So far, nobody has pointed out the connection between the fuss about the selection of Labour's candidate for Falkirk and last month's stock market sell off. Let me remedy this omission.

Both pose the same puzzle: why should there be so much bother about something that's been known for ages?

Everyone's known for years that the processes of selecting parliamentary candidates sees some sharp practice; it's not clear that Unite's behaviour in Falkirk was especially underhand by these low standards. Why then should Falkirk arouse such interest when so many other Machiavellian selections did not? Similarly, everyone's known for ages that the Fed would have to scale back quantitative easing at some time. Why, then, should the markets suddenly take fright at the prospect of it doing so?

The answer to both questions lies in the same fact about human psychology - that our attention is limited. We can't pay attention to everything, so we often ignore lots of things, only to suddenly worry about them when some otherwise perhaps innocuous piece of information strikes us. Such shifts might be motivated by people with agendas, but often they are not but rather arise as emergent processes.

We can call this the Kylie's arse syndrome. Back in 2000, Ms Minogue famously appeared in gold hotpants, prompting everyone to admire her arse. But her arse was much the same as it had been in the previous 13 years. What changed was that the hotpants drew attention to it. The facts didn't change, but people's attention did.

This concept of limited and changeable attention helps explain two things.

One is the waxing and waning of media frenzies and moral panics. These often focus not upon new menaces, but ones that have been around for years, but which suddenly get noticed - perhaps because of a random bunching of events (such as with attacks by dangerous dogs or flesh-eating bugs), or because of an especially egregious example - as when Mick Philpott's murders drew attention to long-term benefit claimants. And they often die down without much change in the underlying prevalence of the phenomena being panicked about.

The other is the apparently excessive volatility of asset prices. Sushil Wadhwani, a former member of the MPC once quipped (pdf) of current account deficits, that they "appear not to matter until, well, they suddenly do!" This hints at the more general fact that the markets' attention towards the "fundamentals" can come and go violently and unpredictably. We can shift from a position in which QE tapering, banks balance sheets, external deficits or whatever don't matter to one in which they do. In this sense, as Mordecai Kurz has shown (pdf), asset price volatility arises not (just) from the external economy but from apparently small shifts in belief equilibria.

You might claim that this is evidence that markets are irrational. whether they are or not is, though, in onme sense irrelevant. Such changes in attention are unpredictable - in part because they can be triggered by tiny signals which are amplified by everyone anticipating everyone else's beliefs - and so you can't make risk-free money from trying to anticipate them.

My bigger point, though, is that the same fact about the limits of human cognition links apparently different behaviour in markets and politics.