Background. Debates about the modifiability of cognitive ability have been largely resolved by reports of successful `thinking skills' interventions. However, such interventions are very diverse and generalization of effects relatively little explored. Aims. This study investigated whether a thinking skills intervention involving collaborative interactive dialogue could lead not only to gains in measured verbal cognitive ability but also generalization to non-verbal and quantitative reasoning ability. Sample. Randomly selected intervention children were aged 10 at pre-test ( N =105, four classes/schools). Controls followed a normal curriculum ( N =72, three classes/two schools). Method. Intervention children engaged in collaborative enquiry for 1 hour per week over 16 months. The control group received normal classroom experiences. The Cognitive Abilities Test was administered before and after the intervention. Results. Intervention pupils showed significant standardized gains in verbal and also in non-verbal and quantitative aspects of reasoning, consistent across intervention schools. Boys and girls made significant gains. The highest quartile of pre-test ability showed the smallest gains. Controls did not gain in any aspect. Conclusions. Philosophical enquiry involving interactive dialogue led not only to significant gains in measured verbal cognitive ability but also generalization to non-verbal and quantitative reasoning ability, consistent across schools and largely irrespective of pupil gender and ability. The effect sizes from this large-scale field trial in one local authority exceeded those reported in the literature. Implications for theory building, replicability and sustainability are addressed.