I like the article.

I remember temperature to be used in this way when speaking about Monte Carlo algorithms. At high temperature the algorithm searches for solutions that are very different from the current solution. At low temperature the algorithm searches smaller variation of the current solution. Lowering the temperature at which such an algorithm operates doesn't change the amount of calculations per minute that might in your model be energy.

Similarly I don't believe that "energy" is a limiting factor for changing the temperature in human belief formation. In Zen Buddhism, keeping the temperature high is labeled as keeping "beginners mind". Human brains are capable of that feat but don't do so automatically.