Fear of movement-related pain significantly contributes to musculoskeletal chronic pain disability. Previous research has shown that fear of movement-related pain can be classically conditioned. That is, in a differential fear conditioning paradigm, after (repeatedly) pairing a neutral joystick movement (conditioned stimulus; CS+) with a painful stimulus (unconditioned stimulus; pain-US), that movement in itself starts to elicit self-reported fear and elevated psychophysiological arousal compared to a control joystick movement (CS-) that was never paired with pain. Further, it has been demonstrated that novel movements that are more similar to the original CS+ elicit more fear than novel movements that are more similar to the CS-, an adaptive process referred to as . By default, movement/action takes place in reference to the three-dimensional space: a movement thus not only involves proprioceptive information, but it also contains spatiotopic information. Therefore, the aim of this study was to investigate to what extent spatiotopic information (i.e., endpoint location of movement) contributes to the acquisition and generalization of such fear of movement-related pain besides proprioception (i.e., movement direction). In a between-subjects design, the location group performed joystick movements from the middle position to left and right; the movement group moved the joystick from left and right to the middle. One movement (CS+) was paired with pain, another not (CS-). between CSs typically reduces differential learning. The endpoint of both CSs in the movement group is an overlapping feature whereas in the location group the endpoint of both CSs is distinct; therefore we hypothesized that there would be less differential fear learning in the movement group compared to the location group. We also tested generalization to movements with similar proprioceptive features but different endpoint location. Following the principle of stimulus generalization, we expected that novel movements in the same direction as the CS+ but with a different endpoint would elicit more fear than novel movement in the same direction of the CS- but with a different endpoint. Main outcome variables were self-reported fear and pain-US expectancy and eyeblink startle responses (electromyographic). Corroborating the feature overlap hypothesis, the location group showed greater differential fear acquisition. Fear generalization emerged for both groups in the verbal ratings, suggesting that fear indeed accrued to proprioceptive CS features; these effects, however, were not replicated in the startle measures.