RSPM performance (M = 62%, SD = 21%) was highly varied across participants, and linear modeling indicated a highly significant relationship between RSPM accuracy and both D’ sensitivity for texts (b = 0.81, p = .0001) and overall judgment accuracy for comments (b = 0.29, p = .0001).
No association was found for empathy, but two empathy subscales approached significance, signaling a possible advantage for individuals who can more capably “read the mind” of the author in human originated works.
While previous work with AI-generated artistic expression found hints of a relationship (that did not reach statistical significance) between human evaluation accuracy and trait-level empathy, the present results offer no further corroboration of any such relationship.