Poster #36 - Statistical Contributions to Knowledge Organization Throughout Development

• In Event: 3-159 - Poster Session 12
  In Poster Session: PS 12 Section - Cognitive Process

Integrative Statement

Fundamental to many cognitive processes, such as memory, reasoning, and learning, is the organization of our knowledge about the world into a lexico-semantic network that consists of concepts linked by relations. For example, our knowledge of chickens does not exist in isolation, but can instead be related to our knowledge of other members of the same stable, “taxonomic” category (e.g., birds), and other concepts with which chickens are associated, such as eggs. Where does this relational knowledge come from?

To date, research on the origins of relational knowledge has focused on the emergence of knowledge about how concepts are related through membership in the same taxonomic category. In contrast, association has received little attention: Many theoretical accounts of the emergence of relational knowledge either omit a role for association, or identify association as only contributing to relational knowledge early in development. Moreover, we know little about how association is learned. In the present research, we focused on how, starting early in development, such relations may be learned through sensitivity to statistical regularities readily available in environmental input with which the labels that denote different concepts co-occur. Furthermore, we investigated the contributions of such co-occurrence regularities in both early childhood and adulthood.

The purpose of the present research was therefore to assess young children (4 years) and adults’ relational knowledge for concepts whose labels reliably co-occur in environmental input (measured from CHILDES corpora), such as “sock” and “foot”, versus concepts that are taxonomically related, such as “cat” and “sheep”. Moreover, to assess the generalizability of the contributions of co-occurrence versus taxonomic relatedness, we measured knowledge for these relations in two separate experiments with qualitatively different paradigms both designed to be appropriate for children and adults. In Experiment 1, we used a “Match Verification” paradigm in which participants identified whether a picture was “of the same thing” as a preceding word (e.g., if the word “chicken” is followed by a picture of a chicken, the correct answer is “yes”, whereas the correct answer for any non-chicken picture is “no”). Relational knowledge in this paradigm was inferred from an Interference Score that captured the degree to which the accuracy (children) or speed (adults) of “no” responses was reduced for related versus unrelated pairs. In Experiment 2, we used a “Cued Recall” paradigm in which participants heard a series of word pairs, then were tested on their ability to recall the 2nd word in each pair when given the 1st. In this paradigm, relational knowledge was inferred from recall accuracy for related versus unrelated word pairs.

The results of both experiments revealed that co-occurrence substantially influenced relational knowledge in adults and children. In contrast, taxonomic relatedness only influenced relational knowledge in adults, and remained weaker than co-occurrence. These results attest to a critical role in the emergence of semantic organization according to relations between concepts for sensitivity to statistical co-occurrence regularities in the environment. Moreover, this substantial influence of co-occurrence regularities continues to persist throughout adulthood.

Authors

• Layla Unger, The Ohio State University
  Presenting Author

• Vladimir M. Sloutsky, The Ohio State University
  Non-Presenting Author