It may be that I'm missing the point (I haven't had my daily ration of chocolate yet), but what's wrong with this way of looking at it? Conscious or not, a thought can't be its own cause -- at least, not given our usual assumptions. That's what I take your premise 1 to entail. And it seems right: I can't be conscious of a particular thought X before I think it. (It may be, for all that, that I can think it before I'm conscious of it.) But why can't my conscious thought X be the cause of a later conscious thought Y ? For example, my thinking now about a jigger of gin might cause me to think, a moment later, that there's a bottle of gin in the freezer. Or maybe the issue is this: if I'm going to consciously cause a thought about Vienna, the content of that thought must already be part of the thought that does the causing. In that case, I'm conscious of the thought before I've thought it, contrary to premise 2. But then premise 2 ends up suspect, doesn't it? Couldn't I be sitting here...

You're right: we shouldn't throw the word "causes" around too casually. Let's fix on depression as our example, and let's keep in mind that simply being sad isn't the same as being clinically depressed. On the one hand, neurochemicals probably aren't just symptoms of depression; they probably have something to do with causing the symptoms -- the listlessness or anxiety, or excessive rumination or protacted feelings of sadness. Perhaps it would be more accurate to say that clinical depression is , at bottom, a malfunction in the neurochemical system, though this may be too reductionistic, and it also may turn out not to get the biology right. But perhaps what you're pointing to is that it still makes sense to ask what causes this malfunction in the first place. That's obviously a very good question. My impression is that sometimes life circumstances can trigger depression, but sometimes there's no clear external cause. The right answer here is likely to be very complicated. At the moment, far as I...

An intriguing puzzle. The first point is that insofar as it's a question about how our minds actually work, it's an empirical matter, and the answer depends on the facts. But there's a design-level issue here (which I'm hoping my better-informed colleagues might chime in on.) Suppose we have a complicated program that's broken up into sub-programs, or modules. And suppose that there is one module whose job it is to monitor what's going on in various other modules and stop them if they appear to be running amok. Perhaps, for example, this monitor module will kill a process if it has cycled through a million iterations without halting. You no doubt get the idea (and may well have thought of it yourself.) If a system is modular enough, and if it has enough safeguards, redundancies, monitoring modules and so on built in, then the chance that it will just go nuts might be small. And so if the mind is essentially a computer, it may be that millions of years of evolution have built it in this sort of...