To date, the majority of neuroimaging studies examining the effects of hunger and satiety have focused mainly on the effect of sensory specific satiety for complex flavor stimuli (Del Parigi et al., 2002a
; Del Parigi et al., 2002b
; Kringelbach et al., 2003
; Gautier et al., 1999
; O’Doherty, et al., 2002
; Small et al., 2001
) and olfactory stimuli (Gottfried, et al., 2003
; O’Doherty et al., 2000
). Sensory specific satiety experiments that use a complex flavor stimulus as the satiety inducer and the stimulus, elicit responses from both taste and olfactory systems and the corresponding reward activation in cortical regions reflects activity in both systems (de Araujo et al., 2003c
). Using flavor stimuli has very strong ecological validity and provides excellent information about cortical processing of complex chemosensory stimuli, yet limits the information regarding the taste system. In contrast, the present study sought to separate taste from flavor and thus investigated brain regions activated in response to pure gustatory stimulation during two physiological conditions, hunger and satiety using six stimuli with different qualities.
While there are clear consistencies in activation across physiological condition, physiological dependent activation is also differentially modulated in different cortical taste regions in response to taste stimuli (, ). The most striking finding, demonstrated by the group and the ROI analyses, is the robust activation in response to sucrose relative to the other stimuli. Specifically, sucrose produced significantly greater global activation, particularly in the hunger state, than other tastes whether they were sweet (the artificial sweetener saccharin), bitter (caffeine) or another taste quality. Frank et al., (2008)
recently reported a similarly greater activation to sucrose than to sucralose (Splenda, another artificial sweetener), in a satiated condition. Artificial sweeteners do not have nutritive value and, therefore, the brain may respond less because the taste of such a stimulus has been conditioned to a stimulus that does not reward with calories. In the present study, it is also possible that the slight bitter side taste that accompanies higher concentrations of saccharin (Schiffman et al., 1995
) contributed to the brain response. It is unlikely due to stimulus intensity differences since perceived intensity ratings were 39.8 ± 3.8 for saccharin, 37.8 ± 5.4 for sucrose (M
± SEM). This effect may be related to the sucrose’s ecological importance to eating behavior. These results suggest that the taste system is sensitive to the internal cues from the body, which may aid in the regulation of feeding behavior.
Fit coefficients corresponding to brain activation for a three way interaction between stimulus (suc: sucrose; caf: caffeine; cit: citric acid; NaCl (sodium chloride), physiology (hunger and satiety), and ROI.
At the group level, various taste stimuli produced significant differences in activation in the hunger minus satiety contrast within the thalamus, hippocampus, and parahippocampus; less consistent activation was found within the insula, OFC BA 47, amygdala, anterior cingulate gyrus and substantia nigra. Globally, brain activation in the hunger condition produced more robust activation to pure taste stimuli relative to water in regions associated with sensory experiences and activation in response to pure taste stimuli in the satiety condition was decreased relative to water in regions associated with emotion and motivation. More specifically, the hunger condition produced consistent activation across taste stimuli that differed in quality (e.g. sweet, bitter, salty, sour, umami) within the insula and thalamus, which are involved in the processing of taste, and the substantia nigra, which is implicated in attentional processing of reward value. This pattern of activation suggests that the motivational state of hunger may elicit activation from brain regions involved in processing sensory information and reward value relative to regions involved in the actual processing of reward per se. From an evolutionary standpoint, activation within these regions may ensure that the tastes being consumed are not potentially dangerous and that what the organism ingests is, in fact, rewarding.
