During the first Pavlovian stage of PIT, subjects are exposed to cues that predict food. For example, in a human version of the task, different colored lights within a computer-simulated image of a vending machine indicate that the participant will receive various food outcomes such as M&M’s, potato chips, or juice. In the rodent version, a loud noise or tone is paired with the delivery of a grain pellet or a sucrose solution. After several training sessions, an instrumental phase begins in which subjects perform an action to earn food outcomes; for example, pressing a left or right button on the vending machine to earn M&M’s and chips, respectively, whereas rodents learn to press left and right levers to earn pellets and sucrose solution. In the final test phase, known as “transfer,” the subject is presented with the cues and their subsequent action selection is measured. In the general version of the task (general PIT; gPIT), subjects will increase the performance of any or all food-seeking actions during presentation of the food-paired cue relative to a neutral, non-predictive cue or when the cues are absent (top panel of Figure 1). In the human example used here, this would comprise presentations of a juice cue elevating overall responding on both left and right buttons, even though they previously predicted M&M’s and chips. In the specific version (specific PIT; sPIT), the subject will increase performance of the action that produced the same outcome as the cue that is currently being presented. For instance, the cue for M&M’s should elicit pressing on the button that earned M&M’s, and the potato chip cue should elicit pressing on the potato chip button (i.e., Same > Different, bottom panel of Figure 1).

Figure 2 outlines how the processes measured in general and specific PIT might influence food-seeking behaviors in the real-world. As can be seen from the examples, PIT is a translationally valid measure of the influence of many types of environmental cues on instrumental behavior (Cartoni et al., 2016). Moreover, each type of PIT measures different facets of food-seeking behaviors. For instance, gPIT reflects the motivational regulation of food-seeking by assessing how food-predictive cues invigorate food-seeking behaviors in a non-specific fashion. In other words as depicted in Figure 2, the predictive cue acts to increase general motivational arousal (Corbit and Balleine, 2005). Outcome value is critical for the expression of gPIT: when outcomes are devalued, the predictive cue loses the capacity to invigorate actions (Balleine, 1994; Corbit et al., 2007; Lingawi et al., 2022). In contrast, sPIT reflects the cognitive control of food-seeking by food-predictive cues. Thus, unlike gPIT, in sPIT the retrieval of the sensory-specific properties of the outcome guides behavior to a greater extent than the general motivational drive to eat (Cartoni et al., 2016). Unlike gPIT, expression of sPIT has been argued to be insensitive to changes in outcome value (Corbit et al., 2007; Lingawi et al., 2022; Sommer et al., 2022), although this conclusion has not been unanimously supported (Panayi and Killcross, 2022).

By measuring distinct aspects of cue-guided food-seeking, gPIT and sPIT represent useful tools for examining the effects of obesity and HFHS diet exposure in controlled settings. Of course, this high degree of experimenter control does come with a trade-off, in the sense that learning in the real world is unlikely to be as neatly partitioned as it is in the laboratory (with separate stages of learning for Pavlovian and instrumental phases). However, whilst keeping this caveat in mind, there are several ways in which a deviation from “normal” PIT performance could indicate the presence of maladaptive behaviors that contribute to excess energy intake (see Figure 1). For instance, an amplified gPIT response – wherein a food-predictive cue invigorates food-seeking in a non-specific fashion – could result in overeating or grazing. By contrast, an amplification of sPIT, wherein a food-predictive cue invigorates behavior targeting a specific food – could become problematic if cues signaling unhealthy foods are more prevalent or more strongly bias our responding toward selecting those foods. On the other hand, an abolishment of sPIT and the loss of the capacity to use cues to guide appropriate choice between actions might manifest as “poorer” food choices, if exposure to cues paired with healthy foods no longer leads to selection of that outcome.

It is worth briefly discussing the neural changes that may underpin dietary effects on PIT. There is a wealth of evidence that HFHS diets alter reward-related brain circuitry, yet direct links to PIT expression have yet to be determined. Lesion, inactivation and imaging studies in rodents and humans indicate that the central nucleus of the amygdala and the core subregion of the nucleus accumbens are required for gPIT, whereas the basolateral nuclei of the amygdala and nucleus accumbens shell subregion are necessary for sPIT (for a comprehensive review of brain regions involved in PIT, see Cartoni et al., 2016). Notably, HFHS diets have been shown to produce inflammation, insulin resistance, glutamatergic dysfunction, and dopaminergic dysfunction across these brain regions, providing candidate mechanisms for the behavioral deficits in PIT that HFHS diets can produce.

