In Progress in Brain Research: Appetite and Body Weight. T.C. Kirkham & S.J. Cooper (Eds.), Academic Press, pp. 191-216, 2007.

CHAPTE R 8

Brain Reward Systems for Food Incentives and Hedonics in Normal Appetite and Eating Disorders

KENT C. BERRIDGE

I. Introduction II. Possible Roles of Brain Reward Systems in Eating Disorders A. Reward Dysfunction as Cause B. Passively Distorted Reward Function as Consequence C. Normal Resilience in Brain Reward III. Understanding Brain Reward Systems for Food “Liking” and “Wanting” A. Measuring Pleasure B. Brain Systems for Food Pleasure C. Pleasure Brain Hierarchy D. Forebrain Limbic Hedonic Hot Spot in Nucleus Accumbens E. Ventral Pallidum: “Liking” and “Wanting” Pivot Point for Limbic Food Reward Circuits IV. “Wanting” without “Liking” A. What Is “Wanting”? B. Addiction and Incentive Sensitization C. Is There a Neural Sensitization Role in Food Addictions? D. Cognitive Goals and Ordinary Wanting V. A Brief History of Appetite: Food Incentives, Not Hunger Drives A. Separating Reward and Drive Reduction VI. Connecting Brain Reward and Regulatory Systems VII. Conclusion Acknowledgments References

What brain reward systems mediate disorders? This chapter identifies motivational “wanting” and hedonic recent discoveries in hedonic “liking” “liking” for food rewards? And what mechanisms, such as the location of opioid roles might these systems play in eating hedonic hotspots in nucleus accumbens

191 192 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE and ventral pallidum that paint a pleasure A. Reward Dysfunction as Cause gloss onto sweet sensation. It also considers First, it is possible that some aspects of other incentive motivation systems that brain reward function may go wrong and mediate only a nonhedonic “wanting” actually cause an eating disorder. Foods component of reward, such as nearby might become hedonically “liked” too limbic opioid “wanting” zones, mesolimbic much or too little via reward dysfunction. dopamine contributions to incentive Or incentive salience “wanting” to eat salience, and other components of brain might detach from normal close association limbic systems, and discusses potential with hedonic “liking,” leading to changes roles in eating disorders. in motivated food consumption that are no longer hedonically driven. Or yet again, I. INTRODUCTION suppression of positive hedonic reward systems or activation of dysphoric stress systems might prompt persistent attempts Obesity, bulimia, anorexia, and related to self-medicate by eating palatable food. eating disorders have risen in recent All of these possibilities have been sug- decades, leading to increasing concern. Can gested at one time or another. Each of them improved knowledge about brain reward deserves consideration because different systems guide us in thinking about eating answers might apply to different disorders. disorders and normal eating? First it is important to recognize that brain reward systems are active partici- B. Passively Distorted Reward Function pants, not just passive conduits, in the act as Consequence of eating. Sweetness and other food tastes are merely sensation, and their pleasure As a second category of possibilities, arises within the brain. Hedonic brain brain reward systems might remain intrin- systems must actively paint the pleasure sically normal and have no essential pathol- onto sensation to generate a “liking” ogy in eating disorders, but still become reaction—as a sort of “pleasure gloss.” What distorted in function as a passive secondary brain systems paint a pleasure gloss onto consequence of disordered intake. In that sensation? And what brain systems convert case, brain systems of “liking” and pleasure into a desire to eat? Answers to “wanting” might well attempt to function these questions requires combining infor- normally. The abnormal feedback from mation from neuroimaging experiments physiological signals that are altered by and studies of eating in humans and infor- binges of eating or by periods of anorexia mation from brain manipulation and phar- might induce reward dysfunction as a con- macological experiments that ethically can sequence of the behavioral disorder that be done only in animals. arose from other causes. This would provide a potential red herring to neurosci- entists searching for causes of eating disor- II. POSSIBLE ROLES OF ders, because brain abnormalities might BRAIN REWARD SYSTEMS IN appear as neural markers for a particular EATINGDISORDERS disorder, but be mistaken as causes when they were actually consequences. However, To begin with, we can sketch several it might still provide a window of opportu- alternative possibilities for how brain nity for medication treatments that aim reward systems might function in any par- to correct eating behavior in part by modu- ticular eating disorder. Let us set out some lating reward function back to a normal alternatives to frame the issue. range. III. UNDERSTANDING BRAIN REWARD SYSTEMS FOR FOOD “LIKING” AND “WANTING” 193

C. Normal Resilience in Brain Reward underlying factors are altering eating, even though it will not reverse those causal Third, it is possible that most aspects of factors. brain reward systems will function even Or instead should treatment be focused more normally than suggested by the pas- entirely on separate brain or peripheral sively distorted consequence model above. targets that are unrelated to food reward? Many compensatory changes can take place That might be the best choice if brain in response to physiological alterations, to reward systems simply remain normal in all oppose them via homeostatic or negative cases of eating disorders, and thus perhaps feedback corrections. The final consequence essentially irrelevant to the expression of of those compensations might restore nor- pathological eating behavior. mality to brain reward functions. In such Placing these alternatives side by side cases, the causes of eating disorder might helps illustrate that there are therapeutic then be found to lie completely outside implications that would follow from a brain reward functions. Indeed, brain better understanding of brain reward reward functions would persist largely nor- systems. Only if we know how food reward mally, and may even serve as aids to even- is processed normally in the brain will we tually help spontaneously normalize eating be able to recognize pathology in brain behavior even without treatment. reward function. And only if we can recog- The answer to which of these alternative nize reward pathology when it occurs will possibilities is best may well vary from case we be able to judge which of the possibili- to case. Different eating disorders may ties above best applies to a particular eating require different answers. Perhaps even dif- disorder. ferent individuals with the “same” disorder will involve different answers, at least if there are distinct subtypes within the major III. UNDERSTANDING BRAIN types of eating disorder. REWARD SYSTEMS FOR FOOD It is important to strive toward discover- “LIKING” AND “WANTING” ing which answers are most correct for par- ticular disorders or subtypes, because those This section turns to some issues answers carry implications for what treat- involved in measuring and understanding ment strategy might be best. For example, components of brain reward function should one try to restore normal eating by (Berridge and Robinson, 2003; Everitt and reversing brain reward dysfunction via Robbins, 2005). At the heart of reward is medications to correct the underlying hedonic impact or pleasure, and so it problem? That would be appropriate if is fitting to begin with the practical pro- reward dysfunction is the underlying blem of measuring pleasure “liking” for cause. food rewards in affective neuroscience Or should one use drugs instead only studies. as compensating medications, not cures? Such a medication might aim to boost A. Measuring Pleasure aspects of brain reward function and so correct eating, even though it may not Fortunately for psychologists and address the original underlying cause? For neuroscientists, pleasure is not just a meta- example, just as aspirin often helps treat physical will-o’-the-wisp. Pleasure is a psy- pain, even though the original cause of pain chological process with neural reality, and was never a deficit in endogenous aspirin, has objective markers in brain and behav- so a medication that altered reward systems ior. These objective aspects give a tremen- might still help to oppose whatever original dously useful handle to neuroscientists 194 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE and psychologists in their efforts to gain a B. Brain Systems for Food Pleasure scientific purchase on pleasure. To identify the brain mechanisms that What brain systems paint a pleasure generate pleasure we must be able to iden- gloss onto mere sensation? Many brain sites tify when pleasure “liking” occurs or are activated in humans by food pleasures: changes in magnitude. A useful “liking” Cortical sites in the front of the brain impli- reaction to measure taste pleasure is the cated in the regulation of emotion, such as affective facial expression elicited by the orbitofrontal cortex and anterior cingulate hedonic impact of sweet tastes in newborn cortex, gustatory-visceral-emotion-related human infants. Fortunately for pleasure zones of cortex such as insular cortex; sub- causation studies, many animals display cortical forebrain limbic structures such as “liking–disliking” reactions elicited by amygdala, nucleus accumbens, and ventral sweet/bitter tastes that are similar and pallidum; mesolimbic dopamine projec- homologous to affective facial expressions tions and even deep brainstem sites (Berns to the same tastes displayed by human et al., 2001; Cardinal et al., 2002; Everitt infants (Grill and Norgren, 1978b; Steiner, and Robbins, 2005; Kringelbach, 2004; 1973; Steiner et al., 2001). These affective Kringelbach et al., 2004; Levine et al., 2003; expressions seem to have developed from O’Doherty et al., 2002; Pelchat et al., the same evolutionary source in humans, 2004; Rolls, 2005; Schultz, 2006; Small orangutans, chimpanzees, monkeys, and et al., 2001; Volkow et al., 2002; Wang et al., even rats and mice (Berridge, 2000; Steiner 2004a). All of these code pleasurable foods, et al., 2001). Sweet tastes elicit positive facial in the sense of sometimes activating during “liking” expressions (tongue protrusions, the experience of seeing, smelling, tasting, etc.), whereas bitter tastes instead elicit or eating palatable foods. facial “disliking” expressions (gapes, etc.). A particular set of taste “liking–dislik- But let us also ask: Which of these many ing” reactions are remarkably similar across brain structures actually cause the pleasure species, and even their apparent differences of foods? Do all generate pleasure “liking” often reflect a deeper shared identity, such or only some? Some activations might as identical allometric timing laws for reflect causes of pleasure, whereas other expression duration that are scaled to the activations might reflect consequences of size of the species. For example, human or pleasure that was caused elsewhere. How gorilla tongue protrusions to sweetness or can causation be identified? Typically only gapes to bitterness may appear languidly by results of brain manipulation studies: a slow, whereas the same reactions by rats or manipulation of a particular brain system mice seem blinkingly fast, yet, they are will reveal pleasure causation if it produces actually the “same” reactions durations in an increase or decrease in “liking” reactions what is called an allometric sense; that is, to food pleasure. each species is timed proportionally to their Recent years have seen progress in evolved sizes and that timing is pro- identifying brain systems responsible for grammed deep in their brains. Such shared generating the pleasure gloss that makes universals further underline the common palatable foods “liked” (Berridge, 2003; brain origins of these “liking” and “dislik- Cardinal et al., 2002; Cooper, 2005; Higgs ing” reactions in rats and humans. That sets et al., 2003; Kelley et al., 2005a; Levine and the stage for animal affective neuroscience Billington, 2004; Rolls, 2005; Wise, 2004; studies to use these affective expressions to Yeomans and Gray, 2002). What has identify brain mechanisms that generate emerged most recently is a connected hedonic impact. network of forebrain sites that use opioid neurotransmission to increase taste “liking” III. UNDERSTANDING BRAIN REWARD SYSTEMS FOR FOOD “LIKING” AND “WANTING” 195 and “wanting” together to enhance food in the brainstem. Normal “liking” reactions reward. are not brainstem reflexes in a whole- Pleasure “liking” appears to be gener- brained individual. This becomes obvious ated by a distributed network of brain when we consider a related example: anen- islands scattered across sites like an archi- cephalic infants cry and vocalize and even pelago that trails throughout the limbic a decerebrate rat squeaks and emits distress forebrain and brainstem (Berridge, 2003; cries if its tail gets pinched. But no one Kelley et al., 2005a; Levine and Billington, would suggest the vocal ability of anen- 2004; Peciña and Berridge, 2005; Smith and cephalic infants to cry and decerebrate rats Berridge, 2005). These sites include nucleus to squeak means that normal human speech accumbens, ventral pallidum, and possibly is merely a brainstem reflex. Obviously amygdala and even limbic cortical sites, neither speech nor normal affective ex- and also deep brainstem sites including the pressions generated by an entire brain is parabrachial nucleus in the pons. These dis- merely a brainstem reflex when forebrain tributed “liking” sites are all connected systems determine them via hierarchical together so that they interact as a single control. integrated “liking” system, which operates When brainstem is connected to the fore- by largely hierarchical control rules. brain, the entire affective system operates in a hierarchical, flexible, and complex C. Pleasure Brain Hierarchy fashion. In a neural hierarchy, forebrain operations overrule brainstem elements, Certain elemental reaction circuits and dictate the output of the whole system. within brain affective systems are contained This means that brainstem reflex aspects of in the brainstem. By themselves, brainstem affective reactions are largely an artifact of circuits have a basic autonomy in the sense brainstem isolation in decerebrates and of functioning as simple reflexes when they anencephalics. In a fully connected brain, have no other signals to modulate them. For affective “liking” and “disliking” reactions example, basic positive or negative facial are determined by an extensive forebrain expressions are still found in human anen- network, and the final behavioral expres- cephalic infants born with a midbrain and sion of affective taste reactivity reflects fore- hindbrain, but no cortex, amygdala, or brain “liking” processes. classic limbic system, due to a congenital defect that prevents prenatal development D. Forebrain Limbic Hedonic Hot Spot of their forebrain. Yet, sweet tastes still elicit in Nucleus Accumbens normal positive affective facial expressions from anencephalic infants, and bitter or One forebrain hedonic island able to sour tastes elicit negative expressions cause “liking” is an opioid hot spot in the (Steiner, 1973). Similarly, a decerebrate rat nucleus accumbens. The nucleus accum- has an isolated brainstem because a surgi- bens contains major subdivisions called cal transaction that separates the brain- core and shell. While the core appears espe- stem’s connections from the forebrain, but cially important for learning about rewards, the decerebrate brainstem also remains able the shell is more important for generating to generate normal taste reactivity expres- actual affective and motivational compo- sions to sweet or bitter tastes placed in the nents of rewards themselves, including the decerebrate’s mouth (Grill and Norgren, pleasure gloss of “liking” for food rewards 1978a). (Fig. 1). However, brainstem generation of basic The shell of nucleus accumbens is an L- reactions does not mean “liking” lives only shaped structure: its vertical back (called 196 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE

