Automaticity of Early Sexual Attention: An Event-Related Potential Study

A promising line of research on forensic assessment of paraphilic sexual interest focuses on behavioral measures of visual attention using sexual stimuli as distractors. The present study combined event-related potentials (ERPs) with behavioral measures to investigate whether detection of a hidden sexual preference can be improved with ERPs. Normal variants of sexual orientation were used for a proof-of-concept investigation. Accordingly, 40 heterosexual and 40 gay men participated in the study. Within each group, half of the participants were instructed to hide their sexual orientation. The results showed that a match between sexual orientation and stimulus delays responses and influences ERP before motor responses. Late ERP components showed higher potential in differentiating hidden sexual preferences than motor responses, thereby showing how ERPs can be used in combination with reaction time measures to potentially facilitate the detection of hidden sexual preferences.

identity scale. Values of 4 and above on the heterosexual identity scale are considered indicative of a heterosexual orientation, and scores of 4 and above on the gay identity scale are considered indicative of a gay orientation. Participants also filled out the SDI-2 (Kuhn et al., 2014based on Spector et al., 1996   Note. HE = heterosexual, Sell-HE-ID = Sell heterosexual identity scale (Sell, 1996), Sell-HO-ID = Sell homosexual identity scale (Sell, 1996), SDI-2 Scale 1 = sexual desire with interaction, SDI-2 Scale 2 = sexual desire without interaction (Kuhn et al., 2014based on Spector et al., 1996, SIS 1 = sexual inhibition due to threat of performance failure, SIS 2 = sexual inhibition due to threat of performance consequences, SES = sexual excitation scale (de Albuquerque, 2012 based on Janssen et al., 2002). Conducting a preliminary study was necessary to verify the emotional valence ratings for the stimuli used in the main study. Male participants recruited for the preliminary study (n = 40) rated all the pictures on three emotional dimensions-valence, arousal, and sexual attractiveness-using a Likert-type scale ranging from 1 to 5 with 1 as the lowest and 5 as the highest rating (programmed with Presentation [Neurobehavioral Systems, Albany, NY, USA]). These ratings were used to ensure that the clothed and naked picture categories mostly differed in sexual attractiveness, whereas, valence and arousal did not differ considerably for these two types of pictures.
Initial analysis showed no rating differences between categories of clothing within each gender category that could be attributed solely to different ratings in the dimension of sexual attractiveness by the appropriate groups (comparisons of clothed females vs. naked females as rated by heterosexuals and clothed male vs. naked male as rated by gay men). As preliminary analyses did not yield the required results, the next step was to select a subset of pictures within each of the four categories so that the new subset would differ between the clothed and naked pictures within a gender category (clothed female to naked female and clothed male to naked male) mostly in terms of sexual attractiveness. For this purpose, the two ratings that should not have differed between clothing conditions (i.e., valence and arousal) were averaged for each picture across all participants from a sexual orientation group. Within each category, this resulted in a list with all pictures and two mean ratings for valence and arousal from one sexual orientation group.
To ensure homogenous ratings in the valence and arousal dimensions across different clothing conditions within a gender category (e.g., between clothed female and naked female), a small range around the median from the average valence and arousal ratings was Note. Results from the rating task in the preliminary study mean and SD for valence, arousal, and sexual attractiveness ratings are depicted for the clothed picture categories and the naked picture categories.

AUTOMATICITY OF EARLY SEXUAL ATTENTION 8
The picture set chosen was rated again in the main study by a part of the main sample.
Only a subset of participants within the lying groups (10 heterosexual participants and 11 gay participants) also gave untrue responses in the main study rating task as well. This resulted in 59 truthful ratings within the main study to replicate the controlled emotional dimensions.
The three ratings for the clothed (sexually non-explicit) and naked (sexually explicit) pictures Supplementary Figure 2 displays the ratings (main study) for the two picture sets.

Truthful and Untrue Responses on the Klein Sexual Orientation Grid
To check if the instructions on lying were understood and executed, KSOG data within the lying conditions were compared. Both groups that were asked to give false responses in the KSOG also gave additional truthful responses. The results are depicted in Supplementary Figure 3. For the two groups within the lying condition, an overall mixeddesign ANOVA with the factors of sexual orientation (2: heterosexual vs. gay) and lying (2: lying vs. truthful) was calculated for the KSOG scores. There was no effect of sexual orientation, F(1, 38) = 2.78, p = .103, ηp 2 = 0.07, or lying, F(1, 38) < 0.01, p = .999, ηp 2 < 0.01. There was, however, an interaction effect between the two factors, F(1, 38) = 1100.00, p < .001, ηp 2 = 0.97. These results show that participants followed the instructions to lie adequately.

Supplementary Figure 3
Note. Mean KSOG scores and SD for the two groups (HEF true = lying heterosexual group giving truthful KSOG responses, Gay true = lying gay group giving truthful KSOG responses, HEF lying = lying heterosexual group giving false KSOG responses, Gay lying = lying gay group giving false KSOG responses).

Trial Composition Within the Dot Probe Task
Supplementary Table 3 gives an overview of the different trials used in the dot-probe task (the colors refer to the picture categories).