Do children use objects to infer the people and actions that created them? We ask how children judge whether designswere socially transmitted (copied), asking if children use asimple perceptual heuristic (more similar = more likelycopied), or make a rational, flexible inference (Bayesianinverse planning). We found evidence that children use inverseplanning to reason about artifacts’ designs: When children sawtwo identical designs, they did not always infer copyingoccurred. Instead, similarity was weaker evidence of copyingwhen an alternative explanation ‘explained away’ thesimilarity. Thus, children inferred copying had occurred lessoften when designs were efficient (Exp1, age 7-9; N=52), andwhen there was a constraint that limited the number of possibledesigns (Exp2, age 4-5; N=160). When thinking about artifacts,young children go beyond perceptual features and use a processlike inverse planning to reason about the generative processesinvolved in design.