Positive hedonic Negative aversive “liking” “disliking”

Nucleus accumbens shell Ventral pallidum Opioid hedonic hot Spots FIGURE 1 “Liking” reactions and brain hedonic hot spots. Top: Positive hedonic “liking” reactions are elicited by sucrose taste from human infant and adult rat (e.g., rhythmic tongue protrusion). By contrast, negative aver- sive “disliking” reactions are elicited by bitter quinine taste. Lower: Forebrain hedonic hot spots in limbic struc- tures where μ-opioid activation causes a brighter pleasure gloss to be painted on sweet sensation. Red/yellow shows hot spots in nucleus accumbens and ventral pallidum where opioid microinjections caused the biggest increases in the number of sweet-elicited “liking” reactions. Modified from Peciña and Berridge (2005) and Smith and Berridge (2005). medial shell), stands against the middle of entire nucleus accumbens, and in quite a the brain in each hemisphere, and its number of other forebrain limbic structures bottom horizontal foot points outward too (Cooper and Higgs, 1994; Kelley, 2004; toward the lateral sides of the brain (the Kelley et al., 2005a; Levine and Billington, core is held in the concave crook formed 2004; Yeomans and Gray, 2002). So if opioid between vertical back and lateral foot). It is activation causes taste pleasure wherever the medial shell, the vertical upright of it stimulates appetite for palatable food, the L, which has received most attention then the widespread brain distribution of in the search for pleasure generators—with appetite-promoting sites means the brain success. has an extensive opioid pleasure network The pleasure generator in the medial that stretches throughout much of the fore- shell runs in part on opioid neurotransmit- brain. That happy possibility would give ters. Opioids are natural brain neurotrans- every brain a really large hedonic causation mitters, such as enkephalin, endorphin, and system for generating pleasure. dynorphin, that act on the same receptors But alternatively, if opioid circuits of as opiate drugs such as morphine or heroin. food hedonic (“liking”) versus motivational Opioid neurotransmitters that activate the μ (“wanting”) functions are organized some- type of opioid receptor appear particularly what differently from each other, then important to causing food reward opioid activation might enhance taste “wanting” and “liking.” pleasure at only some of the sites where it An important fact about opioid neuro- stimulates appetite. In that case, we must transmitters in food reward is that they grapple with a complexity in opioid psy- stimulate eating behavior in nearly the chology and brain function. In other words, III. UNDERSTANDING BRAIN REWARD SYSTEMS FOR FOOD “LIKING” AND “WANTING” 197 brain opioid activation might contribute to positive hedonic impact. That increase in “liking” food and “wanting” food in differ- apparent pleasure-activating quality was ent ways in different brain places. Under- reflected by dramatic increases in sucrose- standing precisely how each opioid brain elicited facial “liking” expressions, which region contributes psychologically is a often doubled or even tripled in number demanding but also interesting task for above control levels. But the “liking” affective neuroscientists. increase was anatomically restricted to a This issue was more recently addressed single hedonic hot spot in the brain’s in a hedonic mapping study by Susana medial shell of nucleus accumbens (Peciña Peciña at the University of Michigan and Berridge, 2005). In rats, the hedonic hot (Peciña and Berridge, 2005). It turns out that spot was roughly just a cubic millimeter in only one relatively small site in the medial size, contained entirely in the rostral and shell may generate food “liking” as an dorsal quadrant of medial shell. Inside the opioid pleasure island or hedonic hot spot. hedonic hot spot, opioid activation made That hedonic hot spot is in the rostral and sweet sucrose taste “liked” even more than dorsal one-quarter of medial shell. Here, μ- normal, and made bitter quinine taste “dis- opioid activation acts as a hedonic genera- liked” less (not as aversive)—shifting both tor to paint a pleasure gloss on sweetness, positive/negative dimensions toward a generating more “liking” for the food it also positive pole. Outside the hot spot, positive makes more “wanted.” hedonic impact was no longer increased, Peciña’s study used a “Fos plume” even though the “disliking” suppression mapping technique to find where micro- zone extended more caudally. In fact, in one injections of a drug that activates opioid posterior site outside the hedonic hot spot circuits causes increased “liking” reactions affective suppression applied to both sweet to the hedonic impact of a pleasant taste. To “liking” and bitter “disliking”—essentially map pleasure generation, Peciña first made an affective cold spot of general suppres- painless microinjections into rats’ brains of sion. Thus Peciña’s mapping of opioid tiny droplets of a drug known as DAMGO, mechanisms indicates that a localized which stimulates μ opioid receptors hedonic island in the medial shell of (natural neurotransmitter receptors for nucleus accumbens helps generate the heroin or morphine). The opioid drug pleasure gloss that opioid circuits paint caused nearby neurons with appropriate onto sweet sensation to make it positively receptors to begin transcribing genes on “liked.” chromosomes in their cellular nucleus. One 1. Larger Opioid Sea of “Wanting” in is a rapid-onset or immediate early gene Nucleus Accumbens called c-fos, which is transcribed to produce a protein called Fos that plays important In addition to causing increased “liking,” roles in subsequent neuronal function. Fos almost all opioid microinjections in nucleus protein stains very dark in appearance accumbens also caused the rats to eat more when postmortem brain slices are appro- food soon afterward, increasing “wanting” priately processed later with chemicals and for food reward too, even outside the antibodies that bind to the protein, and as a hedonic hot spot (and even in the suppres- result the neurons with drug-induced Fos sive cold spot). The appetite-increasing could be seen later as forming a dark plume zone was much larger than the pleasure hot on a slice of brain tissue. The size of each spot: it was as though a large opioid sea of microinjection plume showed how far in “wanting” in the shell of nucleus accum- the brain its drug had acted. bens contains a smaller opioid island of Some opioid drug microinjections “liking” for the same reward (Peciña and caused sweet taste to carry increased Berridge, 2005). 198 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE

