The "Your Brain on Porn" Study
A review of Kühn and Gallinat's (2014) study on pornography supposedly decreasing grey matter (Spoiler alert: very weak evidence).
Typically when discussing the effects of pornography, it’s not uncommon for people to cite Kühn and Gallinat (2014), showing that pornography is associated with lower grey matter volume. News articles have run with such headlines as “Watching pornography rewires the brain to a more juvenile state” (Neuroscience News, 2019), and comments like “It literally shrinks your brain” [emphasis in the original article” (Stone, 2023).
In this commentary, I will critically examine the paper and argue that their weak evidence does not allow for causal claims.
In the original Kühn and Gallinat (2014) paper, which one can access here, the authors looked at 64 male participants, exposed them to 60 explicit images, and measured their brain structure through fMRI. The authors found that individuals who reported watching higher levels of pornography showed decreased levels of grey matter. Indeed, such evidence has been cited in favor of pornography being harmful. For example, Wilson (2014) states:
In May, 2014, the prestigious medical journal JAMA Psychiatry published research showing that, even in moderate porn users, use (number of years and current hours per week) correlates with reduced grey matter and decreased sexual responsiveness. The researchers cautioned that the heavy porn users' brains might have been pre-shrunken rather than shrunken by porn usage, but favoured degree-of-porn-use as the most plausible explanation. They subtitled the study "The Brain On Porn."
Fradd (2017) says, “Researchers have also found that moderate porn use is correlated with shrunken gray matter in parts of the brain that oversee cognitive function” — with other sources repeating similar sentiments (e.g., The Dibble Institute, n.d.; Estephan, 2023; De Sousa and Lohda, 2017).
Of interest is their regression from Table 1, showing a linear relationship between pornography consumption and lower grey matter volume in the brain.
The authors found a significant relationship between pornography use and grey matter volume in the right unclear. The direction of the relationship withheld after adjustments for Internet addiction and sex addiction 1. In other words, pornography seems to play a unique role in grey matter volume.
However, significant issues with this study cast doubt on a causal claim. First off, it’s unknown how valid their measurement of pornography use is as they did not use any validated measurement tool to measure pornography use, instead relying on questions such as “‘How many hours on average do you spend watching pornographic material during a week day?” and ‘How many hours on average do you spend watching pornographic material during a day of the weekend?’” (828). This is an issue as, for example, Scharkow (2016) found that self-reported frequency of internet use was weakly associated with actual log files from the participant’s computers. Ellis et al. (2018) found a weak correlation between self-reported phone frequency use and their actual phone use. It’s unknown if the self-reported pornography consumed by these individuals accurately reflects their pornography use (the sample average was 4 hours a week). Since the study did not adjust for moral incongruence (see Black, 2023 for more information [in Footnote 1]), we do not know if these individuals have moral problems with pornography that could lead them to assume that their pornography use is out of control (Grubbs et al., 2020).
Second, while the researchers found a relationship between pornography use and alcohol use, and depression, the authors did not attempt to adjust for these issues. This is important since frequent pornography users are more likely to consume alcohol (Svedin, Akerman, Priebe 2010), and alcohol consumption is associated with lower grey and white matter (Daviet et al. 2022). Depression has also been shown to be associated with lower grey matter volume, too (Shad et al., 2012). Given that the temporal relationship shows mental health issues lead to high pornography use (Doornwaard et al., 2016), it’s possible that individuals with lower mental health have lower gray matter volume and simply happen to watch more pornography, rather than pornography causally causing lower grey matter volume.
Third, other studies on the issue of Internet domains have not shown them to causally change the structure of the brain. Nivins et al. (2022) found that watching television and videos, using social media, and playing video games do not affect the brain.
Fourth, the relationship found by the authors was a correlation of -0.329. If we square the correlation, pornography consumption predicts 10% of the variance in grey matter. In other words, 90% of the variance is still left unexplained. Even assuming a causal relationship, this means that other variables besides pornography might lead to lower grey matter rather than pornography being the sole variable that’s the issue.
