Unique brain regions involved in positive versus negative emotional modulation of pain

1 Introduction

Pain is a complex and subjective sensory and emotional experience. This experience is affected by psychological factors, including emotions (as manipulated by odors [1], emotional sentences [2], hypnosis [3], films [1], and pictures [4], [5], [6]). According to Motivational Priming Theory, emotions cue our systems to respond to our environment by activating aversive or appetitive drives [7], [8]. The emotional modulation of pain (EMP) effect stems from this theory, such that in response to a noxious stimulus, one will rate a stimulus as more painful when presented in conjunction with a negative emotion-evoking stimulus, and, to a lesser extent, less painful when presented in conjunction with a positive emotion-evoking stimulus [3], [4], [5], [9], [10], [11]. Previous studies have reported a variable effect for EMP depending on whether ratings were gathered for the sensory (intensity) or affective (unpleasantness) construct [12], [13], [14].

A recent study by McIver et al. [15] has shown that the modulatory capacity of positive and negative emotional manipulation is markedly variable – for some individuals, pain ratings were more greatly mitigated by positive emotional stimuli than amplified by negative emotional stimuli, while for others the opposite was true. Despite this variability, a significant difference in pain ratings was found between the positive and negative EMP for ratings of pain intensity and pain unpleasantness. Intriguingly, however, the pain intensity ratings were more heavily influenced by the positive emotional stimuli whereas pain unpleasantness ratings were more heavily influenced by the negative emotional stimuli.

While previous studies have shown that emotion/mood manipulations can influence cortical responses to painful stimulation in regions commonly known to function in pain processing [16], [17], [18], such as the anterior cingulate cortex, insula, thalamus, ventral striatum and periaqueductal gray, and several aspects of the lateral prefrontal cortex [11], [13], [14], [19], they have either omitted the comparison of both positive and negative conditions to an emotionally neutral condition [13], [14], [19], or have not distinguished the effect for EMP on separate sensory and affective components of pain [11]. Examining how the pain response is influenced by both negative and positive emotional contexts compared to a neutral valence would facilitate a more distinct elucidation of the unique neural mechanisms for negative and positive EMP effects. Similarly, that emotional modulation has been shown to affect perception of pain intensity and unpleasantness differently [12], [13], [14] raises the question of how this may be reflected in the neural response to EMP.

The current study sought to further characterize the neural correlates of EMP. Using an identical study paradigm as previously employed in recent behavioural and subcortical studies – displaying negative, neutral, and positive emotion-evoking visual stimuli during the presentation of a noxious thermal stimulus – functional magnetic resonance imaging of the brain was carried out. Behavioural research broadly shows that pain ratings are higher in the context of negative affective stimulus compared to positive. More narrowly, there is evidence to suggest dissociable effects for positive and negative affect on pain modulation relative to an emotionally neutral stimulus pairing, with varied impact on pain intensity and unpleasantness. This context is particularly relevant given recent findings that highlight a varied perceptual susceptibility to either the positive or negative affective influence on pain [15]. The aim of the present study was to expand on previous research and provide further illumination of the biological mechanisms which underlie these processes. We hypothesized that the neural structures involved in the response to a noxious stimulus are different during the presentation of positive versus negative emotion-evoking stimuli.

2 Methodology

It is of note that the current study relates to the third experiment in a three-part study. In the preceding experiments, individual variability and behavioural characteristics associated with EMP perception were examined [15] and the corresponding neural activity in the brainstem and spinal cord were investigated [20]. The current study employs functional magnetic resonance imaging (fMRI) to investigate the cortical correlates of EMP perception in the same sample of participants using an identical study protocol. Participant sessions for each of the three experiments were conducted on different days.