Interestingly, the satiety condition produced less activation for taste stimuli in limbic regions including the parahippocampus, hippocampus, amygdala, and anterior cingulate gyrus. These findings suggest that in the satiety condition, processing taste stimuli produces less activation relative to water in regions involved in emotion and motivation, which may aid in the termination of food consumption that is associated with the motivation for and emotional value of the stimulus. Decreased activation to taste in the limbic regions is in parallel to allesthesia and sensory specific satiety, in that physiologically and behaviorally, food recently consumed influences the motivation to continue with or terminate food consumption. The ROI analysis supports the group level analysis and provides a more precise understanding of the physiological effects of hunger and satiety on the human gustatory system. ROI analysis demonstrated that in the satiety condition, primary and secondary taste regions elicit greater activation in response to taste stimuli relative to regions involved in processing affective, memory and eating behavior. In addition, these same regions (e.g., inferior insula, amygdala, and so forth) have significantly greater activation in the hunger condition than in the satiety condition, with the exception of the superior insula. Examination of insula subregions, based on anatomical and functional distinctions, contributes to the elucidation of the underlying processes that may occur as a result of the physiological condition of the participant and/or reward related contingencies.
The insular cortex is recognized as part of the primary taste cortex in the non-human primate electrophysiology (Yaxley et al., 1988
; Yaxley et al., 1990
; Scott and Plata-Salaman, 1999
) and human neuroimaging literatures (Zald et al., 2002
; Cerf et al., 1998
; Cerf-Ducastel et al., 2001b; de Araujo et al., 2003a
; Faurion et al., 1999
; Frank et al., 2008
; Francis et al., 1999
; Kringelbach et al., 2004
; O’Doherty et al., 2001
Considerable research has also examined the effect of physiological state on activation in response to taste stimuli within the insula. Rolls and colleagues demonstrated that neuronal responses to taste stimuli within the frontal operculum and insular cortex of the macaque monkey were not modulated by physiological condition (Rolls et al., 1988
; Yaxley et al., 1988
). More recently, a number of studies in the human neuroimaging literature suggest that there are motivation/reward related differences in activation within the insular cortex in response to food-related stimuli that include both taste and olfactory components (Small et al., 2001
; Gottfried et al, 2003
; Uher et al., 2006
One possible explanation for the inconsistencies observed in the effect of physiological state on activity of the insular cortex between non-human primates and human primates may be a result of differences in data collection procedures. For example, electrophysiological recordings from non-human primates show activation from single neurons, whereas, human neuroimaging techniques typically average activation over substantially larger anatomically defined regions, such as the insular cortex. Augustine (1996)
has shown anatomically distinct regions in the insula cortex. In line with these findings, human neuroimaging and clinical studies on taste have suggested that sub-regions within the insula may sustain different functions in gustatory processing, such as taste (Cerf et al., 1998
; Faurion et al., 1999
, Prichard, 1999
) somatosensory stimulation (Pardo et al., 1997
; Cerf-Ducastel et al., 2001b), thirst (de Arauo et al., 2003b
; Denton, et al., 1999
) and hunger (Tataranni et al., 1999
). There may also be real across species differences. Interestingly, Scott and Small (2008)
have recently articulated true interspecies differences in pontine taste processing that suggest modulation of taste information by hedonic information at very different levels of the taste system in rodents and primates.
In the current study, at the group level, there was significantly more activation for sucrose and citric acid in the hunger condition than in the satiety condition. Additionally, consistent across five of the six taste stimuli, greater activation was observed in the hunger condition for taste relative to water. For the ROI analysis, activation in the inferior insula was modulated by the physiological state of the participant; however, superior insula did not show similar modulation. A unique strength of the current design is the ability to dissociate responses to different pure taste stimuli that represent a number of taste qualities. The ROI analysis demonstrated that activation between the inferior and superior insulae is modulated by physiological state and taste stimulus.
Taken together, these findings suggest that the insular cortex may be receiving not only sensory signals but also introspective signals of motivation and/or hedonic value and that these effects are more clearly differentiated when examining the insula based on anatomical and functional distinctions. The current findings provide a critical foundation for understanding the multi-modal nature of eating behavior and how motivational/reward value affects brain activation.
