Approach bias for erotic stimuli in heterosexual male college students who use pornography (2019)

Seventy-two male undergraduate students from the University of Connecticut (average age = 19.5 years, SD = 2.4) who self-identified as users of pornography were recruited from the Psychology Department’s online participant pool. Sexual preference was assessed using a question from the Kinsey scale (Kinsey, Pomeroy, & Martin, 1948/1988). Participants received class credit for their participation. The study was approved by the Institutional Review Board at the University of Connecticut.

Participants were seated in front of a computer and asked to complete questionnaires prior to performing a computerized AAT. Questionnaires assessed frequency and severity of pornography use as well as attitudes toward pornography. Scales included the Problematic Pornography Use Scale (PPUS) and the Brief Pornography Screener (BPS), both of which quantify pornography use and related attitudes. The PPUS (Kor et al., 2014) is a 12-item scale that asks individuals to assess statements about their pornography use over the past year on a 6-point Likert scale ranging from “never true” to “almost always true.” The scale includes statements such as, “I was unsuccessful in my efforts to reduce or to control the frequency of my use of pornography” and “I spend too much time thinking about pornography” (Kor et al., 2014). Similarly, the BPS asks individuals to respond to five situations in regard to their use of pornography over the past 6 months on a 3-point Likert scale from “never” to “very often,” and consists of items such as, “You continue to use sexually explicit material even though you feel guilty about it” (Kraus et al., 2017). The BPS is a screening scale that measures only one aspect of problematic pornography use – control or lack thereof over behavior – and may be useful in identifying individuals at risk of problematic pornography use or as a proxy measure. In comparison, the PPUS is a multidimensional scale that assesses four aspects of problematic pornography use and may therefore provide a broader picture of problematic pornography use symptomology (Kor et al., 2014).

We used a modified version of the AAT utilized by Wiers et al. (2011), in which participants are instructed to push or pull a joystick in response to pictures based on irrelevant properties of the picture content (e.g., whether the picture is oriented horizontally or vertically). Each computer was equipped with a standard gaming joystick and headphones and all of the software was custom-written by the author RSA. Prior studies have indicated that the approach–avoidance joystick task is a valid method for activating implicit approach–avoidance behaviors based on the valence of pictorial cues (Krieglmeyer & Deutsch, 2010). Furthermore, Wiers et al. (2009) suggest that when a reliable difference in approach and avoidance movements is found in response to irrelevant picture features (as compared to relevant features), the responses are more likely to be automatic and thus regulated outside of conscious awareness.

The AAT task in this study consisted of 50 erotic pictures of females, heterosexual couples, and female couples and 50 neutral pictures of common household items, such as a lamp or a clock. Erotic stimuli were selected in line with multiple studies reporting that erotica depicting females or couples (male/female and female/female) are subjectively rated as highly arousing and evoke a strong appetitive motivational state, evidenced by physiological response (Bernat, Patrick, Benning, & Tellegen, 2006; Bradley et al., 2001). Half of the images were 600 × 800 pixels and presented vertically (portrait view), and the other half were 800 × 600 pixels and presented horizontally (landscape view).

After providing written informed consent, participants were asked to complete the questionnaires administered through Qualtrics, an online survey service. Following this, the participants were instructed on how to complete the AAT. Participants were seated in front of a computer and were instructed to pull the joystick in response to the images oriented vertically (portrait – 600 × 800 pixels) and to push the joystick in response to the images oriented horizontally (landscape – 800 × 600 pixels). Pulling the joystick made the picture increase in size, generating a sensation of approach; pushing the joystick made the picture decrease in size, simulating an avoidance movement. The initial actual image size was 3 in. × 4 in. for the vertical images and 4 in. × 3 in. for the horizontal images. Approaching resulted in the picture enlarging continuously in size until it filled the screen and disappeared over a 1-s interval. Avoiding resulted in the picture to shrink continuously until it disappeared over a 1-s interval. Half of both types of stimuli were presented as landscape images and the other half were presented as portrait images. Participants were asked to respond as quickly and accurately as possible throughout 2 series of 100 test trials. Response time was calculated as the number of milliseconds from when the image was presented on the screen to when the joystick movement was initiated. The first series began with 20 practice trials using colored rectangles, followed by the 50 erotic and 50 neutral stimuli presented in a pseudorandom order. The second series occurred after a 60-s break and began with 2 practice trials followed by the 100 test trials. Incorrect answers were denoted by a buzzer noise in the headphones. Each block of trials took about 5 min to complete. After completing the AAT, participants were debriefed and dismissed.

