This study aimed to investigate whether the presence (when using physical manipulatives [PMs]) or absence (when using virtual manipulatives [VMs]) of haptic sensory feedback (i.e., open‐ended haptic manipulation of physical materials with the use of the hands) during experimentation can impact preschoolers’ conceptual understanding of concepts concerning three different subject domains (i.e., balance beam, sinking/floating, and springs). The participants were 132 preschoolers (5–6 years old), 44 per subject domain, who were equally divided into two conditions differing in the means of experimentation (PM or VM) they used. The data of this exploratory study were collected through clinical interviews and analyzed both qualitatively and quantitatively. The findings indicated that preschoolers’ mean score improved in both conditions for each subject domain, (probably) as a result of their participation in the experimentation phase of the interviews, across all domains and conditions. No statistically significant difference in preschoolers’ learning between the two conditions was found in the balance beam domain. In the sinking/floating domain, VM were found to be more conducive to preschoolers’ learning than PM, whereas in the springs domain PM were found to have enhanced preschoolers’ learning more than VM did. These findings have important implications for science teaching and learning in the early childhood years. First, we provide information on when PM or VM is conducive to kindergarteners’ science learning. Second, we report on how prior embodied knowledge, established through haptic sensory input and related to the task at hand, affects learning through PM or VM experimentation.