Endogenous precision of the number sense

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Psychophysics

Background:

• Behavioral variability and neural stochasticity indicate imprecise brain representations.
• Numerosity studies suggest an imprecise 'number sense', potentially reflecting optimal resource allocation.
• Efficient-coding models propose that representational imprecision depends on elicitation context.

Purpose of the Study:

• To investigate how representational imprecision scales with contextual factors.
• To determine if imprecision in numerosity perception is context-dependent.
• To elucidate the relationship between task demands, prior distributions, and perceptual precision.

Main Methods:

• Subjects performed numerosity estimation and discrimination tasks.
• The width of the prior distribution for number sampling was manipulated.
• Behavioral data were analyzed to quantify imprecision and its dependence on prior width and task type.

Main Results:

• Imprecision increased sublinearly with the prior distribution width across both tasks.
• The relationship between imprecision and prior width differed significantly between estimation and discrimination tasks.
• This double dependence suggests a resource constraint optimized for task-specific rewards.

Conclusions:

• Perceptual noise is endogenously determined, not solely external.
• Representational precision is dynamically modulated by both environmental context (prior distribution) and task objectives.
• The brain optimizes a trade-off between representational accuracy and neural resource costs.