Locke argued that persons born blind do not possess true knowledge about color. While prior studies find some knowledgeof color among blind individuals, questions remain about the depth of this knowledge. Do blind individuals merely learninferentially shallow verbal associations (e.g., bananayellow)? We hypothesized instead that blind individuals are morelikely to acquire causally-relevant color information. Blind (n=20) and sighted adults (n=20) reported colors of naturalkinds (e.g. banana) and artifacts (e.g. car) and judged the likelihood that two instances of a type have the same color.Relative to the sighted, blind participants were less likely to know specific object colors (e.g. banana-yellow), but madeidentical inferences about color consistency (more consistent colors for natural kinds). Inferences were similar acrossgroups even for novel objects. Further, blind individuals gave detailed and coherent causal explanations of color origins.Inferentially rich knowledge of sensory categories can develop without first-person experience.