Reasoning about social relationships reduces class differences in tests of cognitive aptitude

Class differences in aptitude testing are often attributed to working-class deficiencies in a host of basic cognitive skills. The present research challenges this deficit narrative in the context of logical reasoning. We argue that reasoning tests often privilege abstract styles of reasoning as a "neutral" benchmark and provide a poor match to social-cognitive competencies promoted in working-class contexts. We evaluate this hypothesis in archival data from an international study of mathematical aptitude (N = 158,216), a pre-registered nationally-representative experiment on logical deduction (N = 781), and an integrative data analysis (IDA) combining data from Study 2 with three supplemental studies (N = 2,269). In Study 1, reasoning items higher in social-relational content (i.e., mentions of people) yielded smaller class gaps among eighth-grade students from 46 countries. Study 2 and the IDA experimentally replicate and extend these findings, demonstrating that reframing an abstract reasoning task to recruit social-cognitive skills reduces class-based performance differences among adults. Study 2 and the IDA also highlight that these effects cannot be explained by the mere inclusion of real-world content (e.g., money, food), but are most robustly produced by content invoking social relations. Our results support a cultural-psychological account of reasoning performance, highlight working-class strengths in social reasoning, and point to inclusive approaches to testing as a way to attenuate social-class gaps in aptitude testing performance.