An interference model for visual and verbal working memory

Area of Science:

• Cognitive Psychology
• Neuroscience
• Computational Modeling

Background:

• Working memory (WM) research traditionally divided into two streams: sequential item lists and simultaneous feature arrays.
• Existing models struggle to integrate findings from these distinct WM research traditions.

Purpose of the Study:

• To introduce a formal Interference Model (IM) for working memory.
• To unify benchmark findings from both sequential and simultaneous WM paradigms within a single theoretical framework.

Main Methods:

• Developed the Interference Model (IM) based on context-bound encoding and limited precision bindings.
• Applied the IM to two experiments combining sequential list presentation with similarity-based selection from candidate sets.
• Investigated effects of memory set size, serial position, output position, and spatial/temporal interference.

Main Results:

• The IM accurately predicts error distributions in continuous feature reproduction and their dependence on memory set size.
• The model explains serial position effects, output position effects, and nontarget intrusions based on temporal and spatial proximity.
• Experimental results align with the IM's assumptions regarding encoding strength gradients and interference mechanisms.

Conclusions:

• The Interference Model (IM) provides a unified explanation for diverse working memory phenomena.
• WM contents are encoded via context bindings with limited precision, influenced by attention and interference.
• The model advances our understanding of memory dynamics, including primacy/recency gradients and output interference.