Neural correlates of naturalistic single-trial appetitive conditioning

Background: Prior research suggests naturalistic single-trial appetitive conditioning may be a potent phenomenon in humans, capable of modulating both motivation and attention. In this study, we aimed to characterise the neural correlates of this phenomenon using functional Magnetic Resonance Imaging (fMRI) paradigms Methods: Twenty-three healthy adults (12 males) underwent conditioning during which they ate a novel 3D object made from white chocolate (CS + ) and handled a similar object made from plastic (CS-). Brain activity was recorded before and after conditioning during a passive viewing paradigm Results: A naturalistic CS + was rated as more highly craved, better-liked and elicited greater expectancies for chocolate than the CS (cid:0) after conditioning. An exploration of the interaction between time (pre-and post-conditioning) and CS type (CS + , CS-) during the passive viewing task suggested enhanced activation from pre-to post-conditioning in the right superior frontal gyrus (R.SFG) in response to the CS. Conclusion: Results reveal neural correlates of single-trial appetitive conditioning and highlight a possible role of response inhibition during learning about non-rewards, perhaps optimizing motivated behaviour. These findings contribute to our understanding of the neural mechanisms underpinning rapid reward and non-reward learning, and may inform development of behavioural interventions for reward-driven overeating.


Introduction
Appetite regulation involves a complex interplay between homeostatic, reward and cognitive processes [1].Learning-based models of eating behaviour posit that food cue reactivity is a conditioned response (CR) [2].External cues such as food images or fast food logos, or interoceptive cues such as certain emotional states (conditioned stimuli; CS) become associated with ingestion of food (unconditioned stimulus; US) via Pavlovian conditioning, subsequently eliciting CRs, such as [3] and cravings [4].Whilst these mechanisms were likely adaptive in our evolutionary past, in today's obesogenic environment, replete with palatable calorific foods, these same mechanisms may promote overeating and consequent weight gain.
Humans readily learn reward associations, and researchers have measured a variety of conditioned responses to cues predictive of reward (CS+) but not to those lacking any reward associations (CS-).Physiological measures such as heart rate and skin conductance, and explicit self-report ratings, are among the most commonly used assays of conditioned responses [5] and demonstrate how readily humans can differentiate between objects in their environment based on their reward associations.Furthermore, conditioning has been successfully demonstrated with a range of palatable food rewards, such as sucrose solution [6] chocolate milk [7] or salty pretzels [8].
A combination of animal and human studies has begun to elucidate the neural networks involved in learning, memory, reward and motivation which are associated with food-cue reactivity.A recent review highlights a key appetitive neural network involved in food cue reactivity in humans, which includes the insula, striatum, hippocampus, anterior cingulate cortex (ACC), amygdala, orbitofrontal cortex (OFC), and prefrontal cortex (PFC) [9].Activation of these brain areas is closely associated with desire and food pleasantness [10].Neural networks involved in processing food cues are sensitive to motivational factors and hedonic value; hunger modulates neural activity in response to food cues in the insula, fusiform gyrus and OFC, whereas activity in the hypothalamus and striatum is modulated by energy content [11].In addition, functional connectivity between the insula and hypothalamus, and sub regions of the frontal cortex have been shown to be related to cognitive control of eating in response to alterations in homeostatic energy balance [12].
Previous brain imaging studies have highlighted several brain regions to mediate effects of reward expectation.The OFC is thought to encode the value associated with a rewarding stimulus rather than representing action [13,14].The ACC receives input from the OFC regarding anticipated reward outcomes, and plays a key role in connecting outcome to action [15].The striatum, and particularly the nucleus accumbens (NAcc), is thought to play a key role in cue-elicited approach-motivated behaviours in appetitive conditioning [16,17].Finally, the insula is involved in multisensory integration during processing of food cues and has several roles spanning across gustatory processing, sensory perception, interoception, affective and cognitive processing [18,19].In particular, the anterior insula has been identified as the primary taste cortex [20].Hyper-responsivity in this network of brain regions has been implicated in weight gain and overeating [9,21] and thus research in this area will be important for developing greater understanding of processes related to cue-elicited eating, as well as informing evidence-based interventions for targeting over or undereating.

