capable of reliably taking us well beyond what is presented to us by our senses.
Another approach to understanding knowledge of reality is idealism, which views reality as
dependent on or even constituted by minds. This view is more compatible than is empiricism with the
constructive nature of perception and inference, but grossly overestimates the contributions that minds
make to the world. It leaps from the insight that there is no knowledge of things without construction
of mental representations of them to the conclusion that entities are mental constructions. The
philosopher Immanuel Kant thought that he had accomplished a kind of Copernican revolution by
placing mind at the center of knowledge and reality. But idealism is actually attempting a kind of
Ptolemaic counterrevolution, as implausible as reactionary attempts to return the earth to the center of
the solar system or to deny human evolution. To develop my alternative, brain-based approach to
constructive realism, I will first discuss perception of objects and then move on to how inference
enables us to go beyond perception.
Knowing Objects
There is an old baseball story about three umpires calling balls and strikes. One says, “I call them as
I see them.” The second says, “I call them as they are.” The third insists, “They ain't nothing until I
call them.” These attitudes correspond to the philosophical positions of empiricism, realism, and
idealism. For neuroscience to support realism about objects, I need to show that the structures and
processes used by the brain enable it to represent things in the world as they are, at least
approximately.
It is tempting to think about mental representations of the world by analogy to the linguistic
representations that we use to communicate with each other in speech and writing. The philosopher
Jerry Fodor claimed that there is a language of thought with the same kinds of structures as a natural
language such as English or Chinese. Many contemporary philosophers assume that knowing is a
propositional attitude, which is a relation between a person and some kind of sentencelike entity. But
understanding minds as brains requires us to take a much broader view of representations, with
linguistic structures such as sentences serving as only one way that the brain knows the world. You do
not have to be a linguistically sophisticated adult human to have knowledge of objects. Other
language-limited animals such as rats and lizards have perceptions too, as do human infants well
before they have learned to talk. In the previous chapter, I described how we can think of brains as
functioning by using patterns of activity of firing by interconnected neurons. Now I will go into a bit
more detail about how visual perception of objects works in the brain.
When you see an object—say, a duck—light in the form of wavelike particles called photons is
reflected off the object into your eyes. At the back of your eyes, your retina has photoreceptor cells
that convert light energy into chemical signals that travel to your brain via the firing activity of cells
in the optic nerves. These signals are carried to the back of the brain where multiple areas of the
visual cortex are engaged. Cells in the area called V1 respond to basic features such as color, motion,
and orientation, while other areas contain neural populations that are specialized for more elaborate
representations, such as faces. Different neural populations interact to determine what features can be
grouped together, as when you perceive both the color of the duck and the shape of its bill. These
neural interactions can also fill in gaps in your visual information, as when you can see that the object
is a duck even though you can see only part of its bill. The brain manages to tie various features
together, so that you don't perceive separately the duck's bill, its color, and its motion, but rather you