Prevalence-based decisions undermine visual search

Abstract

In visual search, observers make decisions about the presence or absence of a target based on their perception of a target during search. The present study investigated whether decisions can be based on observers’ expectation rather than perception of a target. In Experiment 1, participants were allowed to make target-present responses by clicking on the target or, if the target was not perceived, a target-present button. Participants used the target-present button option more frequently in difficult search trials and when target prevalence was high. Experiment 2 and 3 employed a difficult search task that encouraged the use of prevalence-based decisions. Target presence was reported faster when target prevalence was high, indicating that decisions were, in part, cognitive, and not strictly perceptual. A similar pattern of responses were made even when no targets appeared in the search (Experiment 3). The implication of these prevalence-based decisions for visual search models is discussed.

Keywords:

• Visual search
• Prevalence
• Low prevalence effect
• Decision theory
• Signal detection
• Perception

JDS was supported by the Howard Hughes Medical Institute's Scientific Teaching Program, under Award No. 52006932. JS was supported by the National Science Foundation through the Graduate STEM Fellows in K-12 Education (GK-12) Program, under Grant No. DGE-0947465. The authors wish to thank Dr. Charles Folk, Dr. Harris Schwark, and three anonymous reviewers for comments on an earlier draft, and members of the PACMANe lab for their assistance in collecting data.

Notes

¹ All post hoc probabilities were corrected using Sidak adjustments.

² It should be noted that correct rejection rates in the 96% condition could only be calculated based on four target-absent trials in the easy condition and two target-absent trials in both the moderate and hard conditions. Standard error of the means for these correct rejection rates were 0.03 (easy), 0.06 (moderate), and 0.09 (hard).

³ Trial difficulty could not be included as a factor with response type (TP click, TP button, or TA button response) in accuracy or RT due to few participants making TP button responses in easy trials. Including both factors would have excluded all but one participant in the 50% condition from analyses.

⁴ Five participants could not be included in the analysis due to never making TP button responses, resulting in n=13 in the 50% condition and n=18 in the 96% condition included in the analysis.

⁵ TP button and click RTs were combined into a general target-present RT. Three participants from the 96% condition could not be included due to making no target-absent responses in the hard trials.

⁶ Trials which timed out before a response was made (< 1% of trials) were excluded from analyses. The treatment of these trials did not significantly impact the results.

⁷ Analyses of total earned points mirrored those of accuracy.

⁸ One participant was removed from RT analysis due to never making a target-present response during the known prevalence block of trials, making it impossible to include them in any analysis of target-present RT. This is the only participant that did not make any prevalence-based decisions during this block, and in a sense their behaviour is contrary to our hypotheses. They did make target-present responses in the second block of trials (active feedback).