Implicit and explicit learning of socio-emotional information in a dynamic interaction with a virtual avatar

Abstract

Implicit learning (IL) deals with the non-conscious acquisition of structural regularities from the environment. IL is often deemed essential for acquiring regularities followed by social stimuli (e.g., other persons' behavior), hence is hypothesized to play a role in typical social functioning. However, our understanding of how this process might operate in social contexts is limited for two main reasons. First, while IL is highly sensitive to the characteristics of the surface stimuli upon which it operates, most IL studies have used surface stimuli with limited social validity (e.g., letters, symbols, etc.). Second, while the social environment is dynamic (i.e., our behaviors and reactions influence those of our social partners and vice-versa), the bulk of IL research employed noninteractive paradigms. Using a novel task, we examine whether IL is involved in the acquisition of regularities from a dynamic interaction with a realistic real-life-like agent. Participants (N = 115) interacted with a cinematic avatar that displayed different facial expressions. Their task was to regulate the avatar's expression to a specified level. Unbeknownst to them, an equation mediated the relationship between their responses and the avatar's expressions. Learning occurred in the task, as participants gradually increased their ability to bring the avatar in the target state. Subjective measures of awareness revealed that participants acquired both implicit and explicit knowledge from the task. This is the first study to show that IL operates in interactive situations upon socially relevant surface stimuli, facilitating future investigations of the role that IL plays in (a)typical social functioning.

Fig. 1

The seven facial expressions used in this study

Fig. 2

The abstract looped numerical sequence. It depicts both the avatar’s possible facial expressions and the participants’ possible response options. Intense anger = position 0, moderate anger = position 1, low anger = position 2, the neutral state = position 3, low joy = position 4, moderate joy = position 5 and, intense joy = position 6

Fig. 3

Depicts a typical trial structure. A = 500 ms. (This is the time interval in which the task first, determined the facial expression that will present in the current trial, based on both, the avatar’s expression and participant’s response in the earlier trial and second, morphed the facial expression of the avatar from the expression that it displayed in the earlier trial to the one that it will display in the current trial); B = 500 ms. (The time interval in which only the facial expression of the avatar at peak amplitude was displayed on the screen); C = 6000 ms. (The seven facial expressions which served as response options were displayed horizontally on the bottom of the screen. After a response has been made, the experiment went ahead to the next trial.); D = 2999 ms. (The counter appeared on the screen if no response has yet been given); E = ∞ (The late response feedback appeared on the screen if no responses were registered)

Fig. 4

The mean proportion of On-target trials (i.e., the trials in which participants managed to bring the emotional facial expression of the avatar in the Neutral state) generated across the acquisition blocks. Error bars depict 95% CIs