This article presents a pilot study with the objective to create and test stimulus material, which consists of photographic portraits of adults and children expressing various emotions. The uniqueness of this work is due to the approach to organizing the creation of stimulus material, in which the models demonstrated emotions not according to an established pattern of facial movements corresponding to a specific emotion, but according to their ideas about the expression of emotions in a certain situation. The inclusion of photographic images of individuals from the Russian population, of different genders and age groups, can be considered an additional advantage of this study. We tested 294 color photographs of 5 models of different age groups (from primary school to the elderly), depicting 14 emotions (pride, anger, joy, irritation, amusement, disgust, pleasure, sadness, relief, despair, interest, fear, surprise, anxiety). Testing took place on the crowdsourcing platform Yandex.Toloka, where 5099 people took part in the study. The answers were analyzed using two approaches. In the first instance, the respondent's answer about an emotion was counted as correct if the emotion s/he chose exactly matched the one depicted in the photograph. The second approach assumed that the respondent’s answer was correct if it corresponded to a group of emotions related to one or another valence. The results of testing stimulus material developed using a non-classical approach showed recognition accuracy in determining the valence of emotions rather than the emotions themselves. An analysis of the possible influence of individual differences (including culture-specific ones) on recognition accuracy may be needed for the interpretation of the results, however, based on the pilot nature of the study and its limitations, this hypothesis still requires additional verification.