Alessandro Castagnetti (University of Warwick) on "Protecting the Ego: Information Selection and Updating"

Speaker:  Alessandro Castagnetti - University of Warwick
  Wednesday, January 15, 2020 at 12:00 PM Polo Santa Marta, Via Cantarane 24, Sala Vaona
This paper aims at identifying how individuals search for ego-relevant information and how they subsequently update their beliefs. In our lab experiment, participants are ranked according to either their performance in an IQ test (ego-relevant treatment) or a random number (control treatment). Subjects are incentivised to report their beliefs about whether their IQ score or their random number is in the top half of the distribution. We ask for both prior and posterior beliefs after three rounds of signals. Before the updating stage, subjects choose between information sources that vary in terms of informativeness, skewness and framing. Moreover, in a further treatment we exogenously assign subjects an information structure to investigate their updating behaviour. In particular, we ask: (1) Do individuals selectively choose less informative and positively framed information structures in order to maintain positive views about themselves? (2) Is the subsequent belief updating process biased by the information structure chosen? Our results show that subjects are significantly more likely to choose information structures that are less informative and positively framed if the rank is based on the ego-relevant task. Subjects subsequent belief formation and updating is also influenced by the information structure chosen. We find that updating differs across information structures with participants reacting not only to the signals' informativeness but also to their framing. Thus, we find that psychological biases (i.e., information structure selection) interact with cognitive biases (i.e., belief formation) in systematic ways that allows people to maintain and exacerbate overconfidence.
 

Programme Director
Maria Vittoria Levati

Publication date
September 9, 2019

Studying