Affective Imagery Boosts Reward Related Delta Power in Hazardous Drinkers

Abstract

The Reward Positivity (RewP) is an event-related potential component with a delta band spectral representation that is elicited by reward receipt. Evidence suggests that RewP is modulated by both reward probability as well as affective valuation (“liking”). Here we determined whether RewP is a marker of enhanced hedonic salience of alcohol images in hazardous drinkers. We recruited 54 participants (Hazardous Drinkers=28, Control=26) who completed a reinforcement learning task with affective versus alcohol imagery during feedback. The learning task used images of puppies vs. alcohol paired with reinforcing feedback. Both groups rated categories of affective images (puppies, scenery, babies, neutral) similarly, but the hazardous drinking group rated alcohol significantly higher. There were no group differences in performance or in RewP amplitudes, even as a function of alcohol imagery. Contrary to prior findings, we did not observe a significant correlation between alcohol image rating and alcohol-specific RewP amplitude, although we did observe this relationship with the alcohol-specific delta band spectral representation of RewP. Within hazardous drinking group, there was significant correlation between hazardous drinking (AUDIT score) and alcohol-specific RewP indicating an inter-individual influence of drinking habits on affect specific RewP. These findings suggest a domain-specific enhancement of reward responsiveness in hazardous drinkers.

1. Introduction

Alcohol use disorder (AUD) is one of the most widespread expressions of substance misuse. It is characterized by impaired control over alcohol consumption and can have severe physical, psychological, and social health impacts. Prior to the onset of AUD, individuals are often characterized by increasingly frequent and/or hazardous consumption of alcohol (Hallgren et al., 2021). This tendency, known as hazardous drinking, refers to a pattern of alcohol use that places an individual at risk for adverse health events (Saunders et al., 1993). Since hazardous drinking is a major public health concern, a vital objective of the scientific study of alcohol research is to assess the biological factors related to aberrant salience, diminished control, and altered reward processing in the transition from problematic alcohol use to alcohol dependence (Gilpin & Koob, 2008).

Hazardous drinkers reliably demonstrate an attentional bias to alcohol-based imagery. This attentional bias is considered to be an indicator of the motivational salience of alcoholrelated cues (Roberts & Fillmore, 2015; Weafer & Fillmore, 2013).When people who engage in hazardous drinking are exposed to images, smells, or other sensory-specific features of alcohol, they may demonstrate enhanced physiological arousal and craving for alcohol. Attentional bias to alcohol-specific cues has been shown to be a pivotal factor in dysregulated drinking (Field et al., 2010; Field & Cox, 2008; Franken, 2003; Robbins & Ehrman, 2016; Roberts & Fillmore, 2015; Wiers et al., 2010).

The Incentive Sensitization Theory of Addiction (ISST) is a neurobiological theory that addresses how drug-associated cues (both incentive and hedonic) affect an individual’s behavior (Berridge et al., 2009; Berridge & Robinson, 2016.; Robinson & Berridge, 1993, 2001). Amongst individuals characterized by hazardous drinking, cues associated with alcohol can attract attention (Gladwin, 2016; Martins et al., 2021; Shin et al., 2010), stimulate appetitive approach behaviors (Fleming & Bartholow, 2014; Wiers et al., 2009), prompt amplified neurophysiological responses (Littel et al., 2012; Schacht et al., 2013) and elicit craving (Field et al., 2009; Ramirez et al., 2015). Hence, addiction may be modulated by an enhanced discrepancy between the motivational salience of drug and the hedonic effects of drugs (Robinson & Berridge, 2001). Although aspects of incentive motivation (“wanting”) of alcohol cues has been extensively studied (Amlung & Mackillop, 2014; Barker & Taylor, 2019; Claus et al., 2011; Filbey et al., 2008; Cofresí et al., 2019; Fleming et al., 2020; Hicks et al., 2012; Martins et al., 2019; Valyear et al., 2017 ), the effects of hedonic salience (“liking”) on psychological processes that are implicated in the development and maintenance of addiction are less well understood.

