In order to assess the authenticity of conservation attainment through an observational learning paradigm, the model's age and presence were manipulated, since these variables are known to induce conformity to modelled behaviour. On the one hand, it was predicted that if, in reaction to social pressure, the observers' performance amounts to their matching the model's responses without comprehension of the involved rationale, conservation scores should differ according to the model's age and presence, and should correlate with readiness to comply to social influence. On the other hand, if the former non-conservers reach some understanding of the conservation demonstration, scores should not be affected by the model's age and presence, and should not correlate with propensity to yield to social influence. Nonconservers were selected during a pretest that comprised an opinion task, with no objective right answers, and conservation tasks varying in degree of difficulty. During the learning session, the children were exposed to either a peer or an adult model who was either present (physically) or absent: on videotape, the model opposed each child's pretest opinion and solved correctly one easy conservation task. On the immediate post-test, the four conditions regarding the model's age and presence were maintained, while on the delayed post-test, the model was absent from all groups. In all these groups, significant but equivalent learning was reported: whereas, for the easy tasks, progress was stable across post-tests, for the more difficult tasks, further improvement occurred on the delayed post-test. Conservation scores did not correlate with adoption of the model's opinion. However, compliance with the model's opinion was higher in the model's presence. Therefore, the failure to find any evidence of a causal or correlational bearing between social influence variables and the observational acquisition of conservation was discussed as being indicative of the genuineness of the provoked achievement.