The large size of the opioid “wanting” lay outside the anterior and dorsal region of zone fits with many earlier results from medial shell. elegant appetite-mapping studies that have 3. Beyond Opioid “Liking”: Other Hedonic shown that many limbic brain structures Neurotransmitters in Nucleus Accumbens? support opioid-increased eating of palat- able sweet or fatty foods (Gosnell and Opioid activation is only one type of Levine, 1996; Kelley et al., 2005b; Levine neurochemical signal received by neurons and Billington, 2004; Will et al., 2003, 2004; in the shell nucleus accumbens. Are any Zhang and Kelley, 2000). But the opioid others able to modulate “liking” reactions “liking” island identified by Peciña indi- to food hedonic impact? The answer cates that only in the rostral/dorsal hedonic appears to be yes. hot spot does opioid activation in medial 4. Cannabinoid “Liking” shell also increase “liking” reactions to food at the same time as increasing “wanting” to Neurons in the nucleus accumbens also eat. manufacture an endogenous cannabinoid neurochemical called anandamide, the CB1 2. Implications for Normal Eating receptors for which are activated by active and Disorders ingredient THC in the drug, marijuana. So it seems that the same μ-opioid neu- Anandamide has been suggested to be a rotransmitter stimulation does different reverse neurotransmitter, which would be psychological things in different spots released by a target neuron in the shell to within the same brain structure. This leads float back to nearby presynaptic axon us to outline several speculative possibili- terminals. ties by which eating disorders might relate Marijuana, in addition to its own central endogenous opioid neurotransmission in rewarding effects, is widely known to cause nucleus accumbens shell. an appetite-stimulating effect sometimes First, one could speculate that patho- called the “marijuana munchies,” raising logical overactivation in regions of the intake especially of palatable snack foods. opioid hedonic hot spot might cause Several investigators have suggested that enhanced “liking” reaction to taste pleasure endogenous cannabinoid receptor activa- in some individuals. An endogenously tion stimulates appetite in part by enhanc- produced increase in opioid tone there ing “liking” for the perceived palatability of could magnify the hedonic impact of food (Cooper, 2004; Dallman, 2003; Higgs foods, making an individual “like” food et al., 2003; Kirkham and Williams, 2001; more than other people, and “want” to eat Kirkham, 2005; Sharkey and Pittman, more. 2005). A second and alternative speculative A more recent taste reactivity study by possibility is that activation in the opioid Jarrett and Parker found that THC causes sea of “wanting” outside the hedonic island increased “liking” reactions to sugar tastes could cause people to “want” to overeat in rats, just as opiate drugs do (Jarrett et al., palatable, without making them “like” food 2005). Focusing on natural endogenous more. If so, the preceding results would brain cannabinoids, a microinjection study predict that this increased appetite would by Stephen Mahler and Kyle Smith in our occur from excessive opioid function in a laboratory pinned the neural causation relatively large region of nucleus accum- of cannabinoid “liking” on activation bens shell. The resulting “wanting” to eat of natural anandamide receptors in the would occur without any concomitant nucleus accumbens shell (Mahler et al., increase in the perceived palatability of 2004). Microinjections of anandamide food if the major locus of elevated activity directly into the medial shell of nucleus III. UNDERSTANDING BRAIN REWARD SYSTEMS FOR FOOD “LIKING” AND “WANTING” 199 accumbens promoted positive “liking” Of these various neurochemical signals, reactions to the pleasure of sucrose GABA at least can potently alter “liking” taste. reactions to the hedonic impact of sugar The ability of natural anandamide to tastes. But the positive/negative valence of magnify the hedonic impact of natural food GABA on “liking” versus “disliking” reward raises interesting questions about depends very much on precisely where in whether natural sensory reward functions the shell the GABA is (Reynolds and will be disrupted in people by taking drugs Berridge, 2002). For example, GABA stimu- that have been proposed to help dieters or lation in the anterior subregion of the shell addicts suppress their excessive consump- can increase “liking” reactions as well as tion by blocking the natural CB1 receptors food intake. GABA in most of the front half for anandamide and other endogenous of the shell increases food intake without cannabinoids (Cooper, 2004; Higgs et al., increasing “liking” reactions to hedonic 2003; Kirkham and Williams, 2004). The impact, a bit like opioid activation in much answer is currently not known: the fact that of the shell described before. And GABA a brain event can cause enhanced pleasure delivered to the posterior half of the shell (i.e., sufficient cause for hedonic impact) dramatically suppresses intake, and makes does not necessarily mean that the brain sweet tastes “disliked,” reversing their mechanism is needed for normal pleasure usual hedonic impact (Reynolds and (i.e., necessary cause for hedonic impact). Berridge, 2002). The two questions must be separately It remains to be known whether similar answered by future research. changes in “liking” are evoked by nucleus In either case, it appears that the cannabi- accumbens glutamate blockade, by noid pleasure zone overlaps with the opioid microinjections of a drug that blocks AMPA hedonic hot spot described previously in receptor signals, and so which may simi- medial shell of nucleus accumbens. That larly hyperpolarize neurons in medial shell. suggests that both natural opioids and However, glutamate AMPA blockade does natural cannabinoids act in overlapping alter eating behavior and food intake in hedonic islands of the nucleus accumbens ways similar to GABA (Reynolds and shell. In those overlapping zones, both neu- Berridge, 2003), which increases the plausi- rochemicals act to enhance “liking” for bility that “liking” might be modulated by the hedonic impact of natural sensory glutamate just as “wanting” is. The role of pleasures. cortico-amygdala-hippocampal glutamate This raises possibilities for potential signals to nucleus accumbens in modulat- interactions between these two neurochemi- ing “liking” reactions to hedonic impact of cal forms of pleasure gloss, opioid and sweetness is a topic of substantial interest. cannabinoid (Vigano et al., 2005). And there are additional “liking” interactions to con- E. Ventral Pallidum: “Liking” and sider too. Beyond neurotransmitters related “Wanting” Pivot Point for Limbic Food to classic drugs of abuse, other neurotrans- Reward Circuits mitters such as GABA are also used by the same neurons in the medial shell. These Leaving the nucleus accumbens, output neurons both send and receive GABA projections head to several destinations but signals. In addition, these neurons receive the single heaviest projections are further neurochemical inputs, such as glu- posteriorly to two nearby neighbors, the tamate from the neocortex, hippocampus, ventral pallidum and lateral hypothalamus. and basolateral amygdala, and dopamine Of these two structures, the lateral hypo- from mesolimbic neurons in the midbrain thalamus has long been famous for roles in ventral tegmental area. food intake and food reward. Lesions of 200 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE the lateral hypothalamus disrupt eating amygdala” for a bit behind that lies and drinking behaviors, sending food and between ventral pallidum and lateral water intakes to zero (Teitelbaum and hypothalamus. Epstein, 1962; Winn, 1995). After electrolytic Ventral pallidum is the chief target of lesions to lateral hypothalamus, rats would the heaviest projections emanating from the starve to death unless they were given nucleus accumbens, and so it is the primary intensive nursing care and artificial intra- output channel through which mesocorti- gastric feeding. colimbic circuits must work (Zahm, 2000). Decades ago, lateral hypothalamic The ventral pallidum is relatively new on lesions were thought not only to abolish the affective neuroscience scene, but there is food “wanting,” but also to abolish food reason to believe this chief target of nucleus “liking” too. Even sweet tastes were accumbens is crucial for both normal reported to elicit bitter-type disliking reac- reward “liking” and for enhanced “liking” tions (Schallert and Whishaw, 1978; Stellar caused under some neurochemical et al., 1979; Teitelbaum and Epstein, 1962). conditions. However, it appears that lateral hypothala- An astounding fact is that the ventral mus may have been blamed through a case pallidum is the only brain region known so of mistaken identity for the effects of lesions far where the loss of neurons is capable of that stretched beyond it in lateral and ante- abolishing all “liking” for sweetness. It is rior directions. Early studies on this topic the only brain site absolutely necessary for indicated that sucrose “liking” would be normal sucrose “liking” in the sense that replaced by sucrose “disliking” only if the damage to it makes “liking” go away (at lesion were in the anterior zone of lateral least for several weeks). Sucrose no longer hypothalamus—and not if the lesion were elicits “liking” reactions from a rat that has in the posterior part of lateral hypothala- an excitotoxin lesion of its ventral pallidum, mus, where it would produce loss of eating a type of lesion that selectively destroys the and drinking, but leave “liking” reactions neurons that live in that structure while pre- essentially normal (Schallert and Whishaw, serving fibers from neurons elsewhere that 1978). Further lesion studies by Cromwell are simply passing through. Instead sucrose mapped more carefully the boundaries of taste elicits only “disliking” reactions after sites where neuron death caused aversion, the ventral pallidal lesion, as though the and found that the “disliking” lesions sweet taste had become bitter quinine might actually have to be so far anterior and (Cromwell and Berridge, 1993). lateral that they actually were outside the Ventral pallidum can also generate lateral hypothalamus itself—and in another enhancement of natural pleasure, at least structure, now called the ventral pallidum when it is intact. Ventral pallidum contains (Cromwell and Berridge, 1993). its own hedonic hot spot where μ opioid Until about 10 years ago the ventral pal- activation can increase the pleasure gloss lidum was known often as the substantia that gets painted on sweetness. In a hedonic innominata, or brain structure without a mapping study, Kyle Smith discovered that name, and earlier than 20 years ago it was opioid DAMGO microinjections into a often mistaken for part of the lateral hypo- hedonic hot spot within the posterior VP thalamus, as we have seen. Today it has a caused sucrose to elicit over twice as many name, actually several names that corre- “liking” reactions than it normally did spond to different divisions of this intrigu- (Smith and Berridge, 2005). Opioid activa- ing part of the ventral forebrain. The chief tion in the posterior ventral pallidum names today are “ventral pallidum” for the increased the hedonic impact of the natural part known to cause “liking” for sensory taste reward, and also caused rats to eat pleasure, and “sublenticular extended over twice as much food. The hedonic hot IV. “WANTING” WITHOUT “LIKING” 201 spot was localized quite tightly within only the posterior one-third of ventral pallidum. pallidum, that change would cause a If the same opioid microinjections were person to experience foods as more pleas- moved anteriorly toward the front of the ant and would stimulate appetite accord- structure, it actually suppressed hedonic ingly. Conversely, if it were possible to “liking” reactions to sucrose and sup- target opioid activation to the opposite pressed food intake too (Smith and anterior end of ventral pallidum, both Berridge, 2005). food pleasure and food intake would be A final reason to suppose ventral pal- expected to be suppressed. Whether either lidum mediates hedonic impact is that the of these events actually occurs in any activity of neurons in the posterior hedonic human eating disorder is unknown. At the hot spot appears to code “liking” for sweet moment, they are simply possibilities to be and other tastes (Tindell et al., 2004, 2005). compared against future observations that Recording electrodes can be permanently may carry the answer. implanted in the ventral pallidum, and In any case, it seems likely that the neurons there fire faster when rats eat a hedonic hot spot in ventral pallidum ordi- sweet taste. The firing of sucrose-triggered narily interacts together with its coun- neurons appears to reflect hedonic “liking” terpart in the nucleus accumbens shell. for the taste. For example, the same neurons There are extensive intercommunications will not fire to an intensely salty solution between these two brain structures. A con- sequence is that the two hedonic islands that is unpleasant (three times saltier than probably actually function together as part seawater). However, the neurons suddenly of a single hedonic system, in close syn- begin to fire to the triple-seawater taste if a chrony with each other. The details of this physiological state of “salt appetite” is interaction remain to be unraveled by induced in the rats, by administering hor- experiments but when they are, an integra- mones that causes the body to need more tive network of hedonic “liking” likely will salt, and which increase the perceived be revealed by which opioid limbic circuits “liking” for intensely salty taste (Smith modulate the final hedonic impact and et al., 2004). Thus neurons in the ventral appetite for foods. pallidum code taste pleasure in a way that is sensitive to the physiological need of the moment. When a taste becomes more IV. “WANTING” WITHOUT pleasant during a particular physiological “LIKING” hunger, in a hedonic shift called “alliesthe- sia,” the ventral pallidum neurons code the A very different product of affective neu- increase in salty pleasure. The observation roscience studies of taste “liking” reactions that those hedonic neurons are in the same is the revelation of a number of false hedonic hot spot where opioid activation hedonic brain mechanisms, which turn out causes increased “liking” reactions to taste to mediate motivational “wanting” to eat suggests that their firing rate might actually without mediating hedonic “liking” for be part of the causal mechanism that paints the same food. False hedonic mechanisms the pleasure gloss onto taste sensation. do not mediate “liking” for sensory pleasures—even though some were once 1. Speculative Implications for thought to do so by neuroscientists. Their Eating Disorders production of “wanting” without “liking” Just as for the hedonic island in nucleus opens up fascinating possibilities for under- accumbens, it is easy to imagine that if a standing particular pathologies of appetite human pathology caused increased opioid and desire, including cases of irrational activation effects in the posterior ventral desire. 202 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE

