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

Highlights

• Animal research showed that is possible to want a reward that is not liked, human research produced contradictory results.

• This discrepancy could be due to inconstancies in concept operationalizations.

• We systematically reviewed methodologies used to assed wanting and liking in humans.

• Expected pleasantness represents a source of confound for wanting and liking operationalizations in humans

Abstract

Animal research has shown it is possible to want a reward that is not liked once obtained. Although these findings have elicited interest, human experiments have produced contradictory results, raising doubts about the existence of separate wanting and liking influences in human reward processing. This discrepancy could be due to inconsistences in the operationalization of these concepts. We systematically reviewed the methodologies used to assess human wanting and/or liking and found that most studies operationalized these concepts in congruency with the animal literature. Nonetheless, numerous studies operationalized wanting in similar ways to those that operationalized liking. These contradictions might be driven by a major source of confound: expected pleasantness. Expected pleasantness underlies cognitive desires and does not correspond to animal liking, a hedonic experience, or to animal wanting, which relies on affective relevance, consisting of the perception of a cue associated with a relevant reward for the organism’s current physiological state. Extending the concept of affective relevance and differentiating it from expected pleasantness might improve measures of human wanting and liking.

Introduction

Psychologists and neuroscientists have long tried to understand how individuals decide to invest their limited resources to pursue a particular rewarding outcome (e.g., Delgado, 2007, Rescorla and Solomon, 1967, Spence, 1956). Common sense suggests that people decide to invest their resources to pursue the outcome they like the most. However, in many situations, individuals invest a considerable amount of effort to pursue an outcome even though after they obtain it, they do not experience it as pleasurable. A clear example occurs in the case of drug addiction, in which individuals are willing to go to extraordinary lengths to obtain a substance that will eventually elicit no pleasurable feelings during its consumption (Robinson and Berridge, 2003). In the nineties, Berridge and Robinson, 1998 proposed the incentive salience hypothesis that challenged the hedonic perspective. Proponents of this hypothesis suggested that the pursuit of an outcome is not always directly proportional to the pleasure experienced during consumption, because reward processing is a process involving multiple distinct parallel components, including the motivation to obtain a reward (i.e., wanting) and the hedonic pleasure felt during its consumption (i.e., liking; see also Berridge, 2009b). These components are typically positively correlated but can also be dissociated, thereby making organisms work for a reward that they will not appreciate once obtained.

This proposal, based on an animal model, has garnered great interest among researchers investigating motivational processes in humans (e.g., Finlayson et al., 2007b, Kringelbach et al., 2012, Mela, 2006, Nawijn et al., 2015). Several scholars have considered the independence of wanting and liking as a potential mechanism underlying a variety of human behaviors that negatively impact well-being such as overeating, pathological gambling and the consumption of addictive substances (Finlayson et al., 2007b, Pool et al., 2015c, Tibboel et al., 2011, Wölfling et al., 2011). However, human experiments have led to contradictory results, opening a debate on the existence of two dissociable components in human reward processing (Havermans, 2011, Havermans, 2012). It has even been claimed that a correct operational definition of wanting and liking as conceived by Berridge and Robinson, 1998 is lacking in human research (Havermans, 2012). Clear operational definitions are particularly important for studies conducted on humans because such studies offer much larger variability in operationalizing psychological constructs–which can be an important source of confound–than animal studies. In the present article, we systematically review the literature that investigates wanting and liking among human populations, as well as systematically describe how these concepts were operationalized regarding the important tenets of the incentive salience hypothesis. We thereby aimed to (1) systematize and quantify, across all kinds of human rewards, the contradictory operationalizations of wanting and/or liking that have been previously highlighted (Havermans, 2011, Havermans, 2012) and (2) identify possible sources of confounds that might be responsible for the contradictory results.