2.2 Materials: noxious stimulus, pain rating scales, and visual affective stimuli

A thorough account of the materials and protocol used in the study are described previously [15]. Briefly, the noxious thermal stimulation was administered by way of a Medoc TSA-II thermal sensory analyzer (Medoc Ltd, Ramat Yishai, Israel). The thermal stimulus was delivered to the thenar eminence of the right hand corresponding to the C6 dermatome. The precise noxious temperature for each participant was determined through a calibration process that took place prior to the imaging portion of the experiment. First, a description of the separate sensory (Intensity) and affective (Unpleasantness) components of pain was provided, as per Price et al. [24], and participants were familiarized with the corresponding pain rating scales for both pain Intensity and Unpleasantness. The pain Intensity scale displayed number ratings from 0 to 100, with verbal anchors at increments of 10 along the spectrum, ranging from “no sensation” at 0, to “intolerable pain” at 100 [25], [26], [27]. The pain Unpleasantness scale was a modified version of the pain Intensity scale, similarly ranging from “neutral” at 0 to “intolerably unpleasant” at 100.

Once participants verbally confirmed that they understood the distinction between the separate rating scales, they were exposed to 1 heat pulse of 47 °C and asked to verbally provide a rating of their perceived pain intensity and unpleasantness for that sensation, using the scales described above. Following this initial pulse and rating, participants were presented with and asked to rate a series of paired-heat pulses (with 5 °C of difference between the temperatures within each pair of pulses) to confirm sensitivity to discriminate fluctuations in temperature. This was followed by sets of 10 consecutive heat pulses, whereby the participant was asked to rate the final pulse, as they felt it occur. These 10-pulse trials were used to calibrate the temperature for the noxious stimulus to the unique pain threshold for each participant. For each participant, this was the temperature that consistently elicited a pain intensity rating of “50” for the 10th pulse. Further detail on this calibration procedure is provided in McIver et al. [15].

The pain stimulus paradigm for the task was presented in an off-on-off design with a 50 s baseline innocuous thermal sensory period, followed by a 30 s period of noxious thermal stimulation, followed by a return to innocuous baseline for 75 s (Fig. 1), with visual stimuli displayed throughout the duration of the task. The baseline innocuous temperature was set to 8 °C below the noxious thermal temperature which was determined for each participant in Experiment 1 [15]. The 30-s noxious stimulation period consisted of 10 pulses, with each pulse lasting 3 s (delivered at a frequency of 0.33 Hz). There was a 2 min break between each run to avoid sensitization and to allow nociceptors in the skin to recover [28], and multiple runs were acquired of each study condition. Visual stimuli relating to one valence (Positive, Neutral, or Negative) were displayed continuously throughout the duration of each run, paired with the off-on-off design of the pain paradigm, to facilitate the dissociation of BOLD signal changes related to the noxious stimulus, while emotional-valence was sustained across the run.

Fig. 1:

The stimulus paradigm for the task was presented in an off-on-off design with a 50 s baseline innocuous thermal sensory period set to 8 °C below the noxious thermal temperature which was determined for each participant, followed by a 30 s period of noxious thermal stimulation which consisted of 10 pulses with each pulse lasting 3 s, followed by a return to innocuous baseline for 75 s.

The visual affective stimuli used in all sessions (including training and imaging sessions) consisted of a selection of images primarily taken from the International Affective Picture System (IAPS) (752 images) [29] and supplemented with 89 high-arousal positive images from publically available sources. The IAPS is a database of images categorized by standardized ratings for various characteristics, including Valence and Arousal. Based on these ratings, images were selected for each of the three conditions (Positive, Neutral, and Negative). Images in the Positive condition were high in valence (6.6±1.7, M±SD), while negative images were low in valence (2.3±1.4), and neutral images were of medium valence (4.9±1.4). Positive and negative images were matched for high arousal (arousal levels of 5.6±2.2 and 5.9±2.1, respectively), while neutral images were (inherently) lower in arousal (3.5±2.0). The 89 images that were supplemented from the internet underwent a pilot rating study, whereby a separate group of 28 healthy women (age of 26.2±4.8, range=18–45 years) voluntarily rated the images in terms of both valence and arousal using the Self-Assessment Manikin (SAM; the same rating tool used by Lang [29] to collect the ratings for the IAPS). Only images that were rated consistently with the positive images selected from the IAPS in terms of both valence (5.7±0.6) and arousal (5.0±0.4) were included in the sample used for the Positive condition. All images were resized to the same dimensions (1,024×768 pixels).

