Pure Pedantry

Over at Economics of Contempt, there is an argument that liberal media bias has to exist because there is evidence that partisanship changes the way that our brains process information. (This is not his only evidence, but it is part of it.)

Now, I don’t want to get into a discussion about the existence or nonexistence of a liberal (or conservative) media bias. What I take issue with is the particular study that Economics of Contempt cites as evidence of this bias. I think that he is misapplying the results of that study.

Economics of Contempt cites the results of Westen et al. 2006. Westen et al. compared 15 partisan Democrats and 15 partisan Republicans in their evaluation of fictitious contradictory statements by John Kerry and George Bush (and also neutral figures). They also examined their brain activation while the subjects evaluated these statements. The conclusion from the study was that the subjects were neurologically biased in their evaluation of contradictory statements. They were less likely to view their guy’s statements as contradictory.

Economics of Contempt uses this as evidence for a liberal bias in the media:

Second, neurological studies have shown that people’s feelings toward a political party dramatically affect the way their brains interpret political news. In his book The Political Brain, Drew Weston described a great study he and two colleagues conducted during the 2004 election. Westen and his colleagues studied the brains of 15 self-identified Democrats and 15 self-identified Republicans as they were presented with a series of slides that showed undeniably inconsistent statements by John Kerry. The partisans were asked to consider whether Kerry’s two statements were inconsistent, and were then asked to rate the extent to which Kerry’s statements were contradictory, from 1 (strongly disagree) to 4 (strongly agree). They then repeated the process with undeniably inconsistent statements by George W. Bush, and again with inconsistent statements by politically neutral males. Here’s how Westen described the results:

They had no trouble seeing the contradictions for the opposition candidate, rating his inconsistencies close to 4 on the four-point rating scale. For their own candidate, however, ratings averaged closer to 2, indicating minimal contradiction. Democrats responded to Kerry as Republicans responded to Bush. And as predicted, Democrats and Republicans showed no differences in their response to contradictions for the politically neutral figures.

The results showed that when partisans face threatening information, not only are they likely to “reason” to emotionally biased conclusions, but we can trace their neutral footprints as they do it.

When confronted with potentially troubling political information, a network of neurons becomes active that produces distress. … The brain registers the conflict between the data and desire and begins to search for ways to turn off the spigot of unpleasant emotion. We know that the brain largely succeeded in this effort, as partisans mostly denied that they had perceived any conflict between their candidate’s words and deeds.

Not only did the brain manage to shut down distress through faulty reasoning, but it did so quickly — as best we could tell, usually before subjects even made it to the third slide [which asked them to consider whether the statements were inconsistent]. The neural circuits charged with regulation of emotional states seemed to recruit beliefs that eliminated the distress and conflict partisans had experienced when they confronted unpleasant realities. And this all seemed to happen with little involvement of the circuits normally involved in reasoning.

But the political brain also did something we didn’t predict. Once partisans had found a way to reason to false conclusions, not only did neural circuits involved in negative emotions turn off, but circuits involved in positive emotions turned on. The partisan brain didn’t seem satisfied in just feeling better. It worked overtime to feel good, activating reward circuits that give partisans a jolt of positive reinforcement for their biased reasoning.

This is basically the root of the well-known “confirmation bias.”

So given that (1) journalists are overwhelmingly Democrats, and (2) party affiliation dramatically affects the way our brains interpret political news, is it really possible that there isn’t a liberal media bias? (Emphasis mine.)

Just to be clear, here is my summary: the study found that when the subjects were asked to rate the contradiction of the statements, Democrats were less likely to rate Kerry’s statements as contradictory and more likely to rate Bush’s statements as contradictory (and vice versa).

With respect to imaging, the study found that the when you looked at the brain activation, areas in the brain that we associate with conflict detection of emotionally-laden stimuli — detecting when a stimulus conflicts with your core beliefs — were activated when considering the contradictory statements of your own party. These regions are the ventromedial prefrontal cortex (vmPFC) and the anterior cingulate (ACC). The argument here is that the activation of this system leads to the suppression of your distaste at finding a contradiction in your guy’s statements. In contrast, this system is not activated when feeling distaste at the other guy’s contradictory statements.

