5 ideas
Vision happens because light from each point of an object travels in straight lines into the eye, not because the eye sends out rays. Ibn al-Haytham supports this with evidence: bright light can injure the eye, and afterimages persist once the eye looks away. Both show that the eye is acted upon by something outside it rather than acting on the world.
Every point on an object sends light in all directions, so the eye would receive a confused jumble unless something selects among the rays. Ibn al-Haytham proposes that only the rays striking the eye's surface perpendicularly pass through without refraction and form the image. This gives a one-to-one mapping between points on the object and points in the eye.
A claim about nature should be settled by building an arrangement that isolates the variable in question, such as tubes, apertures and darkened chambers, and then observing what happens repeatedly. Agreement between the geometry and the observed result counts as proof. Reputation and plausibility do not.
Several candles placed outside a dark chamber with a small hole each cast a separate, inverted spot of light on the inner wall. Covering one candle removes only its own spot. This shows that light from different sources travels in straight lines, crosses at the aperture without mixing, and keeps each source's identity.
Much of seeing is not raw sensation but rapid, unnoticed judgment. The mind compares the incoming image with prior knowledge to infer size, distance, color and identity. Optical illusions, such as the moon looking larger near the horizon, therefore arise from misjudgments in this inference, and they can be classified by which condition of normal seeing has been violated.