In an endogenous cueing paradigm with central visual cues, observers made speeded responses to tactile targets at the hands, which were either close together or far apart, and holding either two separate objects or one common object between them. When the hands were far apart, the response time costs associated with attending to the wrong hand were reduced when attention had to be shifted along one object jointly held by both hands compared to when it was shifted over the same distance but across separate objects. Similar reductions in attentional costs were observed when the hands were placed closer together, suggesting that processing at one hand is less prioritized over that at another when the hands can be “grouped” by virtue of arising from the same spatial location or from the same object. Probes of perceived hand locations throughout the task showed that holding a common object decreased attentional separability without decreasing the perceived separation between the hands. Our findings suggest that tactile events at the hands may be represented in a spatial framework that flexibly adapts to (object-guided) attentional demands, while their relative coordinates are simultaneously preserved.