“…Of course this laboratory task surely overestimates the rate of errors in real discourse, not least because the design deliberately deprives participants of most cues that could enable them to be sensitive to perspective without representing the director's perspective. Nonetheless, these error rates are higher in participants with higher traits for autism and psychosis (Abu‐Akel, Wood, Hansen, & Apperly, ), self‐reported symptoms of depression (Nilsen & Duong, ), and induced positive mood (Converse, Lin, Keysar, & Epley, ), suggesting that they are a valid indicator of variation in social communication performance. Also error rates and/or other indices, such as decision time and rate of change in preference for the target compared with the distractor, are positively influenced by motivation and adversely influenced by concurrent working memory load (Cane et al., ; Lin, Keysar, & Epley, ) and the complexity of the director's instruction (Wang, Frisson, Ali, & Apperly, ).…”