LikeWant: A new methodology to measure implicit wanting for flavors and fragrances

https://doi.org/10.1016/j.foodqual.2019.103829Get rights and content

Highlights

  • LikeWant measures consumers’ effort mobilization to obtain flavors and fragrances as rewards.

  • LikeWant measures wanting for a pleasant odor with odorless air as control.

  • LikeWant discriminates between two fine fragrances from their rewarding properties.

  • Force indicators are more sensitive than squeeze frequency.

  • Implicit wanting analyzed as participants’ irrational effort.

Abstract

According to the incentive salience hypothesis framework (Berridge, 2007, 2012; Berridge & Robinson, 1998, 2003; Berridge, Robinson, & Aldridge, 2009), wanting and liking can be dissociated, suggesting that two products having similar liking levels could trigger different wanting behaviors in consumers. Defined as a motivational state that promotes approach toward and consumption of rewarding stimuli, wanting can be measured through the Pavlovian instrumental transfer (PIT) procedure. Having emerged from animal models, the PIT procedure aims to measure the effort exerted by an organism (consumer) to obtain a particular reward (product). By adapting and optimizing existing human PIT procedures, we developed LikeWant, an innovative behavioral method that measures consumers’ motivation to pursue flavors and fragrances as rewards. Two studies were conducted to test the sensitivity of the LikeWant procedure. In the first experiment, we investigated the sensitivity of the LikeWant procedure to measure wanting for a pleasant odor with odorless air as a neutral control. In the second experiment, we assessed the ability of the LikeWant procedure to simultaneously measure wanting for two competing fine fragrances. The results showed that the LikeWant procedure is able to (1) measure wanting for a pleasant odor with odorless air as a neutral control condition and (2) discriminate between two fine fragrances on the basis of their rewarding properties, potentially enabling the use of the procedure in consumer studies.

Introduction

During the last century, competitive theories emerged to allow a better understanding of the cognitive mechanisms underlying reward processing. Of these theories, the incentive salience hypothesis (Berridge, 2007, Berridge, 2012, Berridge and Robinson, 1998, Berridge and Robinson, 2003, Berridge et al., 2009) is the most prominent. This hypothesis argues that reward processing can be dissociated both psychologically and neurobiologically into two distinct components: liking and wanting. Wanting, or incentive salience, refers to a cognitive process that induces approach toward and consumption of rewarding stimuli, whereas liking refers to the hedonic experience that is immediately gained from its consumption. In general, wanting and liking are positively correlated, meaning that the effort exerted by an organism to obtain a particular reward is proportional to the pleasure experienced from its consumption. Moreover, under particular circumstances, wanting and liking can be dissociated, meaning that an organism (consumer) could irrationally work to obtain a reward (product) that is not liked once consumed. Thus, two products having similar liking levels could potentially trigger different wanting behaviors in consumers. According to Berridge’s theory, incentive salience has both perceptual and motivational features. On the one hand, the salience attribution converts the perceived reward into an object that captures the attention of the organism. On the other hand, the incentive attribution transforms the perceived reward into an object of attraction, making the organism work to obtain it. When perceived with incentive salience, a product has thus the ability to grab consumer’s attention and elicit an appetitive and consummatory behavior. Incentive salience can be triggered not only by the perception of a rewarding unconditioned stimulus (UCS; the product itself), but also by its conditioned stimulus (CS; its brand and packaging), i.e. a formerly neutral stimulus that, after being repetitively associated with the UCS, predicts the presence of the UCS. While a CS encounter generally triggers wanting for its related UCS, the CS often becomes highly desirable and attractive per se (a “motivational magnet”; Berridge, 2007), even to an irrational degree (Berridge, 2012). That is, the CS could trigger the response produced by the UCS itself, with the organism working to obtain the CS even in the absence of the UCS. Notably, incentive salience is an unconscious mechanism that results from the interaction between the Pavlovian system and the physiological state of the organism (e.g. satiety, hunger; Zhang, Berridge, Tindell, Smith, & Aldridge, 2009) and is distinguishable from explicit desires, which involve explicit goals or expectations (see explicit wanting; Berridge & Robinson, 2003).

