When Descartes, in his moment of greatest doubt, consoled himself with the insight that, so long as he experiences himself thinking, which he cannot but do, then at least the fact of his existence cannot be doubted - "I think, therefore I am," in its summative form - he went a giant leap too far. Kant, a century and a half or so later, saw his precursor's mistake: we have no right to conclude the existence of that I from the experience of ourselves in thought, because the I that is concluded to exist is not identical with the I that, in any given moment, thinks. The I that Descartes proposed to exist, the I of "I am," is, as Kant called it, a transcendental I, that is, an I that unites our various moments of thinking, of wishing, doing, believing, intending and so on, in a higher, coherent, unified, self. The I of "I think," is, for its part, what Kant would call an empirical I, that is, an I in the world, now, operating for the moment, subject to contingency, consisting only of the particular thought it is having or the act it is doing, with no expectation of continuation attached to it, certainly no cohesiveness with another moment of thought or action guaranteed. Two different Is: therein Descartes' error in assuming that, because one functions at this moment, the other must exist.
Only so far did Kant depart from the insights of his predecessor, however; for, according to Kant, although we cannot, as Descartes thought we could, be certain that our transcendental I, a unified self in which our various moments of thought and action are gathered together, actually exists, we must continue to think and act as if it exists. In other words, for Kant human being in the world can only realise its potential (in fact, can only make sense of itself) insofar as it assumes that its individual actions, its various thoughts, its beliefs, desires, hopes and dreams, are those of a self that transcends them all and wraps them in its grand, unifying arms. Without the assumption of a transcendental self, empirical thought and action is degraded, inhuman.
But the practice of believing, of faith, which allowed Kant to continue to think and act (and continue to expect others to think and act) as if something exists that we cannot know to exist, has since waned (and Kant is, in many ways, part of the drive for "enlightenment" that has been the great cause of this waning), so that, in Britain at least, the ability to imagine a trancendental self, let alone to think and act as if it exists, is largely in abeyance. Coming from a Catholic country in which the transcendental self is written into every nook and cranny of socio-cultural existence - the experience of conscience, and the guilt with which this experience is famously associated, is impossible unless one regards individual instances of one's thinking and acting in the context of a larger whole to which they contribute and which, in many cases, they function to corrupt and degrade; think of it as Dorian Gray's portrait, which operates as the transcendental self to the empirical actions that are unleashed upon an unfortunate world - what strikes one in Britain is, most immediately, people's lack, in the main, of self-doubt, of self-consciousness, of self-awareness, of self-knowledge. Coming from a culture in which thoughts and actions are almost impossible to experience without the attendant judgment of those thoughts and actions as contributors to the self that is ultimately to appear at the gates for judgment (and whether or not one actually holds to the belief system that encourages this, its effect is now endemic), it is extraordinary to see before one's eyes people (that is, what one takes to be transcendental Is) apparently unattached to what they have just said or done, to the extent that something said in anger or error, even at those times when the anger or error is identified and apologised for (this is, after all, a good country for politeness), appears to have no effect on the person who said it. It is as if what was said has no relation to the sayer, as if she is as unresponsible for it as the person to whom it was said, such that the identification of it as, for instance, unfair or hurtful, is somehow an event in which neither of them really participates.
Disconcerting and all as this is in the moment, it is even more confusing taken as a general condition. For what is absent from a society of empirical Is, in which there is no capacity for belief in trancendental Is, is any real ability to reflect on one's behaviour, to alter it, to learn from it. If one changes, it is only as an animal changes, because one has been beaten again and again, or rewarded again and again; empirical Is respond to circumstances - responses to circumstances is, in fact, all they are - but those circumstances cannot be reflected upon but only reacted to. There is, in a society of empirical Is, so little capacity for self-analysis, that the idea of reasoning with someone, of trying to make them know you (never mind making them know themselves) is utterly misplaced; all one can do is to talk about the weather, or some other standard topic, and hope that, today, in this place at this time, one encounters an I that's not too hot to handle.
Kant was right: we cannot be sure that any such thing as a transcendental I actually exists. Philosophers since Kant are right too: the heuristic benefits of assuming that such an I exists are not as certain as Kant thought they were, and we would do well to loosen our expectation that we, or others round us, must think and act - and understand our and their thoughts and actions - as if they must always cohere. But from the loss entirely of the capacity for reflection, for knowing ourselves and others, that practice at belief in a transcendental self or selves affords, emerges a situation of, as Kant would describe it, deep immaturity, a world of almost literally headless chickens.