Pruning representations in a distributed model of working memory: a mechanism for refreshing and removal?

Area of Science:

• Cognitive Psychology
• Computational Neuroscience

Background:

• Behavioral studies indicate attention is crucial for working memory.
• Attention is often modeled as strengthening memory representations.

Purpose of the Study:

• To propose and test an alternative model of attention in working memory.
• To investigate if attention prunes, rather than strengthens, memory representations.

Main Methods:

• Developed a computational model of working memory incorporating a pruning mechanism.
• Utilized a sampling algorithm to implement selective weakening of memory representations.
• Simulated discrete recall and continuous reproduction tasks.

Main Results:

• Pruning representations improved performance compared to a decay-free baseline.
• The model successfully explained performance enhancements observed with retro-cues.
• Simulations demonstrated benefits for both discrete and continuous memory tasks.

Conclusions:

• Attention may function by pruning less useful memory representations, not solely by strengthening them.
• A pruning mechanism offers a viable computational alternative to pure strengthening accounts.
• This model provides insights into attention's role in working memory efficiency.