Cognitive bias data were calculated in the same manner as found by Wiers et al. (2011) for the AAT, responses that were incorrect/missed and response times longer than three SDs above the mean were discarded based on each participant’s performance. Erotic approach bias scores were calculated by subtracting median reaction times:

Seventy-two participants completed the experiment. Eight participants were excluded due to indicating a sexual preference that was not a heterosexual preference (i.e., they had a score higher than 1) on the Kinsey scale (Kinsey et al., 1948/1988), and an additional six participants were excluded due to incomplete or extreme data (i.e., greater than three SDs above the mean). This resulted in 58 complete data sets.

A one-sample t-test showed that there was a significant approach bias of 81.81 ms (SD = 93.07) for the erotic images, t(57) = 6.69, p < .001, compared to the neutral images (Figure 1). In addition, on examining correlations between the assessments and approach bias scores, we found a significant correlation between BPS and approach bias scores, r = .26, p < .05, indicating that the higher the BPS score, the stronger the approach bias (Figure 2). The correlation between the PPUS and approach bias scores was not significant, r = .19, ns. There was a strong correlation between the BPS and PPUS scores, r = .77, p < .001.

To examine cognitive bias in those at high risk of problematic pornography use, a total score of 28 or more on the PPUS was used as a threshold for problematic pornography use as suggested by the author AK. Accordingly, four participants in our sample were categorized as problematic pornography users based on this criterion. We conducted a one-way analysis of variance (ANOVA) to determine if cognitive bias scores differed significantly between the two groups. The individuals with problematic pornography use demonstrated a significantly stronger approach bias score [186.57 ms (SD = 135.96), n = 4] compared to individuals without problematic pornography use [74.04 ms (SD = 85.91), n = 54], F(1, 56) = 5.91, p < .05 (Table 1). Because of the difference in group sizes, there is some concern about homogeneity of variance between the groups. Accordingly, we ran a Levene’s test of homogeneity of variances and found that there was no significant difference between the groups, hence suggesting that the ANOVA is applicable in this case (Levene’s statistic = 1.79, df₁ = 1, df₂ = 56, p = .19).

Table 1. Average BPS and PPUS scores and RTs for the four conditions for problematic pornography users in the sample (N = 4), defined as individuals that scored 28 or higher on the PPUS

The results support the hypothesis that heterosexual male college students who use pornography are faster to approach than to avoid erotic stimuli during an AAT task. There was a significant approach bias of 81.81 ms for erotic stimuli; that is, participants were faster to move toward erotic images compared to moving away from erotic images. Participants were faster to pull the joystick than to push it in response to erotic stimuli, but this same bias was not present with regard to neutral stimuli. Similar approach biases have been reported in studies using modified AATs, such as that of Stark et al. (2017) using an erotic-AAT and Wiers et al. (2011) using an alcohol-AAT. These findings are also in line with several SRC tasks suggesting that addicted individuals display an action tendency to approach rather than avoid addictive stimuli (Bradley et al., 2004; Field et al., 2006, 2008).

