<p>Our investigation explores the facial feedback hypothesis in relation to the perception and memory of emotional faces. In Experiment 1, participants rated the valence of happy, neutral, and angry faces of different actors while vertically holding a pen either between their teeth (smile-facilitating condition) or between their lips (smile-inhibiting condition). Results indicated that participants in the teeth condition rated happy faces more positively than participants in the lips condition, while no difference was found for angry and neutral faces. Experiment 2 assessed long-term memory for happy, neutral, and angry faces. During encoding, participants saw happy, neutral, and angry faces of different actors. During recognition, while holding a pen either between their teeth or their lips, they had to select the previously seen faces from a set containing the target face (same emotion expression) and two filler faces (other emotion expressions) of the same actor. Results indicated no significant differences both for accuracy and recognition times between the two groups. Taken together, these findings suggest that the pen-in-mouth procedure influences the perception of happy faces only, while the manipulation had no significant effect on memory. Interestingly, angry faces required more time to be evaluated in Experiment 1 and were recognized more accurately in Experiment 2. We discuss these results in relation to the literature on the facial feedback hypothesis and embodied emotion theories.</p>