Only recently there have been a few studies examining the hedonic or “likable” properties of alcohol (Groefsema et al., 2019; Oberlin et al., 2013, 2016). Research studies have shown that alcohol-based cues stimulates activity in brain regions associated with reward (Groefsema et al., 2019; Oberlin et al., 2013, 2016). There is mixed literature on the responsivity to alcohol-based cues in reward learning in hazardous drinkers: some studies show hypersensitivity (Dager et al., 2012; Flagel et al., 2011; Grüsser et al., 2004; Heinz et al., 2004; Oberlin et al., 2013; Vollstädt-Klein et al., 2010) whereas others have argued that hazardous drinking can be defined in terms of a dampened response to reward (Gilpin & Koob, 2008, Blum et al., 2000; Bowirrat & Oscar-Berman, 2005; Schmidt et al., 2001; Volkow et al., 2010). Accordingly, the neural representations of reward and reinforcement are important to understand normal brain-behavior interactions and the pathophysiology of hedonic disorders (Gilpin & Koob, 2008).

One novel measure of neural reward responsivity is the reward positivity (RewP), a positive EEG deflection that appears at approximately 250 milliseconds following reward receipt (Heydari & Holroyd, 2016; Holroyd, Pakzad-Vaezi, et al., 2008; Weinberg et al., 2014). The RewP is sensitive to the degree of reward surprise known as the positive prediction error: more surprising rewards are associated with larger RewP amplitudes (Heydari & Holroyd, 2016; Cockburn & Holroyd, 2018). The RewP has a delta band (1–4 Hz) spectral representation, which differentiates it from the theta band (4–8 Hz) response most often associated with losses and the exertion of control (Cavanagh & Frank, 2014). The RewP has most often been evoked with points (Bellebaum et al., 2010; Wu & Zhou, 2009), pictures such as fruits or symbols such as upward facing arrows (Angus et al., 2015; Holroyd, Pakzad-Vaezi, et al., 2008; Proudfit, 2012), money (Bellebaum & Daum, 2008) or other benign indicators of reward receipt. Yet the RewP is also responsive to affective features of reward, for example, to pleasing pictures (Brown & Cavanagh, 2018; Sabatinelli et al., 2007). We recently developed a task designed to specifically probe how the affective salience of intrinsically rewarding images boosts the RewP (Brown, Jackson & Cavanagh, 2021).

Prior studies have demonstrated that the RewP and associated delta band dominant EEG features are sensitive to reward alterations within various clinically-relevant groups, including depression (Kumar et al., 2008; Proudfit, 2012; Proudfit et al., 2015), schizophrenia (Clayson et al., 2019), and substance misuse (Joyner et al., 2019). Surprisingly, no studies to date have examined how alcohol cue salience affects the RewP or reward-based delta power in a sample of hazardous drinkers.

The objective of the present study was to examine the sensitivity of the RewP and reward-based delta power to salient images (alcohol vs. affective) in individuals characterized by hazardous drinking. Our first hypothesis was that the RewP/delta power would be larger in hazardous drinkers when reinforcement was paired with alcohol-related cues vs affectively salient cues. Our second hypothesis was that individual ratings of image preference would relate to alcohol-specific enhancement of these reward-associated EEG features (c.f. Brown et al., 2021), as would individual ratings of symptom severity in the hazardous drinkers group.

2. Materials and Methods

2.1. Participants

This study was approved by the Institutional Review Board of the University of New Mexico. The data was collected from the Albuquerque community. We recruited individuals from the hazardous drinking group from an ongoing study “ABQDRINQ”. The following inclusion criteria were used for the hazardous drinkers group: (1) age 22–55 years; (2) self-identify as a “moderate to hazardous/binge/weekly drinker”; (3) engage in “hazardous and harmful alcohol use” based on an Alcohol Use Disorder Identification Test (AUDIT) score (Verhoog et.al., 2020) of greater than 8 for males or greater than 7 for females (Higgins-Biddle & Babor, 2018; Smith et al., 2009). The following exclusion criteria was used for all participants at the baseline appointment at ABQDRINQ study: (1) currently seeking alcohol treatment or any form of mutual help for drinking (e.g. Alcoholics Anonymous meetings); (2) history of brain injury or neurological diagnoses; (3) meets criteria for lifetime schizophrenia or bipolar disorder; (4) current substance misuse or dependence other than nicotine or marijuana (i.e. only substance misuse/dependence that occurred within the year prior to the baseline appointment is exclusionary); (5) evidence of recent illicit drug (other than marijuana) use on a urine screen; (6) history of severe alcohol withdrawal (e.g. seizures, tremors, DTs). Following their completion of the ABQDRINQ study, participants from this sample were invited to participate in this EEG experiment. Participants re-completed the AUDIT again upon arrival. Variation in AUDIT scores were expected; notably all participants in the hazardous drinking group remained reliable consumers of alcohol with AUDIT scores higher than all non-drinking controls.

