Notice that dogs see other dogs and any animal in general?  What about how robot brains work?  Do we
robots see cars or animals?

Sep 27th 2007

From The Economist print edition

WHICH is more dangerous, an elephant or a minivan? For most readers of this newspaper, the answer
is going to be a minivan. From childhood, people in motorised civilisations are warned about the
dangers of running into the road, taught the appropriate highway code and—when old enough—
permitted to get behind the wheel only after having undergone a rigorous programme of training that
ends with a formal examination.

You might think, therefore, that such people would be more aware of the movements of vehicles than of
animals. But if you did think that, you would be wrong. An experiment just published in the Proceedings
of the National Academy of Sciences by Joshua New of Yale University shows that people pay more
attention to the activities of animals than to those of vehicles. That applies even among urban
Westerners who rarely see an animal from one year's end to the next.

Dr New was testing a theory of mind originally developed by Leda Cosmides and John Tooby of the
University of California, Santa Barbara, with whom he collaborated on the experiment. Dr Cosmides and
Dr Tooby were among the first to break from the idea that the brain has evolved as a general-purpose
problem-solving machine. They suggested that some tasks are so important and so universal that you
would expect to find specially evolved “modules” to handle them, just as the senses are handled by
specialised areas of the brain's cortex.

Dr Tooby and Dr Cosmides have found evidence to support the existence of such modules in areas of
human relations such as the perception of fairness. Now Dr New has provided some more evidence, in a
completely different area. Building on the observations of other researchers that there seem to be
natural mental categories of objects that are represented separately in the brain (animal, plant, person,
tool and topography are reasonably well-established examples), he wondered if people would respond
in systematically different ways to members of those categories.

His experiment worked by showing volunteers pairs of photographs containing one or more objects from
the five mental categories in question. The photos in each pair were identical except that one object had
changed its orientation or had been removed altogether, and the volunteers had to work out what had

The first thing Dr New looked at was whether the brain pays more attention to the sort of change that
might be expected, or to changes that are unexpected. On the face of it, either might have turned out to
be the case. Paying attention to the expected is probably best for everyday existence. Noticing the
unexpected, though, might save your life.

In this part of the experiment, the expected won out. The volunteers were better at detecting changes
involving things that do routinely move—in other words, people and animals—than of those that would
be expected to be static, such as plants and coffee mugs.

The question Dr New really wanted to address, though, was whether such expectations are learned or
innate. For that, he included a class of object that his subjects would have learned, by experience, have
a tendency to move, but which past evolution could have had no purchase on: motorised vehicles.

The answer was that changes concerning animals were significantly easier to detect than those
concerning cars. In the most telling comparison, 100% of volunteers noticed the movement of an
elephant in the African bush. Only 72% noticed the movement of a minivan in a similar piece of bush.
And that was despite the fact that the image of the van was somewhat larger in the photograph than the
image of the elephant, and that the minivan was red, not grey.

This highly honed ability to notice animal activity (it applies to small familiar animals, such as pigeons, as
well as large unfamiliar ones, such as elephants) argues that an animal-monitoring module is innate in
the brain. As, indeed, might be expected. Animals are important: small ones are supper; large ones are
best avoided, lest they eat you or trample you to death. In other words, you can take the human out of
the savannah. But you cannot take the savannah out of the human.