Perception and Communication

"So suddenly psychology was talking about things like attention, which a behaviourist would not allow. And Donald Broadbent in his first book, Perception and Communication [1958], produced diagrams of what he thought must be happening in the brain. And these consisted of boxes that were linked by arrows. One such box might be a filter, something that filtered out what was happening on the unattended ear. What Broadbent was saying was: ‘We have no way of visualising what is happening in the brain. But my experiment tells me that such and such must be happening in the brain. I don’t know where or how it’s happening but there must be something that essentially acts as a filter.’ So he’s saying yes, we can only study inputs and outputs, but I can still tell you that a particular operation must be happening in between those. It’s the beginning of saying what must be happening in the brain. Now of course there were other experiments being done at the same time that also made one think that things were happening between the inputs and the outputs. If you think of Pavlov’s dogs—the dog hears a metronome and this was followed by meat powder. The dog starts to salivate at the metronome before the food appears. Now, if that happened in your house, you’d say: ‘It’s expecting dinner.’ The question is, can we use words like expect? Well, Donald Broadbent worked at the applied psychology unit, and so did Kenneth Craik, who unfortunately was killed in a bike accident in Cambridge during the war. Craik had a mock-up of an aircraft, and in this the pilot would see enemy aircraft coming in. The question was how does the pilot aim at the enemy aircraft? What Craik found was that you don’t aim at where the enemy aircraft is, you at aim at where it will be. And you can’t explain that without saying that you’re predicting where the enemy aircraft will be. The layman’s word for that is ‘expect.’ “It’s the beginning of saying what must be happening in the brain” If you come to a roundabout or traffic circle, it’s the same problem. There’s a car coming in from the right, and you judge whether it will it be on the roundabout by the time you get there—in which case you have to give way to it. Or, will it not be on the roundabout, in which case you can go. So studying problems like this began to break the ice for words like ‘expect’, ‘attend,’ and so on. At the time, of course, little was known about what was actually happening in the brain during these processes. But when I was a student Hubel and Wiesel in America had just begun recording from individual brain cells in the primary visual cortex in animals, and finding out what the cells responded to. Originally it was thought that they’d respond to spots of light, but they didn’t; they responded to bars. Then it turned out that some responded to more complex stimuli. Those experiments were the most exciting thing that we heard about as undergraduates. Now this was some years after Broadbent produced Perception and Communication , and of course it was very far away from looking at issues like attention. These days there are people working on the physiological mechanisms of attention in animals, and you can use brain imaging to do the same thing as I have done. But at that time you wouldn’t have been able to. I think Donald would have said that. In other words, I think it’s quite a common claim amongst psychologists that you need to have some logical, formal claim of what the operation must be before you look at how it’s actually implemented in the brain. And this idea was put forward specifically by David Marr."