I see this happen, too. I use a cheap analog glass-tube thermometer in my mash; I just drop it in there and put the lid on the cooler. Normally what I do is let it sit in the mash for 10 minutes and then check the temperature. There was only once when I was compelled to adjust the mash temperature based on this initial reading. Just about every time, after I put the lid back on and check the temperature at the end of the mash (40-50 minutes later), the reading on the thermometer is actually higher than the 10 minute reading.
I know there is no reason why the temp should go up, so I figured it was probably due to the slow rise time of the analog thermometer and/or some slow temperature distribution phenomenon. I also stir the crap out of the mash and it doesn’t seem to matter, I see the same rise every time. Now it’s part of my process to add 2 degrees to the 10 minute reading and consider that my “achieved mash temp” and it works pretty much every time.