Psychology has long had a history of fighting over whether we should anthropomorphize animal behavior or whether, instead, we should be careful to not anthropomorphize human behavior. Or whether there really is a difference in underlying form, even when function appears similar.
What does it mean to say someone “deliberates” over choices? What does it mean to say that a person “expresses regret”? How could we tell if these are unique human abilities?
One option is to make our definitions depend on language. We could say that a person “expresses regret” because they say “I regret doing that” (or, conversely, when they say “I have no regrets”). But then, making a computer express regret is trivial. It would take one line of code to write a program that prints “I regret that, Dave” to the screen or to have Siri speak it to you. Somehow, I don’t think that’s a very satisfying option.
The alternative is to carefully define both the situations that should create a cognitive event and the information processing that the subject has to do. But then, we have to trust those definitions, and say that anything that processes information in that way in those situations is cognition.
Deliberation is a process of serially searching through future possibilities based on one’s model of the world, evaluating those possibilities, and then selecting an outcome based on those evaluations. This means that a computer program that searches through those outcomes really is deliberating over choices. It also means that if we find evidence that non-human animals (such as rats) represent their future choices serially, then we have to say that they deliberate. In the 1930’s Meunzinger, Gentry, and Tolman found that in certain situations (in which humans would likely deliberate) rats paused and showed behaviors that anthropomorphically looked like they deliberated. And they got resoundingly attacked for it.
“So far as the theory is concerned, the rat is left buried in thought…”
- Guthrie 1935
The problem was that they did not have access to the information processing going on in the rats’ brain. Modern neuroscience, however, does have that access. In 2007, Adam Johnson and I found that rats really were deliberating over those choices – neurons in the hippocampus were representing the future options serially. Since then, my lab has gone on to find that there are appropriate evaluation steps happening at those times as well. (We’re still looking for the action-selection processes.)
Recently, we’ve found the information processing for regret. In a paper just published in Nature Neuroscience, we report a new task in which rats have a time-budget to spend foraging for food. In this task, rats encounter a sequence of “deals” – delays which they can spend their time budget on to get food. (We call this task “Restaurant Row” because the analogy is How long is the line at the Chinese restaurant?)
Sometimes a rat skips a pretty good deal and then finds itself faced with a really bad deal. That should induce regret because the rat has made a mistake of its own agency. We can contrast this with merely disappointing conditions when the rat simply encounters a bad deal (but hasn’t made a recent mistake). In regret, but not disappointment situations, the orbitofrontal cortical neurons in the rat represent the missed opportunity. Interestingly, the neurons represent the action the rat should have taken in the previous opportunity, not the missed food. But you regret the things you didn’t do, not the things you didn’t get.
The question at hand is whether it is fair to call these processes “deliberation” and “regret”. I argue one has to because the neural signals (in hippocampus during deliberation and orbitofrontal cortex during regret) show that the rat’s brain is actually performing the computations (processing the information) the way that we say deliberation and regret require.