Eventi del

14 2020 12:45 - 14:00
Meeting room E4-SR03, Via Roentgen, 1, 4th floor

Seeing and Comparing: When do Context Effects Occur and What Does That Mean for Decision Processes?

OLEG URMINSKY, The University of Chicago

ABSTRACT

Despite a voluminous literature, the moderators of and the mental processes involved in context effects remain unresolved.  We first investigate recent claims that context effects do not occur for visual stimuli.  We find robust context effects for both visual and non-visual stimuli, but the effects are moderated by visual elements. We then propose a tradeoff-based comparison model that accommodates recent new findings about baseline preference as a moderator of context effects and test the implications of the model on non-visual context effects. 

First, in judgments about sizes of product packages, we find both a robust attraction effect and an orientation bias, such that vertically-oriented packages are seen as smaller and show a stronger attraction effect. We propose that the attraction effect requires detection of the dominance relationship in the choice configuration and application of the relationship when making the decision. When stimuli are presented visually (vs. numerically), identifying the relative attractiveness of alternatives is more challenging, leading to a weaker attraction effect. We find robust attraction effects on choice between options with visually-presented attributes, which are stronger (a) among participants who successfully discern the relative attractiveness of alternatives and (b) when people make more attribute-based trade-offs when examining the options, which increases the detectability of the relative attractiveness.

Next, we propose a simple model (the Comparison Sampling Model), which is based on principles of evidence accumulation and estimable as a random-utility model. We show that this model can shed new light on decision-making by generating testable predictions of the circumstances necessary for context effects to occur. We examine 6 predictions regarding the presence and strength of context effects derived from the Comparison Sampling Model. Results show this model accommodates recent findings and outperforms three prior heuristics-based models of context effects in out-of-sample prediction.  Collectively, our findings suggest that it can be beneficial to think of context effects in terms of comparison-based decision processes.