The Department of Cognitive Science cordially invites you to the public defense of the PhD thesis of
Gabor Brody
Primary supervisor: Gergely Csibra
Secondary supervisor: Agnes Melinda Kovacs
External advisors:
Justin Halberda, Johns Hopkins University, USA
Zsuzsa Kaldy, University of Massachusetts, Boston
with
Jozsef Fiser, Professor, CEU, as Internal Chair.
Venue: Nador street 15, Auditorium B.
Time: 2 pm
ABSTRACT
In this dissertation we explore the representational capacities infants recruit in the process of tokening objects. We argue that infants around the first year of life are equipped with multiple
representational systems that serve to uniquely index entities. The main contribution of the present work is the proposal of a novel indexing system, one that is bound to communicative
discourses. This system is engaged in referential communication, and creates an index for every entity that is construed as under discussion by the communicative agents. Moreover, we claim that this communicative indexing system is largely independent from visual indexing. Using standard methodologies in the object individuation literature we demonstrate that indexing in a discourse is not based on the spatiotemporal characteristics of the represented entities. Although we found that in isolation infants can use spatiotemporal information to create multiple object representations, we consistently found that if the objects were presented in a referential communicative context, infants did not take the spatiotemporal separation between objects as an individuating criterion. We also explored the nature of the discourse-bound system by directly
assessing the process of index creation within this system. We devised a novel individuation paradigm, where infants did not have any direct perceptual access to the objects and had to derive numerical expectations solely from their interpretation of communicative acts. Within this paradigm, infants could recruit different types of information in order to create novel indices.
Our findings suggest that within a discourse context, both distinct communicative agents and distinct referred-to locations generated expectations of multiple objects. Crucially, infants had trouble integrating referential information from disjoint discourse contexts. Together, these results underscore the point that discourse-bound representations are not based on a first-person encoding, but bound to specific communicative contexts. Our proposal and the empirical results have various architectural ramifications for, and raise important questions about, the interface between distinct systems of indexing. Positing a discourse-bound indexing system in infancy is
empirically supported and theoretically productive as it can offer a framework to explore the developmental origins of the displacement property of human communication.