Carl explains this:
After death, brains that do not simply disappear sometimes get smaller. In this particular fish, Sibyrhynchus denisoni the brain must have gotten a lot smaller. Check out this image, in which the braincase is in red, and the brain is in yellow. (The scale bar is 5 millimeters.)
The subject is a paper in PNAS that's available to journalists but no one else so far, yet not still embargoed ... a policy I sort of like for selfish reasons but still can't figure out. This means this link to the paper won't work for a few more days. Paper is on a 3M-year-old fossilized brain, the oldest yet found, and Carl first wrote it up yesterday in a nice post in which he included this flashy movie showing the skull as revealed by synchrotron holotomography, which is a sort of holographic imaging technique that, as you can see, produces some great images:
Carl's post today, which runs the less-flashy figures at top of this post. addresses more directly the shrinkage issue,. Though brief it's quite intriguing, as some argue that the size mismatch was all shrinkage, but one scientist raises the possibility that the fish's brain never really filled the cavity. See the comments for the first post for more on that.)
And I like his take-home:
Why there would be such a mismatch between brain size and braincase size the scientists can't yet say. Which is all the more reason, as Pierre Kerner points out in the comments, for paleontologists to start opening museum drawers and do some more scanning. Let's see what pattern emerges.
There's a lot of data in specimen cabinets around the world, waiting to be revealed with judicious use cutting-edge analysis like this imaging. So go to it.
Which is to say -- and I'm not sure how to say this politely: There's not nearly enough opening of old drawers in science these days.