For recruiting non-drinking controls, we required that the participants not satisfy any of the exclusion criterion and they had to consistenly score between the range of 0–3 on the AUDIT. All control participants completed the AUDIT over the phone during phone screening, as well as again in the lab immediately before EEG. The control participants were matched for age ranges and sex distribution with the hazardous drinkers group. All participants were compensated with $30/hour for their time and the study lasted for an average of 2.5 hours.

2.2. Questionnaires, Neuropsychological Assessments and Method

The participants completed informed consent, and questionnaires for demographic assessment, modified Edinburgh Handedness Inventory, Behavioral Inhibition System / Behavioral Approach System (BIS/BAS), Beck Depression Inventory (BDI) and AUDIT. BIS/BAS and BDI assessments were completed to examine potential personality and mood differences between samples. Data from two of the participants in the hazardous drinkers group were excluded from the analysis due to technical difficulties with EEG, and data of two control participants were removed because they self-reported an AUDIT score>3 during re-assessment in the lab. Thus, for our final analysis, we had n = 28 (16 female) in the hazardous drinkers group and n = 26 (16 female) controls. The hazardous drinkers group and the control group were well-matched and showed no differences in terms of age, sex, years of education, or personality measures of BIS or BAS. However, significant differences were observed between the two groups in self-reported scores of AUDIT and BDI. Table 1 shows the participant demographics and questionnaire scores.

Table 1.

Participant Demographics. Entries are means and SD, except for sample size counts. Asterisks indicate significant (p<.05) group difference.

	Hazardous Drinkers	Control
N (# female)	28 (16)	26 (16)
Age	38.60 (9.43)	37.30 (10.8)
Years of Education	15.00 (2.30)	16.00 (3.01)
AUDIT	10.50 (5.57)*	1.27 (0.96)*
BDI	11.00 (12.10)*	4.58 (7.83)*
BIS Total	14.60 (2.62)	14.70 (2.59)
BAS Total	22.60 (5.88)	24.30 (5.04)
BAS Drive	8.43 (2.46)	7.81 (2.15)
BAS Fun Seeking	6.89 (1.85)	8.23 (2.60)
BAS Reward Responsiveness	7.25 (1.90)	8.27(2.24)

2.3. Tasks

2.3.1. Image Rating Task

We first administered an image rating task to obtain ratings of liking of varied categories of affective imagery (same as in: Brown et al., 2021). We presented the participants with an image from one of the five categories: puppies, sceneries of nature, babies, alcohol, neutral (e.g., a chair), and negative images (e.g., a cigarette butt). We chose these categories of affective images because they were ranked as very positive, neutral, or very negative from the International Affective Picture System (IAPS; Lang et al., 2008). However, for this task, the specific images were chosen from google browsing (e.g., “hd images of puppies”). Six images from each category were presented (a total of 36 images). This task was previously reported in Brown et. al. (2021); here we simply changed the category of cows to alcohol images. Participants ranked the images on the scale of pleasantness ranging from 1 (very unpleasant) to 9 (very pleasant). Irrespective of what they rated as most pleasant, we only used the images of puppies or alcohol in the following reinforcement learning task. Prior comprehensive experimental testing has shown that the image category of puppies is reliably rated most pleasantly (Brown et al., 2021). It took an average of 3.4 minutes to complete this image rating task.

