Abstract: In real-life situations, facial expressions often change dynamically over time. An individual's interpretation of a facial expression may be influenced by its dynamic sequential context. While previous research has shown that simultaneously presented contextual information affects the perception of target expressions, little is known about the effect of sequential changes in facial expressions. The present research aims to systematically investigate how dynamic sequential context shapes the perception of current and past expressions, from both phenomenological and mechanistic perspectives, thereby providing new insights for facial expression processing.
Existing evidence suggests that when dynamic facial expressions transition from a past to a current expression, the perception of the current expression tends to shift in the opposite direction of the valence of the past expression (i.e., a contrast effect; Fang et al., 2021; Hsu & Young, 2004; Russell & Fehr, 1987). However, these studies have primarily used ambiguous current expressions (e.g., neutral faces) or morphed dynamic expressions. It remains unclear whether the past expression would still influence the perception of current expressions that convey clear emotional meaning (e.g., prototypical sad expressions) or when using authentic, human-performed dynamic expressions.
On the other hand, the dynamic sequential context might also influence how individuals perceive the past expression in a dynamic display. Previous research indicates that individuals tend to reconstruct past emotional experiences based on their current emotion state (Levine, 1997; Levine et al., 2018; Levine & Safer, 2002; Van Boven et al., 2009). Yet, little is known about whether similar phenomena occur in the perception of past expressions of a dynamic display (but see Fang et al., 2024). We propose that current expressions might serve as a “recall filter” for the reconstruction of past expressions, such that individuals use the emotional features of the current expression to interpret the past expression, leading to an assimilation effect.
Furthermore, the present research aims to examine the underlying mechanisms of these dynamic sequential effects. We hypothesize that the magnitude of the contrast and assimilation effects may be correlated with the attention individuals allocate to the emotional features of the past and current expressions, respectively. Specifically, we propose that increased attention to the emotional features of the past expression may strengthen the adaptation aftereffects, thereby enhancing the contrast effect in the perception of current expressions. Conversely, increased attention to the emotional features of the current expression may strengthen the "recall filter," leading to a stronger assimilation effect in the perception of the past expression. In addition, the dynamic sequential effects may be related to representational momentum, such that faster changing speeds might enhance representational momentum, thereby increasing the contrast effect when perceiving current expressions and the assimilation effect when perceiving past expressions.
To investigate these issues, the present research will consist of a series of studies. Study 1a will explore the generalizability of dynamic sequential effects on the perception of current expressions by manipulating the emotional ambiguity of current expressions and using both artificially synthesized and human-performed dynamic facial expressions. Studies 1b~1d will investigate the roles of adaptation aftereffects and representational momentum in the dynamic sequential effects when perceiving current expressions. Study 2a will examine how current expressions influence the perception of past expressions, using facial stimuli with varying emotional ambiguity and authenticity. Studies 2b~2d will examine the roles of recall filter and representational momentum in the dynamic sequential effects when perceiving past expressions. The findings of this research will contribute to our understanding of facial expression processing in ecological contexts and provide insights for AI-based dynamic facial expression recognition systems.

Key words: facial expression, emotion perception, dynamic sequential context