Systematic Bias Toward Authentic Painting Styles Revealed Through Style-Transfer Techniques

Poster Presentation 36.312: Sunday, May 17, 2026, 2:45 – 6:45 pm, Banyan Breezeway
Session: Visual Memory: Long-term memory

Xiyan JI1 (), Hiroyuki Tsuda2, Jun Saiki1; 1Kyoto University, 2Shizuoka Institute of Science and Technology

Human memory does not simply reproduce the visual world; it shows systematic biases. The present study investigates whether long-term memory for painting styles shows categorical biases toward authentic artistic styles, and whether styles that elicit stronger biases related to the images’ instinct properties—memorability. To address these questions, we conducted Experiment 1 and a replication, Experiment 2. In Experiment 1, 81 style-transfered images were created by combining 75 landscape contents with four authentic styles—Braque, Kandinsky, Monet, and Rembrandt. Participants remembered 25 study styles and reported it in a continuous-report task. Memory responses did not distribute uniformly: they concentrated on the authentic styles and on mixed styles near the center of style space. According to the analysis of standard memory error, the memory precision for authentic styles was better than mixed styles around the center, showing the memory bias for authentic styles and mixed styles had different mechanisms. Moreover, among the four authentic styles, the strongest bias appeared for Kandinsky’s style, which had the highest memorability score. Experiment 2 repeated this paradigm with four different authentic styles—Gogh, Marc, Hokusai, and Rubens—to test whether the pattern from Experiment 1 reflects a general property of style memory rather than an effect of specific styles. The results also showed higher response frequencies for both the authentic styles and the mixed styles near the center of style space. Moreover, Marc’s authentic style produced the strongest bias, and Marc also had the highest memorability score. Our results reveal two biases in memory for painting styles: attraction toward authentic styles and attraction toward the mean style of the stimulus set. These biases may reflect partly distinct mechanisms. In particular, authentic styles may act as reference points, with more memorable authentic styles producing stronger attraction.