2.3.2. Reinforcement Learning Task

For the learning task, we adapted a close variant of Affective State Reinforcement Learning Task (Brown et al., 2021). In that experiment, we used affective images of puppies and cows (highly liked vs. ambivalently liked affective imagery). To provide alcohol-relevant cues, we replaced images of cows with images of alcohol while retaining the images of puppies as it was in the original task. Figure 1 shows an illustrative representation of trial and timing of the task.

Figure 1. Affective State Reinforcement Learning Task.

During the task, participants were presented with a cue that had different reinforcement rates for a correct button push (Easy: 90% vs. Hard: 70% reinforcement). Prior to feedback, an affective image was presented (puppy vs. alcohol), followed by feedback (reward: green screen vs. punishment: red screen). The RewP was evoked by the green color feedback, but modulated by the affective image (as indicated by the arrow).

In this reinforcement learning task, we first presented the participants with a cue (colored shape). They were given instructions to learn whether a left or right button press was the correct response to the shape based on the subsequent probabilistic feedback. After their button press, an image of a puppy or of alcohol appeared on the screen for 750 to 1000ms, followed by the feedback (reward: green screen, punishment: red screen) for 1000 to 1250ms. All feedback-locked EEG epochs were time locked to this screen color change. This temporal dissociation of the novel image and the feedback effectively controls for a novelty N2 evoked by the image (Brown & Cavanagh, 2020). The stimulus-outcome mapping of left or right for easy or hard alcohol or affective imagery conditions were randomized.

Participants were told that the image presented on the screen did not affect their feedback and that they should only focus on the shape selected and the subsequent color feedback. They were also informed that the feedback was probabilistic. There were four types of cues, split between easy vs. hard conditions (90% probability vs. 70% probability) and image class (puppies vs. alcohol). We label these dimensions DIFFICULTY (90% vs 70% veridical feedback) AFFECTIVE IMAGE TYPE (puppy vs. alcohol) in statistical analyses. There were 40 trials of each condition (easy alcohol, easy puppy, hard alcohol, hard puppy) for a total of 160 trials. All images were of the same size and none of them were repeated. This design effectively orthogonalizes reward prediction error and affective influence on the RewP. The average time taken to complete this task was 19.3 minutes.

2.4. EEG Recording and Preprocessing

Electrophysiological data was recorded using a Brain Vision System (Brain Products GmbH, Munich, Germany) with low and high cut off range of .01–100Hz. An EEG cap embedded with 64 Ag-AgCl electrodes with a sampling rate of 500Hz was placed on the participant’s head. The ground was at FPz and CPz was the online reference electrode. We reassigned four outlying electrodes from the standard cap for alternative purposes. Two electrodes (FT9 and FT10) were placed on the superior and inferior region of the left eye to record the vertical electrooculogram (VEOG). Two other electrodes (TP9 and TP10) were directly placed on the mastoid bones.

We processed all the EEG data using custom Matlab scripts and EEGlab functions (Delorme & Makeig, 2004). We first recreated CPz by computing the average reference using pop.reref.m function of EEGlab. This was followed by re-referencing the data to the linked mastoids, then epoching from −2000 to 6000 ms around feedback onset. We used FASTER (Nolan et al., 2010) to detect and remove artifacts. We removed eye blink activities following ICA (Makeig et al., 1996).

Time-Frequency measures were computed by multiplying the fast Fourier transformed (FFT) power spectrum of single trial EEG data with the FFT power spectrum of a set of complex Morlet wavelets defined as a Gaussian-windowed complex sine wave: e^i2πtfe^{−t^2/(2×σ^2)}, where t is time, f is frequency (which increased from 1–50Hz in 50 logarithmically spaced steps) and the width (or ‘cycles’) of each frequency band were set to increase from 3/(2πf) to 10/(2πf) as frequency increased. Then, the time series was recovered by computing the inverse FFT. The end result of this process is identical to time-domain signal convolution, and it resulted in estimates of instantaneous power (the magnitude of the analytic signal) and phase angle (the arctangent of the analytic signal). Each epoch was then cut in length (−500 to +1000 ms). Averaged power was normalized by conversion to a decibel (dB) scale (10*log10[power(t)/power(baseline)]) from a common cross-condition averaged baseline of −300 to −200 ms, allowing a direct comparison of effects across frequency bands. Reward-associated delta band power was quantified as the average of 150 to 350 ms from 2 to 4 Hz at the FCz electrode. To compute the ERPs, we baseline corrected each epoch (−200 to 0 ms before feedback onset), filtered them between 0.01 to 20 Hz, and then averaged epochs. The RewP was quantified as the ERP component specific to reward and was measured as the average amplitude from 200–400ms at Cz, as in Brown et al. (2021).

