CHAPTER 3 ATTENTION AND INTERTEMPORAL CHOICE
3.1 Existing Literature
There are two main approaches to studying the attention effects on value-based decision making. First, the attention effect can be measured using eye-tracking, and, second, attention can be manipulated directly.
Eye-tracking studies. In eye-tracking studies, measures of visual attention
such as eye movements and gaze duration are directly observed. Most eye-tracking studies focused on the option-wise attention effect (e.g. Fiedler & Glőckner, 2012; Franco-Watkins et al., 2016; Krajbich & Rangel, 2011; Krajbich et al., 2010; Stewart et al., 2016). These studies invariably found an option-wise attention effect on value- based decision making that the more often an option was attended to, the more likely it was to be chosen (when options have positive values). For example, in an eye- tracking study of the attention effect on intertemporal choice between smaller-sooner (SS) and larger-later (LL) options, Franco-Watkins et al. (2016) found that focusing attention on LL increased the likelihood of choosing LL and focusing attention on SS decreased the likelihood of choosing LL.
Fisher and Rangel (2014; Experiment 1) was an exception. They analysed the attention effects on intertemporal choice in a component-wise way: Gaze durations on each of the four components (i.e., SS outcome, SS delay, LL outcome and LL delay) were identified independently, all of which served as the independent variables of a regression with the individual discount rate as the dependent variable. They found that attending to LL outcome decreased the individual discount rate (i.e., increased the likelihood of choosing LL) and attending to SS delay increased the individual discount rate (i.e., decreased the likelihood of choosing LL), while attending to the other two components had no significant behavioural effects. To account for this finding, Fisher and Rangel (2014) proposed a component-wise attention effect: attending to different components in the same option or along the same attribute can contribute
independently to the final choice. However, their conclusion was premature because they did not consider the possibility of the co-existence of multiple ways of the attention effect on value-based decision making, which will be discussed later.
To sum up, eye-tracking generally can be used to test multiple ways of the attention effects on value-based decision making, although this advantage was rarely taken. Despite this advantage, eye-tracking studies can only draw a correlational relationship between attention and choice/preference and thus are silent on whether attention drives or merely reflects preference (Shimojo et al., 2003; Stewart et al., 2016).
Attention-manipulation studies. To draw a causal relationship between
attention and value-based decision making, a few studies experimentally manipulated attention and tested the attention effects on decision making. Again, most of them tested the option-wise attention effect. For example, Shimojo et al. (2003) used the
gaze-manipulation paradigm to study how attention influenced the judgment of facial attractiveness. They presented their participants with a pair of two faces sequentially on a computer screen. The faces were shown alternatively for 300 ms and 900 ms respectively for the same number of repetitions, after which participants decided which one of the faces was more attractive. Option-wise attention was manipulated by varying the duration for which options were directly attended to (i.e., 300 ms vs. 900 ms per repetition), with a longer time inferred as greater attention. Armel et al. (2008) applied the same gaze-manipulation paradigm in choices between food items and choices between art posters respectively. A recent study by Störmer and Alvarez (2016) used the psychophysical attentional-cuing paradigm (Carrasco, Ling, & Read, 2004) to study the option-wise attention effect on perceived facial attractiveness. Participants were shown two faces simultaneously (for 58 ms) and then decided which one was more attractive. Attention was manipulated by a task-irrelevant attention cue (a black dot for 70 ms) at the location of one of the faces preceding the onset of face presentation. They found that the cued face was more likely to be judged as more attractive.
A few studies experimentally manipulated attribute-wise attention. To elicit attribute-wise attention, Hare et al. (2011) explicitly asked participants to attend to the healthiness, the taste or whatever features that automatically came to their minds in different conditions respectively in two-option food choices. This is an attribute-wise attention manipulation because healthiness and tastiness are different attributes or
CHAPTER 3 ATTENTION AND INTERTEMPORAL CHOICE
60 dimensions in choosing between items of food. Fisher and Rangel (2014, Experiment 2) used the gaze-manipulation paradigm to study the attribute-wise attention effect on intertemporal choice. In a standard intertemporal choice task between SS and LL, they presented participants with different attributes (i.e., outcomes and delays) alternatively. Attention was manipulated by presenting one attribute for 2,000 ms, the other for 500 ms alternatively, each repeated twice. They found that longer visual exposure, or greater attention, to the outcomes (i.e., SS and LL outcomes) than to the delays (i.e., SS and LL delays) increased the likelihood of choosing LL and thus decreased impatience. Specifically, when outcomes are exposed longer the proportion of LL choice is 53.6%, opposed to 51.6% when delays longer.
These studies, without an exception, only elicited either the option-wise or the attribute-wise attention effect. Moreover, no previous study elicited the component- wise attention. When the attention effect on value-based decisions operates in different ways, it is of principal interest in this study to synthesize these findings and to provide an integral account of them.