Ensemble statistics accessed through proxies: Range heuristic and dependence on low-level properties in variability...

Area of Science:

• Cognitive Psychology
• Visual Perception
• Computational Neuroscience

Background:

• Humans can accurately estimate both the average size and variability within a set of items.
• The precise cognitive mechanisms underlying the estimation of size variability remain unclear.

Purpose of the Study:

• To investigate how individuals estimate the variability in size of visual arrays.
• To test whether established models of ensemble statistics estimation sufficiently explain variability judgments.

Main Methods:

• Participants compared arrays of circles differing in size variability.
• Experimental manipulations included congruency between range and variance, and low-level visual properties (e.g., filled vs. outlined circles).

Main Results:

• Accuracy in judging size variability was higher when range information was congruent with actual variance, suggesting range is used as a proxy.
• Systematic biases in judgments occurred when comparing arrays with different low-level visual characteristics, indicating sensitivity to these properties.
• Neither parallel access to all items nor random subsampling adequately explains the observed estimation strategies.

Conclusions:

• Individuals flexibly employ range and low- to mid-level visual properties as heuristics for estimating size variability.
• These findings challenge existing models that propose solely parallel extraction of ensemble statistics or random subsampling for variability estimation.