Overall, the findings suggest that approach for addictive stimuli may be a more rapid or prepared response than avoidance, which may be explained by the interplay of other cognitive biases in addictive behaviors. As suggested by the literature (Cousijn et al., 2011; Field & Cox, 2008; Stacy & Wiers, 2010), individuals who display automatic approach tendencies for addiction-related cues also tend to look at them longer (indeed, on average, participants looked at erotic images over 100 ms longer than neutral images before pushing them away; Table 2) and to evaluate them as more positive and arousing than other cues, such as neutral stimuli. Accordingly, findings reported by Mechelmans et al. (2014) indicate that individuals with compulsive sexual behaviors show attentional biases for erotic stimuli. Future studies should explore roles, both separately and together, of approach, attentional, and evaluative biases in problematic pornography use. It is worth noting that this approach bias for erotic stimuli is not necessarily indicative of addiction risk; it is plausible that pornography users have a general positive attitude toward erotic stimuli or simply prefer it over neutral stimuli. Indeed, erotic stimuli have a greater emotional valence than common household items, such as a lamp or desk. Furthermore, erotic stimuli inherently evoke a strong appetitive emotional state (Bradley et al., 2001), suggesting that individuals may be inclined to approach erotic stimuli regardless of addiction risk.

Table 2. Average BPS and PPUS scores and RTs for the four conditions in the entire sample (N = 58)

Moreover, total scores on the BPS were positively correlated with approach bias scores, indicating that the greater the severity of problematic pornography use, the stronger the degree of approach for erotic stimuli. This association was further supported by results suggesting that individuals with problematic pornography use, as classified by the PPUS, showed more than a 200% stronger approach bias for erotic stimuli compared to individuals without problematic pornography use. However, this last finding should be considered particularly cautiously given the small number meeting the criteria for problematic pornography use. These findings resonate with those in studies of addictions indicating that approach bias for addiction-related stimuli is a common feature present in individuals with addictions (Bradley et al., 2004; Cousijn, et al., 2011; Field et al., 2006; Krieglmeyer & Deutsch, 2010; Wiers et al., 2011). In line with research showing that preferential attentional processing for addiction-related cues is associated with severity of addictive behaviors (Field & Cox, 2008; Schoenmakers et al., 2007), we found that approach bias scores were positively associated with scores on the BPS, which may be used as a measure reflecting severity of problematic pornography use.

However, our results differ from those denoting a curvilinear relationship between symptoms of cybersex addiction and approach–avoidance tendencies in student and non-student samples from Germany (Snagowski & Brand, 2015). Similar to the findings of Stark et al. (2017), individuals with greater pornography use problems in this study showed only approach biases for erotic stimuli, not avoidance biases. One possible explanation for this contrast is that Snagowski and Brand (2015) used a task-relevant instruction (i.e., move the joystick according to picture content), whereas this study and that of a study by Stark et al. (2017) used task-irrelevant instructions (i.e., move according to picture orientation or color of the picture frames). The task-relevant instructions may compel participants to process the stimuli more deeply, which may lead to avoidance behavior among users that feel guilt or fear negative consequences related to their behaviors (Stark et al., 2017). While task-irrelevant instructions may not warrant the same level of processing, Wiers et al. (2009) reported that approach movements found in response to irrelevant picture features were more likely to be automatic and unconscious. In sum, given apparent differences across studies performed in different jurisdictions, different samples (student vs. student/non-student), and precise methodologies, further research is needed to understand approach and avoidance behaviors in different populations, using different versions of the AAT. Nonetheless, 4 of 58 subjects (6.89%) met a threshold of 28 points using the PPUS, and this finding is in line with previous studies reporting approximately 10% of cybersex addiction prevalence among male college students (Giordano & Cashwell, 2017).

Taken together, the results suggest parallels between substance and behavioral addictions (Grant et al., 2010). Pornography use (particularly problematic use) was linked to faster approaches to erotic stimuli than neutral stimuli, an approach bias similar to that observed in alcohol-use disorders (Field et al., 2008; Wiers et al., 2011), cannabis use (Cousijn et al., 2011; Field et al., 2006), and tobacco-use disorders (Bradley et al., 2004). An overlap between cognitive features and neurobiological mechanisms involved in both substance addictions and problematic pornography use seems likely, which is consistent with prior studies (Kowalewska et al., 2018; Stark et al., 2018). Nonetheless, additional studies are warranted of cognitive biases, especially in other pornography-using groups (in both clinical and non-clinical populations including women, non-heterosexual individuals, and multiple age groups outside of college-aged), as are studies of the neurobiological and clinical correlates.