Single-trial conditioning
A recent study highlighted the possibility that conditioned responses could be learned through a single hedonic interaction with a food [22].A naturalistic appetitive conditioning procedure was successful in eliciting conditioned cravings for a novel edible object (CS+) but not for a visually similar, but inedible plastic object CS-.The single trial conditioning procedure modulated both early and late event related potentials (ERPs).The authors have argued that late neural responses were reflective of the cognitive motivational processing of these cues, whereas modulation of early neural responses represented heightened attentional processing; vigilance for a newly learned food cue was enhanced after just one learning episode, suggesting that visual attention may be driven by reward processes.
Lender et al. [23] successfully replicated this object learning in an fMRI paradigm.After single trial conditioning, the authors found stronger activation in the dorsal striatum in response to viewing a CS+, relative to the CS-, implicating this area in the early identification or food rewards and, perhaps, representing the strength of the coupling between the CS and US.Moreover, greater evidence of conditioning was evident in the right amygdala, which responds to the biological relevance of a stimulus, for individuals who scored high on subjective cravings.

Study aims and objectives
In an extension of the findings presented by Lender et al. [23] the aim of the present experiment was to investigate neural correlates of single-trial naturalistic appetitive conditioning in humans during a passive viewing task.Importantly, we build on Lender's [23] work by using a pre-post design in which a passive viewing task is completed before and after conditioning.We also aimed to measure subjective conditioned responses acquired through a naturalistic single-trial conditioning procedure, at a behavioural level, outside the scanner.

Methods
The present experiment utilized a within-subjects design to explore how neural and behavioural responses to conditioned stimuli are modulated by a naturalistic single-trial appetitive conditioning procedure.

Participants
Twenty-four participants completed the full experimental procedure.One female was later excluded from all analysis due to detection of a significant brain abnormality.The twenty-three participants (11 females) entered in to the final analysis were, on average, aged 26.78 (SD = 4.68) and had a BMI of 24.68 (SD = 3.24).Participants were required to be aged 18-40, fluent English speakers with normal or corrected-tonormal vision, and have a BMI between 18.5 and 30.0 (normal and overweight).Anyone with a history of neurological disease, eating disorders or diabetes were excluded from participation, as well as anyone declaring a food allergy or intolerance.
A food liking questionnaire was used to select participants who expressed a liking for the US.To hide the importance of the specific US to the experimental aims, participants were required to rate their preferences for nine other filler items as well as white chocolate.Ratings were made on a 9-point hedonic scale with anchors ranging from 'dislike extremely' to 'like extremely'.Participants were required to provide a score of six or above for white chocolate in order to be invited to take part in the full experiment.Data on participants chocolate consumption habits was not collected.
Participants' provided self-reports of their current motivation to eat using four well-validated 100 mm visual analogue scales (VAS): hunger, desire to eat, fullness and prospective consumption.Measures of hunger were repeated across the experiment.Participants arrived at the lab in a state of moderate hunger (see 2.3, below), and self-reported appetite remained unchanged from pre-(M = 63.26,SD = 14.39) to postconditioning (M = 67.24,SD =11.85), t(22) = -1.44,p =.164, suggesting that participants' motivation to eat was consistently high across the experiment.Average taste ratings for the CS+, taken during conditioning, were generally high (M = 70.96,SD = 20.80).
All participants gave full informed consent prior to the start of the experiment, in accordance with the Declaration of Helsinki.The experiment was given ethical approval by the University of Liverpool Research Ethics Committee.