3 Data analysis

4 Results

4.2 fMRI results

Significant activity was detected in the Negative, Neutral, and Positive conditions, as well as for the specific contrasts (See Table 1). When collapsed across emotion-evoking stimulus conditions the efficacy of the pain stimulus to produce a response in pain processing brain regions (ACC, insula, somatosensory cortices, thalamus) was demonstrated (Table 1) and, of the three emotion conditions, the Neutral condition results appear most similar (Fig. 2).

Table 1:

Peak intensity voxel coordinates for the main effect of pain; response to pain during emotion conditions (Neutral, Positive, Negative); contrasts (Neutral>Positive, Negative>Neutral, and Negative>Positive); and post-hoc analyses (unpleasantness scores covaried with the Negative>Neutral map).

Condition	Side	Region	BA	Coordinates			Voxels	t	p-Value
				X	Y	Z
Pain (collapsed across emotion conditions)
	R	Postcentral gyrus	2	63	−25	34	2,911	6.11	<0.001
	R	Insula	13	39	−4	16	2,044	5.96	<0.001
	R	Middle temporal gyrus	37	42	−61	1	2,413	5.28	<0.001
	R	Culmen	n/a	15	−37	−17	3,330	5.49	<0.001
	R	Claustrum	n/a	27	11	−8	4,005	−6	<0.001
	R	Caudate body	n/a	12	14	10	1,996	7.65	<0.001
	L	Precentral gyrus	6	−42	−13	31	140,145	−13.24	<0.001
	L	Anterior cingulate	32	−6	41	−2	9,339	7.22	<0.001
	L	Caudate tail	n/a	−33	−25	−5	1,956	5.17	<0.001
	L	Postcentral gyrus	43	−57	−16	19	13,833	8.66	<0.001
	L	Inferior frontal gyrus	47	−30	20	−5	4,937	−7.09	<0.001
	L	Superior temporal gyrus	41	−54	−19	7	4,368	−6.21	<0.001
Neutral
	R	Postcentral gyrus	43	66	−16	17	1,422	5.27	<0.001
	R	Superior temporal gyrus	41	54	−25	10	3,422	−6.82	<0.001
	R	Precentral gyrus	6	57	−4	28	65,494	−8.35	<0.001
	R	Middle frontal gyrus	46	54	23	22	1,299	−3.86	<0.001
	R	Inferior frontal gyrus	46	33	32	13	2,708	4.81	<0.001
	R	Declive	n/a	18	−58	−14	1,384	−5.92	<0.001
	L	Lingual gyrus	18	−15	−52	4	3,892	−6.46	<0.001
	L	Insula	47	−33	17	−2	1,659	−5.55	<0.001
	L	Postcentral gyrus	43	−57	−16	19	2,366	8.01	<0.001
Positive
	R	Superior temporal gyrus	42	60	−28	10	3,854	−5.66	<0.001
	R	Precentral gyrus	4	36	−16	34	13,261	−8.82	<0.001
	R	Inferior parietal lobule	40	45	−34	34	3,226	4.69	<0.001
	R	Inferior frontal gyrus	9	57	8	25	1,354	5.21	<0.001
	R	Declive	n/a	27	−67	−14	3,157	4.73	<0.001
	R	Parahippocampal gyrus	35	21	−28	−11	4,061	4.64	<0.001