Perhaps the most famous false hedonic reducing the pleasure they get from food. mechanism is dopamine: the mesolimbic By that view, reduced pleasure has been projection of fibers to the nucleus accum- suggested to cause those individuals to eat bens from dopamine neurons in the mid- more food in a quest to regain normal brain ventral tegmental area. Dopamine amounts of pleasure. A difficulty may arise release is triggered by pleasant foods for this account in that it also seems to and many other pleasant rewards, and require that the less people like a food the dopamine neurons themselves fire more to more they will eat it. By contrast, much evi- pleasant food (especially when it is sud- dence from psychology and neuroscience denly and unexpectedly received) (Ahn and indicates that reducing the value of a food Phillips, 1999; Di Chiara, 2002; Hajnal and reward causes reduced pursuit of it as an Norgren, 2005; Montague et al., 2004; incentive, rather than increasing pursuit Roitman et al., 2004; Schultz, 2006; Small et and consumption (Cooper and Higgs, 1994; al., 2003). Everyone agrees that dopamine Dickinson and Balleine, 2002; Grigson, release has important consequences on 2002; Kelley et al., 2005b; Levine and some aspect of reward, but the question is Billington, 2004). One might perhaps rescue which aspect (Berridge and Robinson, 2003; this anhedonia account of D2 decrement by Everitt and Robbins, 2005). Beyond correla- supposing that all other life pleasures are tive activations by rewards, the causal reduced even more by dopamine receptor importance of dopamine is seen in the well- suppression than food pleasure, so that known observation that drugs that are food remains the only pleasure available for rewarding or addictive typically cause consumption. However, we can see that dopamine activation—either directly or by actually getting increased food consump- acting on other neurochemical systems that tion from reduced pleasure out of any in turn cause dopamine activation (Everitt known psychological-brain system of food and Robbins, 2005; Koob and Le Moal, reward may prove trickier than first 2006). Conversely, dopamine suppression appears. So alternatives are worth enter- reduces the degree to which animals and taining too. A reverse interpretation of people seem to want rewarding foods, or reduced dopamine D2 binding in obese rewards of other types (Berridge and people is that the reduction is a conse- Robinson, 1998; Dickinson et al., 2000; Wise, quence of overeating and obesity, rather 2004). than its cause. As a parallel example, over- The suppression of reward “wanting” by consumption of drug rewards that provide dopamine blockade or loss gave rise increased stimulation to dopamine recep- decades ago to the idea that dopamine must tors eventually causes the receptors to also mediate reward “liking” (Wise, 1985). reduce in number, even if dopamine recep- The view of most neuroscientists has tors were normal to begin with—this is a shifted subsequently, although some correla- down regulation mechanism of drug toler- tive evidence collected in more recent ance and withdrawal (Koob and Le Moal, years can still be viewed as consistent with 2006). So it is conceivable that similar sus- this dopamine-pleasure hypothesis of tained overactivation of dopamine systems reward. For example, PET neuroimaging by overeating food rewards in obese indi- studies have suggested that obese people viduals perhaps could cause a similar even- may have lower levels of dopamine D2 tual down regulation of their dopamine receptor binding in their brains’ striatum receptors. In a related vein, other physio- than others (Wang et al., 2001, 2004b). At logical aspects of preexisting obesity states first take, if one supposes that dopamine might also send excessive signals to brain causes pleasure, reduced dopamine recep- systems sensitive to body weight, which tors in obese individuals can be viewed as indirectly cause reduction of D2 receptor as IV. “WANTING” WITHOUT “LIKING” 203 a negative feedback consequence or a sort food or drug reward may better correlate to of long-term satiety signal that down regu- their subjective ratings of wanting the lates incentive systems. These speculative reward than to their pleasure ratings of alternatives are enough to illustrate that liking the same reward (Leyton et al., 2002; possibilities exist by which reduced Volkow et al., 2002). dopamine receptor binding could be a con- Thus, the idea that dopamine is a pleas- sequence, not the cause, of sustained ure neurotransmitter may have faded in obesity. Future research will be needed to neuroscience, with only a few hedonia resolve the fascinating question of whether pockets remaining (though dopamine correlated changes in levels of dopamine seems important to “wanting” rewards, receptors is a cause or a consequence of even if not to “liking” rewards). Separating human obesity. true “liking” substrates from false ones is a If we turn to evidence from animal useful step in identifying the real affective studies in which dopamine’s causal roles neural circuits for hedonic processes in the can be manipulated, then dopamine does brain. not appear to be important for “liking” the hedonic impact of food rewards after all. A. What Is “Wanting”? For example, mutant mice that lack any dopamine in their brains remain quite able “Wanting” is a shorthand term for the to register the hedonic impact of sucrose or psychological process of incentive salience food rewards (Cannon and Palmiter, 2003; attribution, which helps determine the Robinson et al., 2005). Similarly, dopamine motivational incentive value of a pleasant suppression or complete lesion in normal reward (Berridge and Robinson, 1998; rats does not suppress taste “liking” facial Everitt and Robbins, 2005; Robinson and expressions elicited by the taste of sucrose Berridge, 2003; Salamone and Correa, 2002). (Berridge and Robinson, 1998). Instead, the But incentive “wanting” is not a sensory hedonic impact of sweetness remains pleasure. “Wanting” is purely the incentive robust even in a nearly dopamine-free fore- motivational value of a stimulus, not its brain (also, still robust is the ability to learn hedonic impact. some new reward values for a sweet taste, Why did brains evolve separate which indicates that “liking” expressions “wanting” and “liking” mechanisms for the remain faithful readouts of forebrain same reward? Originally, “wanting” might “liking” systems after dopamine loss have evolved separately as an elementary (Berridge and Robinson, 1998). form of goal directedness to pursue par- And conversely, too much dopamine in ticular innate incentives even in advance of the brain, either in mutant mice whose gene experience of their hedonic effects. Later mutation causes extra dopamine to remain incentive salience became harnessed by in synapses or in ordinary rats given evolution to serve learned “wanting” for amphetamine that causes dopamine release predictors of “liking,” following learned or (or that have drug-sensitized dopamine conditioned incentive motivation rules, systems), show elevated “wanting” for guided by Pavlovian associations (Berridge, sweet food rewards, but no elevation in 2001; Dickinson and Balleine, 2002; Schultz, “liking” sweet rewards (Peciña et al., 2003; 2006; Toates, 1986). “Wanting” may also Tindell et al., 2005; Wyvell and Berridge, have evolved as distinct from “liking” to 2000). Supporting evidence that dopamine provide a common neural currency of mediates “wanting” but not “liking” also incentive salience shared by all rewards, comes from PET neuroimaging studies of which could compare and decide compet- humans, which report that dopamine ing choices for food, sex, or other rewards release triggered when people encounter a that might each involve partly distinct 204 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE neural “liking” circuits (Shizgal, 1997). The manipulations in rats (and perhaps by real- important point is that “liking” and life brain sensitization in human drug “wanting” normally go together, but they addicts, see later). For example, electrical can be split apart under certain circum- stimulation of the lateral hypothalamus in stances, especially by certain brain rats, as mentioned before, triggers a number manipulations. of motivated behaviors such as eating. “Liking” without “wanting” can be pro- In normal hunger, increased appetite is duced, and so can “wanting” without accompanied by increased hedonic appreci- “liking.” “Liking” without “wanting” ation of food, a phenomenon named allies- happens after brain manipulations that thesia (Cabanac, 1971). But during lateral cause mesolimbic dopamine neurotrans- hypothalamic stimulation that made them mission to be suppressed. For example, dis- eat avidly, rats facial expressions to a sweet ruption of mesolimbic dopamine systems, taste actually became more aversive, if any- via neurochemical lesions of the dopamine thing, as though the taste became bitter pathway that projects to nucleus accum- (Berridge and Valenstein, 1991). In other bens or by receptor-blocking drugs, dra- words, the hypothalamic stimulation did matically reduces incentive salience or not make them “want” to eat by making “wanting” to eat a tasty reward, but does them “like” the taste of food more. Instead, not reduce affective facial expressions of it made them “want” to eat more despite “liking” for the same reward (Berridge and making them “dislike” the taste. Robinson, 1998; Peciña et al., 1997). Similarly, eating or pursuit of food Dopamine suppression can leave indi- caused by a variety of dopamine, GABA, or viduals nearly without motivation for any other neurochemical manipulations of pleasant incentive at all: food, sex, drugs, nucleus accumbens or ventral pallidum is brain-stimulation reward, etc. (Cannon and not accompanied by enhanced hedonic Palmiter, 2003; Everitt and Robbins, 2005; reactions to the taste of food. Mutant mice Salamone and Correa, 2002; Wise, 2004). Yet, with brain receptors that receive more “liking,” or the hedonic impact of the same dopamine than normal, because of a genetic incentives, remains intact after dopamine mutation that causes released dopamine to loss or suppression, at least in many studies remain in synapses, also show excessive where it can be specifically assessed by “wanting” of sweet reward, while nonethe- either facial affective expressions or subjec- less “liking” sweetness less than normal tive ratings. Intact “liking” in animal affec- mice do (Peciña et al., 2003). Similarly tive neuroscience studies has usually been increased food intake without increased manifest via normal positive affective facial “liking” occurs in rats that have received expressions elicited by the hedonic impact accumbens microinjections of ampheta- of sweet tastes, or by normal learning about mine or a GABA agonist or opioid agonist food reward, after dopamine lesion, block- in certain shell regions, or ventral pallidum ade, or genetic lack (Berridge and Robinson, microinjections of a GABA antagonist 1998; Cannon and Palmiter, 2003; Peciña et (Peciña et al., 2003; Peciña and Berridge, al., 1997; Robinson et al., 2005). Similarly, in 2005; Reynolds and Berridge, 2002; Smith humans, drugs that block dopamine recep- and Berridge, 2005; Tindell et al., 2005; tors may completely fail to reduce the sub- Wyvell and Berridge, 2000). All of these jective pleasure ratings that people give to a brain manipulations make animals “want” reward stimulus such as amphetamine to eat food more, though they fail to make (Brauer and De Wit, 1997; Brauer et al., 1997; the animals “like” food more (and some- Wachtel et al., 2002). times even make them “like” it less). Conversely, “wanting” without “liking” What is “wanting” if it is not “liking?” can be produced by several brain According to the incentive salience concept, IV. “WANTING” WITHOUT “LIKING” 205

