On Vision

One of the modules that I did at university, Graphical User Interface design, involved a certain amount of cognitive science, which interests me greatly, and I have been reading around that subject since. How vision works is particularly complex and interesting. At the moment I'm reading two books on this subject: "Visual Intelligence" by Donald D. Hoffman and "Eye And Brain" by Richard L. Gregory. They cover similar ground: the way that our brain interprets the data which is coming into our brain from our retinas. Rather than vision being a passive process, in which what's 'out there' reflects light through our eyes giving some kind of analogue reflection that directly affects our brain, the visual processing areas of the cortex are interpreting huge amounts of ambiguous data to produce what usually appears to us a consistent and seamless experience of sight.

This is such an important and revolutionary idea, and one that I, for one, find very hard to get my head around, that I will dwell on this for a little longer. It's very natural for us to assume that what we are seeing relates very simply and directly with what is out there; in fact for most of us for our idea of reality is based very strongly on what we think that we're seeing. Interestingly, if we have a vivid dream and then wake up, it doesn't take long for us to dismiss the visual experience that we were having as purely illusory. But in normal waking life, the sighted amongst us usually rely very heavily on our vision. If perspective starts jumping around or colours begin to shift, it is a very unsettling experience. Current scientific thinking is that the reason that these phenomena don't usually happen is that part of the brain is actively preserving the consistency and logic of our perceptions.

So how do we know this? As is often the case in biological matters, we see the underlying structure by implication, when things go wrong. In the case of vision, this is either when part of a person's brain is damaged and we observe the way that they interpret or misinterpret what they see, or in the optical illusions which affect all of us. Brain damage can in fact show us very specific vision modules, by their absence in those with damage to a particular area of the brain.

Optical illusions are a compelling way of demonstrating the ways that we interpret our visual world. There are many different types of illusions, each type demonstrating one way in which we interpret, by showing an anomaly or ambiguity. I'm sure most of us have seen illusions such as those which have two converging straight lines, overlaid onto which are two rectangles of the same size. Where the converging lines are closer together the rectangle looks bigger. Clearly this is something to do with our perception of perspective. If we are looking at receding railway tracks, and on the tracks are two objects which look the same size, we know that in fact the further one must be bigger. What's interesting here though is that is not simply a question of thinking that the shape which appears to be further away is bigger; we actually see it as bigger.

This may seem a minor point, but it immediately undermines the idea that our visual world is a direct analogue to what is going on around us. And it demonstrates that there are certain rules being applied in how we perceive the world: in this case, interpreting lines that converge as representing distance. There are many other optical illusions each of which demonstrates another facet of this active process of perception. We could call these facets rules: rules that govern the way that we interpret the ambiguous data which our eyes present to us. These are hard wired into our brain somehow.

Linguists may notice an interesting parallel here. Noam Chomsky's theories about the Universal Grammar, which is hard wired into our brains, are very similar. To those who are not familiar with these ideas, I'll try to give a brief account here. The acquisition of language by children has been a matter of debate for many years. Noam Chomsky, that most famous of linguists and social activists, proposed that all humans are innately engineered to communicate with language, hence the almost effortless manner in which children acquire language; in other words, that there are a set of rules which govern the way in which people use language constructs, an idea that he, perhaps confusingly, calls Universal Grammar (although some of his ideas are not completely original: many philosophers and mathematicians throughout the ages have proposed some kind of universality of language).

The funny thing is, that we often think that we are teaching our children to talk. We spend hours looking at pictures of animals with them, and are thrilled when they begin to associate the sounds of their names with the pictures in the book. And when a child begins to communicate, it is a magical time. What we might fail to notice, however, is the amazing ability of a child to construct grammatical sentences. The errors that the child makes in the early stages of talking highlight this. Having observed that past participles are constructed by adding a D sound to the end of the word, as in 'looked', 'tried', the child may say "I go'd to the shop". Presented with a certain amount of evidence of how sentences are formed, the child knows how to construct any number of sentences that will conform to the same rules. It is when there is an irregularity in the grammar of that language that the child will make a mistake.