The study also found that when the subjects were given exculpatory evidence about the contradiction, brain regions associated with reward [including the orbitofrontal cortex (OFC)] activated — but only for your guy. Their argument is that when the subjects were given evidence that would solve the contradiction, they felt rewarded.

The take home for the paper is that the subjects were neurologically biased to believe their guy and disbelieve the other guy.

For starters, I do have some nit-picky criticisms of the article itself. For instance, I have argued before that casually associating OFC activation with happy feelings is not accurate and that the underlying systems are more complicated than they are often described. The OFC is involved in the general evaluation of reward for negative and positive stimuli. There is no emotional mood ring in the human brain that can be read by imaging. Further, I take issue with the author’s association of the dorsolateral prefrontal cortex (dlPFC) with a system for “cold” (read: non-emotional, non-biased) reasoning. The author’s argue that the failure of the dlPFC to activate under these circumstances indicates a failure of the “cold” reasoning system to suppress “hot” reasoning and make an objective judgment. This relies on one theory of dlPFC function — response inhibition — but there are many others. The dlPFC does a lot of different things under a lot of different theories, and the simple of assignment of a role in “cold” reasoning belies a lot of complexity. On the whole, however, I think it is a well-done article that presents some interesting results. My criticisms of the article are criticisms about interpreting fMRI evidence in general rather than these author’s conclusions in particular.

That being said, let us presume then that the Westen et al. 2006 is accurate and take the results at face value. Even assuming this, I think that Economics of Contempt misapplies these results for several reasons.

  • 1) You can’t conflate ideological with partisan bias. There is a good deal of difference between someone who is an ideological liberal and someone who always wants Democrats to win. Anyone who watches our political system knows that both our major parties are polyglots. The Democratic and Republican parties are hodgepodges of different ideological groups, some of whom have openly contradictory beliefs. If this weren’t the case, there would be substantially less debate over party platforms. Libertarians and evangelicals often violently disagree, and so do union workers and environmentalists. Thus, you can’t prove that the media has a ideological bias — a “liberal media” — by citing that people are neurologically partisan. It is fair to say that ideology and partisanship are correlated, but they are still different beasts.
  • 2) Economics of Contempt took a study about hardcore partisans and attributed their behavior to journalists as a whole. Now I agree that most journalists are self-identified Democrats. However, this does not mean that they are partisan to the degree that the individuals in the study are partisan. The authors of the study deliberately selected the most rabid partisans they could find, but I don’t think that journalists would necessarily fall into this group. Journalists have much more experience with partisan bullshit than the general public. They are very familiar with being openly lied to by politicians of both parties. Thus, taking information about very partisan lay-people and saying that it applies to journalists as a whole is a stretch.
  • 3) Westen et al. 2006 showed that the participants appeared to be neurologically incapable of suppressing their own biases, but this is not necessarily true of all people. If you remember, the authors cited the dlPFC as a region that if activated could suppress “hot” reasoning and reinstate “cold” reasoning. This region failed to activate in the group of people that they studied. But the fact that we have identified this region as associated with inhibiting this emotional responses suggests that in some people it is possible. People can environmentally conditioned to give into their biases, or they can be trained to suppress them. Take the average scientist. I encounter evidence everyday that disagrees with my biases — whether in my job or writing for ScienceBlogs, but I try (and sometimes succeed) to ignore those biases. Scientists are trained over and over again to ignore the bias and go back to the evidence. And I don’t think that we succeed because we are less emotionally invested in these subjects. It may take practice, but I am quite confident that at least some journalists are capable of suppressing their bias and report the facts as given. Further, there is evidence that even when people give into stereotypes they know they are being biased.
  • 4) There are many types of bias. You can be biased about facts, but you can also be biased about the interpretation of facts. The subjects in this study were given deliberately contradictory statements, but most statements that journalists have to deal with are not so clear cut. It is a fact that Sarah Palin was the mayor of a small town in Alaska. Whether that is a good or a bad thing depends on your interpretation of that fact. Thus, I don’t think that evidence related to how people process artificially concocted lies applies well to the gray areas of reality that journalists regularly have to deal with.

