Imagine No Resources: Attention Is Selection and Normalization for Choice.

Area of Science:

• Cognitive Psychology
• Neuroscience
• Computational Neuroscience

Background:

• Traditional attention research assumes limited capacity.
• Theories include Welford and Broadbent's channel capacity and Kahneman's resource models.
• These models offer limited explanations for attention's role in task computation.

Purpose of the Study:

• To challenge limited-capacity theories of attention.
• To investigate the role and necessity of capacity limitations.
• To propose an alternative computational framework for attention.

Main Methods:

• Analysis of performance under varying cognitive loads.
• Development and testing of computational models with unlimited, limited, and fixed capacity assumptions.
• Review of computational attention models across diverse fields.

Main Results:

• Performance effects attributed to load can be replicated by models without capacity limits.
• Limited-capacity and resource theories provide insufficient answers regarding capacity's function.
• Attention is better understood as goal-driven information selection.

Conclusions:

• Attention is not necessarily a limited resource.
• A computational perspective, viewing attention as choice implemented via gain control and normalization, explains attentional phenomena.
• This framework integrates findings from associative learning, categorization, and artificial intelligence.