It's also a perfectly valid piece of Bayesian reasoning.

It isn't if it is made as a logical deduction - as it was presented in the quote and also how it is frequently implemented in practice. The valid Bayesian reasoning would be a well calibrated calculation of the probability that the speaker happens to think those same thoughts given the remembered behaviours of other individuals of the same species. If only that was how humans behaved!