After the instructions children were asked two things: first, if they really knew which PlayPerson to select, children were told to point to him/her. But if they did not really know which PlayPerson to select, the children were told to point to a ‘mystery man’. Second, children had to tell the experimenter if s/he had given them enough 17-AAG datasheet information to find the PlayPerson or not. Children pointed to the ‘mystery man’ at rates of 68%, showing that in the majority of trials they were aware that they did not know enough

to select a PlayPerson. Nevertheless, subsequently they accepted that the experimenter had said enough at rates of 80%. These findings are straightforwardly in line with our proposal about pragmatic tolerance. Children may choose not to correct their interlocutor when asked to evaluate the instructions in a binary decision task, despite being aware that the instructions are not optimal. Therefore, it is likely that children’s sensitivity to ambiguity in the referential communication task has been underestimated due to pragmatic tolerance4. Additionally, research by Davies and Katsos (2010) using the referential communication paradigm can shed some

light on factors affecting the extent of pragmatic tolerance. Motivated by earlier versions of the present work (Katsos & Smith, 2010), Davies and Katsos (2010) tested English-speaking 5- to 6-year-olds and adults with both under- and over-informative instructions. In a binary judgment task, Anticancer Compound Library order over-informative instructions were accepted at equal rates as the optimal ones by the children, suggesting

lack of sensitivity to over-informativeness. The adults on the other hand rejected over-informative instructions significantly more than optimal instructions, giving rise to a similar child–adult discrepancy as in our experiment 1 for underinformativeness. However, when participants were given a magnitude estimation scale, both children and adults rated the over-informative instructions significantly lower than the optimal ones. Thus, Davies and Katsos (2010) conclude that pragmatic tolerance applies to over-informativeness Demeclocycline as well. Both children and adults rejected underinformative utterances significantly more often than over-informative utterances in the binary judgement task, suggesting that they are less tolerant of underinformativeness than over-informativeness. This makes sense in the referential communication paradigm, as the underinformativeness of the instructions (e.g. ‘pass me the star’ in a display with two stars) precludes participants from establishing the referent of the noun phrase. Hence, these findings suggest that pragmatic tolerance is further modulated by whether fundamental components of the speech act are jeopardized, such as establishing reference and satisfying presuppositions. Finally, we consider whether children are more tolerant than adults, and if so, why.