Procedure and conditioning paradigm
Testing took place between 10:30 and 16:30 in the Liverpool Magnetic Resonance Imaging Centre (LiMRIC at the University of Liverpool. Participants were asked to refrain from eating or drinking (except water) for at least three hours immediately preceding their arrival to create a hunger state that most individuals experience as they approach their next meal.
After eligibility and safety screening, participants completed the three phase experimental procedure (pre-conditioning, conditioning, and post-conditioning; see Fig. 2).During the pre-conditioning phase, participants were shown photographs of the CS+ and CS-to capture baseline pre-conditioning levels of craving, expectancies and pleasantness.The CS stimuli were adapted from those described by Blechert et al. [22].The CS+ and CS-were an edible object made from chocolate, and a plastic, inedible object, respectively.Objects were 3D novel geometric shapes, produced in either yellow or orange, designed to be neutral in valence, so as to have no prior conditioned association for the participants.Colour-object assignment was counterbalanced across participants (see Table 1).
Participants were asked to complete three explicit measures of
conditioning, using self-report 100 mm VAS, adapted from [24].Measures were taken once in the pre-conditioning phase, and once post-conditioning.To capture conditioned cravings, participants were asked, "When presented with this object, how strong is your craving for chocolate right now?" and responded on a VAS scale with anchors "No craving at all" to "Extremely strong craving".Similarly, US-expectancies were assessed with the question, "how strongly do you now expect to be invited to eat chocolate?",and "how pleasant do you find this object?"was asked to explore liking.The order of question type and stimulus type were randomised.Following self-report measures participants completed a preconditioning passive-viewing task in the scanner.
Images of a CS+ and CS-were presented in an event-related design.Four images of the CS+ and four images of the CS-objects were created in Sketch Up (Trimble Inc, CA, USA).Images were 3D, visual representations of the CS+ and CS-objects on a plain grey background.Each image depicted the object from a different orientation to mimic normal viewing (i.e., front view, side view and 45 • left and right).Each image was presented four times, with thirty-two image presentations in total (16 CS+, 16 CS-).Images were presented in a pseudo-random order with each CS type presented consecutively no more than twice.
Each trial began with a 2 s fixation cross, then an image of either the CS+ or CS-was presented for 6 s.Following each image, there was a staggered inter trial interval (ITI) of 7-11 s.The sequence of typical trials is depicted in Fig. 1.Participants were instructed simply to look at the images on the screen for the duration of the task.Participants were then brought out of the scanner to complete the naturalistic conditioning task.
The CS+ and CS-were presented to participants during a conditioning procedure completed outside of the scanner in between the preconditioning and post-conditioning scanning sessions.The CS+ was presented on a small white plate.The CS-was presented in a visually similar small white plastic box, to signal differences in edibility.Participants were instructed to look at, handle and smell the first CS for two minutes (and to eat the CS+), whilst paying particular attention to the sensory characteristics of the objects.They then rated the sensory characteristics of the object on unipolar 100 mm VAS scales, from 'not at all' to 'extremely'.Participants rated the appearance, smell and texture of both objects, as well as the taste of the CS+.The primary purpose of these ratings was to acquire taste ratings for the CS+.After a 90-second interval, they were then presented with the other CS, and the same procedure was followed.
This was followed by the post-conditioning phase of the experiment which involved participants repeating the self-report measures (craving, expectancies, pleasantness) and then returning to the scanner the repeat the passive viewing task.Following the experiment, awareness of the conditioning contingency was measured outside of the scanner by asking participants to recall the colour, shape and edibility of each object.Participants were also asked to write an answer to the question, 'what do you think the purpose of this experiment was?' All participants demonstrated awareness of the CS-US contingencies and were unaware of the experimental aims.Participants received a full debrief of the experimental aims and were reimbursed for their time with high street vouchers worth £30.

fMRI 2.3.1. Image acquisition
Magnetic resonance imaging scans were undertaken using a wholebody Tesla Siemens Trio 3T MRI imaging system (Siemens, Magnetom, Erlangen, Germany) and an 8-channel radiofrequency head coil.
At the start of the pre-conditioning phase scanning session, a highresolution, 3-dimensional, T1-weighted image was acquired using a Magnetization-Prepared Rapid Acquisition Gradient-Echo (MP-RAGE)  This scan was followed by a T2 weighted sequence (48 interleaved axial slices, with no gap, TR = 3000.0ms, TE = 30.0ms, flip angle = 90 degrees, field of view = 192 mm, voxel size = 3.0 × 3.0 × 2.7 mm) whilst task images were presented via Psychopy.The post-conditioning phase fMRI scans had the same parameters as the pre-conditioning fMRI scan and were positioned with a localiser scan.

Data acquisition, reduction and analysis
2.3.2.1.fMRI pre-processing.Spatial pre-processing of functional data was conducted using Statistical Parametric Mapping software package, SPM12 (UCL, UK: www.fil.ion.ucl.ac.uk/spm) running on Matlab version R2018a (MathWorks Inc., Natick, MA).Functional Images were slice-timing corrected, realigned, co-registered to the MP-RAGE structural image (which was then segmented), normalized to EPI (Echo Planar Imaging) template image, and smoothed with an 8 mm Gaussian Kernel with full width half maximum, to improve signal-to-noise ratio.Motion parameters from the realignment pre-processing step were evaluated and a motion artefact threshold (translation > 3 mm, rotation > 1 degree) was employed for exclusion.No participants were excluded on the basis of excessive movement.A temporal high-pass filter was applied to the time series with a 128 second cut-off to remove lowfrequency noise and slow drifts in signal.

