Thirty right-handed healthy volunteers (15 females, Range_age = 20–44 years; M_age = 28.9 yrs, SD_age = 6.0 yrs; M_education = 14.7 yrs, SD_education = 2 yrs) participated in the fMRI study. Handedness was determined by means of the Edinburgh Inventory Scale (Oldfield, 1971). All the subjects had normal or corrected-to-normal visual acuity and gave formal consent to participate in this study. The subjects were paid for their participation or received university credits. This study was approved by the local Ethics Committee.

Thirty-two disparagement jokes (DJ) were chosen from different internet websites. They contained different types of jokes concerning gender, human race, authorities, politicians, religions.

Jokes were modified in a way that each of them consisted of two sentences: a set up line and a funny punch line. Using the same set up lines, an equal number of SI and N stimuli were generated by replacing the punch line:

To select the set of stimuli to be presented during the fMRI sessions, a group of 15 volunteers, different from those subjected to the fMRI experiment, were asked to rate the 96 stimuli (32 stimuli X 3 types: DJ, SI, and N) on a scale of 0 (very offensive) to 6 (very funny). A rating of 3 indicated that the stimulus was neutral. Twenty stimuli reached on average a score of 4.5 or higher and were included in the set of disparagement jokes. Twelve stimuli were rated on average 1.5 or lower and were included in the set of offensive items. Eight stimuli reached a mean score of 3 and were included to form the set of neutral stimuli.

The fMRI experiment was carried out using an event related design (see Figure 1). Each subject performed two fMRI runs, each including the presentation of 10 DJ, 6 SI, and 4 N for a total of 40 stimuli. Each stimulus comprised a setup line lasting 3.5s, followed after 0.5s by the punch line lasting 3.5s (DJ, SI, or N). Subjects were instructed to indicate the level of fun of each stimulus along a four-point scale (0 = not funny; 3 = very funny) at the presentation of the punch line stimulus (see Figure 1), by pressing one of four response keys, using the thumb, the index, the medium, and the ring finger, respectively. For half of the subjects the scale was ascending from thumb (very funny) to the ring finger (not funny at all), the other half the scale was reversed. Stimuli were counterbalanced across runs, and the presentation was randomized. Each run lasted for 5 min.

Stimuli were presented using the IFIS-SA system (MRI Devices Corp., Waukesha, WI). E-prime (Psychology Software Tools, Inc.)¹ was used to control stimulus delivery and to record behavioral responses (type of answer and reaction times). To acquaint subjects with the task, they rehearsed a couple of items before the experimental phase.

Following the acquisition session, out of the scanner, the subjects were asked to rate each of the 40 stimuli on the level of offense they experienced, using a scale ranging from 0 (not offensive) to 3 (very offensive). They were asked to base their answers on how they had perceived the stimuli while they were inside the scanner.

MRI data were obtained on a 3 T Philips Gyroscan Intera MR Scanner (Philips Medical Systems), using the standard setup of body coil transmission and SENSE head coil reception.

BOLD-sensitive fMRI images were acquired using a T2*-weighted gradient-echo echo-planar sequence (TR = 2,000 ms, TE = 35 ms, FA = 80°, FOV = 240 mm, matrix 64 × 64, SENSE factor = 2) to obtain thirty axial slices of 4 mm thickness with no interslice gap, covering the whole brain. A total of 150 volumes were collected in each scan. A high-resolution T1-weighted 3D gradient-echo anatomical image was also acquired to allow anatomical localization (TR = 9,900 ms; TE 4.6 ms; FOV: 256 MM; matrix: 256 × 256; voxel dimension 1.0 × 1.0 × 1.0). Subjects were instructed to refrain from laughter to reduce movement artifacts. Head motion was minimized through foam padding within the head coil.

Behavioral results have been analyzed by comparing the mean scores/response times among the three types of stimuli (DJ, SI, and N).

fMRI data analyses were performed using MatLab version R2020a (Mathworks, Natick, MA, United States) and SPM12 software (The Wellcome Centre for Human Neuroimaging, UCL Queen Square Institute of Neurology, London, United Kingdom). All functional volumes for each subject were slice-time corrected, realigned to the first volume acquired, and normalized to the MNI (Montreal Neurologic Institute) template implemented in SPM12. Then the images were smoothed with a Gaussian kernel having FWHM of 8 × 8 × 8 mm.

Single-subject statistical analysis was performed applying the general linear model, where the time-series data were modeled as a series of events convolved with a canonical hemodynamic response function. Three types of events were implemented as regressors in the single subject-first level analysis: DJ, SI, and N. For this event-related analysis, the appearance of the punch line was considered as the starting time of the stimulus of interest. The six head-motion parameters (translations and rotations) were entered as regressors of no interest.