3. Results

3.1. Group Differences

Chi-square tests of independence did not reveal any statistically significant differences between groups for sex (p =.87), ethnicity (p =.93), or race (p =.23). Independent samples t-tests did not reveal any statistically significant differences between groups for age (p =.64) or years of education (p =.14). Welch’s unequal variances t-tests revealed statistically significant group differences for AUDIT (p < .001) and BDI scores (p =.03), where both were higher in the hazardous drinkers group. The hazardous drinkers group had average AUDIT score suggestive of hazardous drinking (mean=10.50). Within the hazardous drinkers group, AUDIT scores correlated with BDI (rho(26) =.40, p =.03) and years of education (rho(26) =.45, p = .02). BDI and years of education were investigated as alternative regressors in subsequent analyses of AUDIT score.

3.2. Ratings

Both the groups rated all the classes of affective imagery similarly, except for the alcohol imagery, which was rated significantly more pleasant in the hazardous drinkers than the control group (t(52)=5.55, p<.001). Within the hazardous drinkers group, AUDIT score did correlate with pleasantness ratings for alcohol images, but not to a statistically meaningful degree (rho(26=.34, p=.08). Only the puppy and alcohol images here were used for the subsequent study, and there was no between group differences in puppy image ratings (t(52)=0.34, p=.73), see Figure 2.

Figure 2. Rating task.

Both the groups rated the images categories similarly. However, the hazardous drinkers group (ALC) rated the image category of alcohol to be significantly more pleasant than control (CTL) group (p<.001).

3.3. Behavioral Results

To examine how both the groups learned the task, we conducted a 2(DIFFICULTY: Easy vs. Hard) x 2(AFFECTIVE IMAGE TYPE: Puppy vs. Alcohol) x 2(GROUP: Hazardous Drinkers vs. Control) ANOVA for accuracy. This analysis failed to reveal any main or interaction effects between the groups: all ps>.30. This suggests that both the groups learned the task similarly well. There was a main effect of difficulty (F(1,52)=11.13, p=.002: mean accuracies: easy alcohol: 77%; easy puppy: 78%; hard alcohol: 71%; hard puppy: 69%).

3.4. EEG Results

Our first hypothesis was that the RewP would be larger to alcohol-related rewards as compared to puppy-related rewards in the hazardous drinking group (see Figure 3 for ERPs). We conducted a 2(DIFFICULTY: Easy vs. Hard) x 2(AFFECTIVE IMAGE TYPE: Puppy vs. Alcohol) x 2(GROUP: Hazardous Drinkers vs. Control) ANOVA on RewP amplitude. This analysis revealed a significant main effect of affective image type (puppy > alcohol: F(1,52) = 8.455, p=.005) while all other main effects or interactions between the groups were non-significant (ps>.10).

Fig. 3. ERPs time-locked to the color feedback.

The RewP (200 – 400ms) did not differ between image type conditions nor group. Topoplots show averaged amplitude over the 200–400 ms temporal window for each condition. b. Line plots display means +/− SEM for the Win and Loss ERPs for each condition showing no significant amplitude difference in the ERPs.

Next, we conducted a 2(DIFFICULTY: Easy vs. Hard) x 2(AFFECTIVE IMAGE TYPE: Puppy vs. Alcohol) x 2(GROUP: Hazardous Drinkers vs. Control) ANOVA on delta power amplitude, all group main or interaction ps>.39.

