The
binding problem is the problem of explaining how it is that
our brains are able to connect the various features that we perceive
of a single object together. For instance, when you see a T
you are able to tell that it is both shaped like a T and red. The
features redness and T-shape are bound together and
associated with a single object.
The
binding problem is thought to be tricky in part because our brains
utilize distinct neural feature maps to represent the different
features that we perceive.
Sensory
processing is organized hierarchically. Different levels of the
neural hierarchy represent different levels of feature specificity.
For example, one part of the visual cortex represents where in the
visual field there are boundaries, while another part represents
where there are lines/edges, and another part might represent where there
are surfaces. Since some of the details that are represented at the earlier levels of
processing are abstracted away at later levels in the hierarchy, the features that we perceive are represented in
different places. The earlier levels represent very specific features
without representing general features, and the later parts represent
very general features without representing specific features.
What's
more, at least in the visual cortex, some different kinds of properties
are represented in different areas of the brain. There are two
separate processing pathways that information takes after arriving in the visual
cortex. One pathway is primarily involved in object tracking and contains
representations of which parts of the visual field contain objects, and how fast those objects are moving, without fine-grained representations of what
sorts of objects they are. The other pathway contains representations of
what sorts of object's are in the visual field without precisely locating them or representing whether or not they are moving.
The
binding problem is a problem of neural engineering. Somehow, our
brains are able to associate the different features of an object
that are represented in different feature maps together. In order to locate a
red T in a field of black text,
our brains must not only discern the redness and T-shape of the
letter, they must also assign those features to the same object.
There are different proposals for how the brain could do it, and for all we know there are many different ways that would work equally well. The binding problem is that of finding out how our brains actually do
it.
The
unification problem is a separate but related problem.
Here is an intuitive thought: not only does our brain manage to
associate different features of an object as belonging to the same
object, but we sometimes experience them that way. When we see
something and hear it, the sight and the sound occur to (most of) us
as two separate experiences. The experiences appear before our minds
simultaneously, but they are represented in distinct percepts. When we see that
something is a square and watch it move, however, we don’t perceive
the squareness and motion in two separate experiences. We don’t
have a motion experience and a squareness experience that we
intellectually connect as experiences of one object, in the way that
we have connect a sound with an image of the object producing it. The
perception of both properties is unified into a single percept. The
unification problem is to explain why this is the case.
The
binding problem and the unification problem are distinct problems.
Just because some features are bound together in the brain (i.e. associated with the same object) doesn’t
mean that they must be experienced in a single percept. We often
associate sounds with particular objects that we also see without having merged visual and auditory experiences.
It is also
conceivable that we could perceive objects with a unified percept without our
brain being able to represent the unified features as belonging to the
same object. this is made plausible by the observation that it takes our brains some time to bind the features of
objects together. If you flash open your eyes on a complex scene, it is thought that your brain does not have the time to associate the right properties with the right objects. Nevertheless, it is not crazy to think that we have percepts that unify the features belonging to an object together before our brain succeeds in actually binding the
represented features together.
Thus, while solving the binding problem may help us solve the unification problem, and vice versa, they aren't just the same problem.
No comments:
Post a Comment