I don’t disagree with Economics of Contempt that there is very likely bias in the media. That is a simple syllogism. People are biased. Journalists are people. Therefore, journalists are biased. What I disagree with is that this particular study — a study showing neurological bias over clear cut statements for a subset of partisan lay people — suggests that bias is primarily liberal. Further, there is a world of difference between saying that journalists are biased and that coverage is biased. A lot happens between emotions and the production of a printed word. Societies have a variety of institutional constraints designed to reduce bias.

It is tempting to say that because people are neurologically biased, news coverage must be as well. But I just don’t think it is that simple.

Just my opinion about media bias, so take it as you will: yeah, I think that the media is often biased, but I think it is too heterogeneous to be described as liberal bias. Anyone who watches the Daily Show regularly sees clearly contradictory statements by figures in the media of both parties. This may have to do with the fact that some of them — like Pat Buchanan and Chris Matthews — are former pols. They have a horse in the race, even though they are working for the press. But this doesn’t mean that all journalists are biased and that all of them have liberal biases. There are people in the media who are capable of putting their partisan and ideological bias aside. Thus, the statement that there is a “liberal media” or a “conservative media” ignores the heterogeneity of the media.

In addition, while I don’t think that bias is going to go away, I do think that directional bias will. Because of economics changes in the way that media is distributed (read: the Internet), it is becoming nearly impossible to maintain uniformly biased coverage in one political direction (if it is even possible now). What we are looking at is increased diversification of biases with each individual consumer seeking out the bias that suits them best. I agree with Daniel Sutter at Cato that if a cartel of liberally biased media exists now, the free market in news coverage will make the existence of such a cartel more and more untenable.

Oh, and for the record, I think the media both liberal and conservative is populated by a horde of Statists. But I guess that is just my bias…

Hat-tip: Megan McArdle


  1. #1 AtheistAcolyte
    September 9, 2008

    What do you know? You’re just one of the liberal elites, proving Westen’s results.

  2. #2 phisrow
    September 9, 2008

    If one were interested in media bias, wouldn’t it make more sense to look at what the media produces, rather than some unrelated brains?

    Brain imaging is cool and all; but this seems like a problem that could be much more easily solved by various sorts of statistical tests against the media’s output, a laborious but not exactly groundbreaking line of research.

  3. #3 Thai
    September 9, 2008

    Jake, I agree with your criticisms.

    However, I still think the study is wonderful in its consistency with everything else we know about information science-information theory. Further it illustrates yet again how evolution solved a particular data compression problem in our human primate mind.

    For remember the most basic data compression issue, which in a nutshell can be paraphrased along the lines of: “in order to make information useful, we make it wrong” is still entirely consistent with this work. David Hume, Karl Popper, Jean Piaget and all the other giants of epistemology, as well as all the modern pioneers of the artificial intelligence community, would smile if they read this study. On that I am sure.

    In fact, I gave a presentation to a business group about this very issue recently (to help them make better decisions in the work place). I posted that presentation’s Power Point on my blog if you are interested.


    I am pretty sure the relevant part is in Part II

    Anyway, don’t get too caught up in the political mumbo jumbo about liberals and conservative from this study (that part is all bunk), as it distracts from the more interesting features of the study itself.


  4. #4 Dunc
    September 10, 2008

    Even if one were to accept that journalists are liberally biased (which I don’t, except for crazy values of “liberal”), there’s also the fact that their bosses frequently aren’t.

New comments have been temporarily disabled. Please check back soon.