Dynamic Evaluation of High- and Low-Creativity Drawings by Artist and Nonartist Raters

Abstract

How does the quality of artworks change throughout the process of creation? To address this question, artists and nonartists rated the quality of in-progress stages of 20 emerging drawings: 10 whose final state had been rated by artists as highly creative and 10 whose final state had been rated as less creative. Artists associated quality with originality, but nonartists valued realism. Slopes of the quality trajectories of the emerging drawings, based on evaluations of both artists and nonartists, were reliably shallower for ultimately high-creativity final products than low-creativity final products. This suggests that less creative works evolve in a linear, incremental fashion, yet more creative works develop less predictably. No differences in the slopes were found when nonartists' choices of high- versus low-creativity drawings were compared. Thus, unlike artist raters, nonartists' choices of high- and low-creativity drawings do not seem to tap into any meaningful underlying characteristics of the artistic process.

We thank Arthur Reber and Elisabeth Brauner for constructive comments. Thanks also to Abraham Braunstein, Chasya Milgrom, Angelika Seidel, Shira Sheff, Karalyn Shimmyo, Shlomo Stern, and Lillian Sultan, who assisted with data collection and organization. Portions of this research were presented at the 2008 biannual meeting of the International Association of Empirical Aesthetics in Chicago, IL.

Notes

Note. For Intercept (B_0i) and Time (B_1i), df = 18. The dependent variable is average Rasch quality for each still frame. Reliability estimates for Intercept and Time = .947 and .847, respectively.

∗p < .05. ∗∗∗p < .001.

Note. For Intercept (B_0i) and Time (B_1i), df = 18. The dependent variable is average Rasch quality for each still frame. Reliability estimates for Intercept and Time = .252 and .340, respectively.

∗p < .05. ∗∗p < .01. ∗∗∗p < .001. ^t p < .10.

Note. For Intercept (B_0i) and Time (B_1i), df = 18. The dependent variable is average Rasch quality for each still frame. Reliability estimates for Intercept and Time = .878 and .807, respectively.

∗p < .05. ∗∗∗p < .001.

Note. For Intercept (B_0i) and Time (B_1i), df = 18. The dependent variable is average Rasch quality for each still frame. Reliability estimates for Intercept and Time = .774 and .803, respectively.

∗p < .05. ∗∗∗p < .001.