From Timo Budarz:

Re: Rats can reflect on their knowledge, study finds (March 8): I imagine it might be made more clear in the full-text version of the study, but isn’t there an obvious alternative conclusion than the one mentioned? What about the rats choosing option 3 (which was represented as backing out after reflection) in cases where both signals were perceived as equal duration by the rodents? No reflection is required, only a discrimination. Long signal, choose it, no long signal (two roughly equal ones), choose option 3. I imagine at least half of your readers realized this potential flaw in the interpretation.