Adding statistical regularity results in a global slowdown in visual search.

Area of Science:

• Cognitive Psychology
• Neuroscience
• Statistical Learning Theory

Background:

• Statistical learning theories posit that implicit regularities enhance task performance beyond general practice.
• Previous research suggests that consistent patterns in tasks should lead to improved efficiency.

Purpose of the Study:

• To investigate the impact of mixing consistent and random task conditions on visual search performance.
• To challenge the assumption that embedding regularities always improves online performance.

Main Methods:

• Comparing performance in visual search tasks with consistent-mapping, random-mapping, and mixed-mapping conditions.
• Replicating findings with varied stimuli and different types of regularities.

Main Results:

• Performance was unexpectedly better in a purely random visual search compared to a mixed design that included a beneficial regularity.
• Mixing consistent and random conditions resulted in a significant overall slowdown in performance.
• The detrimental effect was linked to the reduced predictive validity of the regularity in mixed designs.

Conclusions:

• The cognitive system may reduce reliance on prior information when its predictive value is low, leading to performance slowdown.
• The human cognitive system appears to prioritize gathering and implementing statistical information, even at the cost of speed.
• Findings challenge existing statistical learning theories by demonstrating a performance decrement in mixed regularity conditions.