	R	Precuneus	31	18	−58	31	3,478	4.23	<0.001
	R	Medial frontal gyrus	6	18	29	37	1,752	−3.94	<0.001
	R	Precentral gyrus	4	18	−28	61	1,388	−4.71	<0.001
	R	Caudate (head)	n/a	6	11	1	1,386	4.65	<0.001
	L	Lingual gyrus	18	−12	−70	4	11,613	−5.32	<0.001
	L	Precuneus	31	−15	−49	34	2,564	−4.22	<0.001
	L	Superior frontal gyrus	10	−9	68	20	1,494	3.86	<0.001
	L	Precentral gyrus	6	−45	−13	34	8,439	−8.17	<0.001
	L	Cingulate gyrus	24	−12	14	28	1,976	−4.73	<0.001
	L	Putamen	n/a	−18	8	7	2,539	5.48	<0.001
	L	Precentral gyrus	4	−15	−31	55	1,812	−4.6	<0.001
	L	Parahippocampal gyrus	19	−30	−40	−5	1,418	4.91	<0.001
	L	Insula	13	−39	−16	16	4,910	6.16	<0.001
	L	Precentral gyrus	6	−54	2	28	1,327	4.74	<0.001
Negative
	R	Precentral gyrus	6	39	−13	34	16,981	−7.49	<0.001
	R	Precuneus	7	3	−55	61	4,509	−5.73	<0.001
	R	Culmen	n/a	6	−70	−8	20,328	−6.12	<0.001
	R	Precentral gyrus	4	18	−28	61	1,433	−5.62	<0.001
	L	Anterior cingulate	32	−6	41	−3	9,156	6.98	<0.001
	L	Superior parietal lobule	7	−18	−49	58	4,977	−5.83	<0.001
	L	Precentral gyrus	6	−45	−13	34	6,850	−8.42	<0.001
	L	Precuneus	7	−18	−70	37	1,220	−4.14	<0.001
	L	Caudate (body)	n/a	−12	11	7	1,290	4.66	<0.001
	L	Middle frontal gyrus	6	−30	2	58	1,553	4.2	<0.001
	L	Postcentral gyrus	43	−57	−16	19	1,562	4.79	<0.001
	L	Angular gyrus	39	−48	−67	28	1,633	3.8	<0.001
Contrast
Neutral>Positive
	R	Superior parietal lobule	7	33	−61	49	10,714	−4.87	<0.001
	L	Postcentral gyrus	5	−12	−43	61	1,074	−3.37	<0.002
	L	Inferior parietal lobule	40	−33	−52	46	2,705	−4.48	<0.001
Negative>Neutral
	R	Hippocampus	n/a	24	−37	−2	1,132	−3.48	<0.001
	R	Lingual gyrus	18	9	−82	1	1,771	−4.44	<0.001
	L	Fusiform gyrus	19	−24	−61	−8	1,659	−4.03	<0.001
	L	Angular gyrus	39	−39	−58	37	2,162	3.77	<0.001
Negative>Positive
	R	Inferior parietal lobule	40	45	−31	46	1,501	−4.43	<0.001
	R	Parahippocampal gyrus	36	33	−31	−14	6,051	−4.36	<0.001
	R	Precuneus	7	21	−70	46	3,087	−4.12	<0.001
	R	Superior frontal gyrus	10	24	50	7	1,063	3.4	<0.002
	R	Cingulate gyrus	31	9	−52	28	1,612	3.4	<0.002
	L	Anterior cingulate	32	−15	44	10	3,275	4.26	<0.001
	L	Parahippocampal gyrus	19	−30	−40	−5	1,628	−4.75	<0.001
	L	Superior parietal lobule	7	−18	−49	58	1,092	−4.33	<0.001
Post-hoc
Unpleasantness covaried in Negative>Neutral
	R	Sub-gyral	4	15	−28	55	1,961	0.76	<0.001
	L	Uncus	28	−24	2	−22	1,035	0.74	<0.001