The incentive salience hypothesis has been conceived as an extension of early models of incentive motivation (Bindra, 1974, Bolles, 1972, Spence, 1956, Toates, 1998). These models challenged the drive reduction theory that accounted for motivated behaviors exclusively in terms of the need to reduce a particular imbalanced physiological state, such as hunger (i.e., drives), in order to reestablish homeostasis. Spence (1956) was the first to propose that the amount of energy invested in an action (e.g., walking toward a restaurant) can be influenced by the perception of external stimuli (e.g., the restaurant logo) that have been associated with a reward through the organism’s experiences. Subsequently, other incentive motivation theorists (Bindra, 1974, Bolles, 1972, Toates, 1998) suggested that the motivational increase subsequent to the perception of the reward-associated cue is proportional to the experienced hedonic pleasure, which is triggered by the consumption of the reward: the more pleasurable the reward, the bigger the increase in motivation triggered by the reward cue. Therefore, according to this suggestion, incentives should influence the organism’s motivation in a logical way: the amount of effort mobilized to obtain the reward is always justified by the hedonic experience during reward consumption. For several years, this intrinsic relationship between motivation and hedonic pleasure has been so deeply integrated in affective neuroscience that the amount of hedonic pleasure for a particular reward has been measured in multiple studies as the amount of effort mobilized to obtain it. This operationalization has been used in research conducted on animals in particular because they cannot verbally report the hedonic pleasure that they experienced (see Bindra, 1974, Bolles, 1972, Toates, 1998). Most researchers assumed that if an organism works to obtain a reward, it must mean that it likes it. In the nineties, Berridge and co-workers challenged this hedonic perspective of incentive motivational theories through a corpus of experiments conducted on rodents (Berridge and Robinson, 1998, Mahler and Berridge, 2012, Pecina et al., 2003, Wyvell and Berridge, 2000, Wyvell and Berridge, 2001). They demonstrated that it is possible to make a rodent work to obtain a reward that it does not like. The most innovative aspect of this series of experiments was the use of two different measures for incentive motivation and hedonic pleasure: the former was measured in a classic way (e.g., the increase in mobilized effort after the perception of a rewarding cue), and, critically, the latter was measured by a distinct dependent variable consisting of prototypical orofacial expressions during reward consumption. These orofacial expressions are elicited by the consumption of pleasant (e.g., sweet taste) or unpleasant (e.g., bitter taste) food and seem to be reliable indexes of hedonic experiences in several organisms (e.g., rats, apes, monkeys, human babies; see Berridge, 2000). Through these measures, Berridge and co-workers showed that two different dissociable neuronal networks underlie hedonic pleasure and incentive motivation in rodents (Berridge, 2000, Pecina and Berridge, 2000, Pecina and Berridge, 2005, Pecina et al., 2003, Wyvell and Berridge, 2000, Wyvell and Berridge, 2001). An important demonstration in this work is that increasing the level of dopamine in the mesolimbic region increases the amount of effort mobilized to obtain a reward without simultaneously modifying the measure of hedonic pleasure experienced during its consumption. From these empirical findings, the investigators formulated the incentive salience hypothesis, which postulates that reward processing involves multiple components, including one that is motivational (wanting) and another that is hedonic (liking), which rely on separate neural networks that can be dissociated under particular circumstances (Berridge and Kringelbach, 2015, Berridge and Robinson, 2003). The interaction between the organism’s brain state (e.g., increased level of mesolimbic dopamine) or physiological state (e.g., hunger/satiety) and the elements present in the environment (e.g., reward-associated cue) is an important tenet of the incentive salience hypothesis. Indeed, computations of wanting dynamically incorporate the current physiological state, reflecting the real internal state of the organism at a particular time with respect to an ideal set point that regulates homeostasis (also called k factor; Zhang et al., 2009). Examples of such physiological states could be satiation, hunger, and thirst, as well as drug effects or stress (Berridge and O’Doherty, 2014). The ability of a reward-associated stimulus to trigger a motivational state is strongly modulated by the relevance of the reward for the physiological state of the individual (Robinson and Berridge, 2013, Zhang et al., 2009). In some cases, the organism’s state can increase both cue-triggered wanting and the liking experience during reward consumption; for instance, hunger increases the relevance of a food reward that becomes both more wanted and liked (Havermans et al., 2009). In other cases, the organism’s state can selectively increase wanting without modifying liking; for instance, stress prioritizes reward relevance (Leyton, 2010), increasing cue-triggered wanting for a particular reward but not liking during reward consumption (see Pool et al., 2015c for a review).