Using statistical maps generated by the single-subject analyses, several random effect second level group analyses (One-Sample t-test) were performed as follows:

To evaluate cortical activations common to DJ and SI three different regions of interest (ROIs) were defined using the SI > N contrast: right thalamus (peak coordinates: x = 2, y = 4, z = 6), left putamen (peak coordinates: x−30, y = −4, z = 2) and left medial frontal gyrus (peak coordinates: x−2, y = 52, z = 10; uncorrected for multiple comparisons, k > 0). From these ROIs we have extracted the beta values using Marsbar² for the DJ and N conditions and we have run a t-test over them.

A double statistical threshold (voxel-wise p < 0.001 and spatial extent) was adopted to achieve a combined significance, corrected for multiple comparisons, of α < 0.05, as computed by 3dClustSim AFNI routine, using the ‘‘-acf’’ option³.

The average scores given by the participants to each stimulus according to the three a-priori categories (DJ, SI, and Neutral) are reported in Figure 2A for the two rated dimensions (level of fun, level of offense). As shown in Figure 2A, on average subjects rated the stimuli according to the a-priori categories.

By considering the three a-priori categories of stimuli and by using data collected inside (level of fun) and outside of the scanner (level of offense), results on skewness and kurtosis revealed that data distribution was normal in each category of stimuli (for response time and for the two levels of rating - fun and offense).

The one-way ANOVA on response time registered in the three a-priori categories did not reach a significant difference [F_{(2, 37)} = 2.034, p = 0.145]. Overall, participants took the same time to respond to DJ, SI and Neutral stimuli.

We run a 3 × 2 ANOVA for repeated measures on the 3 a-priori categories and according to the two levels of rating (fun and offense, see Figure 2B). Results showed a significant effect of the category [F_{(2, 37)} = 173.87, p < 0.001, η_p² = 0.90]. Tukey’s post hoc test showed that the category of Neutral stimuli differed from both DJ and SI (p < 0.001 in both cases). The difference between DJ and SI was not significant. The ANOVA also showed a significant effect of the levels of rating [F_{(1, 37)} = 9.127, p = 0.005, η_p² = 0.20], the rating was higher for the level of offence than for the level of fun; and a significant interaction type X rating [F_{(2, 37)} = 227.6, p < 0.001, η_p² = 0.92].

Tukey’s post hoc test showed a significant difference between level of fun and offense in DJ and SI (p < 0.001). In Neutral stimuli, the level of fun and offense did not differ (p = 1). Furthermore, in DJ the level of fun differed from that registered in SI stimuli (p < 0.001, it was higher in DJ), the same difference was found in the level of offense, which was higher in SI stimuli with respect to DJ (p < 0.001). The level of offense also differed in DJ with respect to Neutral stimuli (p < 0.001, it was higher in DJ). No difference in the level of fun between SI and Neutral was registered (p = 1). In summary, DJ stimuli were funny and at the same time offensive (the level of fun and offense in DJ differed from the two levels registered in Neutral stimuli); whereas SI stimuli were only offensive (no difference between the level of fun registered in SI and Neutral stimuli). DJ stimuli, although offensive, were rated as less offensive than SI stimuli (see Figure 2B).

Overall, the main effect of DJ elicits activations in brain regions implicated in the three stages of humor processing: detection, comprehension, and appreciation. In particular, the activation in the bilateral SMG might be in relation to humor detection (Samson et al., 2008); the activation in the IFG, should be for humor comprehension (Goel and Dolan, 2001, 2007; Moran et al., 2004; Bartolo et al., 2006; Chan et al., 2012, 2013), whereas the activations in the bilateral cerebellum, the hippocampus, the insula, the ACC, and the amygdala bilaterally should be related to humor appreciation (Mobbs et al., 2003; Chan et al., 2016). Additionally, we also found activation in the caudate nucleus and thalamus bilaterally. The activation in the caudate nucleus has been found in gelotophobic participants exposed to hostile verbal jokes (Chan, 2016). Gelotophobia is a type of social phobia characterized by a fear of being laughed; gelotophobics are particularly sensitive to aggressive humor (Samson et al., 2011; Ruch et al., 2014; Chan et al., 2016). Chan et al. (2016) suggested that the activation of the caudate nucleus (a portion of dorsal striatum) might be in relation to its role in modulating “selective attention, planning, and effortful regulation of affective states.” In this view, we found activation of the caudate nucleus when participants perceive SI relative to N sentences, denoting the detection of a feeling of discomfort and the establishment of affective regulation (McGraw and Warren, 2010; Koszałkowska and Wróbel, 2019). Although we did not test gelotophobia and did not include non-hostile jokes, our behavioral data show that mainly DJ were perceived as less offensive than SI stimuli, suggesting that a sort of affective regulation process might take place when socially inappropriate funny stimuli are perceived.