Finally, a more technical critique rests on the use of fMRI data used by the researchers. Weinberg and Radulescu (2015) present critiques of studies utilizing structural MRIs and resting-state functional MRIs, both employed in Kuhn and Gallinat's research. The authors express skepticism, asserting that the evidence supporting the notion that these 'findings' reflect changes in the brain related to pathogenesis is, at best, inconclusive and may potentially represent artifacts or epiphenomena of questionable value. They argue that MRI serves as a physical-chemical measure rather than a direct assessment of brain structure, relying on radio-frequency signals emitted from hydrogen atoms influenced by the magnetic properties of surrounding tissue. According to the authors, the contrasts observed in MRI between tissue compartments, such as gray and white matter, are contingent on the relative signals from these components. MRI contrast hinges on the density of protons (H1) and their magnetic properties, as expressed by the T1 and T2 relaxation constants. While proton density remains relatively uniform across most tissues, variations in T1 and T2 can be significant between tissues and are susceptible to influence by various factors.
The study underscores the reliance of MRI signals on refined parameters, noting their susceptibility to physical-chemical phenomena not necessarily tied to the cellular count or architecture within the tissue. For instance, the widely employed metric of "gray matter volume" in voxel-based morphometry analyses derives from a segmentation algorithm applied to voxel intensity on T1-weighted images. However, a voxel, though typically sized, represents a heterogeneous tissue sample comprising neurons, neuropil, glia, microvessels, and extracellular space. Any of these elements can be altered by diverse biophysical factors impacting the MRI signal. The authors caution against hastily attributing MRI differences between patient and control samples to microstructural abnormalities of pathological significance, emphasizing the influence of nonstructural factors such as psychotropic drug use, body weight changes, blood lipid levels, alcohol and nicotine use, exercise, hydration, pain, and cortisol levels on MRI signals and anatomical measurements.
Before ascribing MRI differences to pathogenic abnormalities, the study suggests considering alternative possibilities. Factors like changes in brain perfusion due to acute drug administration or alterations in magnetic properties affecting signal contrast may masquerade as changes in MRI volume measurements. Even in seemingly unconfounded studies, hidden confounders like temperament or smoking behavior may exist and impact MRI signals.
Concerns related to resting-state functional magnetic resonance imaging (fMRI) studies are also discussed. These studies, examining brain activity without specific tasks or stimuli, lack a within-subject reference state, making it challenging to discern actual brain activity during the imaging protocol. Resting-state fMRI involves subjects lying in the scanner without performing specific tasks, revealing the "default mode network" activity pattern. Initially considered a fundamental marker of brain connectivity, cautionary notes emerged about the influence of cognitive, behavioral, and physiological variables, as well as artifacts and conceptual inconsistencies.
While appealing due to their ease of conduct and ability to yield differences between patient and control groups, resting-state fMRI studies may misinterpret deviations from the default pattern in psychiatric patients. Patients' experiences in the MRI environment, including thoughts, feelings, and reactions, can interfere with observed patterns, challenging the interpretation of perceived "abnormalities." The excerpt emphasizes individual variability in resting-state MRI experiences, questioning the confidence in differences between patient and control individuals not being influenced by variations in their ability to "zone out" during data acquisition. Head movement and motion artifacts, particularly nodding motions, are identified as significant concerns affecting the midline structures of the default mode network, with conventional correction methods deemed inadequate for controlling these confounding factors.
Final Thoughts
Based on the issues with Kühn and Gallinat, it’s hard to know if pornography use causally leads to lower grey matter volume, or if the relationship is simply spurious. Given the lack of adjustment for depression and alcohol use, it’s unknown if the relationship would be biased by such issues. Given that the study can not show causality or a temporal relationship, references to this study require great caution. Individuals who make causal claims about the study itself are wrong, and the conclusions of the study itself seem to be premature given the lack of causality, temporal relationship, and adjustments for other issues 2 .
The authors also found a correlation between hours of pornography watched and the Alcohol Use Disorder Identification Test (r = 0.250) and the Beck Depression Inventory (r = 0.295). In the striatum, there was a negative correlation between pornography use and grey matter (r = -0.329). However, such relationships are unlikely causal (see Black, 2023).
For other comments, see the guest blog on The Knox Student, though it repeats what I said about causality.