Abstracts

Human infants enter the world in a prolonged state of dependency, facing the fundamental challenge of reducing uncertainty about their environment and learning more about the world around them. From this perspective, word learning is not merely the accumulation of labels, but a strategy for resolving gaps in knowledge. Building on Jean Piaget’s notion of the child as an active learner and Lev Vygotsky’s view of the child as a social apprentice, we propose that children’s motivations to learn words are driven by the dynamic interplay between internally generated uncertainty and socially mediated information. The research on active learning and social learning has, however, long run in parallel, often without acknowledgement or due consideration of the other. This talk will synthesise these literatures and propose a developmental framework that captures how children’s motivation to learn may change across early childhood with particular focus on the role of children’s changing metacognitive awareness, sociability and language competence on early language learning. 

TBC

TBC

Why is testing more effective than study for learning and memory? Growing evidence suggests that learning during test (rather than study) is driven by prediction errors (the mismatch between expected and actual feedback), which engage dopaminergic systems to support efficient learning. However, real-world learning environments often lack explicit feedback, yet testing advantages persist even in the absence of feedback. We propose that learners rely on temporal prediction errors (mismatches between successive predictions) during test, a process not available during study. We provide converging computational, behavioural, and fMRI evidence for this account. First, an associative neural network that learns from temporal prediction errors reproduces the feedback-free testing effect. Critically, the model predicts that the testing effect is maximal at intermediate levels of initial learning, where temporal prediction errors are largest. Behavioral and fMRI results confirmed the model’s prediction, showing that both the testing effect and ventral striatal activation (a key region of dopaminergic circuitry) peak at intermediate initial learning levels. Moreover, test (not study) promoted hippocampal pattern separation, potentially reducing interference for long-term retention.

A growing body of evidence links reward-related neural processes to successful language learning in adults, and social interaction might represent a particularly powerful context for these processes. In this talk, I draw on our recent theoretical review (Zappa, Slater & Rodriguez-Fornells, 2025, npj Science of Learning) to argue that social interaction shapes second language learning. Learning a second language in social environments likely engages multiple forms of social reward. For example, native or advanced speakers often provide positive or negative performance feedback on learners' productions, which can function as social reward and punishment. The emotional value carried by this feedback may further engage reward-motivational systems in ways that strengthen the encoding of new linguistic information. I present evidence that dynamic social feedback, compared to matched symbolic feedback, produces broader cortical differentiation between positive and negative outcomes and stronger associations with post-training performance. I argue that social feedback operates both as an informative signal and as an emotionally meaningful one, and that examining how positive and negative social feedback shapes adult second language learning could act as an experimental window into how reward shapes learning more broadly.