A model of human motor sequence learning explains facilitation and interference effects based on spike-timing...

Area of Science:

• Neuroscience
• Computational Neuroscience
• Machine Learning

Background:

• Human brains learn sequential behaviors, but studies on motor sequence learning yield contradictory results regarding interference and facilitation effects.
• The neural mechanisms underlying these diverse findings in sequence learning remain unclear.
• Understanding these mechanisms is crucial for advancing cognitive neuroscience and artificial intelligence.

Purpose of the Study:

• To investigate the neural mechanisms behind varied sequence learning outcomes in humans.
• To determine if a recurrent neural network model can replicate observed interference, facilitation, and transfer effects.
• To elucidate the role of specific plasticity mechanisms in sequence learning and internal model formation.

Main Methods:

• Developed and utilized a self-organizing recurrent neural network (SORN) model.
• Incorporated spike-timing dependent plasticity (STDP), intrinsic plasticity (IP), and synaptic normalization (SN) into the SORN model.
• Trained the SORN model on sequence learning tasks mirroring human experiments to observe its learning dynamics.

Main Results:

• The SORN model successfully reproduced both interference and facilitation/transfer effects observed in human sequence learning.
• The model demonstrated that these effects arise from the network's evolving internal representations of sequences.
• All three plasticity mechanisms (STDP, IP, SN) were found to be essential for effective internal model formation and learning.

Conclusions:

• The SORN model provides a unified explanation for diverse findings in human sequence learning.
• Neuronal plasticity mechanisms, specifically STDP, IP, and SN, are fundamental drivers of sequence learning.
• The interaction between training schedules and task similarity influences learning outcomes through the network's internal representations.