Expectation confirmation results in more conservative perceptual decisions: a statistical learning study
Undergraduate Just-In-Time Abstract
Poster Presentation 56.356: Tuesday, May 19, 2026, 2:45 – 6:45 pm, Banyan Breezeway
Session: Undergraduate Just-In-Time 3
Schedule of Events | Search Abstracts | Symposia | Talk Sessions | Poster Sessions
Zihui Zhao1, Joey Zhou1; 1Shenzhen University
Statistical learning (SL) enables individuals to extract regularities and form expectations. While extensive research has illustrated that SL is prevalent in both humans and animals, it remains unclear how the learned regularities shape perception. Some studies suggest that it shifts the criterion, whereas others indicate that it shapes sensitivity. Here, we tested whether regularities learned via SL modulate sensitivity and/or criterion in simple perceptual decisions. Participants (N = 34) completed a training and a testing session on the same day. In the training session, participants viewed sequences of four gratings with varying spatial frequencies (SF), arranged in either a predictable or random order. In the testing session, three types of grating sequences were displayed: expected, random, and unexpected (i.e., the final grating violated the learned predictable sequence). Participants were required to report if the last grating of the sequence was the same as or different (±3°, ±6° or ±9°) from the previous ones. By modeling the participants’ proportion of “same” responses as a function of the last grating’s orientation offset, we parametrically tested whether the width and/or baseline of the psychometric curves differed across conditions, thereby assessing whether prediction sharpens discriminability and/or shifts the decision criterion. The results showed that accuracy in the expected condition was significantly higher than in the unexpected condition. Furthermore, inspection of the psychometric functions revealed that the expected condition had a significantly lower baseline compared to the other two conditions. In contrast, the unexpected condition was associated with a higher baseline and a narrower width relative to the random condition. Together, these findings demonstrated that participants are able to acquire regularities and form expectations via SL. When stimuli are expected, participants tend to adopt a more conservative criterion. In contrast, violation of expectation leads to a more liberal criterion, accompanied by reduced sensitivity.