The incentive salience hypothesis became increasingly popular among behavioral scientists who were interested in human motivation. Although dissociation between liking and wanting components in humans has been suggested for food (Berridge, 2009, Epstein et al., 2003, Finlayson et al., 2007a, Finlayson et al., 2007b, Finlayson et al., 2008, Lemmens et al., 2009), alcohol (Arulkadacham et al., 2017, Hobbs et al., 2005), nicotine (Brauer et al., 2001, Tibboel et al., 2011), and sex (Dewitte, 2015, Krishnamurti and Loewenstein, 2012), there is ongoing debate regarding how to best dissociate these components in human studies (Havermans, 2011, Havermans, 2012). In an interesting review, Pool, Sennwald, Delplanque, Brosch, and Sander (2016) concluded that an important number of human studies include measures that do not reflect wanting and liking as defined in the animal literature, i.e. within the incentive salience hypothesis framework. Indeed, wanting is often measured through subjective ratings (45% of the studies), by way of single quantitative questions (e.g. “How much do you want to eat this item right now?”, Born et al., 2011; “How much do you want to smell this again?”, Triscoli, Croy, Olausson, & Sailer, 2014), or with questionnaires (e.g. the Partner-Specific Sexual Wanting Scale, Krishnamurti & Loewenstein, 2012; the Desires for Alcohol Questionnaire, Love, James, & Willner, 1998). Although explicit measures have shown high validity for assessing affective experiences, subjective ratings could have limitations, since wanting and liking reactions can occur at an implicit level, without conscious awareness (Berridge and Robinson, 2003, Tibboel et al., 2015). Two promising approaches were proposed to measure human wanting without using introspective methods: (1) by adapting existing implicit methods such as the Implicit Association Task (Dewitte, 2015, Koranyi et al., 2016, Tibboel et al., 2015, Tibboel et al., 2011, Wiers et al., 2002) and (2) by adapting effort measures used in animal models (for a review, see Cartoni, Balleine, & Baldassarre, 2016). The latter has the advantage of measuring wanting in a similar way to that defined in the incentive salience hypothesis framework.

Pavlovian instrumental transfer (PIT) is a key procedure assessing wanting for reward-associated cues. Having emerged from animal models (e.g. Wyvell & Berridge, 2000), the PIT procedure is composed of three main phases: the Pavlovian conditioning, the instrumental conditioning, and the transfer test. During the Pavlovian conditioning, animals are trained to associate neutral stimuli (e.g. auditory cues) with the delivery of the rewarding UCS (e.g. sucrose pellets): after repeated associations with the UCS, the previously neutral stimuli (e.g. auditory cues) become conditioned stimuli (CSs). That is, the rewarding properties of the UCS are transferred to the CSs, making the latter able to trigger the wanting response produced by the UCS itself. During the instrumental conditioning, animals are trained to perform an instrumental action (e.g. press on a lever) to receive the rewarding UCS (e.g. sucrose pellets). Finally, during the critical transfer test, the instrumental task is performed again, but this time, the CSs are presented during the session. The effect of presenting the Pavlovian cues over the instrumentally learned behavior is thus evaluated, which is known as the “Pavlovian-To-Instrumental” transfer (PIT) effect. Notably, the transfer test is normally conducted under extinction, meaning that expected rewards (UCS) are not delivered after the instrumental action. Since the hedonic experience is avoided under extinction, the investigator obtains a pure measure of wanting dissociated from liking.

An example of a successful human PIT study comes from Talmi, Seymour, Dayan, and Dolan (2008). In this study, participants first learned to associate two combined auditory and visual stimuli (CSs) with the presence and the absence of a monetary reward of 20 pence (UCS). The bimodal stimuli presented in the presence of the UCS denoted the positive CS (CS+), whereas the bimodal stimuli presented in the absence of the UCS denoted the negative CS (CS-). During the instrumental conditioning, participants learned to squeeze a handgrip to obtain the same monetary reward (UCS). Finally, during the transfer test, the authors examined the frequency with which participants squeezed the handgrip in the presence of either CS+ or CS- under extinction. Results suggested that the frequency of handgrip squeezes increases in the presence of the CS+ stimuli (i.e. the PIT effect). Pool, Brosch, Delplanque, and Sander (2015) adapted this procedure to measure wanting for a chocolate odor under different stress conditions. Similar to Talmi’s procedure, during the instrumental conditioning, participants were asked to squeeze a handgrip to trigger the release of the reward, i.e. chocolate odor (24 trials). During the Pavlovian conditioning (36 trials), participants learned to associate the presence and absence of the chocolate odor with two neutral images, i.e. CS+ and CS-, respectively. Finally, during the critical transfer test (45 trials); the squeeze frequency was measured in the presence of CSs (CS+ and CS-) and under extinction. The results of the transfer test showed that squeeze frequency increased during CS+ trials, suggesting that the PIT procedure can be successfully used to measure wanting for sweet olfactory rewards in humans.