1. Results were displayed at p<0.05 prior to correction using Monte Carlo simulations at 1,000 iterations.

Fig. 2:

BOLD responses to the presentation of the pain stimulus during the neutral-valence condition (top) and when collapsed across emotion conditions (bottom) presented at p<0.05 and cluster threshold corrected for multiple comparisons. Significant increased activity is displayed in orange and significant decreased activity is displayed in blue.

During presentation of the pain stimulus with the positive-valence images (Fig. 3), there was increased activity in the right inferior parietal lobule, inferior frontal gyri, parahippocampal gyrus, precuneus, caudate, insula, and precentral gyrus and left superior frontal, parahippocampal, and precentral gyri, putamen, and insula. There was a decrease in activity in right superior temporal, precentral, and medial frontal gyri and cingulate, and left precentral, cingulate, and lingual gyri and precuneus.

Fig. 3:

BOLD responses to the presentation of the pain stimulus during the viewing of positive-valence and negative-valence images (p<0.05 cluster threshold corrected for multiple comparisons). Significant increased activity is displayed in orange and significant decreased activity is displayed in blue. Unique to the positive EMP, there was increased activity in the inferior parietal, parahippocampal/perirhinal, precuneus/superior parietal, and the prefrontal and superior frontal cortices. Unique to the negative EMP, there was increased activity in anterior and posterior cingulate and angular gyrus.

During presentation of the pain stimulus with the negative-valence images (Fig. 3), activity was increased in the left anterior cingulate, caudate, middle frontal gyrus, postcentral gyrus and angular gyrus. There was a decrease in activity in the right precentral gyrus, precuneus, culmen, and left superior parietal lobule.

In the Neutral>Positive contrast (Fig. 4), there was decreased activity (i.e. representing greater neural response to pain during the presentation of positive images than neutral images) in right superior and left inferior parietal lobules, and left postcentral gyrus.

Fig. 4:

BOLD response differences for Neutral>Positive (top row), Negative>Neutral (middle row), and Negative>Positive (bottom row) contrasts (p<0.05 cluster threshold corrected for multiple comparisons). The Neutral>Positive contrast detected superior parietal, inferior parietal, and postcentral gyri activity increases during the presentation of positive images in blue. The Negative>Neutral contrast detected increased activity in the angular gyrus in orange and decreased activity in the hippocampus, lingual and fusiform gyri in blue for the negative images. The Negative>Positive contrast detected greater activity in the superior frontal, cingulate and anterior cingulate gyrus for the negative images in orange and in the superior and inferior parietal, parahippocampal gyrus, and precuneus for the positive images in blue. SAG=sagittal; COR=coronal; TRA=transverse; A=anterior; P=posterior; L=left; R=right.

In the Negative>Neutral contrast (Fig. 4), which represents the significant neural response to pain associated with the negative valence of the images above and beyond that of (neutral) images, there was increased activity in the left angular gyrus. There was decreased activity (i.e. greater during the presentation of neutral images than negative images) in the right hippocampus, right lingual and left fusiform gyri.

In the contrast Negative>Positive (Fig. 4), there was increased activity (i.e. neural response to pain significantly stronger for the Negative condition) for the right superior frontal, cingulate and left anterior cingulate gyri. There was decreased activity (i.e. neural response to pain significantly stronger for the Positive condition) in the left superior parietal, right inferior parietal, and bilateral parahippocampal gyri, and right precuneus.

5 Discussion

The fMRI results of this study support the conclusion that positive EMP and negative EMP involve different processes. The main effect of the pain stimulus elicited activity in pain processing regions such as the ACC, insula, somatosensory cortices, and thalamus. The response to the pain stimulus during presentation of the neutral stimuli was largely similar to the main effect of the pain stimulus. Critically, the neural activity elicited during the positive and negative valence conditions included both common and distinct brain regions. The brain regions uniquely active during each condition, when considered together, may point to mechanisms that underlie these processes.

The Positive condition was primarily associated with increased activity in the inferior parietal (BA40), parahippocampal/perirhinal (BA35, BA36, BA19), precuneus/superior parietal (BA7), and prefrontal cortices (inferior PFC, (BA9)). These regions were identified in the Positive condition as well as in Neutral>Positive and Negative>Positive contrasts.

The involvement of the inferior parietal cortex and insula is not surprising given their roles in somatosensory association and judgment of pain intensity, respectively [31], [32]. The involvement of the prefrontal cortex is also consistent with previous research findings. Activity of the functionally defined dorsolateral prefrontal cortex (DLPFC, BA8/9/10) has been shown to be inversely correlated with perceived pain intensity and unpleasantness [33]. However, in the current study these areas are only significantly active during the Positive condition – not during the Negative condition even though the pain stimulus is the same. Considering other functional roles of these brain regions may provide insight into the distinct natures of these two conditions.

The inferior parietal cortex, for example, has been shown to be involved in identifying other people’s emotions through interpretation of posture and gesture [34], and, in the right hemisphere, empathy [35], [36]. Similarly, activity of the insula has been associated with watching pain in others [37], [38]. The left insula in particular has been implicated in emotional intelligence which involves a heightened identification of the emotions of others [39]. It is possible that participants were identifying or empathizing with the emotions of others in the images being presented during our study.