“wanting” is a mesolimbic-generated reward. Fortunately, both usually happen process that can tag certain stimulus repre- together in human life. sentations in the brain that have Pavlovian associations with reward. When incentive B. Addiction and Incentive Sensitization salience is attributed to a reward stimulus representation, it makes that stimulus For some human addicts, however, attractive, attention grabbing, and that drugs such as heroin or cocaine may stimulus and its associated reward sud- cause real-life “wanting” without “liking” denly become enhanced motivational because of long-lasting sensitization targets. Because incentive salience is often changes in brain mesolimbic systems. triggered by Pavlovian conditioned stimuli Addicts sometimes take drugs compul- or reward cues, it often manifests as cue- sively even when they do not derive much triggered “wanting” for reward. When pleasure from them (Everitt and Robbins, attributed to a specific stimulus, incentive 2005). For example, drugs such as nicotine salience may make an autoshaped cue light generally fail to produce great sensory appear foodlike to the autoshaped pigeon pleasure in most people, but still are infa- or rat that perceives it, causing the animal mously addictive. to try to eat the cue. In autoshaping, In the early 1990s, Terry Robinson and I animals sometimes direct behavioral proposed the incentive-sensitization theory pursuit and consummatory responses of addiction, which combines neural sensi- toward the Pavlovian cue, literally trying to tization and incentive salience concepts eat the conditioned stimulus if it is a cue for (Robinson and Berridge, 1993, 2003). The food reward (Jenkins and Moore, 1973; theory does not deny that drug pleasure, Tomie, 1996). When attributed to the smell withdrawal, or habits are all reasons people emanating from a bakery, incentive salience sometimes take drugs, but suggests that can rivet a person’s attention and trigger something else, sensitized “wanting,” may sudden thoughts of lunch—and perhaps it be needed in order to understand the com- can do so under some circumstances even pulsive and long-lasting nature of addic- if the person merely vividly imagines the tion, and especially for why relapse occurs delicious food. even after weeks or months of abstinence But “wanting” is not “liking,” and both from any drugs, and even when addicts do together are necessary for normal reward. not expect to gain much pleasure from their “Wanting” without “liking” is merely a relapse. sham or partial reward, without sensory Many addictive drugs cause neural pleasure in any sense. However, “wanting” sensitization in the brain mesocorticolimbic is still an important component of normal systems (e.g., cocaine, heroin, ampheta- reward, especially when combined with mine, alcohol, nicotine). Sensitization “liking.” Reward in the full sense cannot means that the brain system can be trig- happen without incentive salience, even gered into abnormally high levels of activa- if hedonic “liking” is present. Hedonic tion by drugs or related stimuli. “liking” by itself is simply a triggered Sensitization is nearly the opposite of drug affective state—there is no object of desire tolerance. Different processes within the or incentive target, and no motivation same brain systems can simultaneously for reward. It is the process of incentive instantiate both sensitization (e.g., via salience attribution that makes a specific increase in dopamine release) and tolerance associated stimulus or action the object of (e.g., via decrease in dopamine receptors) desire, and that tags a specific behavior (Koob and Le Moal, 2006; Robinson as the rewarded response. “Liking” and and Berridge, 1993, 2003; Vezina, 2004). “wanting” are needed together for full However, tolerance mechanisms usually 206 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE recover within days to weeks once drugs occasions when they reencounter drug- are given up, whereas neural sensitization associated cues such as drug paraphernalia can last for years. If the incentive- or places and social contexts where drugs sensitization theory is true for drug addic- were taken before. tion, it helps explain why addicts may sometimes even “want” to take drugs that C. Is There a Neural Sensitization Role they do not particularly “like.” The long- in Food Addictions? lasting nature of neural sensitization may also help explain why recovered addicts, Could incentive-sensitization apply to who have been drug-free and out of with- food addictions too? A conclusive answer drawal for months or years, are still some- cannot yet be given. In favor of the possi- times liable to relapse back into addiction. bility, once sensitized, brain mesolimbic Sensitization of incentive salience does systems do overrespond to cues for sugar not mean that addicts “want” all rewards rewards with excessive incentive salience more in a general fashion. “Wanting” (Tindell et al., 2005; Wyvell and Berridge, increases instead are highly specific, in 2001), and repeated exposure to drugs can target object and in time. In target object, produce sensitization-type increases in only specific conditioned stimuli or cues for food intake (Bakshi and Kelley, 1994; Nocjar reward get attributed with higher incentive and Panksepp, 2002). salience after sensitization (Tindell et al., But does sensitization occur to food 2005; Wyvell and Berridge, 2001). Other incentives in the absence of drugs? More unrelated stimuli get no increase at all (e.g., recently, several investigators have sug- CS-). And when multiple reward targets are gested that sensitization-like changes in available, some targets get much more brain systems are indeed produced by increase than others (Nocjar and Panksepp, exposure to certain regimens of food 2002; Tindell et al., 2005). The key to which and restriction that model oscillations stimulus becomes “wanted” may lie in between dieting and binging on palatable Pavlovian learning mechanisms that foods (Avena and Hoebel, 2003a,b; Bell control the direction of incentive salience et al., 1997; Bello et al., 2003; Carr, 2002; attribution, the relative intensities of Colantuoni et al., 2001; de Vaca and Carr, reward unconditioned stimuli, and in other 1998; Gosnell, 2005). These food-sensitization features that differentially distribute ele- studies have generally shown that when vated attribution (Berridge and Robinson, rats are given a number of brief chances to 1998; Robinson and Berridge, 1993). This consume sucrose (sucrose binges) a number directional specificity may relate to why a of accumulating sensitization-like changes drug addict particularly “wants” drug, are seen, especially when binges are whereas someone with an eating disorder separated by periods of food restriction. might particularly “want” food. These include an increasing propensity to In time, sensitized cue-triggered overconsume when allowed, an enduring “wanting” is specific to sudden intense enhanced neural response to the presenta- peaks that rapidly fall off when the cue is tion of food reward and cues, and an over- taken away, only to reappear later when it response to the psychostimulant effects of is reencountered again (Tindell et al., 2005; drugs such as amphetamine (a typical Wyvell and Berridge, 2001). High incentive behavioral marker of drug-induced neural salience comes and goes with cues for the sensitization, which suggests a common particularly “wanted” reward (and in underlying mechanism). people possibly also with vivid reward The possibility of food-induced sensiti- imagery). This may relate to why addicts zation suggests the speculative scenario are particularly vulnerable to relapse on that some diet–binge combinations IV. “WANTING” WITHOUT “LIKING” 207 conceivably might be able to induce incen- facilitates the unconditioned activation of tive-sensitization for food rewards. That is, mesolimbic systems by food, then it may individuals who develop neural sensitiza- also promote stronger incentive condition- tion profiles to food might come to “want” ing, leading to later enhanced mesolimbic to eat, perhaps even with compulsive inten- responses to food conditioned stimuli. sity that ordinary individuals do not Essentially hunger would make food into a usually encounter, that outstrips their better and stronger hedonic reward, which “liking” for the foods they eat. This is an would function as a stronger unconditioned intriguing possible mechanism for a “food stimulus in a Pavlovian sense to promote addiction” that certainly deserve further greater learning. Animals that repeatedly consideration. consume while hungry, in alternating However, several cautions are in order dieting-binge periods, could essentially before concluding that sensitized food develop a stronger learned or conditioned addictions do indeed exist. Repeated food incentive motivation triggered by predic- presentation and repeated food restriction tive cues than animals that are not food- both introduce complications that could restricted between periods of sucrose masquerade as neural sensitization, and access. Those more strongly conditioned these complications could interact. For animals might well be expected to show example, repeated presentation of a enhanced mesolimbic activation later palatable food is likely to induce strong whenever related cues are presented later. Pavlovian conditioning, creating conditioned If these things occurred, the resulting incentive stimuli that strongly activate picture might look a bit like sensitization, mesolimbic systems for incentive salience. without actually being it. It is no surprise that a strong conditioned This is not to discount the possibility food incentive would activate dopamine that real sensitization might occur induced systems more than in an individual that has by diet–binge cycles or similar exposure not undergone the same conditioning expe- regimens, or to deny that the studies men- rience. It would be an error to mistake con- tioned before might be examples of food ditioned incentive activation of mesolimbic incentive-sensitization. Sensitization-like systems via serial experiences for an under- states do indeed seem to be implicated by lying neural sensitization. certain types of physiological deprivation Hunger itself also may promote states (Dietz et al., 2005; Roitman et al., mesolimbic activation under some condi- 2002). However, the previous cautions tions (Ahn and Phillips, 1999; Nader et al., point to the need to better disentangle 1997; Wilson et al., 1995). The use of food direct effects of true food-induced sensiti- restriction can be expected to enhance the zation from masquerading confounds of psychological incentive properties of food unconditioned deprivation states and ordi- directly, with consequences for underlying nary incentive conditioning on mesolimbic neurobiological activations too. During activation. Otherwise these confounds hunger, sucrose tastes more pleasant than could mimic sensitization patterns of when one is full (alliesthesia) (Cabanac, enhanced activation, and mislead us into 1971). When foods become more potent thinking sensitization has occurred when it unconditioned incentives, a variety of has not. unconditioned and conditioned food- Future studies may be able to better related stimuli might be expected to more assess these potential confounds, and strongly activate mesolimbic incentive perhaps separate them from sensitization systems too (Toates, 1986). and set them to rest. If so, it will put food- Further interactions between these induced sensitization of incentive salience potential confounds may ensue. If hunger on a stronger empirical basis, and raise 208 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE incentive-sensitization as a possible mecha- as a result of this difference, as described nism for food addiction. for addiction, excessive incentive salience may lead to irrational “wants” for outcomes D. Cognitive Goals and that are not cognitively wanted, and that Ordinary Wanting are neither liked nor even expected to be liked. Before leaving this topic of incentive Behavioral neuroscience experiments salience and “wanting,” it is useful to note have indicated that these forms of wanting how the incentive salience meaning of the may depend on different brain structures. word “wanting” above differs from what For example, incentive salience “wanting” most people mean by the ordinary sense of depends highly on subcortical mesolimbic the word wanting. A subjective feeling of dopamine neurotransmission, whereas cog- desire meant by the ordinary word wanting nitive forms of wanting depend instead on implies something both cognitive (involv- cortical brain regions such as orbitofrontal ing an explicit goal) and conscious (involv- cortex, prelimbic cortex, and insular cortex ing a subjective feeling). When you say you (Corbit and Balleine, 2003; Dickinson et al., want something, you usually have in mind 2000; Rolls, 2005). The implication is that a cognitive expectation or idea of the some- there may be multiple kinds of wanting, thing-you-want: a declarative representa- with different neural substrates (Berridge, tion of your goal. Your representation is 2001). based usually on your experience with that If eating disorders involve a pathology in thing in the past. Or, if you have never a particular “wanting” system such as before experienced that thing, then, the rep- incentive salience, it would be possible that resentation is based on your imagination of such an individual would “want” to eat what it would be like to experience. In other food at the same time that they cognitively words, in these cases, you know or imagine do not want to eat, and when “liking” cognitively what it is you want, you expect would not be enhanced. In other words, the to like it, and you may even have some idea sight, smell, or vivid imagination of food of how to get it. This is a very cognitive could trigger a compulsive urge to eat, even form of wanting, involving declarative though the person would not expect to find memories of the valued goal, explicit pre- the experience very pleasurable. Neural dictions for the potential future based sensitization of incentive salience systems, on those memories, and cognitive under- if it truly happens in any eating disorder, standing of causal relationships that exist might be one way by which excessive between your potential actions and future “wanting” to eat could generate excessive attainment of your goal. food intake. By contrast, none of this cognition need be part of incentive salience “wants” dis- cussed before (Berridge, 2001; Dickinson V. A BRIEF HISTORY OF and Balleine, 2000). Incentive salience attri- APPETITE: FOOD INCENTIVES, butions do not need to be conscious and are NOT HUNGER DRIVES mediated by relatively simple brain mecha- nisms (Berridge and Winkielman, 2003). To place into a better perspective the role Incentive salience “wants” are triggered by of “liking” and “wanting” mechanisms we relatively basic stimuli and perceptions (not have discussed, it might be helpful to con- requiring more elaborate cognitive expecta- sider here a brief overview of how psy- tions). Cue-triggered “wanting” does not chologists and neuroscientists’ view of require understanding of causal relations appetite has evolved in recent decades. For about the hedonic outcome. Sometimes many years it was thought that eating V. A BRIEF HISTORY OF APPETITE: FOOD INCENTIVES, NOT HUNGER DRIVES 209 behavior is driven directly by hunger ishment effects (if given contingent on the drives, and that food was a reward because animal’s response). Many behavioral neu- it reduces hunger drive. But actually, phys- roscientists of the time believed in drive iological hunger works primarily by modu- reduction theory. So, at first, they expected lating food incentive values, not driving to find that the brain sites where stimula- behavior directly through deficit or drive tion would reduce eating (presumably by signals. reducing drive) would also be the sites And food is rewarding primarily where stimulation was rewarding (again, because of its hedonic value in taste, presumably by reducing drive). Con- texture, and the act of eating, not by its versely, they believed the opposite would eventual drive reduction. A vivid early be true too. They expected to find that pun- example against pure drive reduction is the ishing electrodes would sometimes activate anecdotal medical case of a man named drives like hunger. Tom, from the timber and mining frontier of The best descriptions of these beliefs upper Michigan a century ago, whose come from the experimenters’ own words. esophagus was permanently damaged in As James Olds (the codiscoverer of brain childhood when he accidentally drank stimulation reward; (p. 89, (Olds, 1973)) put scalding soup without knowing it was too it, he was originally guided by the drive hot (Wolf and Wolff, 1943). The burn sealed reduction hypothesis that an “electrical his esophagus and, thereafter, blocked the simulation which caused the animal to passage of food to the stomach. To help him respond as if it were very hungry might live, a surgical opening or gastrostomy have been a drive-inducing stimulus and fistula was implanted in his stomach, and might therefore have been expected to have he was sustained afterwards by placing aversive properties.” In other words, if the food and drink directly through the fistula drive reduction theory were true, an into his stomach. There was no longer any “eating electrode” should also have been a apparent purpose in putting food in his “punishment electrode.” mouth first because food in the mouth Conversely, a “satiety electrode” that could not descend though the closed stopped eating should have been a “reward esophagus. electrode.” As Miller (pp. 54–55, (Miller, Yet, Tom insisted on munching food at 1973)), a major investigator of brain stimu- meals, when he would chew and then spit lation effects, recounted later in describing out the food before placing it in his how drive reduction concepts guided his stomach. Why? Because “introducing research. (food) directly into his stomach failed to If I could find an area of the brain where elec- satisfy his appetite” (p. 8, (Wolf and Wolff, trical stimulation had the other properties of 1943)). normal hunger, would sudden termination of that stimulation function as a reward? If I could A. Separating Reward and find such an area, perhaps recording from it Drive Reduction would provide a way of measuring hunger which would allow me to see the effects of a small The most important evidence against nibble of food that is large enough to serve as a drive reduction concepts came in the 1960s reward but not large enough to produce com- from studies of brain stimulation reward plete satiation. Would such a nibble produce a and related studies of motivated behavior prompt, appreciable reduction in hunger, as elicited by “free” brain stimulation. A single demanded by the drive-reduction hypothesis? electrode in the lateral hypothalamus could both elicit motivated behavior (if just Thus, if the drive reduction theory were turned on freely) and have reward or pun- true, you might actually be able to watch 210 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE hunger drive shrink by recording the hypocretin neurons send projections to shrinking activity of the drive neuron each modulate the nucleus accumbens in ways time a nibble of food reduced that drive and that might allow hunger states to enhance caused reward. food reward (Scammell and Saper, 2005), But the drive-reduction theory was not and even interact with other rewards such true, and nearly all of the predictions as drugs (Harris et al., 2005). In return, based on it turned out to be wrong. In many nucleus accumbens influences hypothalamic cases, the opposite results were found circuits. For example, manipulations of instead. The brain sites where the stimula- nucleus accumbens that cause increased tion caused eating behavior were almost food intake and that modulate reward, such always the same sites where stimulation as GABA microinjections into the medial was rewarding (Valenstein et al., 1970; shell, send descending signals that activate Valenstein, 1976). Eating electrodes were orexin neurons in the hypothalamus (Baldo not punishing electrodes. Instead, the et al., 2004; Zheng et al., 2003). eating electrodes were reward electrodes. Neuroscientists have only begun to Stimulation-induced reward and stimulation- understand the nature and role of interac- induced hunger drive appeared identical tions between mesolimbic reward systems or, at least, had identical causes in the acti- and hypothalamic hunger systems, but vation of the same electrode. This meant more recent developments show that such that the reward could not be due to drive interactions are there and of great impor- reduction. The reward electrode increased tance. They undoubtedly play major roles the motivation to eat, it did not reduce that in modulating the pleasure and incentive drive. Instead reward must be understood value of food rewards during normal as a motivational phenomenon of its own, hunger versus satiety states, possibly also involving its own active brain mechanisms. connecting reward modulation to longer This is a reason why many affective neuro- term body weight elevation and dieting scientists of recent decades have rejected states, and finally perhaps even in allowing drive-reduction explanations of appetite food reward cues to influence the activation and food reward, and instead focused on of hunger deficit systems. These interac- understanding hedonic/incentive brain tions also provide avenues, at least in prin- mechanisms of reward in their own right. ciple, by which eating disorders cause distortion in the function of reward systems, so that they become either exag- VI. CONNECTING gerated or suppressed in function. Such BRAIN REWARD AND interactions will be important to try to REGULATORY SYSTEMS understand better in the future.

