Discounting the Effect of Memory on Repeated Measures of Beauty Judgment

Abstract

The intrinsic variance of beauty judgment is key to modeling beauty ratings. However, in repeated measures of beauty, observers surely make use of what they remember. To test how memory contributes to repeated beauty ratings, we asked participants to rate 75 arbitrarily named images (e.g., Fred). Initially, participants rated (1 to 7) how much beauty they felt from looking at a named image. Then participants completed two conditions. In the memory condition, participants saw only the name of an image and were asked to remember the image corresponding to that name and rate how much beauty they felt. In the repeat condition, they once again rated how much beauty they felt from looking at a named image. Lastly, in a memory check, participants tried to select which image was associated with a name. Only considering the correctly remembered trials (60%), we calculated the distribution of the differences between the initial beauty rating and that from either the memory condition or the repeat condition. The variance for the memory condition was more than double that of the repeat condition. Likewise, the initial beauty ratings predicted 84% of the variance in the repeat ratings but only 30% of the variance in the memory ratings. Cue combination studies report that observers typically combine cues by the optimal Bayesian rule: The combined reliability is the sum of the separate reliabilities for each cue, where reliability is one over variance. Assuming optimal combination of memory and immediate-perception judgment, we can discount the contribution of memory to estimate the variance of the immediate-perception judgment. Thus, in our paradigm the 0.83 variance of the repeated beauty rating corresponds to a 0.97 immediate-perception judgment variance (without memory). Overall, since there also was no significant difference in means, our results indicate that memory contributes little to repeated beauty ratings.