Decision-level processes in rapid numerosity estimation

Area of Science:

• Cognitive Psychology
• Numerical Cognition

Background:

• Human numerical judgments exhibit consistent patterns, including exact enumeration for small quantities and underestimation for large ones.
• Current computational models explain these patterns based on stimulus representation, but decision-level processes remain less understood.

Purpose of the Study:

• To investigate the decision-level cognitive processes individuals use when rapidly estimating numerosity.
• To determine if these processes contribute to observed behavioral patterns in numerical estimation tasks.

Main Methods:

• Combined qualitative interviews, quantitative surveys, and preregistered experiments with participants estimating dot patterns under brief viewing (100 ms) and masking conditions.
• Participants' self-reported strategies (comparing, counting, partitioning) were correlated with task performance and response times.

Main Results:

• Participants reported using strategies such as visual memory comparison, dot counting, and approximate group calculations.
• Decision-level processes including comparing, partitioning, and counting were found to significantly contribute to established behavioral patterns in numerosity estimation.
• These processes systematically correlated with numerosity, response time, and estimation accuracy.

Conclusions:

• Rapid numerosity estimation patterns arise from the flexible deployment of decision-level cognitive processes, influenced by individual factors and task demands.
• Integrating these decision-level mechanisms with representation-level models will advance our understanding of human numerical cognition.