A fascinating final topic is the interaction between brain systems of mesolimbic re- VII. CONCLUSION ward on the one hand and of hypothalamic hunger and body weight regulation on For most people, eating patterns and the other (Baldo et al., 2003; Berthoud, 2002, body weights remain within normally pre- 2004; Kelley et al., 2005b). How do these scribed bounds. Perhaps it is the prevalence brain systems connect and influence each of normal body weights, rather than other? obesity, that should be most surprising in Control signals go back and forth modern affluent societies where tasty foods between mesolimbic reward systems and abound. As is often pointed out, brain hypothalamic systems for hunger regula- mechanisms for food reward and appetite tion. For example, hypothalamic orexin- evolved under pressures to protect us from REFERENCES 211 scarcity. As a result, overeating in the face Avena, N. M., and Hoebel, B. G. (2003b). A diet pro- of present abundance could be an under- moting sugar dependency causes behavioral cross- sensitization to a low dose of amphetamine. standable overshoot inherited from our Neuroscience 122, 17–20. evolutionary past. Bakshi, V. P., and Kelley, A. E. (1994). Sensitization and When eating patterns and body weight conditioning of feeding following multiple mor- do diverge from the norm, questions arise phine microinjections into the nucleus accumbens. concerning the involvement of food reward Brain Res. 648, 342–346. Baldo, B. A., Daniel, R. A., Berridge, C. W., and Kelley, systems in the brain. All eating patterns are A. E. (2003). Overlapping distributions of orexin/ controlled intimately by brain mechanisms hypocretin- and dopamine-beta-hydroxylase im- of food reward, whether those mechanisms munoreactive fibers in rat brain regions mediating operate in normal mode or abnormal arousal, motivation, and stress. J. Comp. Neurol. 464, modes. A primary signpost to help guide 220–237. Baldo, B. A., Gual-Bonilla, L., Sijapati, K., Daniel, R. A., future thinking is to know whether any Landry, C. F., and Kelley, A. E. (2004). Activation of pathological patterns of eating are caused a subpopulation of orexin/hypocretin-containing by identifiable pathologies in brain reward hypothalamic neurons by GABAA receptor- function. Can distorted patterns of eating mediated inhibition of the nucleus accumbens shell, be corrected by medications that alter brain but not by exposure to a novel environment. Eur. J. Neurosci. 19, 376–386. reward mechanisms? Or are the causes of Bell, S. M., Stewart, R. B., Thompson, S. C., and Meisch, eating disorders essentially independent of R. A. (1997). Food-deprivation increases cocaine- brain reward systems? These remain ques- induced conditioned place preference and locomotor tions to guide future research on how brain activity in rats. Psychopharmacology (Berlin) 131, 1–8. substrates of food reward relate to eating Bello, N. T., Sweigart, K. L., Lakoski, J. M., Norgren, R., and Hajnal, A. (2003). Restricted feeding with sched- disorders. uled sucrose access results in an upregulation of the rat dopamine transporter. Am. J. Physiol. Regul. Integr. Comp. Physiol. 284, R1260–R1268. Acknowledgments Berns, G. S., McClure, S. M., Pagnoni, G., and Montague, P. R. (2001). Predictability modulates human brain response to reward. J. Neurosci. 21, This research was supported by grants 2793–2798. from the NIH (DA015188 and MH63649). Berridge, K. C. (2000). Measuring hedonic impact in I thank Kyle Smith, and Steve Mahler for animals and infants: microstructure of affective taste helpful comments on an earlier version of reactivity patterns. Neurosci. Biobehav. Rev. 24, 173–198. this chapter. I am also grateful for the Berridge, K. C. (2001). Reward learning: Reinforce- kind hospitality during preparation of ment, incentives, and expectations. In “The Psy- this manuscript of faculty, staff, and chology of Learning and Motivation” (D. L. Medin, students at the University of Cambridge in ed.), Vol. 40, pp. 223–278. Academic Press, New York. the Department of Experimental Psychology Berridge, K. C. (2003). Pleasures of the brain. Brain Cogn. 52, 106–128. and Downing College, and for fellowship Berridge, K. C., and Valenstein, E. S. (1991). What psy- support from the John Simon Guggenheim chological process mediates feeding evoked by elec- Memorial Foundation. trical stimulation of the lateral hypothalamus? Behav. Neurosci. 105, 3–14. Berridge, K. C., and Robinson, T. E. (1998). What is the role of dopamine in reward: hedonic impact, reward learning, or incentive salience? Brain Res. Rev. 28, 309–369. Berridge, K. C., and Robinson, T. E. (2003). Parsing References reward. Trends Neurosci. 26, 507–513. Berridge, K. C., and Winkielman, P. (2003). What is an Ahn, S., and Phillips, A. G. (1999). Dopaminergic cor- unconscious emotion? (The case for unconscious relates of sensory-specific satiety in the medial pre- “liking”). Cogn. Emot. 17, 181–211. frontal cortex and nucleus accumbens of the rat. Berthoud, H. R. (2002). Multiple neural systems con- J. Neurosci. 19, B1–B6. trolling food intake and body weight. Neurosci. Avena, N. A., and Hoebel, B. G. (2003a). Amphetamine- Biobehav. Rev. 26, 393–428. sensitized rats show sugar-induced hyperactivity Berthoud, H.-R. (2004). Mind versus metabolism in the (cross-sensitization) and sugar hyperphagia. Phar- control of food intake and energy balance. Physiol. macol. Biochem. Behav. 74, 635–639. Behav. 81, 781–793. 212 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE

Brauer, L. H., and De Wit, H. (1997). High dose “Stevens’ Handbook of Experimental Psychology: pimozide does not block amphetamine-induced Learning, Motivation, and Emotion” (C. R. Gallistel, euphoria in normal volunteers. Pharmacol. Biochem. ed.), 3 Ed., Vol. 3, pp. 497–534. Wiley & Sons, New Behav. 56, 265–272. York. Brauer, L. H., Goudie, A. J., and de Wit, H. (1997). Dickinson, A., Smith, J., and Mirenowicz, J. (2000). Dis- Dopamine ligands and the stimulus effects of sociation of Pavlovian and instrumental incentive amphetamine: animal models versus human labora- learning under dopamine antagonists. Behav. Neurosci. tory data. Psychopharmacology (Berlin) 130, 2–13. 114, 468–483. Cabanac, M. (1971). Physiological role of pleasure. Dietz, D. M., Curtis, K. S., and Contreras, R. J. (2005). Science 173, 1103–1107. Taste, salience, and increased NaCl ingestion after Cannon, C. M., and Palmiter, R. D. (2003). Reward repeated sodium depletions. Chem. Senses. without dopamine. J. Neurosci. 23, 10827–10831. Everitt, B. J., and Robbins, T. W. (2005). Neural systems Cardinal, R. N., Parkinson, J. A., Hall, J., and Everitt, of reinforcement for drug addiction: from actions to B. J. (2002). Emotion and motivation: the role of the habits to compulsion. Nature Neurosci. 8, 1481–1489. amygdala, ventral striatum, and prefrontal cortex. Gosnell, B. A. (2005). Sucrose intake enhances behav- Neurosci. Biobehav. Rev. 26, 321–352. ioral sensitization produced by cocaine. Brain Res. Carr, K. D. (2002). Augmentation of drug reward by 1031, 194–201. chronic food restriction: Behavioral evidence and Gosnell, B. A., and Levine, A. S. (1996). Stimulation of underlying mechanisms. Physiol. Behav. 76, 353–364. ingestive behavior by preferential and selective Colantuoni, C., Schwenker, J., McCarthy, J., Rada, P., opioid agonists. In “Drug Receptor Subtypes and Ladenheim, B., Cadet, J. L., et al. (2001). Excessive Ingestive Behavior” (S. J. Cooper, and P. G. Clifton, sugar intake alters binding to dopamine and mu- eds.), pp. 147–166. Academic Press, London. opioid receptors in the brain. Neuroreport 12, Grigson, P. S. (2002). Like drugs for chocolate: separate 3549–3552. rewards modulated by common mechanisms? Cooper, S. J. (2004). Endocannabinoids and food con- Physiol. Behav. 76, 389–395. sumption: comparisons with benzodiazepine and Grill, H. J., and Norgren, R. (1978a). The taste reactivity opioid palatability-dependent appetite. Eur. J. Phar- test. II. Mimetic responses to gustatory stimuli in macol. 500, 37–49. chronic thalamic and chronic decerebrate rats. Brain Cooper, S. J. (2005). Palatability-dependent appetite Res. 143, 281–297. and benzodiazepines: new directions from the phar- Grill, H. J., and Norgren, R. (1978b). The taste reactivity macology of GABA(A) receptor subtypes. Appetite test. I. Mimetic responses to gustatory stimuli in 44, 133–150. neurologically normal rats. Brain Res. 143, 263–279. Cooper, S. J., and Higgs, S. (1994). Neuropharmacology Hajnal, A., and Norgren, R. (2005). Taste pathways that of appetite and taste preferences. In “Appetite: mediate accumbens dopamine release by sapid Neural and Behavioural Bases” (C. R. Legg, and sucrose. Physiol. Behav. 84, 363–369. D. A. Booth, eds.), pp. 212–242. Oxford University Harris, G. C., Wimmer, M., and Aston-Jones, G. (2005). Press, New York. A role for lateral hypothalamic orexin neurons in Corbit, L. H., and Balleine, B. W. (2003). The role of pre- reward seeking. Nature 437, 556–559. limbic cortex in instrumental conditioning. Behav. Higgs, S., Williams, C. M., and Kirkham, T. C. (2003). Brain Res. 146, 145–157. Cannabinoid influences on palatability: microstruc- Cromwell, H. C., and Berridge, K. C. (1993). Where tural analysis of sucrose drinking after delta(9)- does damage lead to enhanced food aversion: the tetrahydrocannabinol, anandamide, 2-arachidonoyl ventral pallidum/substantia innominata or lateral glycerol and SR141716. Psychopharmacology (Berlin) hypothalamus? Brain Res. 624, 1–10. 165, 370–377. Dallman, M. F. (2003). Fast glucocorticoid feedback Jarrett, M. M., Limebeer, C. L., and Parker, L. A. (2005). favors “the munchies”. Trends Endocrinol. Metab. 14, Effect of Delta9-tetrahydrocannabinol on sucrose 394–396. palatability as measured by the taste reactivity test. de Vaca, S. C., and Carr, K. D. (1998). Food restriction Physiol. Behav. 86, 475–479. enhances the central rewarding effect of abused Jenkins, H. M., and Moore, B. R. (1973). The form of drugs. J. Neurosci. 18, 7502–7510. the auto-shaped response with food or water rein- Di Chiara, G. (2002). Nucleus accumbens shell and forcers. J. Exp. Anal. Behav. 20, 163–181. core dopamine: differential role in behavior and Kelley, A. E. (2004). Ventral striatal control of appeti- addiction. Behav. Brain Res. 137, 75–114. tive motivation: role in ingestive behavior and Dickinson, A., and Balleine, B. W. (2000). Causal cog- reward-related learning. Neurosci. Biobehav. Rev. 27, nition and goal-directed action. In “The Evolution of 765–776. Cognition,” pp. 185–204. The MIT Press, Cambridge, Kelley, A. E., Baldo, B. A., and Pratt, W. E. (2005a). A MA. proposed hypothalamic-thalamic-striatal axis for Dickinson, A., and Balleine, B. (2002). The role of learn- the integration of energy balance, arousal, and food ing in the operation of motivational systems. In reward. J. Comp. Neurol. 493, 72–85. REFERENCES 213