In the current study, there was significant positive activation within the thalamus in the hunger condition in response to sucrose, caffeine, NaCl, and citric acid; however, there was no significant activation in the sated condition. Single neuron recordings from non-human primates have demonstrated that the ventroposteromedial thalamic nucleus (VPMpc) responds to gustatory, thermal, and tactile stimulation (Pritchard et al., 1989
). It has been suggested that the VPMpc may play a role in eating behavior given the efferent projections from the VPMpc to regions in the primary and secondary taste cortices (Pritchard et al., 1986
). Komura and colleagues (Komura et al., 2001
) reported that single neurons within the thalamus respond to rewarding stimuli. In addition, human neuroimaging studies have reported activation within the thalamus in response to gustatory stimulation (Faurion et al., 1999
; Kobayashi et al., 2004
; Zald et al., 1998
) and activation that is modulated by hunger and satiety (Gautier et al., 1999
; Tataranni et al., 1999
). Taken in combination, these findings support the role of the thalamus in processing taste stimuli and suggest that activation in response to taste stimuli is partially modulated by reward value.
In the satiated condition, there was a decrease in activation in the parahippocampus for all stimuli, except GMP, in comparison to water and a decrease in activation in the hippocampus for sucrose, saccharin, caffeine, and citric acid. Research in rodents has found that selective lesions in the hippocampus result in an impaired ability to discriminate between interoceptive states of hunger and satiety (Davidson and Jarrard, 1993
). Previous neuroimaging studies have shown that the hippocampal and parahippocampal gyri are implicated in food craving (Pelchat et al., 2004
), the physiological state of hunger (Tataranni et al., 1999
), when motivationally relevant food objects are shown (LaBar et al., 2001
), and when tasting a liquid meal (Gautier et al., 1999
). This suggests that activation in the hippocampal and parahippocampal gyri may be modulated by the interoceptive signaling of satiety, which may aid in the ability to engage in normal feeding behavior (i.e. food consumption and termination). Engagement of neuronal populations in these areas that subserve memory function suggests that taste stimuli are appreciated in the context of previous experience with a stimulus, a fact that may have implications for clinical populations with disordered eating as well as avoiding the ingestion of harmful substances.
In the group and ROI analyses, there were significant differences in activation between the hunger and satiety conditions within the amygdala for both sucrose and citric acid, with greater activation in the hunger condition relative to the sated condition. It is suggested that the amygdala may be responsible for relaying motivational significance, possibly influenced by taste stimuli and physiological state (Yan and Scott, 1996
). As in non-human primate studies, human neuroimaging studies reinforce the notion that the amygdala’s response to gustatory stimuli is modulated by satiety. LaBar and colleagues (2001)
examined brain activation in response to pictures of food when participants were hungry and sated and demonstrated increased activation in the amygdala in the hungry condition relative to the sated condition. Furthermore, the anticipation of a primary taste reward in a hungry state, investigated by O’Doherty et al. (2002)
, resulted in an increase in activation in the amygdala during reward anticipation and not during reward receipt, suggesting that different regions are involved in processing anticipation and receipt of rewarding stimuli. The amygdala is also associated with a number of different affective processes, such as food craving (Pelchat et al., 2004
) hunger-enhanced memory (Morris and Dolan, 2001
), and increased incentive to food items (Arana et al., 2003
Activation within the amygdala in the sated condition was decreased in response to caffeine and citric acid. Both animal and neuroimaging studies have shown that neuronal response within the amygdala is associated with pleasant and unpleasant gustatory stimulation (Scott et al., 1993
; O’Doherty et al., 2001
There were significant differences in activation between the hunger and satiety conditions within the orbitofrontal cortex (OFC) Brodmann area (BA) 47 for sucrose and citric acid. At the group level, brain activation in the OFC 47 was found in response to sucrose during the hunger condition and failed to reach statistical significance in the satiated condition. No other stimulus exhibited as robust a signal within the OFC as sucrose, which may be related to the fact that of all six stimuli, sucrose was the only stimulus that was perceived by all participants as significantly pleasant. Interestingly, citric acid was perceived as moderately unpleasant and there was a significant decrease in activation in the sated condition within the OFC 47.
