Why boys cry and don't cry: The Contextual-Statistical (ConStat) approach to the perceived validity of generics.

People judge the validity of bare plural generic generalizations ('generics' for short, e.g., 'Women are empathic') in a notoriously fickle way. We present the Contextual-Statistical (ConStat) framework to explain these judgments, showing that they follow consistent principles and are therefore logical and predictable. ConStat rests upon the assumption that generics transmit information of relevance for action. It consists of five tenets. (1) People may understand generics as being meant normatively or descriptively. If they deem a generic normative, they endorse or reject it - a decision that may masquerade as a truth judgment even though the concept 'truth' applies only when people deem a generic descriptive. (2) People can understand a descriptively read generic as being primarily about the category or primarily about the feature. (3) People judge whether a descriptively read generic is true by acting as intuitive statisticians who consider, by default, the prevalence of the feature in the target category as compared to an alternative category; only if an alternative category is cognitively unavailable, they base their judgment on the prevalence of the feature in the target category. (4) Both the explicit and implied content of the generic determine the threshold that the prevalence must exceed for the generic to feel true. (5) The processes described in tenet (1) to (4) are all shaped by the generic's context. The ConStat framework explains many fickle patterns in truth judgments, inspires novel directions for research on how people understand generics, and suggests manners to improve the methodology of that research.