The perirhinal/parahippocampal cortices were also consistently represented in the positive condition, with significant activity in both the Positive activity map and the Negative>Positive contrast. Again, activity in this region was unique to the Positive condition and was not elicited during the Negative condition. Although this result is consistent with other reports of parahippocampal response to positive stimuli [40], [41], response to negative stimuli has also been reported [42], [43], [44]. For example, Roy and colleagues [11] observed stronger parahippocampal gyrus response to nociceptive electric shock delivered during presentation of negative- and neutral valence photos compared to positive. Looking to other functional roles of the perirhinal/parahippocampal cortices, it is established that this region is involved in scene recognition [45], [46], [47]. Furthermore, other work has shown that the parahippocampus is specifically involved in context association of scenes [48], [49]. As in, this region is important for processing scenes and is specifically involved in placing a scene into context. Discussed above, the inferior parietal region is involved in identifying or empathizing with other people’s emotions – this may be facilitated by the function of the parahippocampus in inferring or generating contextual information from the scene in which the other is depicted.

The inferior prefrontal cortex, BA9, is also involved in the recognition of emotion in others [50] and in attention to positive emotions [51], [52]. The functional roles of the prefrontal cortex, along with the identified roles of parietal, insular, and perirhinal/parahippocampal cortices listed above, consistently suggest that the participants are attending to, recognizing, and/or empathizing with the emotions meant to be elicited or inferred to be experienced by the people in the scenes. The activity in these brain regions, all elicited during the positive-valence condition only, suggest a mechanism underlying the reduced pain Intensity ratings during presentation of positive stimuli. Common to these functions is an external focus, in that attentional resources are allocated to identifying and empathizing with the emotional state of others. This external focus may explain why participants report lower pain intensity ratings – by focusing outside of the self, experiences within the self are less salient.

Common to both the Positive and Negative condition fMRI results, significant activity was seen in postcentral regions involved in somatosensation and pain perception; the superior frontal cortex (BA10, anterior PFC), a region involved in processing emotional stimuli; the caudate, involved in motor function and inhibitory control; and the premotor cortex (BA6), involved in motor planning. These regions appear to be involved in processing, modulating, and responding to pain and emotion, but do not appear to be valence specific.

Unique to the Negative condition, the cingulate [anterior cingulate (ACC, BA32) and posterior cingulate (PCC, BA31)] was seen in both the Negative activity map and the Negative>Positive contrast. Similarly, the angular gyrus (BA39) was seen in the Negative activity map, as well as the Negative>Neutral contrast. The ACC has been established to be involved in the emotional reaction to pain [53], [54] and the PCC, particularly the rostral PCC, is known to be involved in pain processing [55]. Beyond pain processing, the PCC also appears to play a role in the mediation between emotion and memory, with links to emotional salience [56], the emotional importance of autobiographical memories [57], and self-related processing [58]. The PCC, as well as the angular gyri, are key regions of the default mode network, which is a functionally connected set of brain regions thought to be involved in internal mentation during rest [59], [60]. Additional proposed roles of the angular gyrus include (self-) awareness and theory of mind [61], and autobiographic [62] and subjective memory retrieval [63] – inherently introspective tasks. The results of the post hoc analyses are also consistent with the idea that the negative EMP effect may be rooted in self-focus – pain Unpleasantness ratings covaried with activity in the Negative>Neutral contrast map in motor and limbic cortices, areas involved in memory, emotion and preparation of an emo-motoric response [64]. That the functional roles of the brain regions shown to be active during the negative emotion condition are consistently internally-, or self-, focused suggests that this may be involved in the exacerbation of pain unpleasantness. The participants might have been more keenly aware of their own suffering, or the unpleasantness of their current state, while internally- or self-focused. Taken together, this evidence supports the idea that the act of focusing externally sufficiently distracts us from the physical aspect of pain and thus our ratings of pain intensity decrease, whereas during internal focus the emphasis is on our own experience so our awareness of our own suffering increases, thus ratings of pain unpleasantness increase. The respective functional roles of the brain regions implicated for the Positive and Negative conditions in our study support this.

