Abstract Background: Emotions elicited by advertisements are known to influence what consumers remember; however, prior studies have often relied on averaged measures. We examined how the dynamics of facial expressions during video advertisements relate to ad recall in young adults. Methods: In a laboratory experiment, 40 participants watched branded video commercials while their facial expressions were continuously recorded. From these recordings, we derived features capturing emotional intensity and temporal dynamics (e.g., variability, peaks). We then applied correlation analyses and machine-learning classifiers to predict which ads participants would later recall.
Results: Ads with higher recall consistently elicited greater emotional variability. Dynamic facial emotion features accounted for a substantial proportion of the most informative predictors in the final classification model, which achieved an AUC of approximately 0.867. At the same time, overall levels of positive emotion remained a significant predictor of recall, indicating that emotional dynamics complement rather than replace static affective intensity.
Conclusions: Temporal patterns of emotion during ad viewing substantially enhance the prediction of memorability. Monitoring fluctuations in facial expressions provides unique insight beyond static metrics and offers actionable guidance for designing more memorable advertisements.