Whole-brain analysis of the passive viewing task.
In a standard event-related design, functional scans were entered into a first level design to define effects of condition.Duration parameters were included as regressors.First level (i.e., at the individual level) contrasts were then computed to generate statistical parametric maps of the contrasts of interest (CS+ > CS-, Pre > Post, interaction).As the interaction between CS type and time-point allows for the most meaningful interpretation of the data, analysis is focused on this contrast and further explored, in line with our hypotheses.
At the second level, multiple comparisons were made across the contrast images of the interaction obtained for each participant, which were entered into a univariate t-test.Given the exploratory nature of the research question, a minimum cluster size of 10 voxels (k = 10) was employed in the second level contrasts.A liberal uncorrected statistical threshold for the spatial extent test on the clusters was set at p < 0.05 at the cluster level over the whole brain, with a height threshold of p < 0.001.
Peak voxels at significant clusters were selected as regions of interest (ROIs), and defined as 5 mm diameter spheres using MNI co-ordinates in the MarsBar 0.44 toolbox for SPM12 (http://marsbar.sourceforge.net/;Brett et al., 2002).Contrast estimates for each condition were then extracted.These data were then tested further using a 2 × 2 repeated measures ANOVAs in SPSS, with a confirmatory threshold of P < 0.05.Significant interactions were explored with post hoc paired t-tests.
The coordinates derived from these analyses were converted from MNI coordinates to Talairach coordinates using the Yale BioImage Suite application [25], and labelled using Talairach Client [26] in order to associate findings with an anatomical location [27].Labels were assigned according to the nearest grey matter position.Prior to conditioning, both the CS+ and CS-elicited similar levels of craving, (p = .203).After conditioning, the CS+ elicited significantly greater cravings than the CS-(p < .001),suggesting successful appetitive conditioning.Self-reported cravings in response to the CS+ rose significantly from pre-to post-conditioning (p < .001).Although, there was a slight increase in cravings for the CS-from pre-to post-conditioning as well (p < .035).

US-expectancies
Similarly, there was a significant main effect of time, F( 1 (see Fig. 3B).Expectancies for chocolate were similar for the CS+ and CS-before conditioning (p = .388).After conditioning, the CS+ elicited significantly greater US-expectancies than the CS-, again suggestive of successful differential appetitive conditioning, (p < .001).Expectancies for chocolate in response to the CS+ rose significantly from pre-to postconditioning, (p = .001).However, for the CS-, expectancies for chocolate were consistently low both before and after conditioning, (p = .094).210(see Fig. 3C).The pattern of data was indicative of successful differential appetitive conditioning.Both the CS+ and CS-were equally liked prior to conditioning, (p = .148),but after conditioning the CS+ was liked significantly more than the CS-(p < .001).Liking for the CS+ rose from pre-conditioning to post-conditioning, (p < .001),whereas ratings for the CS-did not change, (p = .927)

Whole brain analysis of the passive viewing task
Whole brain analysis at an uncorrected threshold of P<.001, with a height threshold of 3.50 and an extent threshold of >10 voxels revealed the presence of an interaction effect between time and the type of conditioned stimulus in single significant cluster of 18 voxels in the right superior frontal gyrus (R.SFG) (MNI coordinates 32, 64, 2 mm; T = 4.35, Z = 3.65, P = 0.039) (see Fig. 4).Supplementary ROI analyses can be observed in supplementary materials.
A 2 × 2 repeated measures ANOVA of the contrast estimates in the cluster highlighted in the whole-brain analysis revealed a statistically significant interaction between CS type and time point, F(1, 22) = 7.73, p = .011,η p 2 = .260(there were no main effects of stimulus type, or time point)).Paired t-tests revealed no change in contrast estimates from preto post-conditioning for the CS+, t(22) = .641,p = .528.However, for the CS-, activity in the R.SFG significantly increased from pre-to postconditioning, t(22) = 2.34, p = .029,(see Fig. 5).

Discussion
The aim of this study was to investigate the neural correlates of single-trial appetitive conditioning in healthy subjects using a novel, naturalistic conditioning paradigm, designed to mimic a realistic encounter with a new food.After participants ate a CS+ object (made from chocolate), subjective responses revealed clear evidence for differential appetitive conditioning in a single trial.Participants were more likely to expect to receive chocolate upon subsequent presentations of the CS+, indicating an awareness of the CS_US contingency.The CS+ was also perceived as more pleasant and elicited cravings for chocolate post-conditioning.Similar changes were not observed for a CS-(made from plastic).
This work builds on the original paper by Lender et al. [23].Whilst the previous paper only allowed comparisons between the CS+ and CSat a single time point, our pre-post paradigm allowed us to explore the change in neural activity over time for each of the CS objects, and control for any possible differences in processing between the two similar, yet visually distinct objects, prior to any learned reward associations.