An alternative interpretation derives from studies showing that the dorsal striatum is activated in participants that showed a desire to punish (de Quervain et al., 2004; Crockett et al., 2013). More specifically, de Quervain et al. (2004) conducted a PET study on punishment, in which participants could administer a punishment to unfair participants. Since punishment occurs to establish justice, the authors have interpreted the feeling that accompanied the punishment, as a sort of “sweet revenge,” as the administration of punishment is a “just revenge” toward defectors. As mentioned in the Introduction, disparagement jokes belittle individuals that are portrayed in a ridiculous manner. In some instances, the funny ending, although socially inappropriate, might generate a sense of justice toward the arrogance of the individuals targeted in the joke. Therefore, the activation of the caudate nucleus could be related to this feeling of sweet revenge coexisting with the feeling of mirth in disparagement humor.

This line of reasoning further suggests that disparagement humor might have much to do with the experience of schadenfreude. Schadenfreude is defined as an emotional reaction to misfortunes of others and it “is more likely when the protagonist is disliked, when actors pursue immoral goals and if they are responsible for their misfortunes” (Schindler et al., 2015, p. 1). Interestingly, the assessment of schadenfreude foresees a question like: “how amusing/funny was the situation for you?” In a recent study, Paulus et al. (2018) asked participants to rate the intensity of their amusement at the perception of a series of situations that elicit schadenfreude. They found activations in a network including, among others, the caudate nuclei bilaterally and the left thalamus (Paulus et al., 2018). Such a result might suggest the existence of a relation between disparagement humor and schadenfreude.

Relative to SI, DJ activated the bilateral ACC, the SMA, the precuneus, and the MFG. Relative to N stimuli, DJ activated the anterior insula, the putamen, and the entorhinal cortex. These regions have been previously found to be activated during humor appreciation (Goel and Dolan, 2007; Kohn et al., 2011; Shibata et al., 2014; Caruana et al., 2015, 2016; Liu et al., 2019; Wattendorf et al., 2019; Talami et al., 2020). Interestingly, some neuroimaging studies found activations in the ACC and the anterior insula during the threats to another person’s physical and social integrity (Beeney et al., 2011; Krach et al., 2011; Müller-Pinzler et al., 2016) and the left insula was also found activated in hostile jokes (Chan et al., 2016), suggesting that in DJ a feeling of mirth (and/or a desire to laugh) derived from the joke (e.g., SMA and precuneus) and the perception of the jokes’ social inappropriateness (e.g., ACC and insula) coexist.

Taken together, the anterior insula and the ACC are activated in DJ vs. baseline (both bilaterally), in DJ vs. N stimuli (the left anterior insula), in DJ vs. SI (the bilateral insula). These two areas are known to be related to empathy for pain (see Lamm et al., 2019 for a review). We can speculate that their activations reflected the feeling of empathy perceived in DJ.

SI stimuli activated a large network including frontal (inferior, superior, medial frontal gyrus, ACC), parietal (SMG and AG) temporal regions (STG) as well as the mesolimbic system (parahippocampal gyrus, amygdala, and the insula bilaterally). The activation of the cortical regions might be in relation to sentence comprehension and Theory of Mind for mental state attribution (Amodio and Frith, 2006; Olsson and Ochsner, 2008; Chan et al., 2016), whereas the mesolimbic system should be activated for affective and emotion regulation at the perception of the hurtful emotional content (Hamann et al., 2002; Benuzzi et al., 2004; Chan et al., 2016).

Relative to N stimuli, SI activated the right dorsal striatum (caudate nucleus) and the thalamus. In DJ, where the fun and the offense coexist, activations were found in the dorsal striatum bilaterally. We can speculate that the right caudate nucleus might be specific to negative emotions that arise at the perception of offensive sentences; whereas the bilateral activation in DJ could be related to a feeling of discomfort generated by the perception of something funny that is at the same time offensive.

Intriguingly, the contrast SI vs. DJ did not elicit any significant activation, as if the offense in SI and in DJ produces the same patterns of activations.

This study also aimed at investigating the common brain network in disparagement jokes and socially inappropriate-not funny stimuli. Three functional ROIs were defined by using the contrast SI > N, namely the right thalamus, the left putamen and the left medial frontal gyrus. ROI analyses on the DJ > N contrast revealed a significant activation of these three regions, showing the existence of a common brain activity in DJ and SI with respect to neutral stimuli. In particular, brain activity in the medial prefrontal cortex has been associated with mentalizing processes (Amodio and Frith, 2006), third-person perspective-taking (Vogeley et al., 2004; Takahashi et al., 2009; Zahn et al., 2009; Müller-Pinzler et al., 2016), and empathy (Singer and Lamm, 2009; Bernhardt and Singer, 2012). Furthermore, we found activation of a portion of the left dorsal striatum (the putamen) and of the right thalamus, regions known to be relevant for the processing of socially inappropriate content (Schmahmann, 2003; Chan et al., 2016). Thus, the common regions between DJ and SI relate to a social cognitive complex process that goes from mentalizing to negative feelings arising in response to the perceived offense.