Although existing human PIT procedures were successfully validated to measure wanting for chemosensory stimuli (Chillà et al., 2019, Pool et al., 2015), such methods cannot be applied as is in consumer research for several reasons. First, existing olfactive PIT procedures are too time-consuming (about 40 min) to be used to evaluate wanting for flavors and fragrances on a regular basis. Second, the Pavlovian learning indicators proposed in existing olfactive PIT (reaction time on the key-pressing task and likability of the CSs) cannot be used at an individual level to identify – and potentially exclude – participants who did not learn UCS-CS contingencies. Third, since PIT existing procedures end with a long series of trials performed under extinction, participants should not have previous experience with the experimental procedure. For this reason, existing PIT procedures cannot handle a within-subjects design. Enabling repeated measures is nevertheless crucial in consumer research, where participants could be asked to provide wanting data for a series of products or for a particular product over time. By adapting and optimizing existing human PIT procedures, we developed LikeWant, a new procedure that is likely to suit the following consumer research requirements: reduced experimentation costs, performance assessment at an individual level, and the ability to enable repeated measures design. One of the biggest challenges was to reduce as far as possible the execution time of the whole LikeWant procedure (from 40 to 15 min) while ensuring its sensitivity and validity. This was made possible by (1) drastically reducing the number of trials in the whole procedure, (2) integrating the instrumental learning phase (ILP) in the wanting task, (3) providing shorter instructions to participants, (4) removing the baseline, and (5) optimizing ergonomic aspects, thus facilitating the administration of the task. To assess UCS-CS learning performance at an individual level, we introduced a new task into the procedure: the Pavlovian check phase. Finally, repeated measures design has been made feasible by (1) minimizing as much as possible the number of trials performed in the extinction phase (EP), measuring thus wanting for each CS in a critical trial under extinction; and (2) by introducing a reconditioning phase (RP) immediately after the EP, providing participants with the expected reward.

In the experiments presented here, we aimed to evaluate the sensitivity of the LikeWant procedure by using food and nonfood odors (Experiment 1), extending its application to fine fragrances (Experiment 2). Existing human PIT studies showed that participants invest more effort during the presentation of a cue previously associated with the reward, CS+, than they do during the presentation of a cue associated with its absence, CS-. In particular, Pool and co-workers (2015) found that PIT procedures are able to measure wanting for chocolate by using a neutral condition – i.e. odorless air – as a control. Nevertheless, it is still uncertain as to what PIT procedures can be used to simultaneously measure wanting for two olfactory rewarding stimuli in the same experience. Recent studies showed that PIT procedures are able to discriminate between two odors characterized by a wide range of reported pleasantness – with participants exerting more effort to obtain pleasant odors than to obtain odors having a neutral hedonic value – but they lack sensitivity when discriminating between two pleasant odors (Chillà et al., 2019). In the first experiment, we investigated the sensitivity of the LikeWant procedure to measure wanting for a pleasant odor with odorless air as a neutral control. In the second experiment, we assessed the ability of the LikeWant procedure to simultaneously measure wanting for two competing fine fragrances. Finally, in these experiments we also aimed to assess and compare the sensitivity of different wanting indicators. Following the procedure usually applied in the literature (Chillà et al., 2019, Pool et al., 2015, Talmi et al., 2008), we extracted the number of squeezes surpassing 50% of each participant’s maximal grip strength, i.e. the squeeze frequency. In addition, two additional indicators (expressed in volt-seconds, V⋅s) were extracted by computing the integral of the handgrip signal with trapezoidal numerical integration: the total force (reflecting participants’ energy investment during a particular trial) and the force exceeding 50% of the participants’ maximal grip strength (reflecting participants’ superfluous energy investment during a particular trial). Force indicators have two advantages over squeeze frequency. First, since participants are explicitly instructed to squeeze the handgrip to obtain rewards, they may expect that squeeze frequency is tracked. Conversely, they are likely to be unaware that complex force measurements are being conducted by the investigator. Second, the force exerted on the handgrip is less easy to control than the squeeze frequency. The lack of awareness of what is being assessed and the lack of controllability of the response are often mentioned as requirements for implicit tests (Fazio and Olson, 2003, Greenwald and Banaji, 1995, Petty et al., 2009). For these reasons, force measurements could be valuable implicit wanting indicators. In particular, since an effort of 50% of the maximal grip strength is sufficient to obtain the expected rewards, any effort exceeding this threshold should be considered superfluous, unnecessary, irrational, and contrary to the principle of least effort (Zipf, 1949). This indicator appears to be even more relevant in light of Berridge (2007) consideration: “Conditioned incentive salience (…) is manifested at the CS as a sudden and frenzied burst of effort to obtain the UCS reward” (p. 410). We consider the force exceeding 50% of the participants’ maximal grip strength a privileged implicit wanting indicator.

