Saturday, August 18, 2007

The binding problem (part 2)

I wish to give my own thoughts on the binding problem. It refers to how all our sensory qualia seem to be tied to each other in a consistent way. We all can ascribe spatial attributes to all sensory qualia. In the case of sight, this is pretty obvious. Same is true of sound also (although it is not as localized). Same goes for the somatosensory qualia (touch, pain, etc).

In case of smell, it still resides in the same 3-D space (which I call the CBS), but is spot bound. In other words, the sense is not distributed among a finite region of the CBS, but still belongs to this space (somewhere in the head region).

The brain being an amazingly plastic network, I am not sure how the binding occurs. Is it through a purported higher level association between different types of qualia emanating from the same source (or direction) that the brain figures out automatically over a period of time?

As an experiment, if you wear active goggles that makes you see what you would if you stood facing backwards, and if someone were to clap his hands behind you (so you would now get to see it), you would be confused because the sound comes from the back while your vision is to your front (because your eyes point to the front). But if this experiment were to be performed on a baby monkey, with permanent goggles, it is quite plausible that the senses would recalibrate and "rebind" after a period of time, so that the grown-up monkey would now actually see and hear in the same direction of the CBS. The brain seems to like consistency and keep working at it. I think that even in humans there is a continuous rebinding taking place, since as a person grows up the shape, size and separation of the pinnae (external ears) keep changing, but doesn't really throw off the binding between sight and sound. But I also wonder if a person who is blind from birth can ever afford this kind of recalibration since there is nothing to calibrate against. Another related condition occurs in the case of squint eyed children. When the brain determines that the eyes can never get to focus together, it simply starts discarding the information from one eye, and a condition called amblyopia in which that eye loses its vision results.

But this whole thing seems a bit spooky to me. It seems like there is a hidden observer within our brains that seems to be tweaking our senses without our conscious perception. And this observer needs to be fairly intelligent too to associate sight with the corresponding sound, and may require higher level processing. For example, how does binding take place in a kitten's brain? Let's say, we have a newborn kitten in a cage, and there is a circular arrangement of lights and loudspeakers on the floor a distance of a few feet from the cage, with the cage at the center. Further, let us assume that the sounds of the speakers is all the kitten can hear. Now, if one light is turned on, but at the same time, the speaker on the opposite side is sounded, and this pattern is repeated randomly for all the lights, over some time it is possible that the kitten's hearing binds inverted in its CBS. So the kitten might actually start hearing the sound from the same direction as the bulb. What drives the recalibration in the poor kitty's brain is the fact that its higher cognitive areas somehow associate the light with the sound, detect an inconsistency, and requests the lower areas responsible for the binding to reorient. But what if, instead of a simple association as above, we had a more complex one like as follows- initially the kitten is taught to associate a low-pitched sound to a circular figure and a high-pitched sound to a star figure. Then we show the circular and star figures on opposite sides of the cages, but sound the speakers in the reversed sense (so that the low-pitched sound comes from the star figure side and the high-pitched sound comes from the circular figure side), and keep repeating this for different directions. Now the question is, will the higher-level association of the shape to the pitch compel a rebinding to take place? If the answer is yes, it would seem absolutely creepy! Had the kitten not been taught to associate the shape with pitch, no such rebinding would have taken place. Actually, this is just a hypothesis, and I am not sure if this will happen, even assuming the experiment is practical. Plus it would be difficult to know if the rebinding actually took place, or the cat now consciously associates sound with the opposite direction (the end effect would be the same in both cases).

I call the above examples forward binding. For, the binding takes place from the various sensory inputs in the waking state. But in dreams, there are no external sensory inputs. The different groups of neurons that are responsible for different senses (in dreams) still have to make sure the combined experience is bound properly in the CBS. Therefore, it seems like in dreams, the different sets of neurons don't work independently, but rather get instructions from the binding center to create a consistent sensory experience. I call this backward binding. I think backward binding is more difficult to accomplish than the forward case. In the forward case, once the sensory parameters have been tuned, any external stimuli will automatically lead to binding (assuming the external stimuli are themselves consistent). Exceptions occur with inconsistent stimuli like in the case of motion sickness. In backward binding, all sensory qualia (sight, hearing, smell, somatosense, vestibular sense, etc.) have to kind of mesh with each other, with another portion of the brain orchestrating the whole scene at the higher level (including the theme). And all in real time! That's truly amazing, if you think about it.

Also I wish to bring up the concept of access once again. As I had already discussed in previous posts, we have access limitations with respect to the nature of the hues (eigenhues), as well as intensities. But there is also a spatial variation of intensities (or access), determined by the location in the CBS. For example, it is easy to perceive a sound coming from the back of your head (try snapping your fingers at the back of your head). For the same reason, we can imagine or dream the same. But it is impossible to even imagine or dream seeing anything at the back of your head. This is because of access limitation of sight. The access intensity limit falls off to zero outside of the normal field of vision.

The above observation leads me to believe that the intensity limit of a particular eigenhue as a function of the position in the CBS is decided by the number of corresponding "entities" that are mapped to the same position of the CBS that can be simultaneously excited to produce the corresponding quale. I have left out what the "entity" refers to. It could be the excited state of an "eigenhue molecule" or maybe some fundamental pattern of neuronal firing. For sight, the density of such entities corresponding to any part of the CBS to the back of one's head would be nil (although in the case of hearing, it's obviously not the case).

Also, I believe that this access gateway is common to the waking state, dream state, or the imagination state. It is possible that under the action of psychotropic drugs, the access gateway may get modified somewhat (and this might explain the "heightened experiences" reported by some). But it is very unlikely that drugs can actually cause a new eigenhue to be perceived (such as a fourth fundamental color). It seems like they can only modify the intensity maps.

But does it mean we cannot ever hope to perceive a fourth color? Or feel what it is like to be a bat? I refer to this as the "integration problem" as opposed to the "binding problem". It may well be possible. But that will be for a future post.