The results of the comparison of pain ratings between conditions partially reproduced those found in our previous behavioural [15] and subcortical [20] studies of pain Intensity and Unpleasantness ratings differentially effecting EMP. Consistent with our previous work, we found that Unpleasantness ratings are driven more by negative than positive emotions. The Neutral to Negative difference has more of an effect on pain Unpleasantness than the Neutral to Positive difference. Differences between conditions for pain Intensity ratings, while consistent with our previous work in that the difference between the Negative and Neutral ratings was smaller than the difference between the Neutral and Positive ratings, failed to reach significance. While this finding varies from previous findings for EMP of pain perception from our lab, these results are consistent with other studies that have shown that emotional manipulations differentially affect pain unpleasantness, but not intensity [12], [13], [14].

These findings must be considered in light of certain limitations. Participation in this study was intentionally restricted to females to avoid known sex differences in pain perception and to optimize for the chosen noxious stimulus, as females have been shown to be more sensitive than males to thermal stimulation [21]. Thus, the current study design prohibits generalization to males and the examination of potential gender differences in the EMP processing. Future research should aim to address these questions. It is also of note that previous research has shown mood-related modulation of pain processing in the brain via exposure to pleasant and unpleasant odors. One of these studies highlighted the role of the ventral striatum in the mitigating effect of positive mood on pain perception [14]. The present results showed heightened activation related to positive EMP in both the caudate and putamen (two regions which fuse anteriorly to form the ventral striatum [65]). Activation of these adjacent regions in the present study may align somewhat with this earlier research, though the visual (versus olfactory) nature of the emotional stimuli in the current study may contribute to some differences in results. The mood-odor modality for EMP was also previously used to examine separate mechanisms for mood and attentional modulation of pain [13]. Mood related modulation of pain was correlated with activity in the lateral inferior frontal cortex, whereas “attention-related modulation” correlated with activity in the superior posterior parietal (SPP; BA7) and entorhinal cortex [13]. Other behavioral research has shown that shifting attentional focus from a painful stimulus to simultaneously presented emotional images resulted in modulation of the sensory (intensity), but not the affective (unpleasantness), component of pain perception [66]. While the present study was not designed to disentangle the distinct cortical responses to mood vs. attentional modulation of pain, our findings seem to suggest that a shift in attentional focus may be associated with the neural response involved in emotional-modulation of pain via visual stimuli. Future research may seek to directly compare the effects of positive and negative valence in emotional images on viewer’s innate (i.e. unguided) locus of attention, and how this relates to cortical processes.

Despite these limitations, the present experimental design does offer distinct benefits and further insights into the neural correlates of EMP. The continuous presentation of emotional visual stimuli across the duration of the fluctuating pain stimulus paradigm facilitated the dissociation of the neural response to EMP above and beyond the neural response to the emotional images alone. Additionally, the present study contrasted the neural response to EMP for both the Positive and Negative valence compared to Neutral. While previous studies may have foregone the comparison to neutral images because of the inherently lower arousal compared to Positive and Negative stimuli, Kenntner-Mabiala et al. [66] have suggested that arousal is less influential than valence to affective picture modulation of pain. This inference is rooted in the consistently reported EMP effect for Positive images relating to lower pain ratings than Negative images, despite the high-arousal content of both conditions. Thus, in addition to offering further evidence for the overall effect of EMP contrasting Positive to Negative, the present results provide new insights about the distinct cortical representation of separate mechanisms for Positive and Negative EMP. Lastly, with numerous studies reporting different effects for emotional modulation on the separate sensory and affective components of pain perception, our findings shed further light on the neural correlates of these separate elements by considering differences related to both pain Intensity and Unpleasantness.

In conclusion, this study makes significant contributions toward understanding the neural response underlying the EMP. The results replicate our earlier finding that negative emotions have a greater influence for EMP for pain unpleasantness. Furthermore, we found that different brain regions appear to be involved in the EMP for positive and negative emotions. The results suggest that there may be different cortical mechanisms underlying each process, and that a possible explanation may involve internal (self) versus external attentional focus during negative compared to positive EMP, respectively.