Second Part

0:05:07 I left Berkeley partly because my marriage was breaking up and I had more links back in Cambridge; all my links with people here are long-lasting and deep whereas in America things are constantly changing, and the average time that a person stays in the same place is much shorter; also my mother was old and dying, and I came back to a very nice job as a Royal Society Research Professor where my total commitment was writing a 200 word summary of my activities each year; at that time it was a job for life although I was one of the last; I think it was right to change it as I think that if you want to make the best use of people you should not appoint them for life; confining my work to neurophysiology, I was one of the first people to realize how much, and the type of analysis, has to go on early in vision in order for us to see as we do; this is so particularly with regard to the statistical aspects of vision, the fact that we infer so much from so little, and things like that; of course we already have learnt so much of the statistical facts, these are not things that we say exactly what we have learnt, but we know that if we let go of a penny it drops; as the facts about the environment and the way the world around us behaves, we already have expectations formed; it is because of those expectations we can predict as well as we can; this principle goes right through perception and cognition; this was not the way people thought of the brain when I started in the 1940s, and is very much the way in which people think of the brain now;  I am not talking about high-order patterns like the work of Piaget on infants, but things at a very elementary level; one of the first things showing how extremely sensitive the eye is in this regard, concern the actual detection of single quanta of light, and that you can get down to the graininess of light itself; this limits your perception under a wide variety of circumstances; you have got to have an efficient statistical machinery there otherwise you won't be able to do it; without this you will not have the mechanisms for determining what is associated with what; this is at very much a pre-Piagetian level; some information we receive is permanent and we do not need to do more than note any changes; work on redundancy reduction has often been misinterpreted; what you are really doing is using the redundancy that you have acquired knowledge of in order to understand better what is going on now; the only definition of intelligence that makes sense to me is this using of statistical regularity which you detect in order to make inferences about the present, past, and in some cases, the future; there is a very big difference in the way of thinking about statistics now than was prevalent when I was at school; I was very lucky both here, and later in London, knowing people like Tommy Gold and Donald Mackay who understood the statistical aspects of information; rather few people, even now, understand information theory and what redundancy is and isn't; I think it has a revolutionary effect on what the brain is doing all the time; you can't create new information - that is like believing in spontaneous combustion, which statistical mechanics shows is impossible; what you can do is filter; the fact about the world that is missing from earlier views is what an enormous amount of regularity there is; if you take any object or state of affairs, it is going to be almost exactly the same an instant later; if you look at language, the half a million words in the English language can not be arranged randomly and make any sense at all; the same is true of images; if you take random noise and record it on film you can't see anything there; the only thing one sees in an image are the statistical regularities; in one single frame you don't get very much, but in sequences of frames there is a lot that is shared between them and those are the things you see; if things are arranged in a line, you detect the line immediately, but without the patterned element you don't see the line; in such a case to say that in seeing you remove redundancy is the reverse of the truth, redundancy is what you see; it wasn't until after the war that the full significance of information theory for understanding the brain appeared, because people didn't appreciate the importance of language and all our senses being composed of redundant features; economy of thought was one of the earlier ways of describing this and can be traced back to the nineteenth century; it has now come to fruition with cybernetic devises, which is why it had taken time; even people most deeply involved in trying to do computer pattern recognition, didn't really fully appreciate the significance of this as far as perception was concerned; having been tutored by people like Tommy Gold, Donald Mackay and others, I reckon I am one of the earliest generation of people interested in psychology, who have understood the significance of this

15:25:12 My early work was on frogs' eyes; Hartline's early work had shown there were four different types of ganglion cell; we now appreciate that there are now closer to four hundred types, and each telling the brain something different, about a much larger pattern in a particular region; the simplest example of this is motion; I think it true to say that even though there were lots of illusions associated with motion which have been known since the nineteenth century, the fact that what is crucial in the modern understanding of the subject is that motion is one of the simplest types of patterns we detect; you can't have motion without knowledge at two different positions in time and space; what motion tells you is that something has changed, a coordinate change in position and time of occurrence; once you have got a signal for motion, rather than a separate occurrence of things at two different times and places, then you can get joined motion at different parts - it expands the alphabet that you can use in information theory terms; it is clear that the kind of things that we do use to express our subjective experiences are very complex elements of this type; I still think it is unappreciated how significant this is; I am very lucky having lived and worked in this time as it is such an important principle which applies to so much of perception

19:53:09 Ideas come when thinking to myself, when having a bath or just lying in bed, relaxing; one is always trying to see relations where one hasn't seen them before; novel ideas tend to come when one is not doing anything else; religion is not important to me; I think that saying that you don't believe in God is a very foolish thing as it doesn't explain why so many people talk about it, there has got to be more to it than that; also I think one has to respect what some godly people say and some of the things they do; I wish one could make more sense of it but I don't think the godly people have done a very good job; I was never baptised or confirmed so have never been a practitioner, and I don't miss it; I like all the manifestations; I think that science provides some hope of acting more rationally to handle the social and political problems we have to deal with; if you regard religion as a way of perpetuating a way of thought that might have been lost, then that's fine