Neural correlates of single-trial conditioning
Of particular interest, our fMRI analyses revealed a small yet significant increase in neural activity in the right superior frontal gyrus (R. SFG) in response to the sight of the CS-, although no effects were found for the CS+.
The ability to select appropriate behavioural responses and to suppress action when necessary is an essential skill which can prevent distraction and unnecessary action.Research has begun to uncover distinct neural networks involved in such response inhibition, to which the frontal lobes make a major contribution [28][29][30].In particular, the R.SFG has been implicated in mediating impulsivity and behavioural inhibition [31], and promoting inhibitory control [32].([33]a) confirmed that the R.SFG is part of a distinct neural network involved in action restraint.Disruption of activity in the R.SFG impaired performance on a go/no-go task, which relies on participants inhibiting a response before it is made, whilst the ability to cancel an action already underway was unaffected ([34]b).
The increase in activity at the R.SFG during post-conditioning presentations of CS-images, but not CS+, might suggest that after differential learning, the brain can discriminate between a motivationally relevant CS+, paired with food, and an irrelevant CS-which lacks any reward associations and consequently inhibits responding to this  unimportant stimulus.As a result, further cognitive processing may be inhibited and approach motivated behaviour prevented.Once a neutral stimulus has been found to lack reward associations it would be disadvantageous to continue exploring it further, so post-conditioning activation by CS-may be indicative of an adaptive mechanism ensuring efficient direction of subsequent motivated behaviour toward appropriate stimuli.In the supplementary materials that accompany this manuscript, we also conducted a series of brain-behaviour correlations between change in BOLD signal at peak voxels in the R.SFG in response to a CS+ and CS-pre-and post-conditioning during a passive viewing paradigm and self-report outcome measures (pre-to post-conditioning difference scores (cravings, expectancies and liking) for a CS+ and CS-(see supplementary Table 2).Here a positive correlation between pre-to post-conditioning change in R.SFG activity and change in expectancies for a CS+ was observed (r = 0.45, p = .032).Whilst this significant effect would not withstand partial correction for multiple comparisons (12 correlations were ran in total), it does provide tentative evidence in support of the functional significance of the R.SFG for differential learning.
These novel findings are of particular importance given that modulation of R.SFG activity occurred after just a single trial with naturalistic stimuli.It is important to note that effects were only evident at an uncorrected threshold and thus, results must be interpreted with caution.As neuroimaging is generally more sensitive to changes in processing than behavioural measures, it may be that we were able to detect very subtle changes associated with the very first step of appetitive conditioning.With repeated conditioning trials, these patterns may be more evident.
This pattern of findings is in contrast to the earlier paper by Lender et al. [23].Whereas we found increased activation in the right SFG when viewing a CS-, Lender et al. [23] found greater activation in the left SFG when viewing a CS+ compared to a CS-.The L.SFG is involved in spatial processing and working memory as opposed to inhibition [35].Given that the present study employed a simple design, with just two objects (one of each CS type), while Lender et al. [23] utilised six objects in total (three of each CS type), it is possible that this difference is indicative of greater demand on working memory for differentiation of CS+ and CSwhere there is additional stimulus complexity.
In addition, we did not replicate the increased activation in the amygdala and caudate reported by Lender et al. [23].Prior research has highlighted modulation of the amygdala by incentive salience of the CS+ [36].Thus one possible explanation for the disparate findings may be that individual differences such as food preferences and homeostatic status influenced findings.Whilst these factors were considered in the present study and captured via self-report measures, it is plausible that differences between participant groups in levels of hunger, craving or desire for a CS+ may have an effect on patterns of neural activation.Due to the complexity of human appetite and the high degree of individual variability, Gibbons et al. [37] suggest researchers utilise a robust set of valid and reliable tools to more fully capture motivation to eat and need state.Further research will be beneficial to better understand how homeostatic and hedonic factors could impact neural activity within and between individuals.

