in the mantra 'Oh Indra, you do not deceive those that make offerings to you", it is Indra who is the subject of the mantra.  If that inference prevailed, it would contradict the shruti injunction, which said that it is the gArhapatya agni who is worshipped with the aindri mantra. The inference presupposes that there is an inferred shruti injunction which establishes that Indra is the object of the aindri mantra. However, it is accepted that the actual shruti is stronger than an inferred shruti.

If inference has the ability to overrule perception, the basis upon which the mImAmsaka postulates the relative strength of the actual shruti to the inferred shruti (which the advaitin agrees with) would be defeated.

3) The third shruti would be overruled by the inference that this particular injunction is similar to the general injunction (codanA) in the prakRti (archetypal or model) sacrifice, and therefore would lead one to conclude that it is kusha (darbha grass) that is to be spread as the ceremonial seat, not the shara.

The paurNamAsa sacrifice (prakRti) says that kusha-grass is to be spread as the ceremonial seat, but the somAraudra-caru sacrifice (vikRti) specifically stipulates that shara-grass is to be used. There is a general rule that the vikRti is to be performed like the prakRti, but if there is a specific injunction in the vikRti that differs from the prakRti, the vikRti injunction prevails. If perception could be overruled by inference, the somAraudra injunction which is directly received (heard), would be overruled by the inference that one ought to apply the paurNamAsa injunction instead.

The opponent thus argues that if inference was capable of overruling perception, the rules of ritual exegesis, as laid out in pUrva mImAmsa and accepted by the advaitin, would be contradicted.

The siddhikAra responds to this argument:
- चेन्न; वैषम्यात्, No, for the examples are not applicable in this context.

तथाहि - किमिदमापद्यते, श्रुतित्रयग्राहिप्रत्यक्षमनुमानैर्बाध्यतेति वा, प्रत्यक्षविषयीभूतश्रुतित्रयमिति वा |
What is meant when it is said that "the three directly perceived shruti statements are contradicted by inference"?

a) Is it - "perception, which reveals the three shruti statements, is contradicted by inference"?, or

b) Is it - "the three shruti statements, which are revealed by perception, are contradicted by inference"?

नाद्य: विरोधाभावेन तद्बाध्यबाधकभावस्य शास्त्रार्थत्वाभावात्, It is certainly not the first. As the contradiction of the existence of the shruti-s is not intended by smRti / inference, the denial of the perception of shruti is not the intended meaning of smRti.

अस्माभिरनभ्युपगमाच्च, अनुक्तोपलम्भमात्रत्वे निरनुयोज्यानुयोगापत्ते: |  Nor do we claim that to be the case. By criticising us for something not said by us, you are censuring the uncensurable (niranuyojya anuyoga - one of the 22 nigraha sthAna-s, points of failure in debate, as outlined in nyAya).

अतएव न द्वितीय:, प्रत्यक्षविषयीभूतश्रुतित्रयस्य लिङ्गबाधकत्वपरेऽपि शास्त्रे प्रत्यक्षस्य लिङ्गबाध्यत्वे विरोधाभावात्, Therefore, the second option is also ruled out. Even if it was argued by reductio ad absurdum that the siddhAntin's view would lead to the eventuality that inference would end up contradicting a shruti known through perception, (and as a result, it is argued that the original presumption is wrong), there is no scriptural contradiction in holding that inference overrules perception.

The opponent wants to argue that as inference cannot overrule scripture (which is the object of perception), inference cannot overrule perception. However, even if scripture was known through perception, the inability of inference to overrule scripture does not preclude the ability of inference to overrule perception. The denial of one does not imply the denial of the other because:
नहि शब्दप्रत्यक्षयोरैक्यमस्ति; शब्दस्य च सर्वप्रमाणापेक्षया बलवत्त्वमवोचाम | Scripture and perception are not identical. In fact, we hold that scripture is the supreme means of knowledge, stronger than every other.

तस्मान्मौढ्यमात्रमेतन्मीमांसाविरोधोद्भावनम् | Therefore, it is only out of confusion that one can proclaim that this would lead to the contradiction of the rules of ritual exegesis as laid out in pUrvamImAmsa.

ननु - प्रत्यक्षस्य लिङ्गबाध्यत्वे वह्न्यौष्ण्यप्रत्यक्षं शैत्यानुमानस्यात्मस्थायित्वप्रत्यभिज्ञानं च क्षणिकत्वानुमानस्य बाधकं न स्यात्,
The opponent tries another tack - If perception is capable of being overruled by inference, the following two arguments would end up being contradicted:
1) It is held that the direct perception of the heat of fire, outweighs any spurious arguments made about fire being cold.
2) It is held that the universal direct experience of the same "I" being present from one instant to the next, disproves the Buddhist argument for the momentariness of consciousness
प्रत्युतानुमानमेव तयोर्बाधकं स्यात् Rather, the advaitin's position would lead to such spurious inferences overruling facts established by direct perception.

