How do we interpret questions? Simplified representations of knowledge guide humans’ interpretation of information requests

authors

  • Aguirre Marie
  • Brun Mélanie
  • Reboul Anne
  • Mascaro Olivier

document type

UNDEFINED

abstract

This paper investigates the cognitive mechanisms supporting humans’ interpretation of requests for information. Learners can only search for a piece of information if they know that they are ignorant about it. Thus, in principle, the interpretation of requests for information could be guided by representations of Socratic ignorance (tracking what people know that they do not know). Alternatively, the interpretation of requests for information could be simplified by relying primarily on simple knowledge tracking (i.e., merely tracking what people know). We judged these hypotheses by testing two-and-a-half-year-old toddlers (N = 18), five- to seven-year-old children (N = 72), and adults (N = 384). In our experiments, a speaker asked a question that could be disambiguated by tracking her state of knowledge. We manipulated the speakers’ visuals to modulate the complexity of the ignorance representation required to disambiguate their questions. Toddlers showed no tendency to appeal to representations of Socratic ignorance when disambiguating questions (Pilot S1). Five- to seven-year-olds exhibited a similar pattern of results, and they performed better when information requests could be disambiguated using simple knowledge tracking (Studies 1a-1b). Adults used representations of Socratic ignorance to interpret questions, but were more confident when simple knowledge tracking was sufficient to disambiguate information requests (Studies 2-3). Moreover, adults disambiguated questions as if speakers could request information about things that they were ignorant of, even when speakers had no reason to know about their ignorance (Studies 3-4). Thus, the interpretation of requests for information rests primarily on simple knowledge tracking—and not on representations of Socratic ignorance—a heuristic that reduces processing costs.

more information