The "Blank Slate" Problem in Behavioral Health Research

The near-exclusive focus on extrinsic (and ignoring of intrinsic) drivers of behavior

What Is the Blank Slate?

The term “blank slate” refers to the title of the book The Blank Slate: The Modern Denial of Human Nature by Prof. Stephen Pinker. The book essentially addresses the nature vs. nurture debate in psychology: is it our environment or our heredity that affects our personalities and behavior?

Much of modern academic psychology seems to take the “nurture” side of the debate, I assume as a reaction against harmful and unethical beliefs from a century ago such as eugenics. The “blank slate” viewpoint is essentially that our personalities and behavior are completely undetermined at birth and are shaped only by extrinsic factors such as our upbringing, family environment, society, etc.

Pinker argues, however, that hereditary traits, especially evolved traits, are important for explaining personality and behavior, and that it is a fallacy to dismiss their relevance. (He also argues that simply acknowledging hereditary influences does not necessarily lead to harmful or discriminatory beliefs; and conversely, that if our minds were blank slates, that could also be misused by bad actors).

My aim here isn’t to rehash the nature-nurture debate (though for anyone interested in the evidence for hereditary factors being a strong contributor of behavior, I’d direct you to Steve Stewart William’s Substack, the Nature-Nurture-Nietzsche Newsletter). My aim is to point out parallels to (or maybe downstream effects of) blank-slate thinking in behavioral health research.

Parallels in Behavioral Health Research

What I discuss below doesn’t always exactly map onto the nature/nurture debate, but it’s related in that the focus of a lot of behavioral health research is almost always on extrinsic factors that might explain behavior (e.g. marketing, social norms, or public health messaging), while intrinsic factors (e.g., risk tolerance, curiosity, stress/emotions, or simply enjoying a behavior) are dismissed.

On one hand, it partly makes sense to focus on extrinsic factors, since these are the only things that can possibly be changed by policies or public health efforts (more on that below). But in my experience, it’s vanishingly rare that research papers even acknowledge the existence of intrinsic factors, let alone discuss how they could explain the study’s results.

There is a near-universal focus on extrinsic factors, but this assumption is not made explicitly, and I don’t know if researchers are even consciously aware of this bias. But it’s a bias nonetheless because when researchers don’t even acknowledge intrinsic factors, they can’t possibly test their assumption or hypothesis against obvious competing explanations. This is not a scientific process; it’s a confirmatory exercise.

This will make more sense after going through some examples. As usual I’ll focus mostly on tobacco control research since that’s most of what I read, but I do have other examples from other fields.

Examples in Tobacco Control Research

As in my post that touches on the 10 common flaws that Clive Bates and I see in the science week after week, the examples below are not necessarily the most glaring examples. I’ve simply chosen them as representative examples from recent memory.

Household smoking rules

An analysis of a cross-sectional survey of 7th-12th graders in Hong Kong by Chen et al. 2025 found:

Having SFH [smokefree home] rules without violations was negatively associated with susceptibility to both cigarette (aOR=0.51) and EC use (aOR=0.63). [some stats omitted]

And concluded:

Further research on policies and interventions on SFH rules and practice is warranted for effective tobacco prevention.

This is all framed as household smoking bans preventing adolescents from being “susceptible” to use (side note, “susceptibility” is based on survey questions about curiosity or hypothetical use; it’s often assumed to be a proxy for use, but it poorly predicts use (Selya & Hannon preprint)).

Household bans might deter interest in smoking/vaping, but this is cross-sectional data, so the alternative non-causal explanation (not discussed) is that the families who don’t allow smoking in the house tend to produce kids that are less interested in smoking/vaping. (Note, this is one place that deviates from the nature/nurture issue, because these family-level factors could be environmental or hereditary.)

Such alternative explanations are not discussed in the article, except for a generic statement in the Limitations section that the study’s “cross-sectional design limits causal inference.”

