Did an estimate a couple of years ago that the brains estimated information processinh and the cost of 'erasing' the same amount of information under the assump

Did an estimate a couple of years ago that the brains estimated information processinh and the cost of 'erasing' the same amount of information under the assumption that all processed information is lost, it seemed to imply that the cost of forgetting is somewhere in the range 1/1000 of the amount of energy the actual information processing is thought to consume ( approx 8W ) It seems crazy that it can be that close given the miniscule energy information carries, to me those ~8mW is completely mind boggling big. The most reasonable explanation is that my calculation was wrong, but at the time it did not appear to be the case, although it probablys is anyway :)