Perception is an active process during that we purposively gather sensory information. Haptic perception is the prime example for this principle. When people aim to judge an object by active touch, they first have to appropriately explore the object in order to obtain the relevant sensory information. Often a single exploratory movement is not sufficient, but people explore objects by movements that consist of repetitive segments (e.g. repeated indentations or strokes). In several experiments we studied how humans control multi-segmented natural explorations, and how they integrate the serially gathered information. Results showed that people control and successively fine-tune their exploratory movements in order to hone the gathering of sensory information−using an interplay of closed-loop sensorimotor processes and predictive open-loop processes. Further experiments showed that information from different movement segments is efficiently integrated under conditions of memory decay (as modeled by a Kalman filter), and supplemented by information from other sources such as vision or previous knowledge. Overall, in the talk I will present work suggesting that active touch is performed by a ‘clever’ hand employing processes that serve to optimize perception.