BackgroundAs a part of the set of complex creative behaviours, musical improvisation is a high-level cognitive process characterized by contextually significant generation of new sound and music ideas. It requires sudden elaboration of music components such as melody, harmony or rhythm. Musical improvisation, from a music therapy perspective, is a technique that is widely used with different populations understanding that any person could perform such a creative act (Abrahan & Justel, 2015). Despite its wide use in music therapy, little is known about the direct incidence of musical improvisation in modulating general cognitive processes such as affective memory (Justel & Rubinstein, 2013). Besides, research in music-therapeutic improvisation, particularly from the neuropsychological point of view, is relatively incipient. Recently, some studies indicated that both music perception and musical improvisation could modulate memory in young adults.AimThis study aims to investigate the effect of musical improvisation, as a music-therapeutic technique, on visual emotional memory of old adults without musical training.MethodFifty-four participants without musical training (67% female), over 60 years (M: 70.07, SD: 0.88), participated in this study, from different cultural and social senior? centres?, randomly assigned to three different conditions (improvisation, imitation and silence). Thirty-six images from the International Affective Pictures System (IAPS) were selected: 24 of them were emotionally activating (12 with positive and 12 with negative valence) and 12 were neutral. Firstly, participants had to observe the images and to rated (in a 10-point scale) how emotional the images were for them. Then, they were exposed to a three-minute experimental treatment (a free musical improvisation, with a musictherapeutic orientation. In this condition, participants improvised musical patterns with instruments, their voices or bodies, spontaneously creating some musical feature) or control conditions (no music stimuli or imitation condition). We evaluated the memory through two tasks: Free recall (the participant had to write the images that they remembered) and recognition (the participant had to recognize within a set of 72 images, the original 36 ones), both immediately and deferred (after a week).ResultsThe general result of this study indicate that emotional stimuli are better remembered and recognized than neutral information [F(2, 102)=95.689, p