The current findings in the OFC are consistent with the non-human primate and human literature. Single neuron recordings from non-human primates demonstrated that regions within the caudolateral orbitofrontal cortex (OFC) respond to gustatory, olfactory, and multimodal stimulation (Rolls et al., 1990
) and are modulated by hunger and satiety (Rolls et al., 1989
). Human neuroimaging studies have demonstrated that the OFC responds to aversive and pleasant tastants (Zald et al., 1998
; Francis et al., 1999
; O’Doherty et al., 2002
; Kringelbach et al., 2004
). Moreover, activation within the OFC was found in response to flavor stimuli (Small et al., 2001
; Gautier et al., 1999
; Kringelbach et al., 2003
), food-related olfactory stimuli (O’Doherty et al., 2000
; Cerf-Ducastel and Murphy, 2001a) and was modulated by satiety (Tataranni et al., 1999
; de Araujo et al., 2003b
; Gottfried et al., 2003
). In addition, activation within the OFC is associated with incentive, motivation, and goal selection (Arana et al., 2003
The multi-modal functioning of the OFC substantiates its role in the formation of flavor perception, whereas the modulation of its activity by satiety supports the role of the OFC in the evaluation of pleasantness/motivation. In combination, this provides further evidence that the OFC functions as an integration region where flavor perception and motivation interact.
Anterior Cingulate Gyrus
Activation in the anterior cingulate gyrus in the sated condition was significantly decreased in response to sucrose, caffeine, GMP, and citric acid relative to water. Previous studies have reported that the anterior cingulate responds to aversive and pleasant taste stimuli (de Araujo and Rolls, 2004
; Zald et al., 1998
; Faurion et al., 1998
). The anterior cingulate gyrus has also been shown to be associated with energy content and the palatability of foods (de Araujo and Rolls, 2004
). Additionally, de Araujo and colleagues (de Araujo et al., 2003b
) demonstrated that activation in the anterior cingulate gyrus, independent of the stimulus, was greater in a thirsty condition relative to a sated condition. These findings suggest that activation in the anterior cingulate gyrus may be modulated by motivational states.
In the current study, activation within the substantia nigra was found in response to sucrose, citric acid, and caffeine in comparison to water in the hungry condition. However, activation failed to reach statistical significance in the satiated condition. Research has shown that the substantia nigra is involved in anticipation of reward related stimuli found in both animal (Schultz, 1998
) and human neuroimaging studies (O’Doherty et al., 2002
; Kirsch et al., 2003
). Moreover, Kirsch and colleagues demonstrated that activation within the substantia nigra is modulated by motivation, i.e., the greater the reward the stronger the activation. Therefore, the absence of significant activation within the satiety condition could be interpreted as a decrease in motivation as a result of being satiated in comparison to the motivating state of hunger induced by a 12 hour fast.
Regions of Interest Analysis
Two main analyses were conducted for the ROI analysis. In the first RM-ANOVA, all six stimuli were included. Newman-Keuls Multiple Range Test revealed that, in the satiety condition, the primary (inferior and superior insulae) and secondary (OFC 11 and OFC 47) taste regions exhibited significantly greater brain activation in response to all stimuli than regions involved in processing eating behavior (hypothalamus), affect (amygdala), and memory (hippocampus, parahippocampus and entorhinal cortex), p < .05. These same regions demonstrated significantly greater activation in the hunger condition than in the satiety condition, with the exception of the superior insula. Specifically, for all taste stimuli, brain activation was significantly greater in the hunger condition than in the satiety condition in the hypothalamus, which is involved in metabolic processes (e.g., regulation of hunger and thirst; Elmquist et al., 1998
). In addition, activation was greater in the hippocampus, parahippocampus and entorhinal cortices that have been shown to be involved in hunger, satiety, and various components of memory processing (e.g., encoding, retrieval, and familiarity; for a review, see Squire et al., 2004
). As expected, there was greater activation within the amygdala during the hunger condition relative to the sated condition, which has been shown to be involved in anticipation (Knutson et al., 2001
) and receipt (Elliott et al., 2003
) of rewarding and non-rewarding stimuli.