Berridge and Robinson (2003) proposed that wanting and liking can be further classified depending on whether they are processed at an implicit or explicit level (see also Anselme and Robinson, 2015). Explicit and implicit liking both refer to the hedonic impact of the reward during its consumption and simply differ in terms of explicitness/implicitness; however, explicit and implicit wanting rely on different psychological mechanisms. Implicit wanting, also called incentive salience, relies on a Pavlovian system and refers to cue triggered motivational reactions that can occur without a conscious experience. Explicit wanting, also called cognitive desires, relies on a goal-directed system and often involves the subjective feeling of being attracted toward a desired object. Moreover, cognitive desires rely on expectations individuals have about the pleasantness of the reward, which are built based on past liking experiences. Therefore, cognitive desires are not completely independent from liking, whereas implicit wanting or incentive salience is potentially independent from any hedonic aspect of the reward including expected pleasantness (Berridge and Aldridge, 2008). Please note that in the context of the incentive salience hypothesis and the aforementioned animal literature, the term wanting refers to implicit wanting or incentive salience. Therefore, in the present article the term wanting is used to refer to incentive salience or implicit wanting.

As an extension of incentive motivation theories, the incentive salience hypothesis considers three keys elements when measuring wanting and liking. The first is the rewarding outcome (also referred to as the unconditioned stimulus), the second is the reward-associated cue (also referred to as the conditioned stimulus), and the third is the physiological state of the individual. Wanting and liking depend on different interactions of some of these elements at specific moments in time. Wanting is triggered by the interaction between an individual in a particular state and the perception of a reward cue and can be measured by the effort mobilized in the instrumental action. Notice that the incentive (i.e., the reward-associated cue or the reward) is thus presented before the instrumental action. Timing is particularly important; indeed, if the incentive is presented after the instrumental action, the process no longer relates to incentive motivation but rather to reinforcement learning. In addition, the specific influence of wanting is stronger before reward consumption, since during reward consumption, the hedonic experience is dominant. Similarly, liking is triggered by the interaction between an individual in a particular state and the consumption of a reward that is measured through the hedonic reaction during or immediately after reward consumption. Here again, timing is critical because liking is conceived as a hedonic experience; if the measurement is not taken close to reward consumption, it will reflect the encoded memory of the hedonic experience rather than the hedonic experience itself. This might be particularly problematic because memories of past hedonic experiences are used to build expected pleasantness (Balleine, 2005). Expected pleasantness, which consists of prediction and expectations about how pleasant or unpleasant something is going to be, represents the mechanism underlying cognitive desires that do not correspond to either animal liking or animal wanting, but rather a distinct motivational control system (i.e., goal-directed system) of reward-seeking behaviors (Berridge and Aldridge, 2008, Berridge and O’Doherty, 2014, Dickinson and Balleine, 1994, Wassum et al., 2011b.

Given the importance of the rewards and the reward cues for measures of the incentive salience hypothesis in animals, we decided to systematically describe different aspects of the rewards or reward cues (e.g., kind of reward, format) presented in the methodological procedures that measure wanting and liking in humans.

