Affect Gap

affect-space-webThis post is based on the paper: “The Affect Gap in Risky Choice: Affect-Rich Outcomes Attenuate Attention to Probability Information,” authored by Thorsten Pachur,  Ralph Hertwig, and Roland Wolkewitz that appeared in Decision, 2013, Volume 1, No. 1, p 64-78.  This is a continuation of the affect/ emotion theme. It is more of a valence based idea than Lerner’s Appraisal Tendency Framework.  This is more thinking about emotion than actually experiencing it although the two can come together.

Often risky decisions involve outcomes that can create considerable emotional reactions. Should we travel by plane and tolerate a minimal risk of a fatal terrorist attack or take the car and run the risk of traffic jams and car accidents? How do people make such decisions?  Decisions under risk typically obey the principle of the maximization of expectation.
The expectation expresses the average of an option’s outcomes, each weighted by its
probability. This, of course, underlies expected utility theory and cumulative prospect theory and these models do a good job in accounting for choices among relatively affect-poor

Several authors have hypothesized qualitative differences between decisions involving affect rich prospects (e.g., a possible electric shock) and those involving affect-poor prospects (e.g., a possible $20 fine).  For instance, Rottenstreich and Hsee have argued that decisions regarding affect-rich outcomes are relatively insensitive to probability. Indeed participants’ heart rate in anticipation of an electric shock proved to be impervious to the 50% likelihood of the shock vs. 100%. The authors of this paper devise experiments to try to determine if there is an affect gap and to try to understand the cognitive process that underlie it. They use an experimental paradigm that permits them to compare one and the same person’s choices in affect-rich and affect-poor problems by determining each person’s willingness to pay.

The authors model affect-rich and affect-poor choices using expected value strategy and two noncompensatory strategies that disregard probabilities (Studies 1 and 3: minimax heuristic– according to minimax, the options are compared with regard to their worst outcomes and the option with the more attractive worst outcome is chosen; minimax thus disregards
probabilities altogether; Study 2: maximax heuristic– examines the options’ maximum (i.e., best) outcomes and chooses the one with the most attractive maximum outcome.). To evaluate the contribution of random choice to the affect gap, they also considered those who did not fit the other strategies and were thus classified as guessing.

Three studies are reported. In Studies 1 and 3, they contrast affect-rich and affect-poor choice in a domain with negative affect (medical side effects); in Study 2, they consider a domain with
positive affect (hotel amenities). In Study 3, they gauge the cognitive processes underlying
choice in affect-rich and affect-poor problems using a process-tracing methodology.

Across three studies, Pachur et al, consistently found that individuals often express systematically different preferences in affect-rich and affect-poor choice tasks—a phenomenon they refer to as the affect gap. They found that affect-rich choices were better modeled by heuristics that disregard probabilities and compare options in terms of their outcomes than by the expected value strategy. Assuming that people follow expected value in affect-poor
problems but minimax and maximax in affect-rich problems accounts for 86%–90% of cases
in which people reversed their preferences. Process data further supported the thesis that affect-rich versus affect-poor problems trigger different strategies. In the former, people looked up probabilities less frequently than outcomes and they conducted more intradimensional comparisons than in the latter.

Affect-rich problems and monetarily equivalent affect-poor problems can result in reversed
preferences within the same person. Previous research has found preferences to reverse (a)
when outcomes are framed as gains versus losses; (b) when objects with an attribute that is difficult to evaluate (i.e., whether a given attribute value is good or bad) are presented separately versus jointly; and (c) when preferences are elicited with different methods (e.g., choice vs. matching;  The preference reversals that show up in the affect gap are distinct
from these instances. They occur with outcomes that are framed with regard to the same
reference point, with the same number of objects in a choice set, and based on the same
binary choice task.

The evidence for the use of different strategies in affect-rich and affect-poor problems is consistent with the notion that affect acts “as a spotlight”, focusing people’s attention on specific types of information (e.g., outcomes). In addition, people may treat the amount of affect associated with an outcome as a proximal cue for its subjective value.

People facing rare-event risks with dreaded outcomes, such as becoming a victim in a terrorist
attack, are at risk of making poor decisions.   How can the public’s response to such risks be improved? If the affect gap is caused by the use of heuristics that neglect probabilities, interventions should aim to enhance people’s attention to probability information—for instance, by presenting probabilities as icon arrays.



1 thought on “Affect Gap

  1. Pingback: 2nd Half of 2009 JDM Research Summary - Judgment and Decision Making

Comments are closed.