Item Features Interact With Item Category in Their Influence on Preferences.

Shiran Oren,Tal Sela,Dino J Levy,Tom Schonberg

doi:10.3389/fpsyg.2020.00988

Shiran Oren, Tal Sela + Show 2 more

Open Access

https://doi.org/10.3389/fpsyg.2020.00988

Copy DOI

Abstract

Low-level visual features are known to play a role in value-based decision-making. However, most previous studies focused on the role of only a single low-level feature or only for one type of item. These studies also used only one method of measurement and provided a theory accounting for those specific findings. We aimed to utilize a different more robust approach. We tested the contribution of low-level visual features to value-based decision-making of three item types: fractal-art images, faces, and snack food items. We used two techniques to estimate values: subjective ratings and actual choices. We found that low-level visual features contribute to value-based decision-making even after controlling for higher level features relevant for each item category (for faces, features like eye distance and for food snacks, features like price and calories). Importantly, we show that, overall, while low-level visual features consistently contribute to value-based decision-making as was previously shown, different features distinctively contribute to preferences of specific item types, as was evident when we estimated values using both techniques. We claim that theories relying on the role of single features for individual item types do not capture the complexity of the contribution of low-level visual features to value-based decision-making. Our conclusions call for future studies using multiple item types and various measurement methods for estimating value in order to modify current theories and construct a unifying framework regarding the relationship between low-level visual features and choice.

Highlights

People make value-based choices between different items on a daily basis
As mentioned above in the methods, lab experiments consisted of only preference ratings, whilst in the online and replication experiments, participants performed the binary choice task and the preference rating task for one of the categories
In ratings data obtained from the experiments conducted in the lab, we found that each of the visual features had a different influence on preference ratings and, in some cases, an opposite effect, depending on the items’ category (Figure 2A)

Summary

Introduction

People make value-based choices between different items on a daily basis. In this process, we need to construct a representation of each of the items, assign a value to it, and choose the item we prefer (Rangel et al, 2008). In order to do so, we putatively dissect the item to its low-level characteristics, such as colors (Margaret and Hubel, 1988) and spatial frequency (De Valois et al, 1982; De Valois et al, 2000). The combination of these visual features lead to higher-level representations of the objects, allowing us to identify and assign value to the item (Riesenhuber and Poggio, 1999). More saturated and cooler colors were found to be preferred

Objectives

Methods

Results

Conclusion