In the amygdala, few studies have compared diet-induced inflammatory changes in the central and basolateral nuclei. One such study in rats found that high-fat diet induced increases in markers of microglial function and morphology were more pronounced in the central rather than basolateral amygdala (Spencer et al., 2019). Another study also reported increased markers of inflammation in the central amygdala following high-fat diet exposure (Castro et al., 2013), but did not investigate the basolateral amygdala. Rodent studies looking at whole amygdala have produced additional evidence of dietary-induced neuroinflammation, albeit with a somewhat mixed profile of expression. Exposure to a diet high in fat/or sugar increased levels of the pro-inflammatory cytokine TNFα but not IL-6 or IL-10 in rats (Reis et al., 2023), whereas it increased TNFα and not IL-1β in mice (Akter et al., 2020). In another study, however, the IL-6 and TNFα pro-inflammatory cytokines were both elevated in rats following a high-fat diet (Noronha et al., 2019). Altogether, there is plethora of evidence that inflammation occurs throughout the amygdala following exposure to HFHS diets.

Other studies have reported neuroinflammatory and neuroplastic changes in the nucleus accumbens after HFHS diet exposure. For instance, 6 weeks of HFHS cafeteria diet treatment in mice increased microglial activation and levels of IL-1β and IFN-γ in the core and shell subregions (Gutiérrez-Martos et al., 2018). This was associated with deleterious morphological changes in neuronal spine density and plasticity (Gutiérrez-Martos et al., 2018). Similarly, another study in mice showed that 12 weeks of high-fat diet treatment induced plasticity-related changes in the nucleus accumbens, as indicated by increased BDNF and ΔFosB levels (Sharma and Fulton, 2013). Such diet-induced changes in both the amygdala and nucleus accumbens have been linked to increases in anxiolytic (Noronha et al., 2019; Akter et al., 2020; Reis et al., 2023) and depressive-like (Sharma and Fulton, 2013) behaviors. Importantly, for the current review, direct and indirect projections exist between the nucleus accumbens subregions and nuclei of the amygdala (Heimer et al., 1991; Park et al., 2019) which are critical for PIT (Corbit and Balleine, 2005; 2011; Corbit et al., 2016). Therefore, although links to aberrant PIT and other forms of decision-making have not been interrogated directly, diet-induced changes in these regions is a plausible candidate mechanism for the discussed alterations in PIT expression.

Wallace and Fordahl (2022) proposed one potential mechanism for how HFHS diets could alter cognition. Specifically, they suggested that HFHS consumption promotes neuroinflammation, leading to decreased dopamine synthesis, release, and reuptake in reward-related brain regions. Because nucleus accumbens dopamine activity normally correlates with, and is necessary and sufficient for, PIT (Lex and Hauber, 2008; Peciña and Berridge, 2013; Collins et al., 2016), it is possible that such a dysfunction in dopamine could mediate the dietary effects on PIT. There are several lines of evidence that are consistent with such an account. For instance, there is evidence that glutamate release from the basolateral amygdala targets medium spiny neurons in the nucleus accumbens shell to promote sPIT, and that blocking dopamine receptor D1 (DRD1) receptors on these neurons is sufficient to abolish sPIT (Laurent et al., 2014). It is also consistent with findings showing that high-fat diet consumption reduces DRD1 and dopamine receptor D2 (DRD2) receptor expression in the nucleus accumbens in mice (Carlin et al., 2016). Finally, although 6 weeks of high-fat diet consumption impaired insulin function and dopamine reuptake in the nucleus accumbens of mice, pharmacologically restoring insulin signaling at dopamine terminals restored dopamine reuptake (Fordahl and Jones, 2017). Interestingly, the precise role of DRD2 receptors in PIT is currently unknown but the functional consequences of decreased signaling at these receptors include compulsive food intake, hyperphagia, and compensatory overeating (Johnson and Kenny, 2010).

Overall, the literature shows that HFHS diets induce a broad range of neurobiological alterations, including insulin resistance, dopamine dysfunction, and neuroinflammation. Based on the available evidence we suggest that it is the diet-induced alterations to accumbal dopamine receptor expression (which rely on inputs from the amygdala) that could underlie diet-induced changes to PIT. A priority for future work is to produce direct causal evidence for this hypothesis, as well as identifying the specific mechanisms of other diet-induced alterations in cognition.

