Two experiments dealing with the effects of a majority or a minority source influence, solely on the recognition of a portrait, let us study the generalization of the influence to a portrait symbolically linked to a colour. According to the theory of conversion, the bringing into play of the validation process of the stimulus when the source is a minority should allow such a generalization cognitive association. When the source is a majority, a social comparison process should lead to compliance about the portrait, without any cognitive investigation of the whole stimulus. In the first experiment, four slides were shown successively using material similar to Luchins' (1945) and progressively drawing the portrait of Lenin, with a red‐orange background for each phase. The dependant variables are: (1) the drawing, (2) the colour of the background, (3) the after‐image. On the two last slides for which the answer ‘Lenin’ is given by the source, changes towards red (and the complementary colour green), in the absence of the source under the minority influence, and changes towards orange under majority influence in the absence of the source are registered. Moreoever, the most significant changes of the colour judgment are due to the subjects who refuse to answer ‘Lenin’ during the interaction. In the second experiment, only the fourth slide, on which Lenin's portrait completely appears is shown. The subjects submitted to majority influence answer ‘Lenin’ more than the control group does, only in the presence of the source and change their judgment on the colour of the after‐image towards the complementary of orange in the absence of the source. When the source is a minority a sinificant effect towards the red and its complementary colour is shown. Copyright © 1991 John Wiley & Sons, Ltd