But on the other side of the coin, the Discussion goes into extensive detail about the causal explanation, and speculating about precise pathways:

Previous studies suggest several mechanisms illustrating how SFH rules protect adolescents from smoking initiation: one is through anti-smoking socialisation… and another is the informal social control of youth smoking behavior that parents or caregivers enact by setting rules. The family acts as the primary unit for transmitting social norms that influence smoking behavior.

The paragraph above has an especially heavy dose of “blank slate” thinking because it’s all about socialization and social norms. There is no acknowledgement that adolescents might have thoughts, interests, or desires of their own aside from the social rules they’ve absorbed.

As a result, the Discussion is heavily imbalanced. It overemphasizes the causal explanation, which involves treating household smoking bans as an extrinsic factor that can be manipulated in an intervention. Simultaneously, it completely omits any serious discussion of an alternative explanation, such as factors intrinsic to the family system that explain why families who tend to have lax household rules also have higher interest in substance use.

Exposure to Marketing/Advertising/Online Content

There are dozens of studies that ask survey participants if they remember seeing nicotine/tobacco advertising (from any source), and find that those who do, are more likely to be interested in or use nicotine/tobacco products. Advertising might have some effect in drawing interest, but effects are often weak (e.g. Pasch et al., 2025). The alternative hypothesis, rarely discussed, is that there are attentional/recall biases: people who already use, or are interested in using, tobacco products are more likely to notice and remember relevant advertising or marketing.

For studies looking at tobacco content exposure specifically on social media (e.g. Sawdey et al., 2017), the bias towards extrinsic factors is even worse: social media users are not blank slates; they actively seek out content that they’re intrinsically interested in, and the algorithms continually reinforce content that users interact with. I very rarely see these reverse-causal alternative interpretations discussed in these articles.

One recent article (Pearson et al. 2025) had a quite strong “blank slate” assumption in that authors used common generic search terms (e.g. “vape”) in YouTube and analyzed how often the recommended videos took them down a “pro-” vs “anti-” pathway. But for this work to be relevant, the necessary assumption is that users will passively absorb whatever content is shown to them. I’m not denying that that can happen, sometimes, but it seems to me the bigger factor is that users’ activity on social media drives what they see.

To the authors’ credit, this is the only article I recall that did acknowledge that user behavior influences social media content. But unfortunately this was only an afterthought mentioned in the “limitations” section at the end of the paper. The assumption in the rest of the paper is that extrinsic factors (i.e. algorithm recommendations) are having enough of an effect in driving attitudes towards e-cigarettes that they’re worth worrying about and studying. It ignores the possibility that people might come to YouTube with their own pre-existing opinions and interests, and its that which drives their video consumption and algorithm recommendations.

Expectancies (I.e. How You Expect Nicotine/Tobacco to Make You Feel)

Many other studies examine “expectancies” of nicotine/tobacco use, i.e., the subjective positive and negative experiences they expect to have from using. It’s meant to be a hypothetical question and is treated as an early precursor to, or predictor of, later behavior.

For example, Pinquart & Schuerle 2025 performed a meta-analysis of 215 (!) studies on expectancy and use, finding that:

Results: Positive smoking outcome expectancies were associated with higher smoking levels, with expectancies that smoking reduces negative affect showing the strongest correlation (r = .29) and expectancies that smoking helps with appetite/weight reduction showing the weakest correlation with smoking (r = .14)…

The result itself makes sense and has several possible explanations (which have some overlap and could all be true to an extent). For simplicity, let’s focus on the strongest correlation, about managing stress/anxiety:

1. People who never smoked hear from others that smoking can help them manage stress or anxiety, and these beliefs cause them to try smoking (extrinsic explanation).

2. Many people experiment with smoking, and for some it helps them manage stress/anxiety, and for others it doesn’t. The ones who found it helpful, in the context of their unique physiology, kept smoking, and those who didn’t find a benefit, stopped experimenting (a kind of “survival bias” based on intrinsic factors).

