Deep daxes: Mutual exclusivity arises through both learning biases and pragmatic strategies in neural networks

TL;DR

This paper explores how neural networks can develop mutual exclusivity in word learning, influenced by learning biases and pragmatic strategies, with implications for both AI models and cognitive science.

Contribution

It demonstrates that mutual exclusivity emerges in neural models through constraints in learning and selection, especially when words compete for meaning, across symbolic and visual data.

Findings

Neural models exhibit mutual exclusivity when words compete for referents.

Constraints in learning and selection foster mutual exclusivity in neural networks.

Mutual exclusivity arises in both symbolic and image-based data.

Abstract

Children's tendency to associate novel words with novel referents has been taken to reflect a bias toward mutual exclusivity. This tendency may be advantageous both as (1) an ad-hoc referent selection heuristic to single out referents lacking a label and as (2) an organizing principle of lexical acquisition. This paper investigates under which circumstances cross-situational neural models can come to exhibit analogous behavior to children, focusing on these two possibilities and their interaction. To this end, we evaluate neural networks' on both symbolic data and, as a first, on large-scale image data. We find that constraints in both learning and selection can foster mutual exclusivity, as long as they put words in competition for lexical meaning. For computational models, these findings clarify the role of available options for better performance in tasks where mutual exclusivity is advantageous. For cognitive research, they highlight latent interactions between word learning, referent selection mechanisms, and the structure of stimuli of varying complexity: symbolic and visual.