Dawkins opens by talking how organisms "grow up" to understand the universe around them, which requires certain apparatus, such as a brain. But before brains can become large enough to model the universe they must develop from intermediate forms. Dawkins then discusses the digger wasp and the set of experiments conducted by Nikolaas Tinbergen of how the digger wasp models the local geography around its nest. He then talks about the limitations of the digger wasps' brain and concludes that only the human brain is sufficiently developed to model large-scale phenomena about the world. He then shows a MRI scan of a human brain (later revealed to be his own brain) and describes how an image develops from the eye onto the visual cortex.
Dawkins discusses how the image on the retina is upside-down and in two dimensions but the overlapping images from each of the eyes are composited to form a three dimensional model in the brain. He shows this by asking the audience to focus on him while holding their hand at eye level which causes them to see two images of their hand; one from each eye. He then describes how using his finger to wriggle his eyeball that the outside world appears to move because he is moving the image on his retina. However this does not happen when he voluntary rolls his eyes from side to side. This is due to the brain using the internal model to compensate for the relative change in position of images on the retina. Dawkins gets someone to wear a virtual reality headset and move around in a 3-D computer generated world and draws an analogy between the model of the universe developed in one's head with the virtual reality universe developed in the computer.
He then goes on the show that the brain uses models to describe the universe by looking at how the brain interprets various optical illusions, such as the hollow-face illusion using a rotating hollow mask of Charlie Chaplin, the "impossible" geometry of a Penrose triangle, the shifting interpretations of the Necker cube and the ability of humans to find faces in random shapes.
