The bits 01000001 are just eight on/off switches. Read them as a number and they are 65. Read them as ASCII text and they are the letter 'A'. Read them as the brightness of a gray pixel, on a scale from 0 (black) to 255 (white), and they are a dark gray. Nothing in the bits says which; the program reading them has to know the rule. That rule is the representation.
Each kind of data has its own rule. Text: every character gets a number; ASCII uses 0 to 127, and Unicode stretches past a million so it can cover every alphabet and emoji. Color: a pixel is three numbers for red, green and blue light, each from 0 to 255, so (255, 165, 0) is orange. Sound: the wave from a microphone is measured 44,100 times a second, and each measurement is stored as a number. The rule sets the trade-offs. More measurements per second, or more bits per pixel, mean better quality and bigger files.
A representation is not the thing itself. A photo of a sunset keeps the colors of a grid of points and throws away everything between them, plus the warmth and the sound of the waves. Two files can also represent the same picture differently: a PNG keeps every pixel exactly, while a JPEG drops detail you are unlikely to notice so the file is smaller. Converting a number to base 2 is a different topic, Binary; representation is about what the numbers stand for.
Think of it like this. A representation is like a code shared by two friends: 1 means A, 2 means B, and so on. The numbers carry the message only because both use the same key; hand them to someone with a different key and the message comes out as nonsense. Computers fit this closely, except that their keys, such as ASCII and RGB, are published, not secret.
Core idea. All data in computers is ultimately numbers—representation is the mapping.