On using other measures of preferences

4.3.3.1. Explicit measures

In this thesis, we defined preferences as hedonic evaluative judgments, and consequently measured them with self-reported hedonic evaluations (see section 2.1.5). Note that this approach lies within the scope of classical work regarding cognitive dissonance theory. As explained by Bem (1967), “the dependent variable in cognitive dissonance is, with very few exceptions, a subject’s self-descriptive statement of an attitude or belief” (p. 184).

In chemoreception, and as we did in this work, “preference judgments have been considered to be unidimensional ratings along a continuum such as “like-dislike””

(Schiffman, 1979, p.64). However, Schiffman’s work (1979) has suggested that the hedonic experience induced with smells is multi-dimensional. Since then, several recent studies, conducted in different cultures, have further shown that the valence can be split into several, more refined, categories of pleasant feelings (Chrea et al., 2009; Delplanque et al., 2012;

Ferdenzi et al., 2011; Pichon et al., 2012). These categories, based on verbal reports of emotional experience induced by smells, are, among others, disgust, happiness-well-being, sensuality-desire (Ferdenzi et al., 2011). Could those categories of pleasant feelings be differentially affected by decision-making processes? This question is entirely open.

In a different perspective, another explicit measure of preference one might want to consider, already mentioned while defining preference in the context of economics (section 2.1.2), is the willingness to pay, i.e. the maximum amount one is willing to pay to obtain something desired or to avoid something undesired. People exhibit a higher willingness to pay for preferred objects. The Becker-DeGroot-Marschak method (Becker, DeGroot, &

Marschak, 1964), which is the most accepted method to elicit willingness to pay, could be use to measure preferences, in our adaptations of the free-choice paradigm. It might even

be more directly relevant in the paradigm we created in Experiment 9, where participants were asked to decide, before each odor presentation, whether they wanted to spend a fix amount of money (5 CHF) on this trial or not. It would be easy to adapt this procedure in asking participants how much they are willing to pay to be delivered (or not to be delivered) with a given smell.

4.3.3.2. Implicit measures

Another direction might be to use an implicit measure of preference (for a review, see Bohner & Dickel, 2011; Gawronski, 2007; Gawronski & Bodenhausen, 2006; Petty, Fazio, &

Briñol, 2009). By implicit measures, we mean “measures that assess automatic evaluations without a person’s knowledge of what is being assessed are often referred to as implicit measures” (Briñol, Petty, & McCaslin, 2009, p. 285). Let us illustrate what we mean with the most well known (yet controversial, see e.g. Fiedler, Messner, & Bluemke, 2006) paradigm in the field of implicit measures of preferences (often referred to as attitudes in this literature):

the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998). This paradigm requires participants to categorize a given attribute (e.g., spicy) into two concepts (e.g., pleasant and unpleasant), while their reaction time is measured. The fastest pairings are assumed to reflect a stronger automatic association in memory. Adapted to the question of whether decision-making processes can modulate preference, we could measure the reaction time to associate a smell and the concepts of “pleasant” and “unpleasant”, before and after choice, to test whether the associations between chosen smells and “pleasant” and between rejected smells and “unpleasant” are made faster after versus before choice.

Using implicit measures, different facets of preferences can be measured. An evaluative self-reported response, like the one we used in the different described studies of this thesis, evaluative response which presumably measures likability (Dai, Brendl, & Ariely, 2010), is one aspect of preference. Another aspect of a stimulus’ preference, its “wantability”

or incentive value (the motivational facet of wanting; Berridge et al., 2009), has been assumed to be consciously accessible at the same time as likeability (Dai et al., 2010). Such an aspect is interesting to consider because although liking and wanting are supposed to both refer to preferences, they can be incongruent (Dai et al., 2010). The example of Loewenstein (1999) of mountaineers wanting to reach the top of the mountain at any cost, is particularly illustrative in this respect. The incentive value of a stimulus could be measured in future studies in adapting Aharon et al’s (2001) keypress procedure, where participants have to exert

effort, by pressing keys, in order to increase or decrease the presentation time of a picture.

Participants could perform exactly the same task to be delivered or not deliver with odors in order to measure their wanting for a given smell. This measure can be combined with a liking measure of the odor. Similarly to the paradigm suggested with the IAT, we can compare these measures before and after choice to investigate whether decision-making processes can impact them.

4.3.3.3. Psychophysiological and cerebral measures

Psychophysiological and cerebral measures appear as particularly helpful for tackling two questions related to the research questions developed here. First, they may be useful for investigating whether preference modulation could occur at an implicit level of processing, at least for measures like sniff, whose conscious modulation appears to be difficult (see Bensafi, Sobel, & Khan, 2007). Second, they may be useful for investigating the link between choice’s possible determinants and its consequences for preferences (rather than focusing exclusively our attention, like we did intentionally in the work, focusing instead on the consequences of decision-making processes on preferences). In the different experiments we ran, at the subjective level, we did not find systematic evidence for the possibility that when participants were presented with pairs of similarly pleasant odours, their choices were based on the slight pleasantness differences between the odours within each pairs – in this direction, we found no significant effect in Experiments 1, 5, 6, and merely a statistical trend in Experiments 7 and 8.

This possibility was explored in details by a Sharot et al. (2009) in a study that combined subjective ratings and functional imaging techniques. By investigating predicted utility (Kahneman et al., 1997) about vacation destinations, the authors showed that choices can be predicted on the basis of the Blood Oxygen Level-Dependent (BOLD) activation in the caudate nucleus. The authors proposed that this differential activity is determined by differences in expected hedonic consequences, differences that could not be captured by the subjective ratings of the participants. This intriguing result highlights the interest of using complementary indicators of subjective ratings, such as neural and psychophysiological measures, to better understand the mechanisms underlying post-choice preference changes.

This result should however be taken with caution, as the method is not immune to Chen and Risen’s (2010) critique. As subjective ratings, cerebral measures are indeed also an imperfect measure of preferences (see Chen and Risen, 2010). Psychophysiological and cerebral

correlates of post-choice preference change consequently also need to be controlled for this issue.

Numerous psychophysiological measures exist. Let us mention three that we believe are of particular relevance for our topic of research. First, facial electromyography (EMG) and more specifically, the activities of the corrugator supercilii (responsible for frowning) and of zygomaticus major (responsible for smiling) appear potentially relevant. These activities are related to the hedonic evaluation of stimuli (e.g, Larsen, Norris, & Cacioppo, 2003), including olfactory stimuli (see e.g., Armstrong, Hutchinson, Laing, & Jinks, 2007;

Delplanque et al., 2009). Facial EMG, although non-invasive, however requires the setting up of eletrodes on the face. This is a constraint that the second measure we want to present is free of: thermography. Measuring facial heat patterns with a thermic camera is indeed possible at any distance, with a high temporal and thermal resolution (see Jarlier et al., 2008; Jarlier, 2012). Importantly, facial heat patterns have been linked to the pleasantness of pictures but also of smells (see Jarlier et al., 2012). Third, heart rate variations can also be investigated as an implicit indicator of preferences, since heart rate decreases as smell becomes more pleasant (e.g., Bensafi et al., 2002a, 2002b; Delplanque et al., 2009).

After considering the nature of the stimuli used and the measures of preferences, we will now consider another key aspect of this work – decision-making – in discussing why using consequential choices, rather unconsequential ones like we did, might be an exciting perspective of research.