3. The survey is given to both people who smoked and people who never smoked. For the people who didn’t smoke, explanation #1 might apply. For people who did smoke, their expectancies are directly informed by their lived experience: i.e., their answers to the survey question are because they directly experienced that smoking helps their stress levels or ability to cope with stress) (reverse-causality based on intrinsic factors).

It might seem promising that some researchers are looking at expectancies, which are arguably an intrinsic factor (since they are personal thoughts and beliefs that can influence behavior). But that’s not really how it’s treated: one of the above explanations is given any space in the paper. Can you guess which one?

Conclusions: Expectancies about the reduction of negative affect via smoking are a particularly relevant target for prevention and intervention…

Neither the meta-analysis, nor (I assume) acknowledge the intrinsic explanations, let alone structure the study/analysis in a way that might test the different explanations against each other, or examine whether the dominant explanation differs by smoking status.

This leads to the bizarre conclusion that expectancies can be externally manipulated in a public health intervention, and treated as a lever that can be pulled to reduce smoking. This takes what could be an intrinsic factor and treats it as extrinsic. The recommendation to change people’s expectancies might work for people who have never tried a cigarette, but what would this entail for people who already smoke? Answer: it would mean trying to convince people that their own direct subjective experiences are wrong.

I’m not sure if these researchers simply don’t realize this implication due to the blind spot about intrinsic factors, or if they need more humility regarding how effective they think an intervention could be in overriding people’s direct experiences.

Examples in Other Behavioral Research Fields

• The “calories in, calories out” model simplistically assumes that obesity is a problem of consuming too much and expending too little (extrinsic factors) while ignoring the much-more-impactful intrinsic factors of hormones, metabolism, and satiety. (Gary Taubes provides an excellent and compelling criticism of this model here).

• Motivated by the “calories in, calories out” model and thinking the solution is to simply educate people about how many calories they’re consuming, a large number of small-scale research studies suggested that nutritional labeling (an extrinsic factor) would reduce overconsumption (Miller & Cassidy 2015).

  However, a key flaw in much of this literature (pointed out by Variyam 2007 but seemingly few other studies) was that “Assessing the dietary effects of labeling is problematic due to the confounding of the label effect with unobserved label user characteristics” (i.e. intrinsic factors that determine who is likely to pay attention to the nutritional label, and what their dietary habits tend to be). And indeed, in real-world data (not constrained research studies), putting calorie information on restaurant menus has no evident effect on behavior (Fernandes et al. 2016).

• Health policy researchers studying the WIC program, a US governmental program providing food benefits to low-income Women, Infants, and Children, wanted to increase consumption of healthy foods. They assumed that the low-income women simply needed education (an extrinsic factor) on which foods were healthy and which were unhealthy. To their surprise, when they did a qualitative study interviewing WIC recipients on their food choices, the women knew which foods were considered healthy… but chose not to purchase healthy foods for a variety of reasons some of which were other extrinsic factors (e.g., lack of access to fresh produce) and some of which were intrinsic (e.g., difficulty getting the kids to eat healthy foods, lack of time or capacity for cooking healthier meals; Kharofa et al. 2013).

What’s Wrong with Focusing on Modifiable Factors?

“But Arielle,” you might ask, “intrinsic factors can’t be changed, so what’s wrong with researchers focusing on the things that can be changed, so that we can improve public health?” Fair enough — research should be actionable in some way if it’s to be useful, so it makes sense to focus on extrinsic factors where they exist and are having a measurable impact. I do believe that extrinsic factors have some effect: for example, I believe that research and media articles which exaggerates the harms of e-cigarettes has caused most people to incorrectly believe e-cigarettes are equally or more harmful than cigarettes, which scares people away from using them and may promote or keep people smoking.

