In the first half of this slim book, Braitenberg describes simple hypothetical vehicles, with simple sensors and actuators, and describes how they behave. For example, a vehicle with two light sensors, attached to its wheels, will turn away from the light if the connection is direct, and towards the light if the connection is crossed (right 'eye' connected to left wheel, and vice versa). These delightful vehicles get progressively more complex -- with collections of different sensors (heat, light, smell, sound...), some excitory, some inhibitory, some connected directly, some crossed -- and exhibit progressively more complex behaviours. If we don't know how the vehicles are wired, if we can analyse them only by observing their behaviours, we might naturally describe these behaviours in psychological terms: "this vehicle doesn't like loud noises", "this vehicle is afraid of other vehicles".
With a rather large leap in designs about half way through these vehicle designs, Braitenberg postulates vehicles with simple 'brains', sensors that can detect motion, remember events, predict events, and so on. Viewed from the designer's perspective, from 'inside', these are all relatively simple networks. But viewed only from the 'outside', the vehicle behaviours look ever more subtle and life-like.
In the second half of the book, Braitenberg explains some of the real biological background to his hypothetical designs. I would have liked this to have been interleaved with the vehicle descriptions, to anchor me to the reality a little sooner.
If I had read this classic book a few years ago, closer to when it was originally published , I would probably have given it a higher rating. In the meantime, however, I've come across some of these ideas in other places, but rarely expressed so well as here.