Studied problems of identification, using a meta‐contrast design. The first stimulus (A) was always the word ME. In experiment1, the second stimulus (B) depicted an aggressor and a victim facing each other. The subliminal A was flashed either on the victim or the aggressor or completely withheld. In experiment2, B showed two persons standing against different backgrounds: an open horizon or a closed room. It was assumed that the subject would be “forced” to identify with that figure in B on which A was flashed. Subjects in experiment 1 were 18 psychiatric patients and in experiment2, 32 professional artists. The results showed that the subliminal manipulation had been effective: e.g., when led by A to identify with the aggressor, projective‐sensitive and borderline subjects were confused and even reported perceptual difficulties; when led to identify with the open side, highly creative artists, as opposed to less creative ones, reported more positive impressions of the