Section snippets

Method

All the experience data and analysis scripts used are available on the OpenScience Framework platform at: https://osf.io/6ywk2/.

Method

All the experience data and analysis scripts used are available on the OpenScience Framework platform at: https://osf.io/6ywk2/.

Conclusion

In this paper, we presented and validated the LikeWant procedure, a new methodology to measure consumers’ implicit wanting for flavors and fragrances. Compared with existing PIT procedures, the LikeWant procedure is likely to suit the following consumer research requirements: reduced experimentation costs, performance assessment at an individual level, and the ability to enable repeated measures design. We conducted two studies to advance the validation process of the procedure. The results

Declaration of competing Interest

The author(s) declared no potential conflicts of interests with respect to the authorship and/or publication of this article.

Funding sources

This research was supported by a research grant (EMODOR – project UN9046) from Firmenich SA to David Sander and Patrik Vuilleumier.

Authorship responsibility

The submission is a truthful, original work without fabrication, fraud or plagiarism, and contains no libelous or unlawful statements. It has not been published previously except in abstract form. The manuscript is not under consideration for publication, nor will it be submitted for publication, elsewhere until a final decision has been made by this journal. The undersigned certify that each author has participated sufficiently in the work to take responsibility for its truthfulness and

Acknowledgements

The authors thank Dr. Eric Teillet from SensoStat for assistance with statistical analysis. The authors also thank all the members of the Perception and Bioresponses Department of the Research and Development Division of Firmenich SA for their precious advice. The authors, however, bear full responsibility for the paper.

References (38)

  • R.C. Havermans

    How to tell where ‘liking’ ends and ‘wanting’ begins

    Appetite

    (2012)
  • S.G.T. Lemmens et al.

    Eating what you like induces a stronger decrease of ‘wanting’ to eat

    Physiology & Behavior

    (2009)
  • E. Pool et al.

    Measuring wanting and liking from animals to humans: A systematic review

    Neuroscience & Biobehavioral Reviews

    (2016)
  • H. Tibboel et al.

    Heavy social drinkers score higher on implicit wanting and liking for alcohol than alcohol-dependent patients and light social drinkers

    Journal of Behavior Therapy and Experimental Psychiatry

    (2015)
  • H. Tibboel et al.

    Testing the validity of implicit measures of wanting and liking

    Journal of Behavior Therapy and Experimental Psychiatry

    (2011)
  • H. Tibboel et al.

    Implicit measures of “wanting” and “liking” in humans

    Neuroscience & Biobehavioral Reviews

    (2015)
  • L.J. Arulkadacham et al.

    Dissociation between wanting and liking for alcohol and caffeine: A test of the Incentive Sensitisation Theory

    Journal of Psychopharmacology

    (2017)
  • K.C. Berridge

    The debate over dopamine’s role in reward: The case for incentive salience

    Psychopharmacology

    (2007)
  • K.C. Berridge

    From prediction error to incentive salience: Mesolimbic computation of reward motivation

    European Journal of Neuroscience

    (2012)
  • View full text