The Role of Socioemotional Context on Cognitive and Psychophysiological Mechanisms of Learning in Early Adolescence

• In Event: Novel Approaches to Examining Learning and Self-Regulation Development Across Context and Levels of Analysis

Abstract

Throughout development, we navigate our unique worlds by learning through trial and error in the face of uncertainty. Such learning—or reinforcement learning—is posited to support youth’s ability to engage with and adapt to their increasingly complex environmental demands, with links to long-term social and emotional outcomes (Master et al., 2020). Theory suggests that early experiences calibrate and optimize learning to help individuals encode environmental cues that are most relevant to their social worlds (Frankenhuis et al., 2020). Indeed, a growing empirical base with adults has shown that emotionally arousing cues preferentially engage learning via enhanced neurophysiological activation (Nashiro et al., 2012; Everaert et al., 2020). However, differences in reinforcement learning across varied socioemotional stimuli and levels of analysis remain poorly characterized in developmental samples, as task paradigms are often limited to non-emotional stimuli, and measures of learning are often limited to summary statistics of behavior. Here we explored whether cognitive and psychophysiological mechanisms of learning among youth vary across stimuli ranging in socioemotional salience (i.e., arousal and valence).

In a pre-registered study, a sample of 56 12– to 15-year-olds (51.8% female, 94.6% White, 19.6% Asian, 3.6% Hispanic, 1.8% Other) completed two versions of an adapted computerized learning paradigm while recording continuous cardiac activity from a wireless electrocardiogram. The ‘nonemotional’ version included 24 stimuli of line drawings (e.g., key, chair, shoe) from the original version of the task (Finger et al., 2008). The adapted ‘socioemotionally-salient’ version included 24 IAPS stimuli of emotionally negative (e.g., neighborhood violence, couple fighting) and positive (e.g., happy family, money) images rated similarly in arousal and valence. Across both versions, the task starts with an acquisition phase where youth learn stimulus-outcome associations via reward (where points are gained) or punishment (where points are lost). Subsequently, the outcome associated with half of the stimuli switch to the opposite reinforcement value, forming the reversal learning phase. Each phase consists of 8 blocks (96 trials) and combined make up one run. Each version of the task contains two runs with a new set of 12 stimuli presented in random order. Presentation of task versions was counterbalanced across participants.

Data collection was recently completed and analyses are currently underway. Bayesian computational model parameters indexing reinforcement learning processes from trial-by-trial data, including learning rate (⍺punishment and ⍺reward) and inverse temperature (β), were generated using the hBayesDM R package (Ahn et al., 2017). Preliminary descriptive results (Figure 1) suggest that youth were more sensitive to rewards (⍺reward) and updated their learned associations slightly faster when stimuli were socioemotionally-salient. In contrast, task version did not appear to alter youth’s decision-making strategies (β). Additional analyses will examine how real-time autonomic psychophysiological engagement influences these computationally-derived indices of cognitive learning strategies within and between-participants.

Preliminary and future findings from this study shed light on how youth learn from—and adapt to—different socioemotional demands, with the potential to inform educational programs that support youth’s ability to adjust to their dynamically changing social worlds.

Authors

• Meriah L. DeJoseph, University of Minnesota - Twin Cities
  Presenting Author

• Kathleen M. Thomas, University of Minnesota - Twin Cities

• Willem E Frankenhuis

• Daniel Berry, University of Minnesota - Twin Cities