Happy Travelers Take Big Pictures: A Psychological Study with Machine Learning and Big Data

TL;DR

This study uses a large dataset of travel photos and deep learning to empirically test the psychological 'broaden-and-build' theory, finding that positive emotions are linked to wider visual attention in real-world settings.

Contribution

It demonstrates how big data and deep learning can be used to validate psychological theories outside laboratory environments.

Findings

High-rated tourist sites are associated with more wide-angle photos.

Deep learning effectively classifies photo angles at scale.

Supports the link between positive emotion and broader visual attention.

Abstract

In psychology, theory-driven researches are usually conducted with extensive laboratory experiments, yet rarely tested or disproved with big data. In this paper, we make use of 418K travel photos with traveler ratings to test the influential "broaden-and-build" theory, that suggests positive emotions broaden one's visual attention. The core hypothesis examined in this study is that positive emotion is associated with a wider attention, hence highly-rated sites would trigger wide-angle photographs. By analyzing travel photos, we find a strong correlation between a preference for wide-angle photos and the high rating of tourist sites on TripAdvisor. We are able to carry out this analysis through the use of deep learning algorithms to classify the photos into wide and narrow angles, and present this study as an exemplar of how big data and deep learning can be used to test laboratory findings in the wild.