Link to original paper – “Neuroscience of Internet Pornography Addiction: A Review and Update” (2015)
Note – numerous other peer-reviewed papers agree that Steele et al., 2013 supports the porn addiction model: Peer-reviewed critiques of Steele et al., 2013
Excerpt critiquing Steele et al., 2013 (citation 303):
An EEG study on those complaining of problems regulating their viewing of internet pornography has reported the neural reactivity to sexual stimuli . The study was designed to examine the relationship between ERP amplitudes when viewing emotional and sexual images and questionnaire measures of hypersexuality and sexual desire. The authors concluded that the absence of correlations between scores on hypersexuality questionnaires and mean P300 amplitudes when viewing sexual images “fail to provide support for models of pathological hypersexuality”  (p. 10). However, the lack of correlations may be better explained by arguable flaws in the methodology. For example, this study used a heterogeneous subject pool (males and females, including 7 non-heterosexuals). Cue-reactivity studies comparing the brain response of addicts to healthy controls require homogenous subjects (same sex, similar ages) to have valid results. Specific to porn addiction studies, it’s well established that males and females differ appreciably in brain and autonomic responses to the identical visual sexual stimuli [304, 305, 306]. Additionally, two of the screening questionnaires have not been validated for addicted IP users, and the subjects were not screened for other manifestations of addiction or mood disorders.
Moreover, the conclusion listed in the abstract, “Implications for understanding hypersexuality as high desire, rather than disordered, are discussed”  (p. 1) seems out of place considering the study’s finding that P300 amplitude was negatively correlated with desire for sex with a partner. As explained in Hilton (2014), this finding “directly contradicts the interpretation of P300 as high desire” . The Hilton analysis further suggests that the absence of a control group and the inability of EEG technology to discriminate between “high sexual desire” and “sexual compulsion” render the Steele et al. findings uninterpretable .
Finally, a significant finding of the paper (higher P300 amplitude to sexual images, relative to neutral pictures) is given minimal attention in the discussion section. This is unexpected, as a common finding with substance and internet addicts is an increased P300 amplitude relative to neutral stimuli when exposed to visual cues associated with their addiction . In fact, Voon, et al.  devoted a section of their discussion analyzing this prior study’s P300 findings. Voon et al. provided the explanation of importance of P300 not provided in the Steele paper, particularly in regards to established addiction models, concluding,
“Thus, both dACC activity in the present CSB study and P300 activity reported in a previous CSB study may reflect similar underlying processes of attentional capture. Similarly, both studies show a correlation between these measures with enhanced desire. Here we suggest that dACC activity correlates with desire, which may reflect an index of craving, but does not correlate with liking suggestive of on an incentive-salience model of addictions.”  (p. 7)
So while these authors  claimed that their study refuted the application of the addiction model to CSB, Voon et al. posited that these authors actually provided evidence supporting said model.