When we talk about constraints we talk about a source/owner and the target object. Constraint is applied to the source object. The source object, when it has to pickup the control parameters from another object, then a target object is added. Some constraints of this type are transformation constraint, action constraint, copy location/rotation/scale constraint. Some constraints which do not require a target object are limit location, limit distance, limit rotation and so on.
Track to constraint, in simple terms means a "Look at" sort of constraint. When you have an object say a monkey(target) and the camera (source) on which the constraint is applied, the camera is set to face the monkey, irrespective of the movement or rotation of the Monkey or the camera. If the "Up" and "To" axis setting is the same, the constraint will not work. The influence setting can vary from 0 to 1. Zero means no effect, 1 means maximum effect.
The three images above shows how the Camera is always facing the Monkey.
The settings to activate the Track-to constraint is shown above. Some other examples of using this constraint are given below:
1. Let us say a ball is bouncing across the ground and the player's eye balls are continuously following the ball, then the constraint is applied to the eye ball with the ball as the target.
2. Let us say a singer is moving around on the stage and a focus light is set such that the light beam always follows the singer, then the target will be the singer and the constraint will be applied on the light(source/owner).
Experiment with the settings to get a hang of how it works.
