There's a very interesting item on 'hybrid open access journals' in the ever-useful newsletter on open access publishing produced by Peter Suber. The hybrid OA journal is one that provides open access on the payment of a fee by the author (or someone whom the author persuades to pay!). As Peter says, for the publishers it is a win-win situation - they still have their subscription income, so if the take-up is small (as I suspect it will be), they don't have a problem and if the take-up is large they are being paid twice (once by subscription, once by the author), unless they reduce their subscription rates along with the increased take-up by authors - NOT a likely scenario in my opinion.
Personally, I regard these moves by publishers as nothing more than window-dressing, a PR move against the OA movement, which, eventually, will amount to very little in actual access to the research literature. Pressures from funding agencies and, ultimately, governments (which have so far shown little interest in pitching against the publishers) will demand true OA, not some hybrid, which I don't regards as being OA under any proper definition.