Kelley, A. E., Baldo, B. A., Pratt, W. E., and Will, M. J. Olds, J. (1973). The discovery of reward systems in the (2005b). Corticostriatal-hypothalamic circuitry and brain. In “Brain Stimulation and Motivation: food motivation: Integration of energy, action and Research and Commentary” (E. S. Valenstein, ed.), reward. Physiol. Behav. 86, 773–795. pp. 81–99. Scott, Foresman and Company, Glenview, Kirkham, T. C. (2005). Endocannabinoids in the regula- IL. tion of appetite and body weight. Behav. Pharmacol. Peciña, S., Berridge, K. C., and Parker, L. A. (1997). 16, 297–313. Pimozide does not shift palatability: Separation of Kirkham, T. C., and Williams, C. M. (2001). Endoge- anhedonia from sensorimotor suppression by taste nous cannabinoids and appetite. Nutr. Res. Rev. 14, reactivity. Pharmacol. Biochem. Behav. 58, 801–811. 65–86. Peciña, S., Cagniard, B., Berridge, K. C., Aldridge, J. W., Kirkham, T. C., and Williams, C. M. (2004). Endo- and Zhuang, X. (2003). Hyperdopaminergic mutant cannabinoid receptor antagonists: potential for mice have higher “wanting” but not “liking” for obesity treatment. Treat. Endocrinol. 3, 345–360. sweet rewards. J. Neurosci. 23, 9395–9402. Koob, G. F., and Le Moal, M. (2006). “Neurobiology of Peciña, S., and Berridge, K. C. (2005). Hedonic hot spot Addiction.” Academic Press, New York. in nucleus accumbens shell: Where do mu-opioids Kringelbach, M. L. (2004). Food for thought: hedonic cause increased hedonic impact of sweetness? J. experience beyond homeostasis in the human brain. Neurosci. 25, 11777–11786. Neuroscience 126, 807–819. Pelchat, M. L., Johnson, A., Chan, R., Valdez, J., and Kringelbach, M. L., de Araujo, I. E., and Rolls, E. T. Ragland, J. D. (2004). Images of desire: food-craving (2004). Taste-related activity in the human dorsolat- activation during fMRI. 23, 1486–1493. eral prefrontal cortex. Neuroimage 21, 781–788. Reynolds, S. M., and Berridge, K. C. (2002). Positive Levine, A. S., and Billington, C. J. (2004). Opioids as and negative motivation in nucleus accumbens agents of reward-related feeding: a consideration of shell: Bivalent rostrocaudal gradients for GABA- the evidence. Physiol. Behav. 82, 57–61. elicited eating, taste “liking”/“disliking” reactions, Levine, A. S., Kotz, C. M., and Gosnell, B. A. (2003). place preference/avoidance, and fear. J. Neurosci. 22, Sugars: hedonic aspects, neuroregulation, and 7308–7320. energy balance. Am. J. Clin. Nutr. 78, 834S–842S. Reynolds, S. M., and Berridge, K. C. (2003). Glutamate Leyton, M., Boileau, I., Benkelfat, C., Diksic, M., Baker, motivational ensembles in nucleus accumbens: ros- G., and Dagher, A. (2002). Amphetamine-induced trocaudal shell gradients of fear and feeding in the increases in extracellular dopamine, drug wanting, rat. Eur. J. Neurosci. 17, 2187–2200. and novelty seeking: a PET/[11C]raclopride study Robinson, S., Sandstrom, S. M., Denenberg, V. H., and in healthy men. Neuropsychopharmacology 27, Palmiter, R. D. (2005). Distinguishing whether 1027–1035. dopamine regulates liking, wanting, and/or learn- Mahler, S. V., Smith, K. S., and Berridge, K. C. (2004). ing about rewards. Behav. Neurosci. What is the “motivational” mechanism for the mar- Robinson, T. E., and Berridge, K. C. (1993). The neural ijuana munchies? The effects of intra-accumbens basis of drug craving: an incentive-sensitization anandamide on hedonic taste reactions to sucrose. theory of addiction. Brain Res. Rev. 18, 247–291. Paper presented at the Society for Neuroscience Robinson, T. E., and Berridge, K. C. (2003). Addiction. Conference, San Diego. Annu. Rev. Psychol. 54, 25–53. Miller, N. E. (1973). How the project started. In “Brain Roitman, M. F., Na, E., Anderson, G., Jones, T. A., and Stimulation and Motivation: Research and Com- Bernstein, I. L. (2002). Induction of a salt appetite mentary” (E. S. Valenstein, ed.), pp. 53–68. Scott, alters dendritic morphology in nucleus accumbens Foresman and Company, Glenview, IL. and sensitizes rats to amphetamine. J. Neurosci. Montague, P. R., Hyman, S. E., and Cohen, J. D. (2004). 20026416. Computational roles for dopamine in behavioural Roitman, M. F., Stuber, G. D., Phillips, P. E. M., control. Nature 431, 760–767. Wightman, R. M., and Carelli, R. M. (2004). Nader, K., Bechara, A., and van der Kooy, D. Dopamine operates as a subsecond modulator of (1997). Neurobiological constraints on behavioral food seeking. J. Neurosci. 24, 1265–1271. models of motivation. Annu. Rev. Psychol. 48, 85– Rolls, E. T. (2005). “Emotion Explained.” Oxford Uni- 114. versity Press, Oxford and New York. Nocjar, C., and Panksepp, J. (2002). Chronic intermit- Salamone, J. D., and Correa, M. (2002). Motivational tent amphetamine pretreatment enhances future views of reinforcement: implications for under- appetitive behavior for drug- and natural-reward: standing the behavioral functions of nucleus accum- interaction with environmental variables. Behav. bens dopamine. Behav. Brain Res. 137, 3–25. Brain Res. 128, 189–203. Scammell, T. E., and Saper, C. B. (2005). Orexin, drugs O’Doherty, J. P., Deichmann, R., Critchley, H. D., and and motivated behaviors. Nature Neurosci. 8, Dolan, R. J. (2002). Neural responses during antici- 1286–1288. pation of a primary taste reward. Neuron 33, Schallert, T., and Whishaw, I. Q. (1978). Two types of 815–826. aphagia and two types of sensorimotor impairment 214 8. BRAIN REWARD SYSTEMS FOR FOOD INCENTIVES AND HEDONICS IN NORMAL APPETITE

after lateral hypothalamic lesions: observations in Valenstein, E. S. (1976). The interpretation of behavior normal weight, dieted, and fattened rats. J. Comp. evoked by brain stimulation. In “Brain-Stimulation Physiol. Psychol. 92, 720–741. Reward” (A. Wauquier, and E. T. Rolls, eds.), Schultz, W. (2006). Behavioral theories and the neuro- pp. 557–575. Elsevier, New York. physiology of reward. Annu. Rev. Psychol. Valenstein, E. S., Cox, V. C., and Kakolewski, J. W. Sharkey, K. A., and Pittman, Q. J. (2005). Central and (1970). Reexamination of the role of the hypo- peripheral signaling mechanisms involved in endo- thalamus in motivation. Psychol. Rev. 77, 16–31. cannabinoid regulation of feeding: a perspective on Vezina, P. (2004). Sensitization of midbrain dopamine the munchies. Sci. STKE 2005(277), pe15. neuron reactivity and the self-administration of psy- Shizgal, P. (1997). Neural basis of utility estimation. chomotor stimulant drugs. Neurosci. Biobehav. Rev. Curr. Opin. Neurobiol. 7, 198–208. 27, 827–839. Small, D. M., Zatorre, R. J., Dagher, A., Evans, A. C., Vigano, D., Rubino, T., and Parolaro, D. (2005). and Jones-Gotman, M. (2001). Changes in brain Molecular and cellular basis of cannabinoid and activity related to eating chocolate—From pleasure opioid interactions. Pharmacol. Biochem. Behav. 81, to aversion. Brain 124, 1720–1733. 360–368. Small, D. M., Jones-Gotman, M., and Dagher, A. (2003). Volkow, N. D., Wang, G. J., Fowler, J. S., Logan, J., Feeding-induced dopamine release in dorsal stria- Jayne, M., Franceschi, D., et al. (2002). “Non- tum correlates with meal pleasantness ratings in hedonic” food motivation in humans involves healthy human volunteers. Neuroimage 19, dopamine in the dorsal striatum and methyl- 1709–1715. phenidate amplifies this effect. Synapse 44, 175–180. Smith, K. S., and Berridge, K. C. (2005). The ventral Wachtel, S. R., Ortengren, A., and de Wit, H. (2002). pallidum and hedonic reward: neurochemical maps The effects of acute haloperidol or risperidone of sucrose “liking” and food intake. J. Neurosci. 25, on subjective responses to methamphetamine in 8637–8649. healthy volunteers. Drug Alcohol Depend. 68, 23–33. Smith, K. S, Tindell, A. J., Berridge, K. C., and Aldridge, Wang, G. J., Volkow, N. D., Logan, J., Pappas, N. R., J. W. (2004). Ventral pallidal neurons code the Wong, C. T., Zhu, W., et al. (2001). Brain dopamine hedonic enhancement of an NaCl taste after sodium and obesity. Lancet 357, 354–357. depletion. Paper presented at the Society for Neuro- Wang, G. J., Volkow, N. D., Telang, F., Jayne, M., Ma, science Conference, San Diego. J., Rao, M., et al. (2004a). Exposure to appetitive food Steiner, J. E. (1973). The gustofacial response: observa- stimuli markedly activates the human brain. Neu- tion on normal and anencephalic newborn infants. roimage 21, 1790–1797. Symposium on Oral Sensation and Perception 4, Wang, G. J., Volkow, N. D., Thanos, P. K., and Fowler, 254–278. J. S. (2004b). Similarity between obesity and drug Steiner, J. E., Glaser, D., Hawilo, M. E., and Berridge, addiction as assessed by neurofunctional imaging: a K. C. (2001). Comparative expression of hedonic concept review. J. Addict. Dis. 23, 39–53. impact: Affective reactions to taste by human infants Will, M. J., Franzblau, E. B., and Kelley, A. E. (2003). and other primates. Neurosci. Biobehav. Rev. 25, 53–74. Nucleus accumbens mu-opioids regulate intake of a Stellar, J. R., Brooks, F. H., and Mills, L. E. (1979). high-fat diet via activation of a distributed brain Approach and withdrawal analysis of the effects of network. J. Neurosci. 23, 2882–2888. hypothalamic stimulation and lesions in rats. J. Will, M. J., Franzblau, E. B., and Kelley, A. E. (2004). Comp. Physiol. Psychol. 93, 446–466. The amygdala is critical for opioid-mediated binge Teitelbaum, P., and Epstein, A. N. (1962). The lateral eating of fat. Neuroreport 15, 1857–1860. hypothalamic syndrome: recovery of feeding and Wilson, C., Nomikos, G. G., Collu, M., and Fibiger, drinking after lateral hypothalamic lesions. Psychol. H. C. (1995). Dopaminergic correlates of motivated Rev. 69, 74–90. behavior: importance of drive. J. Neurosci. 15(7 Pt. 2), Tindell, A. J., Berridge, K. C., and Aldridge, J. W. (2004). 5169–5178. Ventral pallidal representation of pavlovian cues Winn, P. (1995). The lateral hypothalamus and moti- and reward: population and rate codes. J. Neurosci. vated behavior: an old syndrome reassessed and a 24, 1058–1069. new perspective gained. Curr. Dir. Psychol. 4, Tindell, A. J., Berridge, K. C., Zhang, J., Pecina, S., and 182–187. Aldridge, J. W. (2005). Ventral pallidal neurons code Wise, R. A. (1985). The anhedonia hypothesis: Mark III. incentive motivation: amplification by mesolimbic Behav. Brain Sci. 8, 178–186. sensitization and amphetamine. Eur. J. Neurosci. 22, Wise, R. A. (2004). Drive, incentive, and reinforcement: 2617–2634. the antecedents and consequences of motivation. Toates, F. (1986). “Motivational Systems.” Cambridge Nebr. Symp. Motiv. 50, 159–195. University Press, Cambridge. Wolf, S., and Wolff, H. G. (1943). “Human Gastric Tomie, A. (1996). Locating reward cue at response Function, an Experimental Study of a Man and His manipulandum (CAM) induces symptoms of drug Stomach.” Oxford University Press, London, New abuse. Neurosci. Biobehav. Rev. 20, 31. York. REFERENCES 215

Wyvell, C. L., and Berridge, K. C. (2000). Intra- tive responding with emphasis on the nucleus accumbens amphetamine increases the conditioned accumbens. Neurosci. Biobehav. Rev. 24, 85–105. incentive salience of sucrose reward: enhancement Zhang, M., and Kelley, A. E. (2000). Enhanced intake of reward “wanting” without enhanced “liking” or of high-fat food following striatal mu-opioid stimu- response reinforcement. J. Neurosci. 20, 8122–8130. lation: microinjection mapping and fos expression. Wyvell, C. L., and Berridge, K. C. (2001). Incentive- Neuroscience 99, 267–277. sensitization by previous amphetamine exposure: Zheng, H. Y., Corkern, M., Stoyanova, I., Patterson, Increased cue-triggered “wanting” for sucrose L. M., Tian, R., and Berthoud, H. R. (2003). Peptides reward. J. Neurosci. 21, 7831–7840. that regulate food intake—Appetite-inducing Yeomans, M. R., and Gray, R. W. (2002). Opioid pep- accumbens manipulation activates hypothalamic tides and the control of human ingestive behaviour. orexin neurons and inhibits POMC neurons. Neurosci. Biobehav. Rev. 26, 713–728. Am. J. Physiol. -Regul. Integr. Comp. Physiol. 284, Zahm, D. S. (2000). An integrative neuroanatomical R1436–R1444. perspective on some subcortical substrates of adap-