3.5. Correlations

3.5.1. Liking Rating Correlations

Even though there were no differences in RewP between groups, previous studies have found inter-individual effects of affective salience on the RewP in the absence of condition-specific effects (Brown et al., 2021; Brown & Cavanagh, 2018). However, across all participants, we did not find a significant correlation between image “liking” rating difference (puppy-alcohol) and RewP amplitude difference (puppy-alcohol), rho(52)=−.02, p=.91, thus failing to replicate prior findings of an affective influence on the size of the RewP. This relationship was not observed for either condition separately (ps>.34) or group (ps>.29).

We did find a significant correlation between this reward-related delta power amplitude difference and image rating difference across all participants (rho(52)= .35, p=.01). See Figure 4 for the spectral plot of reward-related delta activity and ROI-related scatterplot. This pattern was maintained in both the easy (rho(52)=.27, p=.04) and hard (rho(52)=.29, p=.03) conditions. Both the groups showed similar patterns when examined independently as well, with a non-significant trend in control (rho(24)=.34, p=.09) and a significant effect in hazardous drinkers: (rho(26)=.41, p=.03).

Figure 4.

a) Time-frequency delta band power to reward. b) There was a significant positive correlation between affectively-modulated delta band power and relative affective preference across all participants, *p<.01.

3.5.2. ERP-AUDIT Correlations

Since variance in the AUDIT score of the control group is highly restricted (0–3), we did not include this group in symptom-specific correlations. We found that within the hazardous drinkers group, there was a significant positive correlation between the degree of hazardous drinking (total AUDIT score) and RewP amplitude difference at the CPz electrode (alcohol>puppy: rho(25)=0.38, p=.046), indicating an inter-individual influence of drinking habits on affect-specific RewP amplitude, see Figure 5. Since BDI and education both correlated with AUDIT score in the hazardous drinkers group, we tested their correlations with RewP amplitude difference, and they were not significant (rho(26)=.05, p=.82 and rho(26)=.18, p=.37, respectively). In sum, the degree of hazardous drinking (AUDIT score) predicted the alcohol-specific enhancement in RewP amplitude in the hazardous drinkers group, but only at a site slightly more posterior (CPz) to where the RewP is traditionally quantified (usually Cz or FCz).

Figure 5.

Scatterplot between AUDIT score and RewP amplitude difference. There was a significant positive correlation between AUDIT score and alcohol-puppy RewP amplitude difference at CPz electrode in the hazardous drinkers group. The topoplot shows RewP correlations with AUDIT score in hazardous drinkers group from 200 to 400 ms following feedback, *p < 0.05.

This slightly more posterior distribution of the RewP and AUDIT score relationship suggests a potential role of late positive potential (LPP). The LPP is a cortically posterior ERP component elicited by viewing pictures containing affective content (Hajcak et al., 2009; Hajcak & Foti, 2020; Schupp et al., 2000). The LPP typically develops within a few hundred milliseconds after stimulus onset and lasts for the duration of picture viewing (Cuthbert et al., 2000). We examined if this AUDIT relationship captured variance with the LPP rather than with RewP.

Fortunately, our task design allowed a near perfect control for LPP-related affective content: each image was on the screen for 750–1000ms prior to the onset of the reinforcement. We defined an LPP time window of 400–750 ms, and the RewP window remained the same (200–400 ms, even though there was no RewP to the initial picture presentation). Figure 6a–e shows that within the hazardous drinkers group, there were no associations between liking or AUDIT and relative (alcohol minus puppy) LPP amplitudes to picture onset, nor were there group differences in relative LPP amplitude (as shown in Supplemental Figure 1). In contrast, both liking and AUDIT score correlated with relative RewP amplitudes (i.e., alcohol-puppy difference, Fig 6f–i). Surprisingly, AUDIT score also correlated with late relative LPP-like activity (as shown in Figure 7j).

Figure 6. RewP and LPP at Picture and Feedback Stimuli at CPz, only in the hazardous drinkers group.

