Signed reward prediction errors drive declarative learning

Area of Science:

• Neuroscience
• Cognitive Psychology
• Learning Science

Background:

• Reward prediction errors (RPEs) are established learning drivers in procedural tasks.
• Empirical evidence for RPEs in declarative learning, crucial for human cognition, is lacking.

Purpose of the Study:

• To investigate if RPEs influence declarative learning.
• To examine the role of signed RPEs (SRPEs) in acquiring new information.

Main Methods:

• A declarative learning paradigm using Dutch-Swahili word pairs was employed.
• Signed RPEs were coupled with word pair acquisition.
• Recognition memory was tested subsequently.

Main Results:

• SRPEs during learning significantly enhanced subsequent recognition.
• A positive correlation was observed between increasing SRPEs and improved recognition.
• Traditional factors like time-on-task did not predict recognition performance.

Conclusions:

• RPEs play a critical role in declarative learning, a key aspect of human cognition.
• SRPEs act as a beneficial signal for memory consolidation.
• The findings were replicated using visual stimuli, strengthening the generalizability of RPEs' effect on declarative memory.