drupal statistics module

Machines Like Us

One more non-difference between humans and animals

Saturday, 07 July 2012
by Massimo Pigliucci

Humanity has a long history of narcissism, characterized by an endless quest for what makes us “unique,” different from “mere” animals, and – by implication – more likely to be the gods' favorite creatures. Aristotle thought that reasoning was the key to humanity, which led him to suggest that the most perfect life is, of course, that of the contemplative philosopher (hey, at least he didn't go as far as his mentor, Plato, who actually advocated that philosophers should be kings!).

Of course, the perennial hostility generated by the scientific view of humans as animals sharing a common descent with other primates also falls along the same lines: some of us just won't admit that we are animals. Sophisticated, yes, capable of appreciating fine art and killing millions at the push of a button, yes, but animals nonetheless.

A study published in Science (17 February 2006) by Aaron Blaisdell, Kosuke Sawa, Kenneth Leising and Michael Waldmann chips away one more brick in the defense wall allegedly separating us from the rest of nature. They studied reasoning abilities in rats, and came to rather startling conclusions. First of all, it has been well known that rats (and other rodents) can “reason” in the sense that they can make associations between phenomena that guides their behavior. For example, they can note a correlation between event A (say, the appearance of a light) and event C (food will be dispensed), and rapidly come to expect C whenever they see A (such expectation can be measured, for example, by the fact that after seeing the light the rats immediately start poking their nose in the directions where the food is supposed to come from).

This may not sound very impressive, but it is in fact the same sort of inductive reasoning that most of us, scientists included, use every day: one observes a few instances of a phenomenon and generalizes to a broader set of events. Of course, inductive reasoning of this sort is rather unreliable, as summarized in the mantra that “causation is not correlation” (though the quip often associated with it is that “the two are nonetheless highly correlated”).

Humans, it was thought until Blaisdell and coworkers' paper was published, are still the only animals capable of causal inferences, i.e. of a more sophisticated mixture of inductive and deductive approaches that allows us to go beyond simple correlations and actually construct a mental model of the causal connections behind events. For example, we realize that weather conditions, air pressure and barometer readings are all correlated, but we also understand that manipulating a barometer will do precisely nothing to change the weather.

Well, apparently, so do rats. Blaisdell's group carried out a series of elegant experiments in which rats were trained under two situations: first they were exposed to a flash of light (event A), followed by both a noise (event B) and eventually food (event C). This generated the classical type of correlational expectation, so that rats would increase their nose poking in response to either the light or the noise, “thinking” that both events were predictive of the good stuff soon coming their way.

The second situation was more interesting: rats were allowed to press a lever that would cause the noise (event B). Since the food (event C) wasn't coming after they pressed the lever – but was still being delivered after the light flash (event A) -- the rats had the opportunity of “deducing” a more sophisticated causal model, in which A and B were in fact decoupled, and only A would count as a reliable predictor of C. And they did! Once they got used to generating the noise by pressing the lever, they stopped expecting the food in response to it. It was like realizing that changing the barometer's setting isn't going to get you a sunny day after all.

Now, nobody is suggesting that all of this happens consciously, of course, and Blaisdell et al. do not expect rats to start publishing papers on the philosophical nature of causality. But by the same token, most of us don't reach similar conclusions consciously either. We are capable of articulating a reason for why we behave in a certain way, but much of our thinking is in fact unconscious. We ain't that different from rats, as it turns out, yet another little blow to our innate narcissism. Hey, perhaps it is the latter that is a truly unique human character? I wonder what sort of experiment we could carry out to test that hypothesis...