________________
THEORY OF KNOWLEDGE
113
need separate treatment :
(1) Thus in Dharmakīrti's view a piece of cognition has impressed upon itself the form of its object; in other words, identity of form (sārūpya) is the relation that obtains between a piece of cognition and its object. In the case of sensory experience, the object is active on its part and so it is this object that is supposed to impress its own form on the corresponding piece of cognition; but in the case of thought this much alone can be said that the cognition concerned has somehow got impressed upon itself the form of its object. Here Dharmakīrti has polemised at length against the philosophers who maintain that a piece of cognition is devoid of form in the sense that it has got impressed upon itself no form of its object. His point is that one piece of cognition cannot be distinguished from another unless two happen to bear the form of their respective objects.!' Dharmakīrti specially emphasises that on the rival view all cases of memory should turn out to be identical; for, so runs his arugment, a piece of memory cognises a past cognition, but if one piece of past cognition does not differ from another, one piece of memory too should not differ from another.20
(2) Then Dharmakīrti is of the view that a piece of cognition is necessarily self-cognitive, and this is how he argues his case : 'All our dealing with the things of the world is based on our cognition of these things, but our cognition cannot play this role if it itself remains uncognised. And it will not do to say that one piece of cognition is cognised by a subsequent piece of cognition, for this subsequent piece of cognition too cannot play its role if it itself remains uncognised. So one is bound to face the contingency of an infinite regress unless one concedes that a piece of cognition is necessarily self-cognitive,21 Here too Dharmakīrti has polemised at length against the rival philosophers and here too he has made special reference to the phenomenon of memory. Thus, he points out that the cognition of a word takes place by way of recalling the earlier heard cognitions of the different letters that go to constitute this word, but that no such recall can take place unless the cognitions in question were cognised at the same time they took place; but, so runs his argument, if each of these cognitions is cognised not by itself but by another cognition, then what should take place, though it never does take place, is that the cognition of the first letter of the word is followed by the cognition of this cognition; then follows the cognition of the second letter to be followed by the cognition of this cognition, and so on and so forth.22