(a). The ERPs following the picture presentation, showing an LPP. The early vertical lines highlight the RewP time window (200–400 ms) and the later vertical line are the LPP time window (400–750 ms). All EEG measures were differences (alcohol minus puppy). There were no significant correlations between (b). liking difference and RewP time window (c). liking difference and LPP (d). AUDIT score and RewP time window, nor (e). AUDIT score and LPP. (f). The ERPs following the the feedback presentation. There were no significant correlations between (g). liking difference and RewP nor (h). liking difference and LPP. As shown in (i). there was a significant correlation between AUDIT score and RewP (same as Figure 5), (j). as well as in in AUDIT score and LPP. Together, it is clear that phenotypic expressions in hazardous drinkers are not related to an LPP per se, but rather they relate to cue-relevant RewP enhancements and any subsequent reward-evoked salience changes in reward-following LPP *p<.05.

While these analyses rule out the alternative explanation that the LPP and not RewP was affected, there was clear influence of picture type on posterior slow wave activity only following reward. This finding suggests that this symptom-dependent relationship not only affects the RewP, but also an interesting amalgamation of both these ERP phenomena in reward-following late parietal activities.

4. Discussion

This is the first study to demonstrate that aberrant salience of alcohol imagery boosts cortical reward processing in the relevant population of hazardous drinkers . In this study, we first demonstrated that the hazardous drinkers group rated the pictures of alcohol to be more pleasant than controls, providing important face validity. Our first hypothesis was not supported though: there was no difference in the RewP to affective imagery type as function of group, suggesting that the across group dynamics of the RewP is not affected by alcohol cue-relevant imagery. However, there was a significant main effect of affective imagery on RewP amplitude across all participants (puppies > alcohol), indicating a robust relationship between hedonic salience and RewP amplitude. The lack of a group effect was not entirely surprising since our foundational work on the affective influence on RewP has also failed to reveal condition-specifc effects (Brown et al., 2021; Brown & Cavanagh, 2018, 2020). This suggests that the effect of hedonic or “liking” salience on RewP may be highly idiosyncratic in nature. For example, existing literature suggests that RewP may be influenced by the variance in subjective differences in motivational factors related to task performance (Cockburn & Holroyd, 2018). Furthermore, RewP is also sensitive to individual differences in personality (Cherniawsky & Holroyd, 2017; Cockburn & Holroyd, 2018; Schmidt et al., 2016; Umemoto et al., 2015), emotions (Cockburn & Holroyd, 2018; Foti et al., 2011; Hewig et al., 2010) and psychiatric distress (Baker et al., 2010; Cockburn & Holroyd, 2018; Hewig et al., 2010; Holroyd, Baker, et al., 2008; Umemoto et al., 2014.).

We did not find a significant correlation between RewP amplitude difference vs. rating difference across all participants, thus failing to replicate the major findings of Brown et al. (2021) of interindividual differences in RewP due to emotional influence. Yet we did reveal a significant correlation between reward-related delta power and this rating difference. The fact that RewP and delta power findings were not aligned in the current study is surprising, but this delta power finding does align with the previously established conceptualization that event-related EEG responses to reward are characterized by a sensitivity to a hedonic dimension.

We obsered a significant correlation between hazardous drinking and alcohol cue-modulated RewP amplitude, suggesting that heavier drinkers are more responsive to a cue-relevant enhancement of intrinsic reward. Interestingly, our findings showed this correlation was only present at the more posterior electrode cite CPz. We would like to note here that a study by Cavanagh (2015) used CPz as the main site for quantifying RewP, so this is not an entirely novel quantification. To follow up this surprising outcome, we examined the role of emotional salience in EEG features responsive to simple affective salience (i.e. LPP) vs. reward processing (i.e. RewP). All major correlational associations with liking ratings or AUDIT were only found in the presence of reward, demonstrating that these effects could not be explained by a simple LPP. Yet, within the hazardous drinkers group, there was a significant correlation between AUDIT score and late LPP-like activity to alcohol cues only following reward presentation. Taken together, these results suggest that there indeed was a robust influence of affective image type on posterior slow wave activity, but this was dependent on to the presence of rewarding feedback. We can’t fully explain this unexpected finding at this point, but it may be interpreted as a reward punctuated LPP.

