Neural and behavioural differences in multisensory statistical and reinforcement learning across development and task...

Area of Science:

• Neuroscience
• Developmental Psychology
• Cognitive Science

Background:

• Multisensory processing and integration are crucial for development, relying on pattern detection and feedback-based behavioral adjustments.
• The precise developmental trajectories of these multisensory learning mechanisms remain incompletely understood.

Purpose of the Study:

• To investigate neural and behavioral differences in multisensory statistical and reinforcement learning between adults and children.
• To examine these learning processes across varying task difficulties in children.

Main Methods:

• Utilized discriminative choice and match recognition tasks with audio-visual or tactile-visual stimuli and embedded statistical regularities.
• Compared learning in 28 adults (19.0-30.9 years) and two groups of children (N=28 and N=29, 8.5-12.8 years).
• Employed computational modeling and neuroimaging (fMRI) to analyze behavioral and neural data.

Main Results:

• All groups showed improved accuracy and faster reaction times over time; adults outperformed children, and audio-visual learning surpassed tactile-visual.
• Adults exhibited more sophisticated learning strategies, higher sensitivity to value differences, and more deterministic choices.
• Statistical surprise modulated specific brain regions (lateral prefrontal cortex, intraparietal sulcus, anterior cingulate cortex), while reward prediction errors engaged other areas (striatum, medial frontal, hippocampus).
• Reward prediction errors were consistent across groups and tasks, but neural processing of statistical surprise showed developmental differences, suggesting immaturity in children's statistical regularity detection.

Conclusions:

• Task design significantly influences the comparison of learning differences across developmental groups.
• Reinforcement learning and statistical learning can be studied concurrently, revealing distinct developmental patterns in their neural underpinnings.
• Children demonstrate less mature abilities in detecting statistical regularities compared to adults, highlighting developmental changes in multisensory learning.