Limitations and future directions
The relatively poor reliability and reproducibility across fMRI experiments is garnering increased attention [10,38].Single-trial appetitive conditioning research is still in its infancy, and so it was considered appropriate to use analyses of an exploratory nature, in line with prior research in this area [23].Neural activity effects suggest that fMRI is more sensitive to detecting rapid reward/non-reward learning than behavioural measures alone.Given the transient nature (single trial conditioning) of the conditioning procedure on the novel stimuli, other thresholding procedures were considered as potentially too conservative to detect subtle effects.However, it is noteworthy that the whole-brain analysis results presented here are from liberal (p < .001uncorrected) statistical thresholds.Coupled with the relatively small sample size, the current experiment was perhaps underpowered to detect small effects.Increased trial number may have also improved detection of subtle effects.The relatively low reproducibility across individual brain based studies may also be overcome in future research through the application of novel methods, such as activation network mapping which has been shown to improve reliability in similar research areas [39].
Limitations of the task design are also important to note.To avoid lengthening the experimental procedure and introducing potential confounds, we limited visual stimuli to the CS+ and CS-, presented from different angles.Therefore, the variety and visual complexity of the presented stimuli were reduced considerably, compared to more conventional tasks using more varied food and non-food images.According to the meaning-and-attention-components (MAC) model, boredom will arise when attentional demands are too high or low, and/or when a task lacks meaning for the individual [40].It is likely that the passive viewing element lacked meaning for participants, and also that attentional demands were very low.It is therefore, highly likely that boredom was an unintended confound in the present study.Inclusion of more engaging cognitive tasks can reduce boredom, as well as associated issues such as head motion, thus improving data quality [41].Design changes such as, the inclusion of more heterogenous stimuli or requiring behavioural responses via a button box or similar could significantly increase task engagement.
Similarly, Franssen et al. [42] highlight the importance of considering participants' mental process while completing passive viewing tasks.The authors demonstrate that attentional focus during passive viewing of palatable food images has a significant effect on neural activity and consequently is a limitation of such designs.Whilst we expected participants would focus on the reward value of the stimuli post-conditioning, we cannot confirm participants' attentional focus.Consequently, future research should consider task selection carefully, considering task demands and engagement, as well as attentional focus to allow clearer interpretation of findings.
Despite limitations, the flexibility of the present paradigm allows for many different shapes, colours, tastes and perhaps smells or textures to be introduced, allowing for designs that capture the acquired motivational value of various stimuli differing in valence [43].The fact that the CS and US form a compound also reduces the complexity of the conditioning procedure, removing the need for complex, precisely timed laboratory experiments which carefully control CS and US presentation.As a result, this paradigm could be easily adapted for many experimental designs, both in the laboratory, as well as in real-world settings such as the home, enhancing ecological validity of research in this area [3].
Although, as discussed by Lender et al. [23], the single-trial naturalistic appetitive conditioning paradigm only informs about the immediate responses during the first step of conditioning.In real-life, it is unlikely that such brief encounters would occur in isolation; palatable foods are typically sought out and consumed repeatedly.As a result, exploring neural and behavioural conditioned responses beyond this initial encounter, and after multiple CS-US pairings may provide further insights and perhaps highlight involvement of more brain regions.Similarly, recording activity during acquisition, while evaluative judgements are being made, and throughout extinction might provide additional, critical information regarding the neural basis of appetitive conditioning, as well as inform the development of interventions such as cue exposure therapy (CET) for reducing cue-elicited cravings [44].
Furthermore, the present study highlights the functional significance of the R.SFG for naturalistic appetitive conditioning, which may represent inhibitory control processes.Given the importance of inhibitory control for making deliberate food choices [45], the potential relationship between differential appetitive conditioning and inhibitory control warrants further attention.While some research suggests stronger trait impulsivity is associated with stronger conditioned responses and food cue-reactivity [46], findings are inconsistent [47].More research is needed to better understand how inhibitory control processes may be related to the acquisition and expression of appetitively conditioned responses, and how these may differ within and between individuals depending on state and trait levels of cognitive control.

Conclusion
The present experiment extended the investigation of the neural mechanisms underlying single-trial appetitive conditioning.We report a novel finding suggesting a possible role for the right superior frontal gyrus in differential appetitive conditioning that may be specific to learning about a CS-with no reward associations.The pattern of activity we observed indicates particular involvement of this structure in inhibiting action to irrelevant stimuli, which may help to optimize motivated behaviour toward specific goal objects.Future research incorporating methodological advances may be useful to better understand contrasting findings between studies and develop a more thorough understanding of the neural correlates underpinning single trial appetitive conditioning.

Fig. 1 .
Fig. 1.Schematic representation of two consecutive trials on the passive viewing task.

Fig. 2 .
Fig. 2. Flowchart outlining the three phases of the experimental procedure.

Fig. 4 .
Fig. 4. A contrast showing the interaction effect between stimulus type (CS+/CS-) and session (pre-and post-conditioning) in the superior frontal gyrus.'L' and 'R' represent left and right hemisphere respectively.