These findings bridge the gap in the existing literature involving hedonic salience of addiction-specific stimuli and reward processing. No studies have examined how hedonic salience to addiction-specific stimuli relates to RewP. We did this by utilizing a unique approach to examine the neural correlates of RewP and reward sensitivity to alcohol-based cues in individuals with hazardous drinking with respect to affective salience of intrinsically rewarding images. Additionally, this study conceptually replicates the results of Brown et al. (2021) by showing that individual rating of relative liking relates to affect-specific enhancement of reward-related EEG activity. Furthermore, we describe a novel finding that individual ratings of drinking severity in the hazardous drinkers group related to cue-specific boosting of the RewP.

5. Limitations and Future Directions

Hazardous drinking is a complex, multimodal phenotype. This suggests that there could be several supplementary factors leading to hazardous drinking that may not be associated with incentive salience sensitization that needs to be examined. In our sample, the AUDIT score was significantly correlated with the BDI score which is consistent with the results of previous studies that describe how depressive symptoms are common among individuals with high levels of alcohol and other drug use (Compton et al., 2007; Hasin et al., 2007). People who use alcohol and other drugs are two to four times more likely to have major depression as compared to other individuals in the general (Compton et al., 2007; Hasin et al., 2007). Depressive symptoms impact 30–45% of people pursuing treatment for substance misuse (Grant et al., 2004). Yet notably, in the current analysis, BDI scores did not account for variance in the relationship between alcohol-induced RewP change and hazardous drinking (AUDIT). We predict that changing the affective image class from a generally likable theme (i.e. puppies) to a neutral theme (e.g. chairs) may lead to increased relative salience of alcohol imagery and thus may elicit a stronger effect of affective imagery on the RewP.

6. Conclusions

In conclusion, our findings showed that reward-related delta power is sensitive to liking of alcohol-based stimuli in individuals who consume alcohol heavily. Additionally, there was an interesting consolidation of LPP and RewP following rewarding feedback, at a slightly more posterior brain region than usually observed. This may suggest that an increased salience follows reward administration to alcohol-based cues in hazardous drinkers group. To our knowledge this is the first study to show an association between LPP and RewP feature of reward processing: this compels replication and further examination. Furthermore, this study replicated the findings of Brown et al. (2021) with the absence of a main effect of affective imagery but strong interindividual differences related to liking. In sum, our findings suggest that alcohol-specific imagery boosts the reward-related EEG activity in hazardous drinkers, thus demonstrating a potential mechanism for linking biased attention with reward integration for addiction-specific stimuli.

Supplementary Material

1

Supplemental Figure 1. RewP and LPP at Picture and Feedback Stimuli at CPz, only in the Control group. (a) The ERP following the picture presentation, showing an LPP. The early vertical lines highlight the RewP time window (200–400 ms) and the later vertical line are the LPP time window (400–750 ms). All EEG measures in scatterplots are differences (alcohol minus puppy). There were no significant correlations between (b). RewP time window amplitude difference and liking difference (c). nor between LPP amplitude difference and liking difference, (d). there were no group differences in RewP time window amplitude, nor (e). relative LPP amplitude, (f). The ERP following the the feedback presentation. There were no significant correlations between (g). RewP amplitude difference and liking difference, nor between (h). LPP amplitude difference and liking difference. (i). There were no group differences in RewP amplitude (j). nor in relative LPP amplitude, (all p’s>.42).

Supplemental Figure S2: Difference waves (reward minus loss) for each group and condition.

Highlights.

In this report we address a gap in the literature involving hedonic salience of addiction-specific stimuli and reward processing.

We utilized a unique task specifically designed to examine the influence of affective images on the ERP component of the Reward Positivity (RewP) and its time-frequency spectral reflection in the delta and. There was a significant main effect of affective imagery on RewP amplitude across all participants (puppies > alcohol), indicating a robust relationship between hedonic salience and RewP amplitude. Across all participants, ratings of affective liking alcohol images correlated with affect-specific RewP delta band power. Within the heavy drinking group, hazardous drinking severity (AUDIT score) correlated with alcohol cue specific RewP amplitudes.