Digging Into the “Oreos Are as Addictive as Cocaine” Study

I have to agree with Keith’s post below: the Connecticut College study does not show that “Oreos are as addictive as cocaine.” The college should not have put out a press release making that claim, and respectable news outlets (plus Fox and HuffPo) should not have published that claim uncritically.

But I wouldn’t just put this in the “junk science” folder next to vaccines-and-autism or supply-side economics. Understanding what the study found, and what those findings mean and don’t mean, seems to me worth some effort. Eoin O’Carroll’s careful reporting in the Christian Science Monitor is both a rebuke to the shoddy journalism – mostly based on that story – that Keith justly derides and a valuable contribution to public understanding of the phenomenon of habituating behaviors.

The study was straightforward: maze-running rats were offered pairs of alternative outcomes: cocaine v. saline, Oreos v. rice cakes.

1. In each case, they overwhelmingly chose the more rewarding alternative over the less rewarding alternative.

(No sh*t, Sherlock!)

2. The preference for Oreos over ricecakes, measured in terms of the fraction of the time rats chose one rather than the other, was quantitatively greater than the preference for cocaine over saline.

(So what? The relevant measure would have been a choice between cocaine and Oreos, and that choice wasn’t offered. Or, as Edythe London points out, experimenters could have measured how much work the rats would do for each type of reward. Those would have been valid quantitative measures of reward intensity, with the caveat that the results would probably have been strongly dose-dependent, and there’s no obvious equivalence between milligrams of cocaine and grams of Oreo. The actual measurement doesn’t seem to measure anything interesting.)

So far, then there’s less here than meets the eye. It’s a cute science-fair project, but not much more than that.

However, Schroeder and his students then took another step. They killed the rats and measured c-Fos in the nucleus accumbens. The brains of the junk-food junkie rats produced even more of it than the brains of the cocaine-junkie rats.

Of course London is right: all that provies is that Oreos produce pleasure, which we already knew. But once you’ve said that, much of the NIDA brain-centric theory of addiction starts to fall apart.

In some ways, the more illuminating finding about food and addiction came from the Johnson and Kenney study (published in Nature Neuroscience three years ago) showing that rats habituated to a high-fat diet, but not other rats, would pursue fatty food even in the fact of electric shocks. Now that’s starting to sound like “addiction” rather than mere “reward-seeking.”

Of course none of that means that Oreos are as dangerous as cocaine. But surely the life-years lost to junk-food eating habits must exceed the life-years lost to illicit drug abuse, even counting the violence in illicit markets. And eating disorders are much more chronic, and much harder to cure, than drug dependency. In a natural population, two-thirds of the people with diagnosable substance abuse disorders recove within three years, almost all of them without treatment, while most people who try to lose weight on their own can’t keep it off, and even aggressive interventions produce only a few pounds’ worth of sustained improvement. (Whoever wrote the Washington Post headline implying that people habituated to sweets don’t go through withdrawal has obviously never been on a diet.) Food, like alcohol, is a familiar problem, while “drugs” is an exotic problem. That fact has been used both to mystify “drug abuse” or “addiction” and to make those categories in to bogey-men, while minimizing the extent of the more familiar risks.

Bad habits and unwise decisions are integral to the human condition, and they exist on a continuum between fully self-aware, self-controlled, and rational actions on the one hand and intractable habitual behaviors on the other, not in neat boxes labeled “normal behavior” and “addiction.” And that means – contrary to the mantras of the brain-scan fetishists – that addictive behaviors respond to contingencies, and that therefore categories of moral responsibility apply. Having impaired volitional control is not the same as acting involuntarily; a meth addict is not a zombie. And it’s quite possible to be a little bit addicted, or to be dependent at one time and a controlled user at a later time.

Note also that the Connecticut College study, though it hasn’t been through the peer-review mill yet, isn’t just “press-release science.” Apparently the paper has been accepted for presentation at the forthcoming Society for Neuroscience meetings, though I’m not sure how high a bar the Society for Neuroscience sets. Perhaps Keith and I differ on this point, but it seems to me that once a paper has been presented at a scientific meeting – even as a poster – it’s fair game for reporting, though a careful journalist should note whether the study has or has not been peer-reviewed and published.

[Cross-posted at The Reality-Based Community]

Mark Kleiman

Mark Kleiman is a professor of public policy at the New York University Marron Institute.