But their actions should still count as a precursor to empathy. I believe these things are a continuum and divisions along the way are pretty arbitrary. For example does a parrot talk or is it just pattern recognizing? To that I say, what is the difference? Can we say talking is more than simply pattern recognition recursed to the next level?
Consider Searle's Chinese Room, where he argues that since a human following a program to translate Chinese does not understand Chinese then a machine running the same also does not understand and hence is not 'strong AI'. I do not agree with his conclusion as I find such a distinction to be meaningless. You can't define what a mind is, human or not, by whether its aware of its operations. That is too slippery a concept and not necessarily an advantage. You can only observe and enumerate what properties the mind does (not) exhibit.
Read Peter Watts' Blindsight if you want to see this line of thinking handled incredibly well.
http://en.wikipedia.org/wiki/Blindsight_(science_fiction_nov...
If I had to guess, people are taking things a bit personally as an attack on their pets, as though requesting further evidence of higher order processing is insinuating that their pets don't really have the feelings they believe their pets do. (Nobody here commenting on the experiment's conclusions is claiming this, as far as I can tell; they're just asking for more conclusive and objective evidence before they believe such conclusions can be made.)
Some of the 'reasoning' being used ("Just ask any dog owner") is eerily familiar to the kind of reasoning people use when justifying homeopathy ("It worked for my cousin's ailment", which in the extreme leads to lovely things like Rhino poaching and whale hunting, something I'm sure these people abhor), the paranormal and supernatural ("I sensed something [and my senses are obviously infalible]", which in the extreme leads to a reduced quality of life), religious extremism ("Just ask anybody who's heard God's voice") etc.
There are plenty of possible ways things could work out. For instance, if the rats have figured out that the humans can identify individual rats and treat them different (which they might figure out if the humans have done several kinds of experiments on the same rats because each experiment would have a control group that gets different treatment), the rat outside might think "the humans but this other rat in the cage--if I let it out, then when they come back they'll be distracted putting it back in, and might leave me alone".
Or maybe it would be good to let the other rat out so that if danger arrives, you can go hide in the cage.
Or it could make sense to have the other rat out so that the two of you can more quickly explore the environment and find a way to escape.
The main point is that you can't really attribute letting the other rat out to empathy, as there are many other possible explanations for why a rat might let another rat out.
Do you really think this is the simpler explanation?
Also, I believe it's implicit in parent's post that evolution would produce more rats whose behaviour increases their likelihood of survival. The rats don't need to understand how their behaviour ensures their survival, or that it can be modelled with game theory, etc.