The siddhikAra responds:
इति चेन्न; अर्थक्रियासंवादेन श्रुत्यनुग्रहेण च तत्र प्रत्यक्षयो: प्राबल्येनानुमानबाधकत्वात् | No, in each of the two examples cited, perception is supported by validation in performance (fire's heat leads to specific measurable outcomes which validates the experience of heat) and shruti (which says that the self that is recognised as the object of the "I" cognition is not momentary). As a result, such a validated perception overrules inference.

अपरीक्षितप्रत्यक्षं हि परीक्षितानुमानापेक्षया दुर्बलं, 'नीलं नभ' इति प्रत्यक्षमिव नभोनीरूपत्वानुमानापेक्षया, अतो न सामान्यतो दृष्टमात्रेण सर्वसङ्करापत्ति: | Whereas, it is an unvalidated perception that is said to be weaker than validated inference. Like the perception of blueness in the sky is said to be weaker than the logical argument that sky is colourless, because the latter has various supporting arguments that validate it. Therefore, we cannot stipulate a rule of universal strength of perception based on its application in a few examples. 

The opponent argues against the rule that a validated inference can overrule an invalidated perception.
नन्वेवं - पशुत्वेन शृङ्गानुमानमपि स्यात् ; लाघवात् पशुत्वमेव शृङ्गवत्त्वे तन्त्रं,
By this much, you could argue that all animals have horns, because it is more parsimonious to say that being an animal is sufficient to imply being endowed with a horn. 
नतु तद्विशेषगोत्वादिकम्; अननुगतत्वेन गौरवात् इत्येतत्तर्कसघ्रीचीनत्वेन प्रत्यक्षापेक्षया प्राबल्यात्, अनुकूलतर्कसाचिव्यमेव हि अनुमाने बलम् | You would be unable to qualify your inference to say that it is a particular kind of animal, ie "cow", etc., that has horns, because the genus of animals is more pervasive than the cow species. Thus, one could say that such an inference, which is supported by the principle of parsimony, is stronger than perception, for the strength of inference is dependent on some supportive logic (such as this).
एवंच येनकेनचित् सामान्यधर्मेण सर्वत्र यत्किञ्चिदनुमेयम् |Thereby, any inference in the world can be argued to be stronger than perception, based on the application of some generic attribute
लाघवतर्कसाचिव्यस्य सत्त्वात्, तावतैव प्रत्यक्षबाधकत्वादिति व्यावहारिक्यापि व्यवस्था न स्यात्,  for every inference can be validated citing support from the principle of parsimony. As a consequence, there would be no room for the empirical invalidation of any inference on account of it being contradictory to perception.
नह्यत्र प्रत्यक्षबाधादन्यो दोषोऽस्ति Because, apart from the one defect of being contradictory to perception, there is no other defect in any of the inferences mentioned previously. 

Therefore, one must necessarily accept that inference can be invalidated, merely on the basis of it being contradictory to perception.

The siddhikAra responds:
 - इति चेन्न; अयोग्यशृङ्गादिसाधने प्रत्यक्षबाधस्यासंभवेन No, because if the inference seeks to prove the existence of an imperceptible horn, even "being contradictory to perception" cannot be an invalidating defect.

तत्र व्याप्तिग्राहकतर्केष्वाभासत्वस्य त्वयाऽपि वक्तव्यत्वेन Therefore, in such an instance, even you will have to admit that the failure of the inference is not because it is contradicted by perception, but because there is no example to prove the concomitance "every animal has a horn, whether perceptible or otherwise", and it is this lack of a valid example that invalidates it, and not because it is contradictory to perception. 

व्यवस्थाया उभयसमाधेयत्वात्, न हि तर्काभाससध्रीचीनुमानम् प्रमाणमिति केनाप्यभ्युपेयते ; This accommodation being something that has to be necessarily accepted by both of us, no one can claim that an inference where the concomitance is flawed is a valid cognition. 

अत उपपन्नं सत्तर्कसचिवमनुमानं प्रत्यक्षस्य बाधकमिति Therefore, it is quite appropriate to say that the inference which is supported by a valid argument is the one capable of overruling perception.

|| इति प्रत्यक्षस्यानुमानबाध्यत्वसिद्धि: || Thus, the invalidation of perception by inference has been established.