ABSTRACT
Objective:To investigate the protective effect of warning at different time points on eyewitness memory during repeated memory retrieval.Methods:A pilot test and artificial intelligence technology were used to develop a set of witness memory experiment procedures applicable to Chinese people. A total of 46 healthy participants were included in the pilot test and 101 healthy participants were included in the formal experiment. A mixed 3 (group of warning) ×4 (memory test time points) design was applied to this study and SPSS 23.0 software was used for data analysis.The memory of 101 healthy participants (33 in the pre-warning group, 30 in the no-warning group, and 38 in the post-warning group) using two-factor repeated measure ANOVA. The memory accuracy rate, memory misleading rate and confidence of the three groups in different memory test time points were taken as dependent variables, and the group of warning and the test time points were taken as independent variables to analyze the differences in memory among the groups in different test time points.Results:(1)In the early two repeated memory tests, the accuracy of memory in pre-warning group ((0.67±0.11), (0.67±0.12)) were significantly higher than those in no warning group ((0.58±0.15), (0.60±0.15)) (both P<0.05); at the third repeat memory test, the accuracy of post-warning group (0.63±0.12) and no warning group (0.60±0.13) were significantly lower than that of pre-warning group (0.69±0.12) (both P<0.05). (2)In the three repeated memory tests, the memory misleading rates of the pre-warning group (0.25(0.13, 0.38), 0.25(0.13, 0.38), 0.25(0.13, 0.38)) and the post-warning group (0.25(0.13, 0.38), 0.25(0.13, 0.38), 0.25(0.13, 0.38)) were significantly lower than those of the no warning group (0.63(0.34, 0.78), 0.63(0.34, 0.75), 0.63(0.25, 0.75)) (all P<0.05). (3)There were no significant difference in the answer confidence between any pair of groups at any corresponding memory test (both P>0.05). Conclusion:The protective effect of warning before misleading information is better than the warning after misleading information, and the protection of the two warnings are still working.