While the rodent evidence suggests that diet composition is a key determinant of changes to food-seeking behavior, the effects of diet on food-seeking have not been well explored in human studies. This could be achieved by incorporating measures of habitual diet and/or acute food intake manipulations in studies of PIT and related behaviors. These options are not always practical, however, and measuring energy intake may be prone to human error.

Careful consideration should be given to methods of assessing obesity in future studies. The human PIT studies discussed here all categorized participants by BMI. Yet, obesity is a heterogenous condition and there is high variability within behavioral datasets when participants are categorized exclusively by BMI. Indeed, not all people with an elevated BMI are metabolically “unhealthy” (Rask-Andersen and Johansson, 2023) and people of normal weight can develop the metabolic syndrome (Stefan et al., 2017). While BMI is a quick and practical option with established norms, it is a poor indicator of adiposity and does not consider the distribution of body fat, which is relevant for several pathologies (Fox et al., 2007; Nuttall, 2015; Rask-Andersen and Johansson, 2023). Because evidence from the rodent literature indicates that altered sensitivity to PIT is more likely a moderator that drives overconsumption and weight gain, rather than a consequence of it (Derman and Ferrario, 2018, 2020; Kosheleff et al., 2018), identifying a link between obesity and PIT calls for a finer resolution of the prognostic factors – beyond BMI. Accordingly, the role of individual differences in neurobiology and reward motivation on responsivity to food-related cues and subsequent vulnerability to weight gain was recently outlined (Ferrario, 2020). Future studies could endeavor to directly assess body fat mass, abdominal adiposity, and other metabolic variables as opposed to BMI alone (Davillas and Benzeval, 2016).

Methodological differences between PIT studies are also important to consider, including the palatability and energy density of food rewards used. For example, studies have used high versus low calorie foods (Watson et al., 2017), palatable and highly caloric pellets (Kosheleff et al., 2018; Derman and Ferrario, 2020), snacks selected by participants (Lehner et al., 2017), and equally palatable juices (Meemken and Horstmann, 2019) as rewards. These options provide good ecological validity in understanding food-seeking and choice, as in day-to-day life people might face choices between two similarly palatable options or choices where a more preferred, palatable option is available. On the other hand, the different food types used across studies might explain their incongruent results. Even though the hedonic properties of food should not be essential for intact choice during sPIT (Corbit et al., 2007; Lingawi et al., 2022; Sommer et al., 2022), they could still alter the strength of sPIT expression (Corbit et al., 2007; Watson et al., 2014).

Finally, the use of behavioral tests involving food rewards should be employed carefully when studying HFHS diets, which can alter the motivational value and palatability of the rewards used in conditioning (Harb and Almeida, 2014; Tracy et al., 2015). A recent 8-week trial of daily HFHS snack intake presented an elegant solution by assessing associative learning in a paradigm that paired neutral auditory and visual cues instead of food-paired stimuli or food rewards (Thanarajah et al., 2023). fMRI analyses taken during the neutral associative learning task revealed that daily consumption of a HFHS snack elevated neural responses in brain regions involved in cue-outcome learning. This indicates that unhealthy diets can indeed incur neural adaptations that might impact how we form sensory associations, which is of course relevant in the context of food-paired stimuli and rewards. Refining the clinical methodologies and reproducing previous findings is necessary to gain insight into the power that cues have over food-seeking, whether it depends on food type, and how this might be altered in the development of obesity.

There has been significant research into the effects of obesogenic diets on higher-order cognitive functions and memory, yet there is considerable need for greater exploration and replication of the effects of these diets on food-seeking behaviors. This is because changes to these behaviors might perpetuate a cycle of unhealthy dietary choices and health problems. As reviewed here, the evidence so far indicates that predisposition to metabolic disturbance and the consumption of HFHS diets dysregulates the ability to use food-predictive cues to moderate food-seeking actions. There is, however, no clear link between excess body weight and changes to cue-driven food-seeking. Together, these studies suggest that obesity and/or weight gain is not necessary, but HFHS diets are sufficient, for food-cue related cognitive deficits. In other words, the relationship between overweight and obesity and cognitive impairment appears correlational, with an unhealthy diet perhaps the common cause. This is difficult to parse out in human studies when BMI is used to categorize participants without more detailed measurement of dietary patterns. Therefore, future research in both rodents and humans will help to continue to clarify the causal effects of obesogenic diets on cue-driven food-seeking behaviors such as PIT and the mechanisms that underlie them.