Rene Descartes' basic position can be summarized in the following manner:
i) I can only know (or be certain) as to whether or not some statement or proposition (p) is true if I already know (am certain of) that God exists and is not a deceiver.
ii) I can only know (or be certain) that (i) God exists and is not a deceiver if I already know (am certain of) that every proposition I clearly and distinctly perceive is true.
That encapsulates the "Cartesian circle".
In other words, i) depends on ii) and ii) depends on i). Neither i) nor ii) can get off the ground without the prior existence of i) or ii). Or, in other words, i) can't get off the ground without ii) and ii) can't get off the ground without i). It seems to follow, then, that neither can get off the ground.
In more detail, one would need to ask why Descartes has already assumed God's existence, let alone the fact that He's not a deceiver. It could be said that written into God's nature and existence, as it were, is the fact that He can't be a deceiver. But since Cartesian Doubt is supposed to begin at rock bottom, Descartes could hardly have assumed any of these things.
In terms of ii), we would need to know what Descartes mean by 'clear' and 'distinct'. How does clarity and distinctness guarantee truth? How did he know that he clearly and distinctly perceived certain propositions? And what did he mean by “perceive” here? (He wasn't talking here about external objects such as trees or chairs.)
Descartes, or at least philosophers defending Descartes, have said that metaphysical or logical certainly isn't required at the beginning of Cartesian Doubt. Or, more correctly, metaphysical or logical certainty isn't required to take one to the existence of God. Instead only psychological certainty is required: presumably to take you to God and be certain about the aforesaid propositions.
From psychological certainly you can reach metaphysical or logical certainty about both God's existence and the truth of clear and distinct propositions. This, of course, leaves out entirely what the nature of this psychological certainty is and whether or not it's meaty enough to take us to both God's existence and nature and the truth of these propositions.
It also leaves out the Cartesian circle: why talk about taking oneself to God's existence at all when God hasn't been proven in the first place? Surely even the assumption of God's existence can't be accepted in Cartesian Doubt. Yet here we are assuming God's existence in the very process of arguing about psychological certainly taking us to the existence of God. That means that God's existence is assumed not only before metaphysical certainty: it's also assumed before psychological certainty (whatever that is).
Nonetheless, philosophers have defined what metaphysical certainty is. This is one definition:
A proposition P is metaphysically certain if and only if there is no other proposition R that is a reason for doubting P.
Firstly, this definition is completely psychological in nature. The Cartesian wouldn't deny that. What I mean by that is that there could be a proposition R (or even many propositions) that would give S reasons to doubt P if he were aware of them. However, S isn't aware of them. In addition, R could be a logical reason to doubt P even though S doesn't take R as a logical reason to doubt P. Nonetheless, taken non-psychologically, R is logically a contradiction of P which S may not be aware of at all.
Indeed, in the post-Cartesian world, the very locution “metaphysically certain” has a strange psychological ring to it. Some philosophers or metaphysicians would ask: What has certainty (psychological or otherwise) got to do with metaphysical reality? In addition, R could contradict P without having any metaphysical reality as such; just as contradictions can be formulated in logical systems from symbols or formula which don't have any (metaphysical or empirical) content.
However, none of that is strictly relevant if all we're talking about is the Cartesian's move from psychological certainty to metaphysical certainty because that metaphysical certainty can be deemed to be just as psychological as the prior psychological certainty. That is, the metaphysical certainty that is derived from the psychological certainty is done so in an equally psychological manner – in what other manner could it be derived?
In an case, most of that is a sidetrack.
The Cartesian move is that, yes, we begin with only psychological certainty as to various clear and distinct propositions. The first thing we do with these psychologically certain propositions is prove that God exists. Then, with God's help, as it were, they become metaphysically certain as well. The question here, however, is that even if we accept psychological certainty, how is it, exactly, that such certainty “proves” God's existence? After all, psychological certainty alone can't prove God's existence or anything else for that matter. Only valid and true arguments can do that.
Then the Cartesian seems to move in another circle within the general Cartesian Circle. By proving God's existence from psychologically certain and therefore true propositions, we're giving a reason, apparently, for doubting all possible reasons for doubting those very propositions which prove God's existence. The propositions which took us to God are then blessed by God and therefore become certain. Thus:
God + psychological certainty = metaphysical certainty.
This could be seen as a virtuous, rather than a vicious, circle. Certain propositions lead us to God and God leads us the those very same propositions. And that, surely, is a classic case of the Cartesian Circle. I simply can't see how anyone has broken out of this very vicious Cartesian Circle. All that's has happened is we've expressed and codified it, not solved it.
Perhaps it all hinges on how psychologically certain propositions (if that's what they truly are) somehow “prove” God's existence. Clearly, simple psychological certainty of one or a hundred propositions can't prove anything, as I said. It all depends on what these propositions are. It also depends on whether or not these propositions assume the existence of God as well as the ability of God's existence to somehow give metaphysical legitimacy to the prior psychologically-certain propositions.
Another part of this Cartesian Circle is that reasons for doubting P also have to be certain. That is, not-P, or R, also has to be clearly and distinctly perceived. Clearly, if not-P is clearly and distinctly perceived to be true, then P is false, according to the Cartesian. This seems to imply that not-P or R is not clearly and distinctly perceived to be true. Therefore P must be true. Again, all this depends on what P is and what not-P (or R) is. In addition, the Cartesian requires more than one P and must defend itself against more than one not-P: there will be many not-Ps.
Of course one not-P is that God is a deceiver. Can I clearly and distinctly perceive that God is a deceiver? Not according to the Cartesian. (Really? Even if I'm an atheist?) That can't be true: there are many reasons for doubting clear and distinct propositions. One reason is their very existence as adumbrated by Descartes himself. Another thing we can doubt is that psychological certainty leads to metaphysical certainty, God's existence or anywhere else (as I have said).
Again, the Cartesian only accepts any not-P if it is certain. Alternatively, the Cartesian must be certain that not-not-P is certain (or -–P). Here again, and not surprisingly, it's all about psychological certainty, even if that can lead us to all sorts of other things. And, again, the only not-P that Descartes appears to consider is that God is a deceiver.