My criticism is how extreme the focus on extrinsic factors is, to the point where intrinsic factors are very rarely even acknowledged or thought about. Ambiguous results are often interpreted in favor of an extrinsic factor having a causal effect (e.g. household bans, tobacco content exposure), when instead results could be explained by associations or reverse-causality involving an intrinsic factor (e.g., personality traits, subjective experiences, interests).

This leads to several problems:

• Low-quality science. The failure to consider intrinsic drivers of behavior leads to incomplete or flawed understanding of why people engage in certain behaviors. Focusing on extrinsic factors is so ingrained that alternative explanations involving intrinsic factors are not even considered. Most behavioral science papers, therefore, are not truly science: they are not structured to test the hypothesis against other alternative hypotheses, or to advance our understanding of what drives behavior. They are only low-quality exercises in hypothesis confirmation.

• Ineffective public health efforts. Policies, interventions, and other public health efforts that are based on an incomplete or faulty understanding of behavior (i.e. exclusively based on extrinsic factors) will be unsuccessful. When all you have is a hammer, everything looks like a nail. If an intervention aims to change people’s expectations about smoking helping them to manage stress, it will not work if it involves overriding people’s direct prior experiences.

  Focusing on extrinsic factors for public health efforts is a lost opportunity to work with intrinsic factors towards the same goal. For example, if it were accepted by the research community that, say, 20% of people enjoy using nicotine in one form or another, then the natural course of action to recommend is to facilitate moving (at the individual level and at the population level) down the continuum of harm, so people can still use nicotine but in a less lethal form than cigarettes… rather than chasing after diminishing returns related to banning public use, regulating marketing and sales, and messaging campaigns.

• Undermining personal agency. When a large majority of the research focuses on extrinsic factors and ignores intrinsic ones, I think that promotes a paternalistic attitude. If it’s believed that people’s sole reasons for smoking are extrinsic (e.g., that they’ve absorbed a social norm around smoking, or are victims of being exposed to tobacco marketing, or have heard from others that smoking can relieve stress, or that they don’t know the health harms of smoking) then it follows that there should be aggressive tobacco control efforts such as banning smoking in public places (to change social norms), restrict advertising, marketing, and packaging (to prevent would-be-never-smokers from becoming interested), and rolling out messaging campaigns on the physical and mental health effects (to change expectancies and risk perceptions).

  But this denies personal agency — that people might choose to engage in certain behaviors despite public health informing them of the risks and structuring their extrinsic environment to deter the behavior — and denies the intrinsic reasons for engaging in the behavior. It’s counterproductive, since these policies don’t address the underlying intrinsic reasons. It’s even insulting, e.g. to think that an overwhelmed low-income mother who struggles to feed her children simply needs better education about healthy food choices.

Conclusions

I see a lot of parallels to the “blank slate” fallacy in behavioral research, in which the research field essentially ignores important intrinsic factors that drive behavior (e.g. prior subjective experience, interests, personality traits) and exclusively focuses on extrinsic factors that could possibly be targeted in a policy or intervention (e.g. marketing, social norms, or educational campaigns).

Going hand in hand with this is the more general flaw in the research about failing to consider alternative hypotheses that could explain a study’s findings. When it comes to extrinsic vs. intrinsic factors, the missing explanation is almost always intrinsic. Meanwhile, the extrinsic factor is overemphasized as not only a causal factor, but one that can be used as a lever and manipulated by policies or interventions to change behavior. Many of these studies don’t advance science in any way because they only serve to confirm the prior hypothesis about extrinsic factors, while failing to test it against competing hypotheses.

On one hand, extrinsic factors do play a role and it makes sense to study how they can be changed to facilitate less-harmful behaviors. But on the other hand, the near-complete absence of even acknowledging intrinsic factors compromises the science, produces ineffective public health efforts, and leads to paternalistic attitudes.

What intrinsic or extrinsic factors have I missed? Is this an unfair criticism? Share your thoughts below.