Is a small apple more like an apple or more like a cherry?

Two experiments investigate the role played by either visual online information or long- term memory stored information in categorization. Participants were required to categorize object photographs into artefacts or natural kinds (e.g., hammer and carrot). In Experiment 1 hands in either a grasping posture (precision vs. power grip) or a neutral posture (fist) were presented closed to the objects. Participants had to refrain from responding when they saw the fist. The objects (manipulable with either a power or a precision grip) were presented in either their real or modified size (e.g., strawberries were presented in either their actual size or that of an apple). In Experiment 2, the hand was displayed as a prime before the object (Borghi et al., 2007). Reaction times were faster with the power than the precision posture, suggesting that visual hand stimuli activated a simulation of grasping. Moreover, in both experiments participants categorized objects on the basis of their appearance rather than of their real size. The role played by online information is remarkable if we consider that the categorization task requires accessing the semantic knowledge. Results are in line with recent studies suggesting that the distinction between the dorsal and the ventral stream (Milner & Goodale, 1995) is certainly useful, but probably too dichotomous (Gallese et al., 1999; Derbyshire et al., 2006). Namely, some form of object representation is also encoded in the dorsal route (Gentilucci, 2003), and action information is crucial for concepts and language (Barsalou et al., 2003; Buxbaum et al., 2003).