Knowledge as such is always an immediate awareness, but of what? (Guidelines p. 220)Once again, there seems to be a confusion here between knowledge (pratyaksa, presumably, also from what follows) and consciousness.
He goes on to outline Prabhakara's tri-puti-pratyaksa-vada. For P, knowledge is not eternal but is a self-luminous (svaprakasa) event wich arises and vanishes as a moment in the cognitive process. [Very good.] It is always a triple revelation: it reveals itself (svasamvitti of jnana); the knowable as its object (visaya-vitti of jneya); and the knower as 'I' (aham-vitti of jnatr). Thus it has simultaneously three terms.
The note that follows is interesting:
Note that the self is not self-luminous but known only in cognition and not in its absence as in deep sleep. Dream cognition, though derived only from memory, is of the same general type. To hold, like the N. Vaisesikas, that knowledge is known through another cognition would lead to infinite regress... (Guidelines 220)K. Bhatta's theory is called jnatata-vada, because he holds, in contrast to P., that the only immediate datum in cognition is the cognizedness (jnatata) or illuminedness (prakatya) of the object., i.e. the object as cognized. Both the act/process of knowing and the knowing ego have to be inferred.
De Smet's judgment:
Prabhakara's theory seems to be closer to the fact. But he fails to distinguish the implicit (awareness of the ego and the knower) and the explicit (awareness of the object as known) in the immediate content of cognition which he erroneously considers as totally explicit. Hence, Bhatta wants to correct him but for the same reason he reduces the immediate to the explicit and thinks that the implicit is not immediate but mediate and inferable only. (Guidelines p. 221)