Neural Models of the Psychosemantics of `Most'

TL;DR

This paper explores neural network models, specifically CNN and RAM, to understand how the meaning of the quantifier 'most' is processed in visual verification tasks, revealing similarities and differences with human performance.

Contribution

It introduces neural models for psychosemantic tasks involving 'most', demonstrating their potential to mirror human performance and offering insights into cognitive processing of quantifiers.

Findings

Neural models show sensitivity to set size ratios, similar to humans.

Models exhibit different patterns based on image type, highlighting nuanced cognitive mechanisms.

Results suggest neural networks can serve as cognitive models for quantifier understanding.

Abstract

How are the meanings of linguistic expressions related to their use in concrete cognitive tasks? Visual identification tasks show human speakers can exhibit considerable variation in their understanding, representation and verification of certain quantifiers. This paper initiates an investigation into neural models of these psycho-semantic tasks. We trained two types of network -- a convolutional neural network (CNN) model and a recurrent model of visual attention (RAM) -- on the "most" verification task from \citet{Pietroski2009}, manipulating the visual scene and novel notions of task duration. Our results qualitatively mirror certain features of human performance (such as sensitivity to the ratio of set sizes, indicating a reliance on approximate number) while differing in interesting ways (such as exhibiting a subtly different pattern for the effect of image type). We conclude by discussing the prospects for using neural models as cognitive models of this and other psychosemantic tasks.