Attractor models of working memory and their modulation by reward

Area of Science:

• Theoretical Neuroscience
• Cognitive Neuroscience
• Computational Neuroscience

Background:

• Working memory (WM) is susceptible to distracters.
• Reward influences cognitive functions, including attention and memory.
• Previous models explored distractibility and reward dependence in isolation.

Purpose of the Study:

• To empirically examine the interaction between distractibility and reward in working memory.
• To parameterize an observation model reflecting nonlinear interactions between distractibility and reward.
• To investigate how distracters affect target memory under varying reward conditions.

Main Methods:

• Empirical examination of distractibility and reward dependence in working memory.
• Parameterization of an observation model based on a neuronal network model.
• Analysis of memory corruption by distracters during the delay period.

Main Results:

• Distracters partially corrupted target memory, resulting in a compromise rather than complete overwrite.
• A trend towards reduced distractibility was observed under high reward conditions.
• Findings contrast with initial theoretical predictions regarding complete memory overwrite.

Conclusions:

• The interaction between distractibility and reward in working memory is complex and nonlinear.
• Findings inform theoretical models of basal and dopamine (DA)-modulated neural bump-attractor networks.
• Partial memory corruption suggests a more nuanced mechanism than simple overwriting.