In the second RM-ANOVA, activation to the four prototypical tastants (sweet, sour, salty, and bitter) were included in the analysis. There was a significant three-way interaction between physiology, stimulus, and ROI (). As can be seen in and supported statistically by Newman Keuls Multiple Ranges Test, p < .05, similar patterns of activation emerge from caffeine, NaCl and citric acid. There is statistically greater activation in the hunger condition relative to the satiety condition within regions involved in primary taste regions and those involved in higher order gustatory processing (affect, eating behavior, memory). Interestingly, the pattern of brain activation in response to sucrose was atypical in comparison to the other stimuli. Specifically, brain activation was significantly greater in response to sucrose than to the other stimuli.
The data suggest that a number of regions involved in taste processing are modulated by the physiological state of the participant. It should also be noted that qualitatively different stimuli produce variable patterns of activation that are physiologically dependent. These findings underscore the importance of including qualitatively different stimuli in order to extract the functional role of regions involved in processing gustatory stimuli as well as flavor stimuli.
In the present experiment, water was used as the baseline comparison. Water was chosen as the baseline for a number of reasons. First, there exists a significant body of research for comparison; second, because water was used as the solvent for the tastants; and finally, water was used as the rinse between stimuli. There is a very large psychophysical literature on the taste of solutions produced in this way. However, we note that previous fMRI research has demonstrated significant activation in primary gustatory cortex during the presentation of water (Zald and Pardo, 2000
; de Araujo et al., 2003b
). Additionally, de Araujo and colleagues found that brain activation in response to water yields patterns of activation that are similar to those evoked during the presentation of prototypical tastants (e.g., glucose and NaCl). Activation of the cortical gustatory regions is not limited to tastants. For example, olfactory stimuli delivered retronasally produce patterns of activation in cortical gustatory areas, similarly to those produced by taste stimuli (Cerf-Ducastel and Murphy, 2001
; Small et al., 2005
), which may also suggest that these regions are involved in somatosensory and motor processing during oral stimulation, irrespective of the stimulus. One could hypothesize then, that even a tasteless solution would engage the same gustatory regions. In fact, a recent study by Veldhuizen et al., 2007
, demonstrated that paying attention to a tasteless solution administered to the oral cavity produced activation within the insula and operculum. These findings are concordant with behavioral observations of taste-smell confusions when odorants are presented in aqueous solution in the oral cavity (Murphy et al., 1977
). In the present experiment, activation to taste stimuli was contrasted with activation to water. This does present a limitation. However, any influence of using water as the baseline in the analysis can be expected to underestimate the size of effects in those cases where water alone would produce activation.
The physiological state of the subject modulated brain activation within specific regions in response to a range of pure taste stimuli. In particular, the inferior insula was modulated by physiological condition, whereas, the superior insula was not. The most striking finding was the robust activation for sucrose relative to other stimuli, particularly in the hunger state. Differences in activation across physiological condition and stimuli suggest that other motivational (i.e., wanting) and hedonic (i.e., liking) factors, as well as memory for previous experience with a stimulus, result in the recruitment of additional neuronal populations and thereby contribute to the pattern of brain activation. These findings highlight the importance of using stimuli that vary in quality when examining neural correlates of reward and emotion, particularly in clinical populations that may have complex relationships with taste stimuli. Understanding how the physiological states of hunger and satiety modulate the taste system provides a necessary foundation for the complete elucidation of the role of hunger and satiety in the more global perception that is flavor. The current experiment also provides a foundation on which comparisons can be made regarding neural correlates of chemosensory function and hunger and satiety in healthy young adults and clinical populations with eating disorders such as anorexia and bulimia.