Dawson Margin Notes On
Pfeifer and Scheier, Understanding Intelligence
This chapter basically provides an condensed account of the highlights of Braitenberg’s book. In his book, Braitenberg describes 14 different thought experiments. Each experiment is a vehicle with known properties. Each experiment adds a new component to previous vehicles, making them more specific. Braitenberg uses these thought experiments to show that simple components can produce complicated, emergent behaviors. Clearly that is why this chapter is in Pfeifer and Scheier’s text.
Interestingly, they use Braitenberg’s stuff to talk about a “cognitive science” spin on embodied agents, and to illustrate the frame of reference problem. They take this latter perspective instead of focusing on Braitenberg’s law of uphill analysis and downhill synthesis.
They describe in a fair amount of detail Vehicle 1, 2, 3, 4, 5, and 6, and allude to the others. Interestingly, they end with the evolution vehicle, and don’t spend much time on the neural network vehicles that Braitenberg develops.
They introduce a term called the “extended Braitenberg architecture”, which they do not explain very well. Hopefully this term will be made clearer with examples in Chapter 11.
For an alternative look at Braitenberg vehicles, you can take a look at my Margin notes for that book. Better still, if you are interested, then borrow the book from me. It can be read in an evening.