The incentive salience hypothesis, which has been formulated on the basis of an animal model (Berridge and Robinson, 1998, Berridge and Robinson, 2003), has garnered great interest among researchers who are investigating motivational processes in humans. Several lines of research have subsequently been launched to investigate the effect of dopamine deregulation on motivation and hedonic pleasure for a particular reward (e.g., Brauer et al., 2001, Evans et al., 2006, Volkow et al., 1997), the role of wanting and liking in addictive behaviors (e.g., Goldstein et al., 2010, Tibboel et al., 2011, Wachtel et al., 2002), or the role of these two components in the normal processing of rewards related to different needs, such as offspring caretaking (with babies) or nourishment (with food) (e.g., Finlayson et al., 2007a, Parsons et al., 2011). This corpus of experiments provided evidence supporting the idea that the same processes found in rodents could potentially exist in humans. More particularly, in clinical disorders involving dopamine deregulation (Evans et al., 2006, Volkow et al., 1997), it has been argued that the level of mesolimbic dopamine influences the motivational processes without necessarily modifying the hedonic experience of reward consumption. Moreover, it has been suggested that different brain regions are activated by a motivational state such as the expectation of a reward (e.g., amygdala; O’Doherty et al., 2002, Small et al., 2008) and by a hedonic state such as the consumption of a reward (e.g., orbitofrontal cortex; O’Doherty et al., 2002, Small et al., 2008). Although the aforementioned findings support the existence of two distinct components in human reward processing that reflect wanting and liking, other experiments have provided contradictory evidence against a dissociation between wanting and liking in humans (Havermans, 2011, Havermans, 2012, Tibboel et al., 2011). Specifically, Havermans, 2011, Havermans, 2012 highlighted that in studies investigating wanting and liking for food reward in humans, construct operationalizations are far from the original incentive salience hypothesis and often contradict each other: in some cases, a similar operationalization is used to measure wanting in one study and liking in another (Finlayson et al., 2007a, Lemmens et al., 2009). He argued that, in research conducted on humans, measures of wanting and liking still need to be validated by dissociating them under precise circumstances that are clearly predicted by the incentive salience hypothesis. In the absence of such a validation, the differential contributions of wanting and liking found in studies conducted on humans are likely to reflect poor construct validity rather than real effects. He also proposed abandoning the distinction between wanting and liking in the investigation of food reward in humans, claiming that wanting and liking are so intrinsically related that they cannot be considered as two distinct components having separate influences.

Several researchers agree that the incentive salience hypothesis has important explanatory power in the understanding of various human behaviors, in particular problematic behavior such as overeating, addictive consumption of substances or pathological gambling (Finlayson et al., 2007b, Goldstein et al., 2010, Pool et al., 2015c, Wölfling et al., 2011). Nonetheless, results of studies investigating the incentive salience hypothesis with food reward in humans led to skeptical conclusions being drawn concerning the existence of wanting and liking as two distinct components with separate influences (Havermans, 2011, Havermans, 2012). These criticisms raised two important aspects that seem to be problematic for the hypothesis: (1) operationalizations of wanting and liking are often far from the original incentive salience hypothesis and (2) measures of wanting and liking are inconsistent across studies and often contradict one another.

Here, we systematically review studies on wanting and liking for all rewards in humans to estimate the extent to which these aspects represent a problem. We systematically describe (1) how wanting and liking were measured across these studies and (2) how the methodological procedures integrated the key elements (i.e., reward cue, reward consumption and their respective timing) of the incentive salience hypothesis. From the results of this systematic description of the existing literature, we argue that the majority of studies seem to correctly integrate the main tenets of the incentive salience hypothesis; however, numerous studies operationalize the concepts of wanting and liking in contradictory ways. We claim that these contradictory operationalizations are often derived from confusion over the concept of expected pleasantness that is sometimes considered liking, but at other times considered wanting. Finally, we suggest that clarifying the distinctions between (1) expected pleasantness and affective relevance and (2) experience and memory could improve the conceptual clarity of the mechanisms involved in wanting and liking, thereby reducing sources of confusion when operationalizing these constructs.